THE  DICTATOR’S  DILEMMA  AT  THE  BALLOT  BOX:     ELECTORAL  MANIPULATION,  ECONOMIC  DISTRIBUTION,  AND   POLITICAL  ORDER  IN  AUTHORITARIAN  REGIMES   By   Masaaki  Higashijima             A  DISSERTATION   Submitted  to     Michigan  State  University   in  partial  fulfillment  of  the  requirements     for  the  degree  of     Political  Science  –  Doctor  of  Philosophy   2015 ABSTRACT   THE  DICTATOR’S  DILEMMA  AT  THE  BALLOT  BOX:     ELECTORAL  MANIPULATION,  ECONOMIC  DISTRIBUTION,  AND  POLITICAL   ORDER  IN  AUTHORITARIAN  REGIMES   By     Masaaki  Higashijima              This   dissertation   explores   the   causes   and   consequences   of   authoritarian   elections.   When  holding  an  election,  the  authoritarian  leader  faces  a  dilemma:  In  order  to  maintain   his  rule,  he  needs  to  win  big  in  elections.  Yet,  the  manipulation  of  election  results  risks   losing  some  of  the  informational  benefits  of  authoritarian  elections  –  credibly  showing   regime   strength   and   knowing   the   distribution   of   political   support   from   the   citizenry.   Under   the   constraint   of   the   electoral   dilemma,   the   authoritarian   leader   designs   authoritarian   elections.   The   manners   in   which   elections   are   designed,   then,   have   important  implications  on  post-­‐electoral  political  order  in  autocracies.              Specifically,  this  study  asks  the  following  three  questions.  (1)  Under  what  conditions   do   authoritarian   rulers   refrain   from   using   serious   electoral   fraud   such   as   election   violence,   electoral   cheating,   and   the   manipulation   of   electoral   law?   (2)   When   do   authoritarian   leaders   decide   to   change   their   electoral   system   from   Single-­‐Member   Districts  (SMD)  systems  to  Proportional  Representation  (PR)  systems?  And,  (3)  when  do   authoritarian   elections   backfire   on   dictators   in   the   form   of   protests   and   leadership   turnover.                                                                                                                  In   this   dissertation,   I   argue   that   the   power   distribution   between   the   dictator   and   political   elites   determines   to   what   extent   the   dictator   manipulates   authoritarian   elections.   “Strong”   dictators,   who   can   mobilize   regime   supporters   by   using   their   financial  resources  in  efficient  ways,  do  not  have  an  incentive  to  manipulate  elections  by   resorting   to   extensive   electoral   fraud   and   maintaining   SMD   systems   that   may   bias   election   results   in   their   favor.   By   refraining   from   serious   manipulation   of   election   results,   dictators   can   take   advantage   of   elections   to   overcome   the   shortages   of   information   under   authoritarian   rule.     On   the   other   hand,   “weak”   dictators,   who   lack   financial   resources   or   face   strong   oppositions,   need   to   rely   more   on   electoral   manipulation   because   revealing   their   de   facto   weakness   through   election   results   may   lead  the  elections  to  exert  destabilizing  effects  on  the  political  order.              In   order   to   test   this   theory,   I   conduct   cross-­‐national   statistical   analyses   and   comparative   case   studies   of   the   two   Central   Asian   republics,   Kazakhstan   and   Kyrgyzstan.   My   empirical   analyses   demonstrate   that   autocrats   with   rich   financial   resources   are   more   likely   to   refrain   from   extensive   electoral   fraud   and   shift   electoral   systems   from   SMD   to   PR.     I   also   show   that   if   the   dictator   fails   to   manipulate   elections   strategically  when  dealing  with  the  electoral  dilemma,  then  authoritarian  elections  may   backfire  on  him.  If  the  dictator  employs  excessive  electoral  fraud,  then  he  is  more  likely   to   face   popular   protests   because   political   elites   are   unable   to   make   sense   of   de   facto   strength   of   the   regime.   On   the   contrary,   if   the   autocrat   fails   to   use   sufficient   levels   of   fraud,   then   election   results   may   reveal   the   weakness   of   the   dictator,   leading   to   leadership  turnover  via  a  post-­‐electoral  coup  or  an  opposition’s  victory  at  elections.                       Copyright  by   MASAAKI  HIGASHIJIMA   2015                                               To  My  Parents,  Kumiko  and  Toshiharu  Higashijima                               v   ACKNOWLEDGEMENTS                        I   am   deeply   indebted   to   the   many   people   who   have   helped   me   complete   this   dissertation.  First  and  foremost,  I  thank  the  members  of  my  dissertation  committee  for   their   guidance   and   feedback   during   my   years   at   MSU.   The   chair   of   my   doctoral   dissertation,   Eric   Chang,   has   been   a   tireless   supporter   since   I   began   working   on   this   project.   Eric   always   encouraged   me   to   ask   interesting   questions   and   solve   the   puzzles   with   provocative   theory   and   rigorous   empirical   methods   while   drawing   a   big   picture.   Eric’s  mentoring  made  it  possible  for  me  to  complete  my  graduate  work  and  I  owe  much   of   my   success   thus   far   to   his   guidance.   Cristina   Bodea   provided   helpful   comments   on   dissertation  manuscripts  to  sharpen  my  arguments.  I  was  also  very  fortunate  to  have  the   opportunity   to   work   with   her   on   multiple   research   projects   and   could   learn   an   important  part  of  academic  life  -­‐-­‐  how  to  publish  work  in  journals.  Ani  Sarkissian  agreed   to  be  my  mentor  when  I  arrived  at  MSU  and  continued  to  serve  as  a  committee  member.   She  gave  practical  tips  and  useful  suggestions  for  me  in  how  to  conduct  fieldwork  and   interviews  in  Central  Asia,  as  well  as  good  advice  to  write  competitive  funding  proposals.   Also   serving   on   my   dissertation   committee,   Christian   Houle   has   always   pushed   me   to   think  harder  about  the  logic  behind  my  arguments  and  provided  useful  feedback  on  my   empirical  analyses.              Beyond   my   dissertation   committee,   my   gratitude   goes   to   other   members   of   MSU   faculty   and   staff.   Ben   Appel,   Ravi   Bhavnani,   Mike   Bratton,   Mike   Colaresi,   Jeff   Conroy-­‐ Krutz,   Eric   Freedman,   Brian   Silver,   and   Jakana   Thomas   have   been   important   teachers   who   provided   invaluable   feedback   and   suggestions   on   my   research   projects.   The     vi   Department  Chair,  Charles  Ostrom,  and  three  Graduate  Directors,  Steve  Kautz,  Melinda   Hall,   and   Thomas   Hammond,   provided   me   with   generous   financial   support   to   conduct   several   field   trips   and   make   presentations   at   various   academic   conferences.   Karen   Battin   helped   me   with   every   administrative   need,   whereas   Rhonda   Burns   and   Sarah   Krause  processed  my  NSF  grant  by  patiently  making  a  long  list  of  receipts  that  I  received   in  the  field.  I  would  also  like  to  express  my  special  thanks  to  my  MSU  colleagues  for  their   friendship,   thoughts   and   careful   proofreading   on   my   papers:   Fang-­‐Yu   Chen,   Hyun   Jin   Choi,   Tolgahan   Kinay   Dilgin,   Ebru   Eren-­‐Webb,   Sung   Min   Han,   Dan   Hansen,   Petra   Hendrickson,   Shih-­‐Hao   Huang,   Brian   Kennedy,   Nicholas   Kerr,   Alon   Kraitzman,   Bob   Lapton,  Jerry  Lavery,  Helen  Lee,  Hsin-­‐Hsin  Pan,  Chunho  Park,  Peter  Penar,  Wen-­‐Chin  Wu,   and  Fangjin  Ye.                                Outside  of  MSU,  I  would  like  to  thank  Allyson  Benton,  Annette  Fath-­‐Lihic,  Rich  Frank,   Rob  Franzese,  Olli  Hellmann,  Sarah  Hummel,  Yusaku  Horiuchi,  Pauline  Jones  Luong,  Tom   Le,   Mike   Miller,   Svend-­‐Erik   Skaaning,   Daniel   Stockemer,   Netina   Tan,   Jan   Teorell,   Matt   Wilson,  and  Matt  Winters  for  their  invaluable  comments,  feedback,  and  encouragements   on   my   dissertation   research   at   various   stages.   The   Electoral   Integrity   Project   (EIP)   led   by   Pippa   Norris   kindly   invited   me   to   contribute   one   of   my   dissertation   chapters   to   a   forthcoming   edited   volume,   as   well   as   gave   me   several   chances   to   present   part   of   my   dissertation  project  at  EIP  workshops,  for  which  I  really  appreciate.            Without   my   professors   in   Japan,   I   would   not   have   even   thought   of   studying   in   the   United  States  for  my  doctoral  degree.  I  would  like  to  thank  Hideko  Magara,  Takayuki  Ito,   Masaru  Kohno,  and  Ikuo  Kume  for  their  moral  support  and  mentoring  since  I  began  to   study   Political   Science   at   Waseda   University.     I   would   also   like   to   thank   Takeshi   Iida,     vii   Yuko  Kasuya,  Takeshi  Kawanaka,  Keiichi  Kubo,  Ryo  Nakai,  Katsunori  Seki,  Kengo  Soga,   Atsushi   Tago,   Shin   Toyoda,   and   Yuki   Yanai   for   their   comments   and   suggestions   on   my   dissertation  project.                  Over  the  years,  I  have  been  supported  by  the  following  fellowships  and  grants  which   allowed   me   to   conduct   language   study,   field   research,   and   dissertation   writing:   National   Science  Foundation  Dissertation  Improvement  Grant  (#1323671),  Fulbright  Scholarship,   Akino   Yutaka   Award,   Tokyo   Foundation   SYLFF   Research   Abroad   Award,   Suntory   Foundation   Research   Grant   for   Young   Scholars,   Konosuke   Matsushita   Memorial   Foundation   Research   Grant,   Waseda   University   GLOPEII   Research   Grant,   MSU   Political   Science  Department  Fellowship,  and  MSU  Dissertation  Completion  Fellowship.              I   conducted   fieldwork   in   Almaty   and   Astana,   Kazakhstan   and   Bishkek,   Kyrgyzstan.   I   am  grateful  to  Department  of  Public  Administration  at  KIMEP  University  in  Almaty  and   the   Social   Research   Center   at   American   University   of   Central   Asia   in   Bishkek   for   sponsoring   me   while   I   was   a   researcher   in   the   field.   I   would   like   to   thank   Meruert   Makhmutova,  Ellina  Malina,  Nurseit  Niyazbekov,  and  Medet  Tiulegenov  for  introducing   me   to   politicians,   researchers   and   NGO   activists   whom   I   should   interview.   Kamila   Mustafina,   Gulden   Suleimenova,   Aizhamal   Sydykova,   Bermet   Zhumakadyr,   and   Kalys   Zhumakadyr   provided   valuable   research   assistance.   I   am   grateful   as   well   to   many   politicians,  researchers  and  opposition  leaders  who  agreed  to  have  interviews  with  me,   as   well   as   a   lot   of   citizens   in   both   countries   who   talked   frankly   about   what   they   think   about   politics   in   their   countries.   This   research   was   approved   by   the   MSU   Office   of   Regulatory  Affairs  (IRB#x13-­‐590e)           viii            Last  but  not  least,  I  would  never  have  completed  this  dissertation  without  my  family’s   support.   Soon   after   having   married   me,   my   wife,   Yuko,   came   with   me   to   the   States   for   my   PhD   study.   Since   then,   she   has   sacrificed   everything   for   me.   When   facing   the   numerous   difficulties   during   this   journey,   she   was   always   far   more   optimistic   and   positive   than   myself   and   enjoyed   life   in   Michigan   on   her   own   way.   Shuh   and   Soh,   our   sons   who   were   born   in   Michigan,   have   been   an   important   source   of   energy   as   well   as   good   timekeepers   when   I   worked   on   this   dissertation.   My   parents-­‐in-­‐law,   Akiko   and   Takehide,  have  been  very  supportive  and  helped  our  endeavor  to  the  States  in  various   ways.  My  grandmother,  Kimie,  cheerfully  welcomed  us  every  time  when  we  came  back   to  Fukuoka  and  played  with  our  kids.  Finally,  my  parents,  Kumiko  and  Toshiharu,  have   always  believed  in  me  and  let  me  do  what  I  really  want  to  do  since  I  was  little.  I  would   like   to   dedicate   this   dissertation   to   my   parents   for   their   love   and   steadfast   support   over   the  past  three  decades.                                           ix   TABLE  OF  CONTENTS       LIST  OF  TABLES……………………………………………………………...…………………....xiii     LIST  OF  FIGURES………………………………………………..…………………………..…….xvi     CHAPTER1  INTRODUCTION…………………………………………………...........................1            1.1  Three  Puzzles  of  Authoritarian  Elections…………..…………………………….8                    1.1.1  Electoral  Fraud……………………………………………..……………………….....8                    1.1.2  Electoral  System  Change……………………………….………………………..10                    1.1.3  Post-­‐Electoral  Political  Order………………………….………………………11              1.2  Theory  and  Argument  in  Brief……………………………………………………...13            1.3  Methods  and  Research  Design……………………………………………………...19            1.4  Contributions………………………………………………………………………………22            1.5  Outline  of  the  Dissertation……………………………………………………………25       CHAPTER  2  ELECTORAL  DILEMMA  AND  THE  MANIPULATION  OF   ELECTIONS………….................................................................................................................27            2.1  Introduction………………………………………………………………………………...27            2.2  The  Electoral  Dilemma  in  Dictatorship………………………………………….30            2.3  Electoral  Fraud  and  the  Manipulation  of  Policy  Instruments………….35              2.4  Power  Distribution  between  the  Dictator  and  Elites………………………38                      2.4.1  Financial  Resources……………………………………………………………….40                      2.4.2  Organizational  Bases…………………………………………………………......43                      2.4.3  Opposition’s  Strength…………………………………………………………….45                          2.5  Cross-­‐National  Statistical  Analysis  of  Electoral  Fraud……………………47                      2.5.1  Data  and  Methodology…………………………………………………..……….47                        2.5.2  Results……………………………………………………………………………….....57                      2.5.3  Robustness  Check  and  Additional  Data  Analysis……………………..61                  2.6  Conclusion…………………………………………………………………………………..67     CHAPTER  3  THE  CHOICE  OF  ELECTORAL  SYSTEMS  IN   DICTATORSHIPS………………………………………………………………..………………….69              3.1  Introduction………………………………………………………………………………..69                3.2  Literature  Review………………………………………………………………………..73              3.3  Electoral  System  Change  in  Electoral  Authoritarianism.………………..79              3.4  Cross-­‐National  Statistical  Analysis….…………………………………...……….86     x                      3.4.1  Sample:  Electoral  Authoritarianism………………………………………..86                      3.4.2  Dependent  Variable:  Effective  Electoral  Threshold……...………….88                      3.4.3  Key  Independent  Variable:  Natural  Resource  Wealth………………90                      3.4.4  Statistical  Method……………………………………………………...…………..92                      3.4.5  Results………………………………………………………………………...………..94                      3.4.6  Robustness  Check………………………………………………………………….95                3.5  Testing  Causal  Mechanisms…………………………………………….…………..96                      3.5.1  Natural  Resource  Wealth  and  Authoritarian  Regime  Support.....97                      3.5.2  Electoral  Systems  and  the  Pro-­‐Dictator  Bias………….……….….….102                      3.5.3  PR  Systems  Inhibit  Pre-­‐Electoral  Opposition  Coalitions….……..108                3.6  Conclusion…………………………………………………………………...…………..111       CHAPTER  4  POLITICAL  MANIPULATION  OR  FISCAL  MANEUVERING?  THE   CASE  OF  KAZAKHSTAN….....…………………………………………………………………114                    4.1  Introduction…………………………………………………………………………....114                  4.2  Limited  Liberalization……………………………………………………………...116                  4.3  Growing  Dominance  of  Ruling  Parties  Since  1995……………………..121                  4.4  Electoral  Manipulation  in  Kazakhstan……………………...……………….126                          4.4.1  Electoral  Fraud…………………………………………………………………..127                          4.4.2  Electoral  System  Change:  From  SMD  to  PR……………………….…135                                  4.5  The  Dictator’s  Mobilization  Power……………………………………………140                          4.5.1  Natural  Resource  Wealth  as  a  Source  of  Patronage……….……..143                          4.5.2  Centralization  of  Governing  Institutions………………...…….……..152                            4.5.3  The  Dominant  Party,  Nur  Otan………………………………...…………157                                    4.5.4  Weakening  of  Opposition……………………………………...……………159                      4.6  Political  Business  Cycles  in  Kazakhstan……………………...…………….164                    4.7  Conclusion………………………………………………………...……………………172     CHAPTER  5  PROTESTS  AND  LEADERSHIP  TURNOVER  AFTER   AUTHORITARIAN  ELECTIONS……………………………………………………………..173                      5.1  Introduction……………………………………………………………..……………173                      5.2  Literature  Review…………………………………………………………………..178                      5.3  Dictator’s  Calculus  over  Electoral  Manipulation……...……………….181                      5.4  Backfiring  at  the  Ballot  Box…………………………………………………….184                      5.5  Cross-­‐National  Statistical  Analysis…………………………………………..189                                5.5.1  Data  and  Modeling  Strategies………………………...…………………189                                  5.5.2  First  Model  Specification:  A  Mobilization  Model…...……….….191                                5.5.3  Second  Model  Specification:  Turnover  and  Protests……….…196     xi                                5.5.4  Results…………………………………….…………………………...…………199                                5.5.5  Robustness  Check...………………….……………………………..………..202                        5.6  Conclusion……………………………………………….……………………………203     CHAPTER  6  THE  FAILURE  OF  STRATEGIC  FRAUD  AND  POPULAR   PROTESTS:  THE  CASE  OF  KYRGYZSTAN……………………………………………….204                        6.1  Introduction……………………………………………………………………...…..204                          6.2  From  Electoral  Democracy  to  Electoral  Authoritarianism………..206                        6.3  The  Weakening  of  Akaev’s  Mobilization  Power……………………….211                                6.3.1  Financial  Resources:  Gold  and  Aid……………………...………….…212                                6.3.2  Decentralization  and  Lack  of  Organizational  Bases……………215                                6.3.3  Emerging  Opposition…………………………………………...…………..220                          6.4  Electoral  Fraud  and  the  Rise  of  Massive  Protests:  Comparing                                          Elections  in  the  Akaev  Regime………………………………...…………….222                                6.4.1  The  1995  Elections:  Relatively  Fair  Elections  and  the  Durable                                          Akaev  Regime……………………………………………………………...……….222                                6.4.2  The  2000  Elections:  Electoral  Manipulation  and  the  Absence  of                                            Large-­‐Scale  Protests…………………………………………………...………...224                                  6.4.3  The  2005  Elections:  Electoral  Fraud,  Popular  Protests  and  the                                          Collapse  of  the  Akaev  Regime………………………………………………..226                              6.5  Conclusion………………………………………………………………...………...229     CHAPTER  7  CONCLUSION……...…………….…………………………………...………….231       APPENDIX….…..………………………………………………………………………………...…238     BIBLIOGRAPHY….………………………………………………………………………………..270                           xii   LIST  OF  TABLES         Table  2.1:  Determinants  of  Electoral  Fraud  in  Dictatorship……...……...………57     Table  3.1:  Determinants  of  Electoral  Systems  in  Electoral   Authoritarianism…………………………………………………………………………………..94     Table  3.2:  Dictators’  Performance  at  the  Ballot  Box……………...………...…..…100     Table  3.3:  The  Pro-­‐Dictator  Bias  under  Majoritarian  Electoral  Systems.…104     Table  3.4:  Electoral  Systems  and  Pre-­‐Electoral  Coalitions  in  Electoral   Authoritarian  Regimes…………...……………………………………………………………109     Table  4.1:  Time-­‐Series  Change  in  Electoral  Fraud  in  Kazakhstan  (1995-­‐ 2007)………………………………………………………………………………………………….127     Table  4.2:  Electoral  System  Change  and  Effective  Electoral  Threshold  in   Kazakhstan………………………………………………………………………………………….135     Table  4.3:  Percentage  of  Central  Government  Transfers  in  Total  Revenues  of   Regional  Governments…………………………………………………………………..…….152     Table  4.4:  Political  Business  Cycles  in  Kazakhstan  (1995-­‐2008)………….…167     Table  4.5:  Electoral  Cycles,  Organizational  Strength,  and  Fiscal   Resources................................................................................................................................168         Table  5.1:  The  First-­‐Stage  Model  Predicting  Electoral  Fraud………………….194     Table  5.2:  Probit  Analysis  of  Post-­‐Electoral  Turnover  and  Protests………..199     Table  C2-­‐1:  Descriptive  Statistics  of  Chapter  2……………………….……………..238     Table  C2-­‐2:  List  of  Authoritarian  Countries  (1977-­‐2004)………………………239       xiii   Table  C2-­‐3:  Additional  Analyses…………………………………………………….…….240     Table  C2-­‐4:  Alternative  Definition  of  Political  Regimes  –  Boix,  Miller  and   Rosato  (2012)……………………………………………………………………………………..241     Table  C2-­‐5:  Alternative  Definition  of  Political  Regimes  –  Polity  IV…………242     Table  C2-­‐6:  Alternative  Measure  of  Natural  Resources  –  Change  in  Oil-­‐Gas   Value  per  capita…………………………………………………………………………………..243     Table  C2-­‐7:  Alternative  Measure  of  Natural  Resources  –  3  Years  Moving   Average  of  Oil-­‐Gas  Value  per  capita………………………………………………………244     Table  C2-­‐8:  Alternative  Measure  of  Natural  Resources  –  Haber  and  Menaldo   (2011)………………………………………………………………………………………...………245     Table  C2-­‐9:  Region-­‐Clustered  Robust  Standard  Errors………………………….246     Table  C2-­‐10:    Additional  Controls…………………………………………………...……247     Table  C2-­‐11:  Alternative  Methods  –  Fixed  Effects  and  Regional  Specific   Effects…………………………………………………………………………………………………248     Table  C2-­‐12:  Alternative  Methods  –  Random  Intercept  Models……………..249     Table  C3-­‐1:  Descriptive  Statistics  of  Chapter  3……………………………………...252         Table  C3-­‐2:  List  of  Electoral  Authoritarian  Countries  (1946-­‐2007)…..……253     Table  C3-­‐3:  Including  Polity  IV  as  a  Control  Variable……………………………..254     Table  C3-­‐4:  Region-­‐Clustered  Robust  Standard  Errors………………………….255     Table  C3-­‐5:  Using  a  Different  Measure  of  the  Effective  Electoral  Threshold   …………………………………………………………………………………………………………...256     Table  C3-­‐6:  Limiting  Sample  into  1970-­‐2007………………………………………..257       xiv   Table  C5-­‐1:  Descriptive  Statistics  of  Chapter  5…………………………………...…259     Table  C5-­‐2:  Alternative  Measure  of  Political  Regimes……………………………261     Table  C5-­‐3:  Using  the  Electoral  Fraud  Variable  per  se…………………...………262     Table  C6-­‐1:  Political  Business  Cycles  in  Kyrgyzstan  (1995-­‐2013)…………..268                                                                     xv     LIST  OF  FIGURES         Figure  1.1:  Democracies  and  Autocracies  in  the  World  (1945-­‐2010)…………3     Figure  1.2:  Elections  in  Authoritarian  Regimes  (1946-­‐2008)……………………..4     Figure  2.1:  Variation  in  the  Level  of  Electoral  Fraud  in  Authoritarian   Regimes  (1977-­‐2004)………………………………………………………………………...….48     Figure  2.2:  The  Different  Effects  of  Natural  Resource  Endowments  between   Dominant-­‐Party  and  Non-­‐Dominant-­‐Party  Regimes……………………………..…59     Figure  2.3:  The  Marginal  Effect  of  Natural  Resource  Endowments   Conditional  upon  Ethnic  Organizational  Power……………………………………....60     Figure  2.4:  Additional  Analysis……………………………………………………………....65     Figure  3.1:  Hypothesis  and  Causal  Mechanisms  for  the  Origins  of  Electoral   Systems  in  Electoral  Authoritarianism………………………………………...……..…..83      Figure  3.2:  Effective  Electoral  Threshold  in  Electoral  Authoritarian   Regimes………………………………………………………………………………………………..89     Figure  3.3:  Regional  Variations  in  the  Effective  Electoral  Threshold  in   Electoral  Authoritarianism………………..……………………………………………...……90     Figure  3.4:  The  Magnitude  of  the  Pro-­‐Dictator  Bias  Conditional  upon   Regime  Strength  (Vote  Shares  of  Ruling  Parties)…………………….…………….106     Figure  3.5:  Effect  of  Electoral  Systems  on  Pre-­‐Electoral  Opposition   Coalitions.…………………………………………………………………………………….…..…110     Figure  4.1:  Increasing  Dominance  of  Regime  Parties  in  Kazakhstan……….123       xvi   Figure  4.2:  Effective  Electoral  Threshold  in  Kazakh  Parliamentary  Elections   (1994-­‐2007)……………………………………………………………………………………….136     Figure  4.3:  Economic  Growth  and  Fiscal  Revenues  in  Kazakhstan  (1993-­‐ 2008)……………………………………………………………………………………………...…..143     Figure  4.4:  Natural  Resource  Wealth  in  Kazakhstan  (1985-­‐2008)……….…144     Figure  4.5:  Coherence  of  Ruling  Parties  and  Proportion  of   Independents………………………………………………………………………………………154     Figure  4.6:  Time-­‐Series  Change  in  Consumer  Price  Index………………………165     Figure  4.7:  Political  Business  Cycles,  Organizational  Strength,  and  Financial   Resources……………………………………………………………………………………………170     Figure  5.1:  Gap  in  Electoral  Fraud  under  Dictatorship………………......………196     Figure  5.2:  Predicted  Probabilities  of  Turnover  and  Protests……...…………200     Figure  C5-­‐1:  Histogram  of  Predicted  Values  of  Electoral  Fraud……………...260     Figure  6-­‐1:  Time-­‐Series  Change  in  Inflation  rates  in  Kyrgyzstan……………267   xvii   CHAPTER  1   INTRODUCTION                This   project   explores   the   causes   and   consequences   of   elections   in   authoritarian   regimes.1  Since   the   end   of   World   War   II,   political   scientists   have   explored   various   aspects   of   political   institutions   in   democracies.   Some   scholars   did   draw   their   attention   to   the   politics   of   non-­‐democracies   (e.g.   Arendt   1951;   Friedlich   and   Brzezinski   1965;   Moore   1966;   O’Donnell   1973;   Linz   1977;   Collier   ed.   1979;   Tullock   1987),2  yet   their   focus   was   largely   on   non-­‐institutional   aspects   of   autocratic   rule,   implicitly   assuming   that   political   institutions   under   dictatorships   are   no   more   than   just  “window  dressing”  (Gandhi  2008).              After   the   "third   wave   of   democratization”   (Huntington   1991)   that   swept   over   authoritarian   regimes   in   Southern   Europe,   Latin   America,   sub-­‐Saharan   Africa   and   Eastern   Europe,   scholars   shifted   their   interests   to   determinants   of   democratic   transitions  and  consolidation  (Dahl  1972;  O’Donnell  and  Schmitter  1986;  Przeworski                                                                                                                   1 Throughout   the   dissertation,   I   use   terms   autocracies,   dictatorships,   authoritarianism,  and  non-­‐democracies  interchangeably.  By  dictators  and  autocrats,  I   refer   to   top   political   leaders   who   hold   de   jure   supreme   authority   in   authoritarian   regimes.  I  use  the  male  pronoun  to  refer  to  authoritarian  leaders,  given  the  fact  that,   according   to   Archigos   Version   2.9,   99.7   percent   of   political   leaders   in   authoritarian   regimes  after  World  War  II  have  been  male.           2  Important   exceptions   are   Hermet,   Rose   and   Rouquie   eds.   (1978)   and   Friedgut   (1979)   in   which   they   explore   elections   in   communist   countries   and   other   authoritarian   regimes   in   Europe   and   Africa.   Huntington   and   Moore   (1969),   on   the   other  hand,  focus  on  one-­‐party  rule  that  were  proliferated  in  the  1960s.             1   1991;   Haggard   and   Kaufmann   1995;   Linz   and   Stepan   1996;   Bratton   and   van   de   Walle   1997;  Przeworski  et  al.  2000;  Boix  2003;  Acemoglu  and  Robinson  2006).  By  the  end   of  the  1990s,  researchers  began  to  notice  that  there  were  still  considerable  numbers   of   authoritarian   regimes   surviving   in   the   era   of   democratization.   This   observation   encouraged   researchers   to   rethink   how   authoritarian   politics   truly   operates   in   the   contemporary   world   (Wintrobe   1998;   Geddes   1999;   Schedler   2002;   Levitsky   and   Way  2002;  Magaloni  2006;  Gandhi  2008;  Blaydes  2010;  Svolik  2012).                  Figure   1.1   shows   time   series   change   in   percentages   of   democracies   and   autocracies   around   the   world   between   1945   and   2010.   Although   the   number   of   democratic  countries  has  grown  since  the  late  1970s,  there  are  still  a  large  number  of   authoritarian   countries,   comprising   more   than   40   percent   of   all   countries   in   the   world.   Figure   1.2   looks   more   closely   at   time-­‐series   changes   in   three   types   of   authoritarian   regimes:   autocracies   with   no   elections,   autocracies   with   façade   elections,   and   autocracies   with   semi-­‐competitive   elections.   The   number   of   autocracies   holding   semi-­‐competitive   elections   rapidly   increased   particularly   after   the   end   of   the   Cold   War.   Although   semi-­‐competitive   autocracies   consisted   of   20   percent   of   all   authoritarian   regimes   as   of   1975,   they   were   more   than   65   percent   in   2008.  On  the  other  hand,  autocracies  with  either  façade  elections  or  no  elections  at  all   decreased   during   the   same   period   of   time.   In   1975,   each   of   the   two   authoritarian   regime  types  constituted  40  percent  of  all  authoritarian  regimes,  respectively.  But  the   end   of   the   Cold   War   terminated   the   dominance   of   these   two   regime   types.   In   2008,     2   only   20   percent   of   autocracies   did   not   hold   elections,   whereas   only   10   percent   of   authoritarian  elections  did  not  allow  multiple  parties  and  candidates  to  participate  in   electoral  processes.                       0 Percent of Countries (%) 20 40 60 80 100 Figure  1.1:  Democracies  and  Autocracies  in  the  World  (1945-­‐2010)   1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Autocracy Democracy   Note:   The   dichotomous   distinction   between   democracy   and   autocracy   is   based   on   Cheibub,   Gandhi   and   Vreeland   (2009).   According   to   their   definition,   a   country   is   regarded  as  democratic  if  the  country  satisfies  the  following  four  conditions:  (1)  the   executive  is  elected,  (2)  the  legislature  is  elected,  (3)  there  is  more  than  one  political   party,  and  (4)  an  incumbent  regime  has  lost  power.  Otherwise,  the  country  is  deemed   authoritarian.             3   Proportion of Authoritarian Countries (%) 0 20 40 60 80 100 Figure  1.2:  Elections  in  Authoritarian  Regimes  (1946-­‐2008)   1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Year No Elections Semi-Competitive Elections Facade Elections   Note:   A   country   is   regarded   as   holding   semi-­‐competitive   elections   if   either   legislative   or   executive   elections   allow   multiple   political   parties   and   candidates.   If   legislative   and/or  executive  elections  are  held  with  only  one  single  party  or  candidate,  then  the   country   is   defined   as   an   authoritarian   regime   holding   façade   elections.   If   a   country   holds   neither   legislative   nor   executive   elections,   it   is   an   authoritarian   regime   with   no   elections.   Source:  Svolik  (2012)                Subsequent   to   the   proliferation   of   regimes   that   are   now   referred   to   as   “electoral   authoritarian   regimes”   (Schedler   2013)   or   “competitive   authoritarianism”   (Levisky   and   Way   2010),   researchers   started   to   explore   the   roles   of   political   institutions   in   authoritarian   politics.     Many   argued   that   parties,   legislatures   and   elections   play     4   various  roles  in  benefiting  authoritarian  rule.3  In  founding  a  dominant  party,  dictators   can   institutionalize   ruling   elite’s   career   promotion   and   create   long-­‐lasting   power   sharing   between   dictators   and   ruling   elites.   Thus,   the   establishment   of   a   dominant   party  helps  authoritarian  regimes  survive  longer  (Magaloni  2008;  Svolik  2012;  Boix   and  Svolik  2013).  Dominant  party  organizations  also  allow  dictators  to  mobilize  mass   support   (Huntington   1968)   by   constructing   well-­‐developed   patronage   networks   which  contribute  to  the  efficiency  of  economic  distribution  (Geddes  2006;  Brownlee   2007).   The   legislature   serves   as   a   useful   access   point   for   political   elites   to   various   privileges,   spoils   and   policy   concessions   conferred   by   the   regime   (Gandhi   and   Przeworski   2007;   Lust-­‐Okar   2008;   Blaydes   2011).   As   a   result,   such   co-­‐optation   mechanisms   allow   dictators   to   have   a   firm   grip   on   power.   From   the   perspective   of   citizens,   authoritarian   parliaments   may   become   a   main   channel   of   political   accommodation  and  an  important  source  of  material  favors  (Lust-­‐Okar  2008).              Of   all   the   political   institutions   under   dictatorship,   this   dissertation   analyzes   authoritarian   elections.   Compared   to   legislatures   and   dominant   parties,   the   impacts   and   implications   of   authoritarian   elections   are   still   hotly   contested   among   researchers.   Thus   far,   there   are   two   broadly   conflicting   views   on   authoritarian   elections   in   the   growing   literature   of   authoritarian   politics.   One   strand   of   research   argues   that   authoritarian   elections   help   dictators   hold   onto   power,   in   three   ways.                                                                                                                   3  For  exhaustive  reviews  on  the  roles  of  political  institutions  under  dictatorships,  see   Gandhi   and   Lust-­‐Okar   (2009),   Magaloni   and   Kricheli   (2010)   and   Ezrow   and   Frantz   (2011).       5   First,   by   winning   elections   with   large   margins,   authoritarian   leaders   can   show   that   their   regimes   are   invincible   (Geddes   2006;   Magaloni   2006;   Simpser   2013).   The   demonstration   of   regime   strength   at   the   ballot   box   prevents   political   elites   from   launching   coups,   assassinating   the   dictators,   and   leading   popular   uprisings.   Second,   election   results   may   also   provide   dictators   with   useful   information   on   (1)   ruling   elites'  loyalty  toward  the  regime  and  their  capability  of  mobilizing  a  large  number  of   supporters,   (2)   the   areas   where   voters   are   more   enthusiastically   supporting   the   current   regime,   and   (3)   the   strongholds   of   opposition   parties,   where   these   parties   may   be   able   to   mobilize   their   supporters   against   the   regime.   Based   on   such   information   gleaned   from   election   results,   dictators   can   more   discriminately   apply   repression   and   more   efficiently   allocate   economic   resources   to   cement   political   support   and   prevent   defection   (Ames   1970;   Shi   1999;   Malesky   and   Schuler   2010;   Reuter   and   Robertson   2011;   Miller   2013).   Third,   authoritarian   elections   may   effectively   “divide-­‐and-­‐rule”   the   opposition.   Since   moderate   opposition   parties   participate   in   regime-­‐sponsored   elections   whereas   radical   opposition   parties   boycott   the   elections,   elections   may   enable   authoritarian   leaders   to   generate   coordination   problems  among  the  opposition  camp    (Lust-­‐Okar  2004;  Beaulieu  2006).   On   the   other   hand,   other   scholars   have   long   contended   that   elections   in   dictatorships  may  pave  the  way  for  democratization  or  result  in  violent  conflict  and   protests   against   the   regime.   Observing   the   so-­‐called   third   wave   of   democratization,   Huntington  (1991,  174)  suggests  that  holding  elections  under  authoritarian  rule  is  an     6   important  first  step  for  a  dictatorship  in  its  transition  to  democracy.    More  recently,   Lindberg  (2006)  has  argued  that,  the  more  elections  a  country  holds,  the  more  likely   it  is  to  democratize  and  improve  the  quality  of  democracy  particularly  in  the  context   of   Africa.   Similarly,   Roessler   and   Howard   (2009)   and   Brownlee   (2009)   argue   that   competitive   authoritarian   regimes   are   more   likely   to   democratize   than   both   hegemonic   and   closed   authoritarian   regimes.   Elections   not   only   encourage   democratization   but   also   popular   protests   and   civil   war.   Tucker   (2007),   Kuntz   and   Thompson   (2009),   and   Kalandadze   and   Orenstein   (2009)   all   assert   that   fraudulent   elections   are   more   likely   to   be   followed   by   anti-­‐government   opposition   movements   and   massive   popular   protests,   as   exemplified   by   the   Color   Revolutions   in   the   post-­‐ Soviet   region   and   electoral   revolutions   in   some   African   countries.   Such   increased   tension   may   also   lead   to   a   violent   confrontation   between   the   government   and   the   opposition,   and   can   even   trigger   civil   war   (Snyder   2000;   Cederman,   Gleditsch,   and   Hug  2012).                        How  can  we  understand  the  two  contrasting  conclusions  that  extant  work  derives   on   elections   under   authoritarian   rule?   This   dissertation   attempts   to   reconcile   these   two   literatures   on   authoritarian   elections   both   theoretically   and   empirically.   In   doing   so,  I  first  provide  a  new  theory  of  authoritarian  elections  and  aim  to  shed  light  on  a   dilemma   that   dictators   face   at   the   ballot   box:   If   dictators   refrain   from   manipulating   elections,   they   can   obtain   the   informational   benefits   described   above.   Free   and   fair   elections,  however,  decrease  the  probability  that  dictators  win  big,  which  may  result     7   in   revealing   their   weaknesses,   or   unpopularity.   After   centering   on   this   dilemma   in   autocratic   elections,   what   this   dissertation   refers   to   as   “the   electoral   dilemma   in   dictatorships,”   I   then   derive   observable   implications   and   empirically   test   the   hypotheses   by   employing   cross-­‐national   statistical   analyses   as   well   as   comparative   case  studies  of  two  authoritarian  regimes  in  post-­‐Soviet  Central  Asia.                This   dissertation   first   investigates   the   causes   of   autocratic   elections,   i.e.,   why   dictators   determine   the   degree   of   electoral   fraud   in   the   way   they   do   and   why   they   choose   one   type   of   electoral   system   over   others.   I   then   explore  the   consequences  of   authoritarian   elections   by   looking   at   the   question   of   whether   elections   have   a   significant  effect  on  political  stability  in  authoritarian  regimes  by  deterring  leadership   turnover   and   popular   protests.   More   specifically,   this   dissertation   attempts   to   solve   three  puzzles  on  elections  and  political  order  in  autocracies.                                     1.1  Three  Puzzles  of  Authoritarian  Elections     1.1.1  Electoral  Fraud            The   first   research   question   addressed   in   this   dissertation   is   what   explains   variation  in  electoral  fraud  under  authoritarian  regimes.  I  here  define  electoral  fraud   as   a   series   of   illegal   measures   that   bias   election   results   in   favor   of   the   political   leader   (Lehoucq   2003).   Based   on   this,   I   analyze   the   following   three   subcomponents   of   electoral   fraud:   (1)   election   violence,   (2)   election   cheating,   and   (3)   undemocratic     8   restrictions   on   electoral   law.   Election   violence   is   physical   intimidation   during   elections   exercised   largely   by   incumbent   parties   (Straus   and   Taylor   2012;   Hafner-­‐ Burton,   Hyde,   and   Jablonski   2013).   Governments   tend   to   use   electoral   violence   to   threaten  opposition  candidates  and  citizens.  In  so  doing,  dictators  aim  to  undermine   the   efficacy   of   oppositions’   political   campaigns   and   discourage   opposition   supporters   from   going   out   to   the   polling   stations.   Electoral   cheating   allows   dictators   to   affect   the   number   of   votes   during   campaign   periods   and   election   days   through   nonviolent   fraudulent   measures   such   as   constraints   on   the   freedom   to   campaign,   media   bias,   ballot  stuffing,  vote-­‐buying,  and  intimidation  (Kelley  2012).  Restrictions  on  electoral   law   refers   to   a   series   of   regulations   that   prevent   citizens   and   electoral   candidates   from   influencing   politics,   including   limits   on   voting   rights   by   certain   social   categories   like   gender   and   ethnicity,   intentional   flaws   in   the   complaints   procedures,   high   thresholds  for  new  parties  to  get  registered  and/or  to  gain  seats,  constraints  on  the   right  to  run  for  office  such  as  language  and  educational  requirements  (Kelley  2012).   These   manipulation   techniques   are   considerably   different   in   the   way   in   which   they   take   place,   but   all   “boost   up”   regime   support   artificially,   and   contribute   to   the   continuation  of  dictators’  overwhelming  majority  (Simpser  2013).                    Although  authoritarian  elections  do  not  allow  government  alternation  and  thus  are   not   fully   fair   and   free,   there   tends   to   be   a   high   degree   of   variation   in   the   electoral   fairness  across  the  authoritarian  regimes  around  the  world.  This  is  puzzling,  since  it   is  often  assumed  that  authoritarian  leaders  are  always  rigging  elections  seriously  by     9   using   their   dominant   political   power.   This   raises   a   crucial   question   in   our   understanding   of   authoritarian   politics:   Why   do   some   dictators   refrain   from   using   extensive  electoral  manipulation,  despite  the  fact  that  they  are  entitled  to  such  power  in   their  regimes?                   1.1.2  Electoral  System  Change            As   already   discussed,   electoral   fraud   is   an   illegal   strategy   to   bias   election   results   favorable  to  authoritarian  leaders.  Authoritarian  rulers,  however,  can  also  use  other   manipulation   techniques   to   bring   an   overwhelming   electoral   victory   to   themselves.   The   manipulation   of   electoral   systems  is  one  of  such  measures  by  which  dictators  can   increase   the   likelihood   of   winning   big   without   resorting   to   relentless,   blatant   electoral   fraud.   Many   researchers   have   focused   on   electoral   systems   in   democratic   countries   or   transitioning   states,   implicitly   presuming   that   electoral   systems   in   dictatorships   may   not   have   any   meaningful   effects   (e.g.   Boix   1999;   Cusack,   Iversen,   and  Soskice  2006;  Calvo  2009).  Therefore,  extant  work  has  not  sufficiently  explored   when  dictators  are  more  inclined  to  adopt  one  electoral  system  over  others.  As  this   dissertation   shows,   however,   electoral   systems   are   considerably   different   across   authoritarian   regimes.   Particularly   in   regimes   where   authoritarian   leaders   permit   opposition   parties   to   participate   in   electoral   processes,   authoritarian   rulers   strategically  choose  their  electoral  systems.  Hence,  there  is  wide  variation  in  electoral   system   types:   Some   electoral   autocrats   choose   Single   Member   Districts   (SMD)     10   systems  in  which  only  one  candidate  with  a  plurality  of  votes  is  elected  in  an  electoral   district.   Others   adopt   more   Proportional   Representation   (PR)-­‐based   systems   where   voters  cast  their  votes  for  a  political  party  in  a  relatively  larger  electoral  district  and   parties  are  given  seats  more  or  less  proportionally  according  to  the  votes  they  obtain.              More   importantly,   this   variation   poses   an   intriguing   puzzle   on   electoral   system   designs.  If  we  assume  that  SMD  systems  bring  a  large  seat  premium  to  big  parties,  as   often   happens   in   democracies,   authoritarian   leaders   should   always   choose   majoritarian  electoral  systems.  Since  the  ruling  party  is  a  major  party  in  most  cases  of   electoral  authoritarianism,  the  party  should  be  able  to  enjoy  such  a  seat  premium.  In   reality,  however,  some  authoritarian  countries  continue  to  adopt  PR  systems  or  even   decide   to   change   their   electoral   systems   from   SMD   to   PR.   This   brings   about   an   important   puzzle   that   this   thesis   aims   to   answer:   Why  do  dictators  often  dare  to  adopt   a  less  advantageous  electoral  system,  given  that  majoritarian  electoral  systems  typically   help  them  score  an  overwhelming  electoral  victory?       1.1.3  Post-­‐Electoral  Political  Order            The   final   puzzle   of   authoritarian   elections   that   I   will   explore   is   more   directly   concerned   with   the   different   views   on   authoritarian   elections:   Why   do   autocratic   elections,   which   must   have   been   designed   by   the   dictators   themselves,   sometimes   backfire  on  them?  In  particular,  this  dissertation  focuses  on  the  following  two  political   consequences  of  authoritarian  elections  -­‐-­‐  popular  protests  and  leadership  turnover.         11            Post-­‐election   popular   protests   are   often   led   by   opposition   leaders   as   well   as   ruling   elites   that   defected   from   the   regimes.   Taking   to   the   streets,   people   protest   against   electoral  fraud  by  the  ruling  party.  Protesters  often  demand  that  dictators  nullify  the   election   results,   call   for   another   election,   and/or   step   down   from   the   government.   Either   way,   protesting   against   authoritarian   governments   is   a   costly   behavior   for   participants  because  autocrats  often  brutally  repress  post-­‐election  protests.  Yet,  once   launched,   protests   may   threaten   authoritarian   stability   in   both   short   and   long   time   span.   In   the   short   run,   popular   protests   may   invigorate   opposition   forces   and   forge   armed   rebellion,   thereby   dragging   the   country   into   civil   war,   like   the   cases   of   Libya   and  Syria  during  the  Arab  Spring.  Or,  protests  may  lead  to  authoritarian  breakdown,   which   was   evident   in   the   cases   of   Georgia,   Kyrgyzstan   and   Ukraine   in   the   Color   Revolutions.   Even   if   governments   successfully   manage   to   crack   down   on   post-­‐ electoral   protests,   the   repressive   reaction   to   public   dissidence   may   threaten   authoritarian  stability  in  the  long  run,  with  international  actors  tightening  economic   sanctions   and/or   adopting   coercive   diplomacy.   Because   popular   protests   are   never   good   news   for   authoritarian   rulers,   they   require   a   more   thorough   investigation   of   scholars:   Under   what   conditions   do   popular   protests   occur   after   authoritarian   elections?                      The   second   consequence,   leadership   turnover,   also   sometimes   comes   about   in   the   aftermath   of   authoritarian   elections.   In   some   cases,   leadership   turnover   is   brought  by  post-­‐election  coup  d’état  or  assassination,  which  leads  to  transition  from     12   one   authoritarian   regime   to   another,   as   in   the   cases   of   Algeria   (1991)   and   Burundi   (1993).   In   other   cases,   election   results   announce   that   opposition   parties   score   a   surprising  victory  and  the  dictator  subsequently  steps  down  from  the  office,  resulting   in   democratic   transition,   like   the   cases   of   Chile   (1988)   and   Poland   (1989).   These   two   forms  of  authoritarian  breakdown  are  different  to  a  large  extent,  but  both  scenarios   are   common   in   the   sense   that   authoritarian   elections   often   unexpectedly   reveal   the   weaknesses   of   the   incumbent   regime,   rather   than   its   strength.   This   raises   another   important  question:  When  do  autocratic  elections  oust  dictators?                                                               1.2  Theory  and  Argument  in  Brief              In  order  to  address  the  foregoing  puzzles  concerning  authoritarian  politics  and  its   elections,  I  construct  a  theory  of  elections  in  dictatorships.  Much  like  extant  work  on   authoritarian  politics,  I  start  with  the  assumption  that  authoritarian  rulers  and  their   potential   opponents   face   information   shortages   in   autocracies.   In   democracies,   the   media,   the   free   expression   of   public   opinion,   and   the   presence   of   civil   liberties   provide   political   actors   with   valuable   information   on   the   intentions   and   abilities   of   the   government   or   citizens,   whereas   these   rights   and   freedoms   are   not   fully   institutionalized  (and  are  often  repressed)  in  autocracies.  Because  both  the  dictator   and   his   potential   opponents   have   difficulties   in   knowing   about   the   capabilities   and   preferences   of   each   other,   dictators   may   face   two   major   problems.   First,   in   the     13   presence   of   uncertainty,   potential   opponents   may   fail   to   correctly   estimate   the   dictator’s  de  facto  strength  and  popularity.  Thus,  they  may  challenge  the  dictator  by   trying   to   assassinate   him,   plot   a   coup,   or   mobilize   protests   after   a   miscalculation   (Tullock   1987).   Second,   information   shortages   may   make   it   more   difficult   for   the   dictator   to   know   about   other   actors’   true   preferences   or   intentions   (Wintrobe   1998).   The   failure   to   correctly   understand   people’s   preferences   may   lead   the   autocrat   to   resort   to   indiscriminate   violence   to   preempt   the   threats   that   he   believes   exist.   This   creates  a  problem  regarding  the  efficient  use  of  regime's  limited  resources  on  existing   threats  and  may  also  magnify  people’s  grievances  against  the  regime.  Put  differently,   without  reliable  sources  of  information  on  the  dictator  and  other  political  actors,  the   dictator  is  likely  to  face  serious  obstacles  in  governing  the  country  efficiently.            For   the   purpose   of   improving   the   quality   of   information,   autocrats   can   take   various   measures;   they   may   relax   tight   control   over   the   media   (Egorov,   Griev,   and   Sonin   2009);   allow   opposition   parties   to   attend   the   parliament   (Gandhi   and   Przeworski   2007);  become  more  tolerant  of  criticism  against  government  policies  on  the  internet   (King,  Pan,  and  Roberts  2013);  and  decentralize  the  governance  in  order  to  accurately   grasp   preferences   and   demands   among   local   elites   and   citizens   (Montinola,   Qian,   and   Weingast  1994;  Treisman  1999).                Besides   these   measures,   autocratic   leaders   can   also   take   advantage   of   authoritarian  elections.  As  discussed  above,  semi-­‐competitive  elections  in  autocracies   allow   dictators   to   credibly   demonstrate   their   strength   or   popularity   to   potential     14   opponents   via   election   results.   Signaling   the   invincibility   of   their   regimes   via   overwhelming   election   victories,   autocrats   can   deter   political   elites   from   rebelling   against   them.   In   addition,   semi-­‐competitive   elections   also   enable   dictators   to   know   the   distribution   of   popular   support   for   the   regimes   and   competence   of   ruling   elites   through   election   results.   Based   on   this   information,   dictators   can   improve   the   efficiency  of  their  governance.         If   dictators   manipulate   elections,   they   can   easily   maintain   supermajorities.   Such   manipulated   elections,   however,   do   not   render   the   benefits   that   dictators   are   expected  to  enjoy  through  elections.  On  the  flip  side,  less  manipulated  elections  allow   dictators   to   exploit   the   above-­‐mentioned   advantages   of   authoritarian   elections.     However,  less  manipulated  elections  should  also  decrease  the  likelihood  of  dictators   to  win  big.  The  failure  to  maintain  an  overwhelming  victory  is  more  likely  to  reveal   weaknesses  of  authoritarian  rulers,  which  may  threaten  authoritarian  rule.                                              Under   the   constraint   of   the   electoral   dilemma   in   authoritarian   rule,   the   dictator   decides   how   much   they   manipulate   elections   through   electoral   fraud   and/or   the   adoption  of  a  certain  electoral  system.  The  dictator  wants  to  stay  in  power  (Bueno  de   Mesquita  et  al.  2003;  Chiozza  and  Goemans  2010).  In  order  to  maximize  his  survival   prospect,  the  dictator  needs  to  win  an  election  overwhelmingly,  but  at  the  same  time   he  also  needs  to  make  election  results  believable  for  himself  as  well  as  other  political   actors.   In   so   doing,   he   can   overcome   the   problems   of   information   shortages   and   stabilize  authoritarian  rule.         15            This   dissertation   argues   that   the   distribution   of   power   between   the   dictator   and   political   elites   determines   the   ways   in   which   the   dictator   designs   authoritarian   elections.   Here,   political   elites   refer   to   both   ruling   and   opposition   elites   who   have   potential  abilities  to  topple  down  the  regimes  by  using  their  political  resources  such   as  their  supporters,  money,  organizations,  and  social  status.  In  particular,  I  argue  that   differences   in   capability   of   mobilizing   popular   support   between   the   dictator   and   political   elites   influence   the   extent   to   which   the   dictator   engages   in   electoral   fraud   and   electoral   system   change.   If   the   dictator   is   stronger   than   potential   opponents   among  political  elites  -­‐-­‐  that  is,  when  there  is  a  large  gap  in  the  amount  of  resources   to   mobilize   popular   support   -­‐-­‐   the   dictator   can   garner   more   supporters   than   other   elites  in  the  elections.  Thus,  he  can  win  big  while  exploiting  the  informational  benefits   of   elections   without   manipulating   the   results.   Conversely,   if   the   elites   possess   greater   resources   vis-­‐à-­‐vis   the   dictator,   the   dictator   may   not   be   able   to   organize   a   large   scale   of  mobilization  by  himself.  In  this  situation,  the  dictator  tries  to  manipulate  elections   and   secure   an   overwhelming   majority   at   elections   through   electoral   fraud   and/or   electoral  system  reform.                  The   theory’s   empirical   implications   are   twofold   with   regard   to   the   manner   of   electoral   manipulation   in   dictatorships.   Broadly,   there   are   two   ways   to   manipulate   election  results.  First,  dictators  can  employ  electoral  fraud  such  as  electoral  violence,   electoral   cheating,   and   the   manipulation   of   electoral   law.   These   methods   are   undemocratic   measures   to   bias   election   results.   The   second   technique,   the  choice  of     16   electoral   systems,   is   more   of   an   indirect   way   of   electoral   manipulation;   yet,   it   still   facilitates   the   dictators’   obtaining   an   overwhelming   majority   through   the   manipulation   of   the   seats-­‐votes   elasticity.   SMD   systems   enable   dictators   to   obtain   supermajorities  with  fewer  votes,  whereas  PR  systems  require  them  to  mobilize  more   regime  supporters  to  secure  the  same  supermajorities.                  My   theory   predicts   that   dictators   who   have   large   financial   resources   to   distribute   are   more   likely   to   refrain   from   using   extensive   electoral   fraud   and   adopt   PR-­‐based   electoral  systems.  Both  electoral  fraud  and  SMD  systems  bias  election  results  in  favor   of   dictators   and   ruling   parties.   Both   tactics   help   autocrats   score   an   overwhelming   victory,  yet  instead  such  supermajorities  do  not  necessarily  reflect  de  facto  strength   of   dictators.   To   the   extent   that   election   results   are   artificially   produced,   the   informational   effects   of   elections   are   undermined.   Only   “strong”   dictators,   who   can   garner   political   support   from   the   citizenry   via   public   spending,   will   be   able   to   hold   “less   advantageous”   elections   by   abstaining   from   the   use   of   serious   electoral   malpractices  and  adopting  more  proportional  electoral  systems.                My   empirical   analysis   of   electoral   fraud   demonstrates   that   dictatorships   with   abundant   natural   resources   and   weak   opposition   permit   elections   to   be   more   free   and   fair   by   relying   less   on   electoral   fraud.   Further,   the   analysis   also   shows   that   the   positive   effect   of   natural   resources   on   electoral   fairness   greatly   increases   when   dictators  have  strong  organizational  bases  that  effectively  discipline  ruling  elites  and   thus  streamline  public  spending.  With  regard  to  the  choice  of  electoral  systems,  I  find     17   that  there  is  a  robust,  negative  correlation  between  effective  electoral  threshold  and   dictators’   financial   strength   measured   by   their   natural   resource   wealth.   It   suggests   that  rich  dictators  are  more  likely  to  choose  a  proportional  representation  system  to   credibly  show  his  strength.              If   dictators   fail   to   deal   with   the   electoral   dilemma,   authoritarian   elections   may   backfire   on   them.   With   this   in   mind,   the   latter   half   of   my   dissertation   investigates   political   consequences   of   autocratic   elections.   There   may   be   two   scenarios   where   elections   destabilize   authoritarian   rule:   When   dictators   use   excessive   electoral   manipulation   in   light   of   their   mobilization   power,   elections   cannot   bring   credible   information.   This,   in   turn,   makes   it   difficult   for   opposition   leaders   and   their   supporters   to   correctly   estimate   the   dictator’s   true   strength,   provoking   popular   protests.  On  the  other  hand,  when  dictators  are  unable  to  resort  to  electoral  fraud  at   an  extent  where  they  would  be  able  to  complement  their  mobilization  power  through   economic   distribution,   election   results   then   credibly   reveal   the   weaknesses   of   the   authoritarian   regimes.   Updated   information   on   de   facto   regime   weakness   via   elections  encourages  ruling  elites  to  defect  from  the  regime  or  stage  coup  d’état.  Or,  it   may   simply   bring   a   surprising,   perhaps   landslide,   electoral   victory   to   opposition   parties,   leading   to   leadership   turnover.   My   theory   expects   that   if  dictators  resort  to   excessive  electoral  fraud,  then  authoritarian  elections  are  more  likely  to  spark  popular   protests.   If   dictators   with   less   mobilization   power   fail   to   rig   elections   extensively,   authoritarian  elections  are  more  likely  to  lead  to  leadership  turnover.                 18    1.3  Methods  and  Research  Design              In  examining  the  observable  implications  derived  from  the  theory  of  authoritarian   elections,   this   dissertation   mainly   employs   two   methods.   The   first   method   is   cross-­‐ national   statistical   analysis.   For   the   analysis   of   electoral   fraud   and   post-­‐election   conflicts,  my  dataset  covers  78  authoritarian  regimes  from  1977  to  2004.  To  analyze   electoral   system   choice   in   electoral   authoritarian   regimes,   I   use   newly   collected   cross-­‐national   datasets   on   electoral   systems   and   election   results,   which   cover   95   electoral   authoritarian   countries   from   1946-­‐2007.   The   primary   purpose   of   the   cross-­‐ national   statistical   analyses   is   to   test   the   expected   correlations   between   the   variables   of  interests,  thereby  confirming  whether  the  testable  hypotheses  will  hold  in  general   among   authoritarian   regimes.   In   addition   to   testing   the   direct   implications   for   the   hypotheses,   I   also   carry   out   additional   cross-­‐national   investigations   to   explore   to   what   extent   assumptions   and   causal   mechanisms   proposed   in   theory   is   pertinent   cross-­‐nationally.              Cross-­‐national   statistical   analysis   using   country-­‐level   aggregated   data   is   a   highly   powerful   tool   for   testing   statistical   correlations   between   the   variables   of   interests.   Cross-­‐national   comparison   with   a   large   number   of   countries   and   macro-­‐level   data,   however,   makes   it   difficult   to   illuminate   how   the   key   independent   variable   causes   a   particular   outcome,   namely   causal   mechanisms.   In   order   to   process   trace   mechanisms   through   which   dictators’   mobilization   power   is   linked   to   the   political     19   manipulation   of   elections   and   its   post-­‐election   consequences,   I   also   conduct   case   studies.  First,  as  a  typical  case  in  which  the  dictator  strategically  controlled  the  level   of  election  fraud  and  changed  electoral  systems  depending  on  his  mobilization  power   in   a   successful   way,   I   investigate   the   case   of   Kazakhstan   (1991-­‐2008).   Although   during   the   1990s   there   were   numerous   electoral   malpractices   in   Kazakhstan,   President   Nursultan   Nazarbaev   became   less   inclined   to   use   blatant   electoral   fraud   by   the  middle  of  the  2000s  and  even  decided  to  shift  the  country’s  electoral  system  from   SMD  to  PR  in  2007.  Why?  Focusing  on  the  distribution  of  mobilization  power  between   the   dictator   and   political   elites,   I   argue   that   the   president   decided   to   shift   electoral   strategies   from   serious   electoral   fraud   to   extensive,   pre-­‐electoral   economic   distribution  to  win  elections  overwhelmingly.              Second,   I   carry   out   a   case   study   of   Kyrgyzstan   to   illustrate   how   the   failure   to   strategically   manipulate   elections   brings   popular   protests.   In   stark   contrast   to   Nazarbaev,   President   Akaev   of   Kyrgyzstan   faced   massive   popular   protests   in   the   2005   parliamentary   elections,   which   ended   up   forcing   him   to   leave   office.   Why   did   the   elections   backfire   on   him   in   2005?   Similar   to   the   case   of   Kazakhstan,   I   describe   how   the   power   relationships   changed   from   independence   until   the   collapse   of   the   Akaev   regime   in   2005.   I   first   show   that,   while   Akaev   was   relatively   successful   in   mobilizing   support   by   using   financial   resources   and   clan   alliances   under   relatively   free   and   fair   elections   during   the   1990s,   the   weakening   regime   encouraged   him   to   be   more   inclined   to   electoral   fraud   by   the   early   2000s.   Then,   I   show   that   excessive     20   electoral   fraud   and   the   president’s   electoral   victory   in   the   2005   parliamentary   elections  sparked  popular  protests  by  opposition  supporters  who  formed  a  belief  that   the   regime   should   not   be   strong   enough   to   bring   such   an   overwhelming   winning   in   elections.              I  choose  the  Nazarbaev  regime  in  Kazakhstan  (1990-­‐2008)  and  the  Akaev  regime   (1990-­‐2005)   in   Kyrgyzstan   for   my   case   studies   because   both   regimes   share   many   commonalities   like   the   same   historical   legacies   of   the   Soviet   Union,   similar   cultural   backgrounds,   formal   political   institutions   (presidentialism   and   administrative   systems),   regime   types   (civilian   dictatorships   with   semi-­‐competitive   elections),   ethnic   diversity,   the   implementation   of   radical   economic   reforms,   and   serious   economic   crisis   during   the   1990s.   All   of   these   factors   are   seen   as   important   variables   influencing   the   odds   of   democratization   and   authoritarian   breakdown.     The   case   studies   of   Kazakhstan   and   Kyrgyzstan   are   based   on   my   fieldwork   in   Central   Asia   in   the  summers  of  2011,  2012  and  2014.  I  stayed  in  Almaty  and  Astana  in  Kazakhstan,   and  in  Bishkek  in  Kyrgyzstan  for  about  seven  months  in  total.  In  the  case  studies,  in   addition   to   various   secondary   sources   and   election   monitoring   reports   on   the   two   Central   Asian   republics,   I   use   unique   quantitative   data   gathered   at   the   Statistical   Agencies   of   Kazakhstan   and   Kyrgyzstan,   various   local   publications,   and   qualitative   resources  collected  through  34  semi-­‐structured  interviews  in  the  countries.                  Comparing   these   two   regimes   enables   me   to   control   for   these   factors   while   focusing   on   the   variable   of   my   interests   –   power   distribution   between   the   dictator     21   and   political   elites.   In   fact,   from   the   same   reason,   comparativists   studying   Central   Asia  have  often  picked  up  these  two  countries  to  make  rigorous  comparisons  under   the  most  similar  systems  design  (Jones  Luong  2002;  Schatz  2009;  McGlinchey  2011).   Besides   the   between-­‐country   comparison,   there   are   rich   variations   with   regard   to   the   extent  of  electoral  fraud,  electoral  systems  and  popular  protests  within  each  country.   Such   within-­‐country   variances   permit   me   to   control   for   additional   confounding   factors   that   I   am   unable   to   control   for   by   the   between-­‐country   comparison,   such   as   the  level  of  economic  development,  population  size,  territorial  size,  and  so  on.  In  sum,   combining   the   two   comparative   perspectives   (time-­‐series   and   cross-­‐country),   the   case  studies  of  Kazakhstan  and  Kyrgyzstan  make  it  possible  to  provide  rigorous  case-­‐ study  evidence  on  causal  pathways  in  theory.                     1.4  Contributions              Broadly,   this   dissertation   aims   to   contribute   to   the   following   three   strands   of   literatures.   Most   importantly,   this   dissertation   contributes   to   our   understanding   of   authoritarian  politics  and  its  elections.  As  discussed,  there  have  been  two  conflicting   views   on   elections   in   autocracies.   Scholars   have   taken   emphasis   on   the   informational   benefits   of   elections   on   the   one   hand   (e.g.   Magaloni   2006;   Gandhi   and   Lust-­‐Okar   2009;   Blaydes   2011),   and   the   destabilizing   effects   of   elections   on   the   other   (e.g.   Lindberg  2006,  2009;  Tucker  2007;  Bunce  and  Wolchik  2011).  Introducing  the  idea  of     22   the  electoral  dilemma,  my  theory  provides  a  general  framework  to  reconcile  the  two   different  assessments  of  authoritarian  elections.                Second,   the   dissertation   investigates   theoretically   and   empirically   relationships   among   “the   menu   of   manipulation”   (Schedler   2002)   at   elections.   Scholars   have   suggested   that,   in   order   to   hold   onto   power,   political   leaders   use   a   variety   of   techniques  at  the  ballot  box,  such  as  electoral  malpractices  (Hyde  2007;  Kelley  2012;   Birch   2013;   Simpser   2013),   electoral   violence   (Straus   and   Taylor   2012;   Hufner-­‐ Burton,   Hyde   and   Jablonski   2013),   pre-­‐electoral   economic   distribution   (Nordhaus   1975;   Tuft   1978;   Shi   and   Svensson   2003;   Brender   and   Drazen   2006),   and   the   manipulation   of   electoral   systems   (Rokkan   1970;   Boix   1999;   Cusack,   Iversen,   and   Soskice   2007;   Calvo   2009).   Researchers,   however,   have   tended   to   individually   consider   each   of   these   different   strategies   without   drawing   much   attention   to   how   each  relates  to  others.  Thus  far,  researchers  have  not  posited  a  systematic  theoretical   framework   to   understand   when   political   leaders   are   more   inclined   to   use   one   manipulation   technique   over   others   in   elections.   We   also   know   little   about   when   political  leaders  are  more  likely  to  combine  particular  manipulation  tools  to  maintain   supermajorities  at  the  ballot  box.  Focusing  on  the  distribution  of  power  between  the   dictator   and   political   elites,   this   dissertation   suggests   that   there   may   be   a   trade-­‐off   between  the  “political”  manipulations  of  elections  such  as  electoral  cheating,  election   violence   and   electoral   system   change   and   “economic”   maneuvering   before   elections   through   public   spending.   I   assert   that   political   leaders   do   not   necessarily   choose   a     23   particular   electoral   strategy   at   random,   but   may   strategically   decide   which   strategies   to   be   taken   under   the   interactions   with   political   elites   and   citizens.   In   so   doing,   I   provide   a   general   theory   predicting   which   manipulation   tools   dictators   are   more   likely  to  use.                    Lastly,   the   dissertation   makes   a   contribution   to   the   growing   literature   on   the   political  economy  of  natural  resources.  Scholars  have  debated  over  whether  natural   resource  wealth  prolongs  authoritarian  rule  (Ross  2001;  Smith  2004;  Ulfelder  2007;   Morrison  2009;  Wright,  Frantz,  and  Geddes  2013),  prevents  democratization  (Haber   and   Menaldo   2010;   Ross   2012),   damages   political   transparency   (Mahdavi   1970;   Beblawi   and   Luciani   1987;   Paler   2013),   slows   economic   growth   (e.g.   Sachs   and   Warner   1995;   Jones   Luong   and   Weinthal   2011;   Ross   2012),   undermines   state   capacity,   and   fuels   civil   conflict   (Fearon   and   Laitin   2003;   Collier   and   Hoeffler   2004;   Humphreys  2005).  Resonating  with  recent  studies  that  begin  to  rethink  the  oil  curse   on   democratization   and   transparency   (Haber   and   Menaldo   2010;   Paler   2013),   this   dissertation   also   suggests   that   natural   resource   wealth,   boosting   up   the   dictator’s   resource   mobilization   and   hence   increasing   regime   supporters,   may   not   necessarily   contribute   to   strengthening   electoral   manipulation.   Findings   in   my   dissertation   are   consistent   with   those   in   the   previous   literature,   which   contends   that   natural   resources   prolong   authoritarian   rule   by   enhancing   the   dictators’   ability   to   increase   public   spending.   Yet,   I   suggest   that   due   to   the   very   fact   that   dictators   can   buy-­‐off   extensive   popular   support   through   the   distribution   of   natural   resource   wealth,     24   natural   resources   permit   dictators   to   hold   less-­‐manipulated   elections.   Since   petroleum   keeps   dictators   popular   among   the   citizenry,   authoritarian   leaders   take   advantage   of   the   informational   benefits   of   authoritarian   elections   without   letting   elections   backfiring   on   them.   Therefore,   elections   in   oil-­‐rich   dictatorships   stabilize   autocratic  rule.           1.5  Outline  of  the  Dissertation              This   dissertation   is   organized   as   follows.   Chapters   2-­‐4   investigate   the   causes   of   authoritarian  elections.  Chapter  2  introduces  an  information  theory  of  authoritarian   elections,   focusing   on   the   electoral   dilemma   and   the   distribution   of   power   between   the  dictator  and  other  political  elites.   Then,   I  test   empirical   implications  by   analyzing   electoral  fraud  in  authoritarian  regimes  through  cross-­‐national  statistical  analyses  of   78   authoritarian   countries   (1977-­‐2004).   With   the   theory   in   mind,   Chapter   3   turns   attention   to   the   logic   of   electoral   system   change   in   electoral   authoritarian   regimes.   Employing   original   datasets   of   electoral   systems   and   election   results   in   95   electoral   authoritarian  countries  (1946-­‐2007),  I  explore  the  conditions  under  which  electoral   autocrats   prefer   SMD   systems   to   more   PR-­‐based   systems,   and   vice   versa.   Chapter   4   conducts  an  in-­‐depth  case  study  of  Kazakhstan  (1991-­‐2008),  in  which  I  explore  how   an   increasingly   powerful   dictator   –   President   Nazarbaev   -­‐-­‐   shifted   his   electoral   strategies   from   blatant   electoral   fraud   to   extensive   pre-­‐electoral   economic     25   distribution  over  the  two  decades  as  he  succeeded  in  obtaining  rich  natural  resource   wealth  as  well  as  constructing  centralized  political  organizations.  I  also  analyze  how   the  Kazakh  government  adopted  a  Proportional  Representation  system  with  a  nation-­‐ wide   electoral   district   in   2007   and   its   consequences,   comparing   to   the   precedent   SMD-­‐based  systems.              Chapters  5  and  6  then  investigate  political  consequences  of  authoritarian  elections.   Based  on  estimation  results  obtained  in  Chapter  2,  Chapter  5  examines  the  conditions   under   which   authoritarian   elections   backfire   on   dictators   in   the   form   of   popular   protests  and  leadership  turnover  by  conducting  cross-­‐national  statistical  analysis  of   78  authoritarian  countries  (1977-­‐2004).  Chapter  6  illustrates  the  cases  of  Kyrgyzstan   (1990-­‐2005)  to  show  that,  when  President  Akaev  failed  to  strategically  deal  with  the   electoral   dilemma,   he   suffered   post-­‐election   protests,   which   resulted   in   the   breakdown   of   his   regime   in   2005.   Lastly,   Chapter   7   concludes,   while   presenting   policy  implications  and  future  research  agenda  on  elections  in  dictatorships.                                         26   CHAPTER  2     ELECTORAL  DILEMMA  AND  THE  MANIPULATION  OF  ELECTIONS       2.1  Introduction              Modern   dictatorships   hold   elections   even   though   citizens   hardly   expect   to   peacefully   replace   political   leaders   via   the   ballot   box   (Przeworski   et   al.   2000;   Hyde   and   Marinov   2012).   Although   not   truly   fair   and   free   (Schedler   2006;   Levitsky   and   Way  2010),  there  is  remarkable  variation  with  respect  to  how  much  political  leaders   rig   authoritarian   elections:   Some   dictators   resort   to   whatever   fraudulent   measures   they  can  take  to  secure  an  overwhelming  electoral  “victory,”  while  others  are  loath  to   commit   electoral   malpractices   and   even   willingly   make   efforts   to   keep   the   electoral   field   fairer   through   electoral   reforms.   In   the   heyday   of   Mexico’s   Institutional   Revolutionary  Party,  stunned  by  the  fact  that  the  main  opposition  party  boycotted  the   1976   presidential   election,   the   authoritarian   government   increased   electoral   transparency  thereby  enabling  opposition  parties  to  gain  some  presence  in  Mexican   politics  (Eisenstadt  2004,  32-­‐44).  Nursultan  Nazarbaev,  the  president  of  Kazakhstan,   announced  shortly  before  the  2012  election  that  he  would  “reform”  electoral  laws  so   that  moderate  opposition  parties  could  gain  some  seats  in  parliament  (Mutlu  2012).   Cross-­‐national  evidence  shown  later  in  the  section  of  empirics  suggests  that  election   fraud   varies   substantially   in   authoritarian   regimes.   There   are   some   authoritarian     27   states   where   election   violence,   electoral   cheating   and   legal   restrictions   on   electoral   participation   are   not   as   severe.   These   include   Bongo’s   Gabon   (especially   1990s-­‐ 2000s),   Mahathir’s   Malaysia,   Singapore   under   People’s   Action   Party’s   rule,   Kuomintang’s  Taiwan  (1980s  through  1995),  and  Kuwait.  On  the  other  hand,  in  other   autocracies   including   Mugabe’s   Zimbabwe   (2000s),   Karimov’s   Uzbekistan   (1990s-­‐ 2000s),   Togo   (1990s-­‐2000s),   Suharto’s   Indonesia   (1990s),   and   Paraguay   (1970s-­‐ 1980s),   dictators   used   extreme   manipulation   techniques   in   an   attempt   to   fabricate   election   results.   Such   differences   in   the   level   of   electoral   fraud   under   authoritarian   regimes   remain   incompletely   addressed   by   previous   work.   What   explains   this   variation  in  electoral  manipulation  under  authoritarian  regimes?                In   solving   this   puzzle   of   elections   in   dictatorship,   this   chapter   explores   theoretically   and   empirically   the   logic   of   electoral   fairness   in   authoritarian   regimes.   First,   I   suggest   that   authoritarian   leaders   face   a   serious   dilemma   between   the   credibility   of   election   results   and   the   certainty   of   winning   big   at   the   ballot   box.   Dictators   can   stay   in   power   by   rigging   elections   through   vote   stuffing,   violent   repression,   and   the   manipulation   of   election   rules.   Pro-­‐regime   bias   in   electoral   results,  however,  makes  it  difficult  for  dictators  to  not  only  convey  a  credible  signal  of   their  regime  strength  to  potential  opponents,  but  also  to  collect  reliable  information   on   key   actors’   strength.   On   the   other   hand,   if   dictators   allow   excessive   electoral   reforms,   then   they   may   fail   to   obtain   an   overwhelming   victory,   which   may   result   in   increased  anti-­‐regime  sentiment  and  threats  their  dictatorial  rule.       28              Then,   I   go   on   to   argue   that   the   balance   of   power   between   the   dictator   and   its   potential   opponents   (from   both   ruling   elites   and   opposition   forces)   explains   the   level   of  electoral  fraud  under  the  constraint  of  this  electoral  dilemma.  More  specifically,  if   there   exists   a   large   gap   between   the   dictator   and   the   elites   in   their   ability   to   mobilize   voluntary   popular   support,   the   former   can   win   a   supermajority   without   relying   on   heavy  electoral  fraud.  Thus,  strong  dictators  should  be  able  to  signal  their  strength  by   producing  an  overwhelming  majority  without  electoral  manipulation.  By  contrast,  if   other  elites  possess  greater  political  resources  to  mobilize  supporters,  fair  elections   are  less  likely  to  bring  a  big  victory  to  dictators,  who  have  a  strong  interest  in  using   extensive  fraud  to  bias  election  results.                I   conduct   cross-­‐national   statistical   analysis   to   empirically   assess   my   theory   of   authoritarian  elections.  There  are  three  key  findings.    First,  natural  resource  wealth,   which   I   use   to   measure   dictators’   financial   resources,   is   negatively   correlated   with   election  fraud.  Second,  political  organizations  that  strongly  bind  ruling  elites  tend  to   magnify  the  effect  of  natural  resources  on  reducing  electoral  fraud.  More  specifically,   both   dominant-­‐party   regimes   and   less   fractionalized,   large   dominant   ethnic   groups   have  such  conditional  effects.  Finally,  pre-­‐electoral  anti-­‐government  collective  action   significantly  increases  the  propensity  of  dictators’  employing  electoral  fraud.              The   contribution   of   this   chapter   to   the   existing   literature   is   twofold.   First,   introducing  the  idea  of  the  election  dilemma  in  dictatorship,  this  research  attempts  to   illuminate  how  political  leaders  design  elections  under  authoritarian  regimes.  Second,     29   different  from  the  conventional  wisdom  that  focuses  on  political-­‐institutional,  socio-­‐ economic  or  international  factors  influencing  election  fraud,  my  research  shows  that   government’s   ability   to   garner   popular   support   via   public   spending   is   an   important   factor  to  reduce  electoral  manipulation.                This   chapter   proceeds   as   follows.   Section   2   introduces   the   idea   of   the   electoral   dilemma,  while  reviewing  the  previous  literature  of  authoritarian  elections.    Section  3   discusses   electoral   fraud   in   authoritarian   regimes.   In   Section   4,   I   theorize   how   the   dictator   decides   the   level   of   electoral   fraud   under   the   constraint   of   the   electoral   dilemma.   Section   5   empirically   tests   observable   implications   of   the   theory   with   a   series  of  cross-­‐national  statistical  models.  Finally,  conclusions  and  policy  implications   follow.       2.2  The  Electoral  Dilemma  in  Dictatorship              In   dictatorships,   the   political   leader   and   potential   opponents   both   are   likely   to   suffer   information   shortages   because   political   transparency   is   seriously   circumscribed  (Wintrobe  1998;  Egorov,  Gariev,  and  Sonin  2009).  This  is  problematic   for   the   dictator   because potential   opponents   likely   fail   to   correctly   estimate   the   dictator’s   de   facto   strength   and   thus   may   challenge   the   dictator   after   such   miscalculation,   resulting   in   political   turmoil   (Tullock   1987).   Strengthening   the   military   is   one   of   the   most   frequently   used   measures   to   credibly   demonstrate   the     30   dictator’s   power   to   potential   opponents,   yet   heavily   relying   on   the   sword   directly   threatens  leader  tenure  via  coup  d’état  (Svolik  2012).  Thus,  it  is  not  a  perfect  solution.   Second,   information   shortages   in   dictatorship   make   it   difficult   for   the   dictator   to   know   the   accurate   distribution   of   popular   support   among   the   citizenry.   Without   reliable   sources   of   information   on   popular   support,   the   dictator   faces   difficulties   in   governing   the   country   efficiently.   In   authoritarian   regimes   where   political   leaders   retain  large  potential  to  use  violent  repression,  people  are  more  tempted  to  conceal   their   true   preferences   for   fear   of   being   targeted   by   state   repression   (Kuran   1991;   Wintrobe  1998).  Therefore,  the  dictator  needs  to  pursue  alternative  measures  other   than  violent  means,  by  which  to  solve  these  information  problems.                            Scholars   have   argued   that   authoritarian   leaders   use   elections   to   overcome   the   information   problems   in   dictatorships.   First,   some   researchers   point   out   that   elections   in   dictatorships   help   dictators   show   regime   invincibility   to   potential   opponents  (Hermet  1978,  12;  Magaloni  2006;  Geddes  2006;  Simpser  2013).  Winning   elections   while   obtaining   an   overwhelming   majority,   dictators   are   able   to   demonstrate   that   the   regime   is   unshakable.   In   so   doing,   they   can   prevent   potential   opponents   from   launching   coups,   defecting   from   the   regime,   and   leading   popular   uprisings.   Second,   elections   provide   dictators   with   knowledge   of   the   accurate   distribution  of  political  support  among  population.  Through  election  results,  dictators   can   be   informed   about   local   popularity   of   ruling   elites,   which   becomes   one   of   the   criteria  to  pick  up  competent  and  loyal  politicians  and  screen  out  others  (Ames  1970;     31   Shi  1999;  Magaloni  2006;  Boix  and  Svolik  2007;  Blaydes  2010;  Malesky  and  Schuler   2010;  Reuter  and  Robertson  2011).  Election  results  may  also  render  information  on   the   distribution   of   popular   support   for   the   regime   and/or   popularity   of   opposition   parties  (Magaloni  2006;  Cox  2009;  Miller  2012),  thereby  enabling  dictators  to  decide   the  targets  of  repression  or  make  efficient  allocation  of  economic  resources  to  cement   political  support.                          Which   benefits   dictators   want   to   exploit   the   most   might   be   different   across   authoritarian   regimes.   Regardless   of   which   electoral   benefits   each   dictator   emphasizes,   however,   the   crucial   fact   is   that   authoritarian   elections   do   not   always   bring   informational   benefits   to   dictators.   The   probability   that   dictators   obtain   an   overwhelming   majority   at   elections   will   increase   as   they   rig   elections   in   more   heavy-­‐ handed  ways  (Donno  2013,  33).  In  fact,  using  a  comprehensive  cross-­‐national  data  on   electoral   manipulation   and   electoral   margins,   Simpser   (2013,   67-­‐72)   shows   that   manipulated   elections   tend   to   bring   larger   margins   of   victory.   The   credibility   of   election   results,   however,   will   decrease   as   authoritarian   leaders   resort   to   more   serious   electoral   fraud.   When   elections   are   rigged,   dictators   lose   the   potential   informational   benefits   of   elections   in   the   following   two   ways.   First,   results   of   seriously  rigged  elections  can  no  longer  send  a  credible  signal  on  regime  strength  and   legitimacy.  For  instance,  if  it  is  obvious  that  dictators  can  receive  100  percent  of  seats   and  votes  easily  by  not  allowing  oppositions  or  using  too  much  election  violence  and   cheating,   then   election   results   do   not   reflect   true   popularity   of   the   dictator.   By     32   contrast,   if   dictators   do   not   resort   to   manipulation   but   they   still   win   by   a   large   margin,   elections   can   send   a   costly   signal   of   the   dictator’s   strength   to   potential   opponents.  In  fact,  using  World   Values   Survey   (2010-­‐2014),  Norris  (2014,  Chapter  6)   demonstrates   that   electoral   malpractice   tends   to   undermine   people’s   confidence   in   governments   as   well   as   legal   compliance.   Second,   if   they   commit   serious   electoral   fraud,  then  dictators  can  no  longer  gather  reliable  information  on  the  distribution  of   popular  support.  When  fraud  is  used  to  bias  election  results  toward  the  dictator,  the   results   do   not   reflect   the   true   preferences   of   voters   (Wintrobe   1998).   Therefore,   in   addition   to   making   themselves   invincible   during   elections,   dictators   simultaneously   have  to  consider  how  they  can  make  these  election  results  believable.            When  dictators  refrain  from  using  extensive  electoral  manipulation,  elections  will   set   a   higher   hurdle   to   obtain   an   overwhelming   majority.   The   failure   to   craft   a   landslide   victory   is   likely   to   activate   the   defection   of   ruling   elites   and   invigorate   opposition   parties.   In   fact,   Kuhn   (2012)   shows   cross-­‐national   empirical   evidence   suggesting  that  post-­‐electoral  protests  are  more  likely  to  occur  after  close  elections.   The  cases  of  Mexico  (2000)  and  Ukraine  (2004)  are  illustrative  examples.  Conversely,   when   dictators   resort   to  excessive   election   fraud,   then   their   governance   may   become   more   inefficient   because   (1)   dictators   can   no   longer   know   the   distribution   of   popular   support,  and  (2)  potential  opponents  cannot  accurately  estimate  the  strength  of  the   dictator.   Subsequent   misinformation   and   miscommunication   among   political   actors   will   increase   the   possibility   of   political   conflict   (Beaulieu   2014):   dictators   may   be     33   forced   to   use   indiscriminate   state   repression,   whereas   potential   opponents   may   be   tempted  to  stage  coups  and  protests  either  for  preemptive  purposes  or  as  a  result  of   underestimating   the   dictator’s   strength.   Indeed,   Gandhi   and   Vreeland   (2004)   empirically   demonstrate   that   authoritarian   legislatures   without   multiparty   systems   contribute  to  increasing  the  risk  of  violent  conflict.                  In   his   study   of   electoral   manipulation,   Simpser   (2013)   argues   that   committing   serious   electoral   fraud   itself   enables   governments   to   show   their   strength   because   election   fraud   is   costly   in   that   it   needs   vast   amounts   of   human   and   economic   resources.  I  do  not  deny  façade  elections  entail  some  costs  in  resource  mobilization.   There   are   a   number   of   reasons,   however,   why,   as   long   as   dictators   can   retain   an   overwhelming   majority,   partially   transparent   elections   become   more   useful   than   façade   elections.   First,   as   discussed,   election   results   from   free   and   fair   elections   are   more  informative  and  overcome  information  shortages  in  dictatorships  –  increasing   the   credibility   and   reliability   of   election   results.   Second,   some   empirical   studies   suggest  that  excessive  and  blatant  electoral  manipulation  may  contribute  to  revealing   the   dictator’s   weakness   rather   than   his   strength   potentially   leading   to   protests   and   leadership   change   in   the   aftermath   of   elections   (Tucker   2007;   Bunce   and   Wolchik   2010;  Hafner-­‐Burton,  Hyde,  and  Jablonski  2013;  Higashijima  2015).  Further,  although   Simpser’s   theory   tells   us   why   political   leaders   use   excessive,   blatant   electoral   manipulation,   it   does   not   provide   an   adequate   answer   as   to   why   some   political   leaders   use   serious   electoral   fraud   while   others   do   not.   My   theory   is   built   on   Simpser     34   (2013)   in   assuming   that   dictators   try   to   obtain   an   overwhelming   majority,   yet   I   draw   more  attention  to  the  credibility  of  election  results  in  authoritarianism.             2.3  Electoral  Fraud  and  the  Manipulation  of  Policy  Instruments                  Electoral  fraud  is  defined  as  a  series  of  illegal  measures  that  bias  election  results  in   favor   of   the   political   leader   (Lehoucq   2003).   In   light   of   this   broad   definition,   electoral   fraud   consists   of   the   following   three   subcomponents:   (1)   election   violence,   (2)   election   cheating,   and   (3)   undemocratic   restrictions   on   electoral   law.   Election   violence   is   physical   intimidation   during   elections   exercised   largely   by   incumbent   parties   (Straus   and   Taylor   2012;   Hafner-­‐Burton, Hyde,   and   Jablonski   2013).   Governments  use  electoral  violence  to  threaten  and  deter  opposition  candidates  and   citizens,   thereby   undermining   oppositions’   effective   campaigns   and   decreasing   turnout   of   opposition   supporters.   Electoral   cheating   allows   dictators   to   affect   the   number  of  votes  during  campaign  periods  and  election  days  with  nonviolent  but  still   illegal   measures   such   as   undermining   of   oppositions’   freedom   to   campaigns,   media   bias,  ballot  stuffing,  vote-­‐buying,  nonviolent  intimidation (Kelley  2012).  Restrictions   on   electoral   law   refer   to   a   series   of   regulations   that   prevent   citizens   and   electoral   candidates  from  influencing  politics,  including  limits  on  voting  rights  by  certain  social   characteristics   like   gender   and   ethnicity,   flaws   in   the   complaints   procedures,   high   thresholds  for  new  parties  to  get  registered  and  gain  seats,  constraints  on  the  right  to     35   run   for   office   such   as   language   and   educational   requirements   (Kelley   2012).   These   fraud   techniques,   though   considerably   different   in   how   they   manipulate   elections,   contribute   to   providing   the   dictator   a   margin   of   victory   that   would   otherwise   be   impossible.                The   burgeoning   literature   on   electoral   fraud   has   focused   on   various   factors   to   explain   electoral   fraud.   First,   placing   emphasis   on   domestic   political   factors,   researchers   find   severe   political   competition   (Lehoucq   and   Molina   2002),   single-­‐ member   districts   (Birch   2007)   and   the   absence   of   domestic   election   monitoring   (Ichino   and   Schuendeln   2012)   encourage   politicians   to   use   serious   electoral   fraud.   Second,   socio-­‐economic   variables   such   as   poverty   (Lehoucq   and   Molina   2002),   the   size   of   population   (Lehoucq   and   Molina   2002;   Fukumoto   and   Horiuchi   2011),   and   economic   inequality   (Ziblatt   2009)   have   also   been   regarded   as   important.   Third,   scholars   find   that   international   factors   such   as   trade   openness,   foreign   direct   investment   (Birch   2011)   and   international   election   monitoring   (Hyde   2007;   Kelley   2012)  reduce  electoral  fraud.              Though   these   studies   have   significantly   increased   our   understanding   of   election   fraud,  I  suggest  that  there  are  at  least  two  problems  that  still  need  to  be  addressed.   First,   most   importantly,   since   they   disregard   the   importance   of   the   election   dilemma,   the   existing   studies   fail   to   consider   relationships   between   “illegal,”   undemocratic   strategies  at  elections  such  as  election  violence,  election  cheating  and  manipulation  of   electoral   law,   and   “legal,”   legitimate   ones   including   manipulation   of   fiscal   and     36   monetary   policies.   Put   differently,   previous   literature   does   not   consider   the   possibility   that   political   leaders   may   willingly   refrain   from   using   fraud   to   mitigate   information   shortage   especially   when   they   can   garner   popular   support   using   available   financial   resources.   Previous   studies   have   exclusively   focused   on   the   electoral   fraud   side   without   drawing   much   attention   to   the   distribution   strategies   that   governments   may   alternatively   take. 4  Second,   the   existing   studies   do   not   consider   possible   differences   in   theoretical   scope   conditions   between   democracies   and   autocracies.   Cross-­‐national   studies   include   all   the   countries   in   their   samples   assuming   that   election   fraud   should   play   the   same   roles   both   in   democracies   and   dictatorships,   namely,   producing   a   victory   at   elections.   Most   single   case   studies   primarily  focus  on  election  fraud  in  democratic  countries  where  party  competition  is   strong   enough   to   generate   government   alternation.   As   discussed,   however,   the   purpose   of   holding   authoritarian   elections   is   not   to   get   reelected,   but   to   obtain   an   overwhelming   majority   to   show   regime   strength   via   elections   (Magaloni   2006;   Geddes   2006;   Simpser   2013).   This   notable   difference   in   the   governments’   purpose   of   holding  elections  may  change  the  motivations  for  committing  election  fraud.                                                                                                                           4  An  important  exception  is  Hyde  and  O’Mahoney  (2011).         37   2.4  Power  Distribution  between  the  Dictator  and  Elites              Under   the   constraint   of   the   electoral   dilemma,   the   dictator   decides   the   level   of   electoral   fraud.5  After   observing   the   level   of   electoral   fraud   and   election   results,   potential   opponents   among   political   elites   determine   whether   they   revolt   or   not.   The   dictator  aims  to  achieve  an  overwhelming  majority  in  as  credible  a  way  as  possible,   thereby   trying   to   exploit   the   informational   benefits   and   thus   maximize   their   prospects  of  survival.  Potential  opponents  opportunistically  target  the  seat  of  the  next   political   leader   in   the   sense   that   they   try   to   rebel   against   the   current   ruler   only   when   they  think  that  the  current  regime  is  too  weak  to  withstand  their  challenge  after  the   election.   More   specifically,   when   elections   are   either   too   transparent   to   obtain   a   supermajority  or  too  rigged  to  solve  information  shortage,  the  dictator  is  more  likely   to   face   such   challenges   from   potential   opponents,   compared   to   when   he   wins   a   credible  overwhelming  majority.                                                                                                                     5  My  theory  does  not  consider  when  dictators  begin  to  hold  elections.  This  is  because   most  modern  dictatorships  hold  periodical  elections  other  than  a  handful  of  countries   such   as   China   and   Saudi   Arabia,   and   thus   there   is   hardly   any   significant   variation   (Gandhi  and  Lust-­‐Okar  2009;  Hyde  2011,  262;  Hyde  and  Marinov  2012,  3).  It  is  also   reasonable   to   think   that   when   to   start   elections   may   be   determined   by   different   factors   such   as   colonial   heritages,   the   end   of   the   Cold   War,   international   pressures,   and   civil   war   onset.   Further,   coding   when   dictators   begin   to   hold   elections   is   empirically  intractable.  Since  dictators  can  easily  manipulate  election  timing,  there  is   often  a  surprisingly  long  period  of  time  between  elections,  which  makes  it  difficult  for   researchers  to  identify  when  to  hold  the  first  election.         38              In   this   strategic   interaction   between   the   dictator   and   potential   opponents,   the   dictator   decides   the   level   of   electoral   fraud.   I   argue   that   differences   in   power   of   mobilizing  popular  support  between  the  dictator  and  political  elites  are  important  to   determine   to   what   extent   the   dictator   resorts   to   electoral   fraud.   If   the   dictator   is   stronger  than  the  elites,  that  is,  when  there  is  a  large  gap  between  them  in  the  amount   of   economic   resources   to   mobilize   popular   support,   the   former   can   garner   more   supporters   who   willingly   vote   for   the   dictator   than   potential   opponents   do.   Thus,   the   stronger   the   dictator   is,   the   more   able   he   is   to   signal   his   strength   by   producing   an   overwhelming  majority  without  using  electoral  fraud.  By  contrast,  if  the  elites  possess   greater  political  resources  vis-­‐à-­‐vis  the  dictator,  the  dictator  is  not  able  to  organize  a   large  scale  of  mobilization  by  himself.  In  this  situation,  less-­‐manipulated  elections  are   more  likely  to  produce  surprising  results,  so  that  the  dictator  has  a  strong  interest  in   stealing  elections  in  his  favor.                  Dictators   with   rich   resources   can   mobilize   voters   through   large-­‐scale   economic   distribution   by   adopting   expansionary   economic   policies.   Loosening   fiscal   policies   and   strengthening   pork-­‐barrel   politics,   authoritarian   leaders   can   create   public   employment,   adopt   tax   exemption   for   party   supporters   and   the   poor,   give   bonuses   to   public   employees,   construct   infrastructure   and   implement   other   forms   of   public   goods   provision.   These   tactics   are   not   illegal   but   garner   “voluntary”   political   support.   In   fact,   numerous   studies   on   authoritarian   politics   demonstrate   that   durable   dictatorships   are   more   likely   to   manipulate   economic   policies   prior   to   elections   in     39   countries  like  Russia,  Mexico,  Malaysia,  and  Egypt  to  win  semi-­‐competitive  elections   with  large  margins  (e.g.  Akhmedov  and  Zhuravskaya  2004;  Magaloni  2006;  Pepinsky   2009;   Blaydes   2011;   Wright   2011).   Importantly,   relying   more   on   manipulation   of   economic  policy  instruments,  dictators  can  increase  the  credibility  of  election  results,   which   helps   them   send   clearer   signals   of   regime   invincibility.   In   sum,   prioritizing   economic   distribution   instead   of   political   fraud,   dictators   are   more   able   to   keep   the   credibility  and  certainty  of  election  results  simultaneously.                This   research   observes   the   balance   of   mobilization   power   between   the   dictator   and   elites   in   three   ways:     (1)   natural   resources   that   dictators   can   control,   (2)   disciplinary   ruling   organizations   that   prevents   ruling   elites’   opportunistic   behavior   and   thus   streamline   economic   distribution   to   citizens   and   (3)   opposition’s   anti-­‐ government  collective  action  capability.       2.4.1  Financial  Resources            Financial  resources  that  the  dictator  controls  enhance  his  power  of  popular  support   mobilization  through  extensive  economic  distribution.  In  particular,  natural  resource   wealth  such  as  oil,  natural  gas  and  minerals  enriches  state  coffers  and,  thus,  increases   public   spending   to   co-­‐opt   citizens   (Jensen   and   Wantchekon   2004;   Desai,   Olofsgard,   and   Yousef   2009;   Morrison   2009).   Since   authoritarian   leaders   can   use   natural     40   resources   to   placate   citizens’   economic   dissatisfaction,6  a   large   body   of   previous   studies  has  demonstrated  that  abundant  natural  resources  are  more  likely  to  prolong   authoritarian  rule  (Ross  2001;  Smith  2004;  Ulfelder  2007;  Wright,  Frantz,  and  Geddes   2013).7            Compared   to   other   sources   of   revenues   such   as   foreign   aid   and   taxes,   natural   resources   can   be   more   useful   sources   of   pork   barrel   for   authoritarian   regimes.   The   use   of   tax   revenues   tends   to   be   carefully   monitored   by   citizens   and   thus   tax   revenues   are   difficult   to   be   squandered   by   political   leaders.   On   the   other   hand,   budgetary   processes   of   natural   resources   are   highly   opaque.   This   secrecy   of   natural   resource   wealth   allows   dictators   to   distribute   them   for   their   political   gain   more   easily   (Ross   2012:   59-­‐62).   Foreign   aid   is   also   a   non-­‐tax   revenue   akin   to   natural   resources   that   are                                                                                                                   6  One  may  think  that  oil  resources  are  not  necessarily  available  only  by  authoritarian   leaders.   Especially,   if   oil   resources   are   regionally   dispersed   and   opposition   elites   control   the   resources   in   their   local   strongholds   independent   of   the   dictator,   windfalls   may  also  enhance  mobilization  power  of  them.  For  example,  Mahdavi  (2014)  presents   evidence   from   Iran   that   incumbent   parliamentarians,   be   ruling   or   opposition   elites,   tend   to   be   reelected   if   they   possess   rich   oil   resources   in   their   electoral   districts.   To   what   extent   the   dictator   can   control   natural   resources   are   very   important   and   I   consider   this   issue   by   taking   into   account   the   conditional   effect   of   oil   upon   the   strength   of   centralized   organizational   bases   that   enable   dictators   to   control   the   resources   effectively.   The   case   study   of   Kazakhstan   in   Chapter   4   also   provides   evidence  that  the  centralized  management  of  oil  is  crucial  to  measure  how  much  oil   contributes  to  the  dictator’s  mobilization  power.             7  Recent   literature   argues   that   natural   resources   per   se   do   not   necessarily   undermine   political   accountability   (Paler   2013)   and   retard   democratization   (Dunning   2008;   Haber   and   Menaldo   2011).   Although   whether   petroleum   is   detrimental   to   political   transparency  is  still  controversial,  much  cross-­‐national  evidence  at  least  compellingly   shows   that   natural   resources   do   prevent   authoritarian   breakdown   and   help   incumbent  dictators  survive  by  increasing  public  spending.         41   not   exposed   to   the   public   views.8  Yet,   when   using   foreign   aid,   autocrats   need   to   be   sensitive   to   international   reputation   about   how   efficiently   they   utilize   it   to   deepen   political   and   economic   reforms   (e.g.   Bermeo   2011;   Dietrich   and   Wright   2014).   Contrastingly,   most   natural   resources   are   owned   by   state   companies9  and   thus   not   constrained   by   international   scrutiny.   Such   absence   of   international   monitoring   mechanisms   makes   it   much   easy   for   dictators   to   dissipate   oil   money   for   short-­‐term   electoral  purposes.                Thus,   dictators   in   countries   with   greater   natural   resource   do   not   need   to   institutionalize  the  electoral  arena  in  ways  that  are  extremely  advantageous  to  them.   By   mobilizing   regime   supporters   through   resource   distribution,   dictators   can   win   elections  with  large  margins  without  relying  extensively  on  overt  fraud.       Hypothesis  1:  If  natural  resources  are  abundant,  dictators  are  less  likely  to  rig  elections.                                                                                                                           8  In   fact,   some   studies   show   that   foreign   aid   contributes   to   strengthening   authoritarian   rule   (e.g.   Morrison   2009;   Ahmed   2012).   On   the   other   hand,   other   research   contends   that   foreign   aid   is   positively   associated   with   democratization   especially  after  the  end  of  the  Cold  War  (Dunning  2004;  Wright  2009;  Bermeo  2011;   Dietrich   and   Wright   2014).   To   take   into   account   the   impacts   of   foreign   aid   on   electoral  fraud,  I  include  it  as  a  control  in  subsequent  empirical  analysis.         9  According   to   Ross   (2012,   37-­‐39),   natural   resources   in   the   developing   world   had   begun   to   be   nationalized   since   the   1950s   and   the   nationalization   of   petroleum   industries  had  been  completed  by  1980.  This  historical  process  suggests  that  natural   resource   wealth   can   be   used   as   a   good   proxy   to   measure   the   dictators’   ability   of   economic   distribution   after   the   1970s.   In   empirical   analyses   later   shown   in   this   chapter,  my  dataset  begins  from  1977  when  the  nationalization  of  natural  resources   were  almost  accomplished  in  developing  countries.             42   2.4.2  Organizational  Bases              Authoritarian   leaders   need   to   outsource   everyday   governance   to   the   members   of   ruling   coalitions   including   ministers,   lawmakers,   bureaucrats,   and   local   politicians.   Put   differently,   there   is   a   power-­‐sharing   contract   between   the   dictator   and   ruling   elites   (Svolik   2012;   Boix   and   Svolik   2013).   To   make   such   deals   successful,   dictators   need   to   discipline   the   members   of   ruling   coalitions   to   make   them   loyal   to   the   regime.   If  dictators  cannot  get  ruling  elites  to  comply  with  deals,  then  ruling  elites  are  more   likely   to   be   tempted   to   commit   serious   political   corruption   and   accumulate   power   resources   secretly   by   abusing   the   authority   delegated   by   the   dictator   (Haber   2006;   Magaloni  2008).                According   to   previous   work,   strong   organizational   bases   that   bind   ruling   elites   allow  dictators  to  prevent  ruling  elites  from  engaging  in  such  opportunistic  behavior   and   thus   to   promote   cooperation   of   ruling   elites.   First,   since   organizational   bases   such   as   dominant   parties   allow   dictators   to   guarantee   institutionalized   career   promotion   to   party   cadres,   dictators   can   make   inter-­‐temporal   power   sharing   deals   credible   through   the   creation   of   such   party   institutions   (Magaloni   2008;   Svolik   2012;   Boix   and   Svolik   2013).   Such   disciplined   elites   work   for   the   dictator   loyally   while   engaging   in   less   political   corruption,   leading   to   enhance   dictators’   mobilization   potential.   Second,   organizational   bases   may   enable   dictators   to   police   ruling   elites   effectively   for   detecting   politicians’   corruption   and   conspiracies,   thereby   easily   deterring   ruling   elites’   corrupt   behaviors   and   anti-­‐regime   actions.   In   African     43   countries,   when   dictators   are   facing   strong   coup   threats   from   potential   rivals   belonging  to  other  ethnic  groups,  they  often  allow  only  members  of  their  own  ethnic   groups   to   get   access   to   power,   aiming   to   increase   the   coherence   of   ruling   coalitions   (Roessler  2011).              These   disciplinary   organizations   may   have   different   implications   on   the   level   of   electoral  fraud,  depending  on  financial  resources  available  to  the  dictator.  When  the   dictator  does  not  hold  abundant  financial  resources,  he  is  more  likely  to  order  ruling   elites   to   engage   in   serious   electoral   fraud   because   he   is   unable   to   garner   voluntary   support  from  citizens.  In  such  resource-­‐scarce  scenario,  if  dictators  are  equipped  with   strong  political  organizations,  ruling  elites  are  more  likely  to  follow  orders  loyally  and   engage   in   systematic   electoral   fraud   within   local   strongholds   (Levitsky   and   Way   2010:   63).   In   fact,   dominant-­‐party   dictatorships,   when   suffering   serious   decline   in   popular   support,   systematically   resorted   to   electoral   cheating   and   violence   through   local  party  organizations  (Eisenstadt  2004;  LeBas  2006).  This  was  what  happened  in   countries  like  Mexico  from  the  late  1980s  until  the  late  1990s  and  Zimbabwe  in  the   2000s.   On   the   other   hand,   when   the   dictator   has   rich   financial   resources,   he   can   confidently  delegate  to  ruling  elites  the  task  of  collecting  popular  support  in  locality   through  economic  distribution,  rather  than  fraud.  In  this  case,  financial  resources  for   electoral   mobilization   will   be   used   more   efficiently   to   garner   popular   support   with   local   elites   committing   less   political   corruption.   In   fact,   Chang   and   Golden   (2010)   show   that   party-­‐based   regimes   tend   to   suffer   less   political   corruption   than     44   personalist   regimes   that   under-­‐institutionalize   ruling   coalitions.   In   addition,   numerous   studies   also   show   that   dictators   with   rich   financial   resources   and   dominant   parties   such   as   PRI’s   Mexico   (Magaloni   2006;   Greene   2007),   UMNO’s   Malaysia   (Pepinsky   2009),   Mubarak’s   Egypt   (Blaydes   2011)   have   all   successfully   engineered   strong   pre-­‐electoral   economic   distribution   during   their   heydays,   assuring   election  victories  and  demonstrating  regime  invincibility.10              Because   of   these   reasons,   I   hypothesize   that   strong   organizational   bases   enhance   the   dictator’s   power   of   voluntary   support   mobilization   only   when   he   holds   rich   financial   resources.   As   strong   organizations   discipline   ruling   elites,   dictators   can   effectively   streamline   the   distribution   of   public   resources.   Therefore,   financial   resources   with   strong   organizations   should   further   reduce   the   need   for   dictators   to   manipulate  elections  in  their  favor.               Hypothesis  2:  The  positive  impact  of  natural  resources  on  electoral  fairness  will  become   larger  if  dictators  possess  stronger  organizational  bases.       2.4.3  Opposition’s  Strength            The  strength  of  opposition  groups  is  also  an  important  factor  that  determines  the   extent  to  which  dictators  rely  on  electoral  manipulation.  Especially,  the  capability  of   opposition   groups   launching   collective   action   against   dictatorial   governments   is   a                                                                                                                   10  See  also  Greene  (2009)  for  cross-­‐national  evidence  regarding  the  impact  of  public   resources  on  longevity  of  dominant  party  regimes.       45   good  indicator  to  observe  to  what  extent  they  can  mobilize  their  supporters  against   the   dictator.   In   authoritarian   regimes,   challenging   dictators   is   an   extremely   costly   political  behavior  because  in  most  cases  anti-­‐regime  protests  are  brutally  repressed   (Davenport  2007).  On  the  other  hand,  however,  once  initiated,  this  costly  action  will   result   in   credibly   showing   the   dictator   that   a   considerable   number   of   people   are   extremely   unsatisfied   with   the   regime   and   opposition   leaders   have   enough   mobilization  power  to  lead  citizens  to  take  to  the  streets  (Kuran  1991;  Kricheli,  Livne,   and  Magaloni  2011;  Weiss  2012).  Therefore,  after  observing  such  popular  collective   action,   dictators   should   consider   that   they   might   not   be   able   to   bring   an   overwhelming  victory  at  the  next  election  without  serious  electoral  manipulation.  In   Zimbabwe,   when   facing   growing   power   and   protests   organized   by   the   opposition   party  Movement  for  Democratic  Change  (MDC),  President  Robert  Mugabe  resorted  to   brutal   election   violence   and   serious   vote   stuffing   in   the   2008   presidential   election.   Consequently,   the   MDC   presidential   candidate   Morgan   Tsvangiarai   withdrew   from   the  presidential  race  in  the  second  round,  which  resulted  in  Mugabe’s  victory  with  85   percent   of   the   vote   (Bratton   and   Masunungure   2008,   41).   Conversely,   if   opposition   parties   are   weak   so   as   to   not   be   able   to   mobilize   citizens   and   coordinate   collective   action,  dictators  are  willing  to  open  the  electoral  field,  because  it  is  highly  likely  that   they   can   win   big.   Prior   to   the   Kazakhstan   2012   elections,   President   Nazarbaev   changed   the   electoral   law   to   make   it   easier   for   moderate   opposition   parties   to   obtain   seats.   One   of   his   motivations   behind   this   electoral   reform   was   in   the   fact   that     46   opposition  parties  were  too  unpopular  among  citizens  to  threaten  his  overwhelming   victory   in   the   election.   Despite   the   most   “transparent”   elections   in   over   a   decade   in   Kazakhstan   (Olcott   2012),   the   dominant   party   Nur   Otan   obtained   80.99   percent   of   total  votes  whereas  the  main  opposition  National  Social  Democratic  Party  only  gained   a  mere  1.4  percent  of  votes.    Therefore,  contrary  to  Magaloni  (2011)  who  expects  that   strong   opposition   urge   the   autocrat   to   hold   less   fraudulent   elections,   Hypothesis   3   can  be  expressed  as  follows:           Hypothesis  3:  The  higher  the  oppositions’  capability  of  collective  action,  the  more  likely   dictators  are  to  use  serious  electoral  fraud.     2.5  Cross-­‐National  Statistical  Analysis  of  Electoral  Fraud     2.5.1  Data  and  Methodology              The  unit  of  analysis  is  country/election  year  in  authoritarian  regimes.  First  of  all,  I   identify   authoritarian   regimes   by   using   Geddes,   Wright   and   Frantz’s   (2014)   Autocratic   Regime   Dataset   (GWF   Data),   which   codes   each   regime   failure   and   the   subsequent  regime  government  as  democracy  or  dictatorship.11                                                                                                                     11  For  countries  and  election  years  included  in  empirical  analysis,  see  Appendix  C2.  I   do  not  use  Cheibub,  Gandhi  and  Vreeland’s  (2010)  dichotomous  measure  of  political   regimes,   since   in   their   dataset   countries   are   regarded   to   have   democratized   in   the   country-­‐year  when  governments  lost  elections.  This  is  problematic  for  this  research   because  it  arbitrary  censors  the  dependent  variable  and  excludes  the  possibility  that     47   Figure  2.1:  Variation  in  the  Level  of  Electoral  Fraud  in  Authoritarian  Regimes   (1977-­‐2004)             Note:   The   data   source   is   Kelley   (2012).   VCL   consists   of   pre-­‐electoral   violence,   election-­‐day  violence  pre-­‐electoral  cheating,  election-­‐day  electoral  cheating  and  legal   structural  problems.  VC  only  includes  electoral  violence  and  cheating  while  excluding   legal   problems.   Each   component   has   4   scales   (0   =   no   problems,   1=minor   problems   only,  2=moderate  problems,  3=major  problems).                      To   measure   the   dependent   variable,   Electoral   Fraud,   this   study   relies   on   The   Quality   of   Elections   Dataset   (QED)   constructed   by   Kelley   (2012). 12  The   dataset                                                                                                                                                                                                                                                                                                                                                 the  dictators  hold  fair  elections  enough  to  bring  government  turnover.  In  this  respect,   as  the  GWF  dataset  codes  that  democratization  occurs  one  year  after  the  election  that   brought   government   turnover,   it   has   an   advantage   in   avoiding   this   censoring   problem.       12  There   are   two   other   global   datasets   that   are   publicly   available   and   globally   measure   electoral   fraud.   The   first   is   Birch’s   (2011)   Index   of   Election   Malpractice   (IEM).  The  second  is  Hyde  and  Marinov’s  (2012)   National  Elections  across  Democracy   and   Autocracy   (NELDA).   IEM   is   similar   with   QED   in   many   respects,   yet   only   covers   between   1995   and   2007.   NELDA   covers   more   extensive   period   (1945-­‐2010),   yet   focuses   more   on   the   competitiveness   of   elections   and   does   not   evaluate   the   magnitude  of  election  cheating  and  election  violence.       48   evaluates   to   what   extent   each   election   is   characterized   by   electoral   fraud   using   country   reports   published   by   the   U.S.   Department   of   State.   In   light   of  the   definition   of   electoral   fraud   presented   above,   I   use   the   following   five   variables   to   measure   the   extent   of   election   fraud   from   this   dataset:  13  (1)   Was   the   legal   framework   not   up   to   standards,   limits   on   the   scope   and   jurisdiction   of   elective   offices,   and   unreasonably   limits  of  who  can  run  for  office  etc.?  (“Legal  problems”)  (2)  Were  there  restrictions  on   freedom   to   campaign,   media   restrictions,   intimidations,   and   improper   use   of   public   funds?   (“Pre-­‐electoral   cheating”)   (3)   Was   there   any   violence   or   unrest   before   an   election   day?   (“Pre-­‐electoral   violence”)   (4)   Did   any   vote   padding,   tampering   with   ballot  box,  voter  impersonation,  double  voting,  vote  buying,14  intimidation  etc.  occur?                                                                                                                                                                                                                                                                                                                                                   13  Previous   cross-­‐national   studies   (Birch   2011;   Kelley   2012)   use   the   overall   evaluation   on   the   quality   of   elections   to   measure   electoral   fraud,   yet   this   treatment   is   problematic   especially   for   this   research.   The   overall   evaluation   includes   election   administrative   capacities   as   a   subcomponent   measuring   electoral   fraud,   yet   this   does   not   necessarily   come   from   dictators’   eagerness   to   commit   electoral   fraud   but   often   stems  from  the  quality  of  bureaucracy  and  central  electoral  management  bodies.       14  Election   day   cheating   includes   vote-­‐buying   practices   and   thus   one   may   think   that   this   component   of   fraud   might   be   overlapped   with   pre-­‐electoral   manipulation   of   fiscal  policy  that  I  conceptualize  not  as  a  strategy  of  electoral  fraud.  Yet,  vote  buying   and   manipulation   of   economic   policy   instruments   are   theoretically   different   techniques   of   distributing   economic   favors.   The   former   assumes   direct,   illegal   exchanges   between   votes   and   (relatively   small)   goods   under   monitored   circumstances,   while   the   latter   is   delivered   more   or   less   legally   through   programmatic  fiscal  policy  and/or  pork-­‐barrel  politics,  not  necessarily  presupposing   clientelistic   structures   (Stokes   et   al.   2013,   6-­‐18).   In   addition,   if   the   data   wrongly   codes   pre-­‐electoral   manipulation   of   economic   policy   instruments   as   vote   buying   practices   in   a   serious   way,   election-­‐day   cheating   should   be   positively   associated   with   financial  resources.  Yet,  oil-­‐gas  value  per  capita  is  negatively  correlated  with  election-­‐ day  cheating,  suggesting  the  possible  measurement  errors  are  not  that  serious.           49   (“Election-­‐day   cheating”)   (5)   Did   any   violence   and   unrest   occur   on   an   election   day?   (“Election-­‐day  violence”).  Each  variable  takes  4  values  between  0  (no  problems)  and   3  (serious  problems).  I  aggregate  the  five  variables  and  create  a  dependent  variable   Election  Fraud  (Violence,  Cheating  and  Legal  Problems;  VCL)  that  ranges  from  0  to  15.   As  an  alternative  measure,  I  use  another  dependent  variable  that  takes  into  account   only   election   violence   and   cheating   (Violence   and   Cheating;   VC).15  Since   voters   can   recognize   relatively   easily   to   what   extent   dictators   commit   electoral   violence   and   cheating  at  the  ballot  box  compared  to  implicit  legal  restrictions,  dictators  should  be   more   careful   about   whether   they   use   election   violence   and   cheating   to   maintain   credibility  of  election  results.              My   dataset   includes   324   elections   in   78   authoritarian   countries   between   1977-­‐ 2004.  One  may  think  dictatorships  with  no  electoral  competition  should  be  excluded   from   the   sample   because   dictators   have   no   incentive   to   resort   to   fraud   in   such   plebiscite  elections.  Yet,  my  theory  suggests  that  whether  to  allow  opposition  parties   to   join   the   electoral   battle   is   part   of   dictators’   toolbox   of   electoral   manipulation.   Indeed,  this  aspect  is  captured  in  a  subcomponent  of  electoral  fraud,  “legal  problems.”   I   also   conduct   a   difference   of   means   t-­‐test   to   see   if   there   is   a   meaningful   difference   between  the  two  types  of  dictatorships  with  regards  to  the  level  of  electoral  fraud  and                                                                                                                                                                                                                                                                                                                                                                             15  This  variable  ranges  from  0  to  12  aggregating  the  values  of  questions  (2)  through   (5).       50   do   not   find   a   statistically   significant   difference.16  Figure   2.1   shows   two   histograms   on   the  distribution  of  the  dependent  variables,  VCL  and  VC.              To  test  Hypothesis  1  about  natural  resource  abundance,  I  use  Ross’  (2012)  oil-­‐gas   value   per   capita   in   constant   2000   dollars   to   operationalize   economic   resources   that   dictators  can  use  at  their  disposal.  The  variable  is  calculated  multiplying  a  country’s   total   oil-­‐gas   production   by   the   current   oil-­‐gas   price   and   then   divided   by   total   population.              To   test   Hypothesis   2   about   the   conditional   effect   of   resource   abundance,   I   introduce   an   interaction   term   of   Ross’   (2012)   oil-­‐gas   variable   and   dictators’   organizational   bases.   Organizational   bases   are   empirically   derived   from   two   perspectives:   (1)   party-­‐based   organizations   and   (2)   ethnicity-­‐based   networks.   Levitsky   and   Way   (2010,   60-­‐66)   conclude   that   dominant   party   organizations   and   ethnic-­‐based   identity   are   two   of   the   most   important   aspects   to   measure   dictators’   organizational   power.   Dominant-­‐party   regimes   are   likely   to   solve   commitment   problems   between   the   dictator   and   ruling   elites   (Magaloni   2008).   Increasing   coherence   of   ruling   coalitions,   dominant-­‐party   regimes   contribute   to   reducing   the   need   for   fraud   by   increasing   the   efficiency   of   economic   distribution,   but   only   when   the  dictator  has  rich  financial  resources  to  distribute.  I  use  a  dummy  variable  based   on  Geddes,  Wright,  and  Frantz  (2014)  to  code  party-­‐based  regimes.  This  measure  of                                                                                                                   16  I  rely  on  Hyde  and  Marinov’s  (2012)  empirical  definition  of  electoral  competition:   elections   are   competitive   if   they   satisfy   that   (1)   opposition   is   allowed,   (2)   multiple   parties  are  legal,  and  (3)  more  than  one  candidate  compete.               51   dominant-­‐party   regimes   is   appropriate,   because   the   variable   defines   party   regimes   based   on   whether   party   organizations   control   selection   of   officials,   organize   distribution   of   benefits,   mobilize   citizens   to   vote   and   show   party   support   (Geddes   2003;   Wilson   2013:   5)   “though   other   parties   may   exist   and   compete   as   minor   parties   in  elections,”  (Geddes  2003:  51)17              The  size  and  cohesiveness  of  dominant  ethnic  groups  is  another  good  measure  to   capture  a  different  aspect  of  organizational  bases  —  the  ethnic  organizational  power   (EOP)   of   dictators.   If   an   ethnic   coalition   consists   of   many   ethnic   groups,   it   is   more   difficult   to   monitor   ruling   elites’   opportunistic   behaviors   through   intra-­‐ethnic   policing   mechanisms   (Fearon   and   Laitin   1995).   Indeed,   cross-­‐national   studies   have   demonstrated   that   ethnically   fractionalized   ruling   coalitions   are   more   exposed   to   coup   risk   (Wimmer,   Cederman,   and   Min   2009;   Roessler   2011).   Furthermore,   as   the   size   of   ethnic   groups   gets   larger,   it   becomes   easier   for   dictators   to   distribute   patronage   to   broader   citizens   by   using   extensive   ethnic   networks   than   when   their   ethnic   groups   are   small   (Fearon   1999;   Chandra   2004;   Habyarimana   et   al.   2008).   A   large   literature   demonstrates,   in   fact,   that   ethnic   diversity   prevents   efficient   public   good  provisions  (Alesina,  Baqir,  and  Easterly  1999;  Habyarimana  et  al.  2008;  Franck   and   Rainer   2012).   In   measuring   the   size   and   cohesiveness   of   dominant   ethnic   groups   simultaneously,   I   construct   an   ethnic   organizational   power   index,   which   is   made   by                                                                                                                   17  Other   datasets   such   as   Hadenius   and   Teorell   (2007)   and   Svolik   (2012)   focus   primarily  on  the  number  of  parties  in  legislature  to  define  single-­‐party  regimes.       52   multiplying   the   proportion   of   politically   dominant   ethnic   groups   in   total   population   by   the   reversed   fractionalization   index   of   dominant   ethnic   groups.18  The   variable   has   a  lower  value  as  ethnic  groups  in  ethnic  coalitions  occupy  a  smaller  portion  of  total   population  and  as  ethnic  coalitions  are  composed  of  more  ethnic  groups.  I  expect  that   the  effect  of  financial  resources  will  be  conditioned  by  both  regime  types  and  ethnic   organizational   power:   Financial   resources   make   autocratic   elections   less   obviously   fraudulent   if   dictators   have   party-­‐based   regimes   and   large,   cohesive   ethnic   coalitions   as  supporting  bases.              For  Hypothesis  3,  I  operationalize  the  opposition’s  capabilities  of  collective  action.  I   follow   Bueno   de   Mesquita   and   Smith   (2010)   and   Howard   and   Roessler   (2006:   372)   and  code  “revolutionary  threats”  or  anti-­‐government  collective  action  by  counting  the   number  of  demonstrations,  riots  and  strikes  from  Arthur  Bank’s  Cross-­‐National-­‐Time-­‐ Series  Data  Archive.   One   possible   problem   with   this   measure   is   that   protests   in   the   immediate   past   may   be   highly   correlated   with   those   in   the   present,   and   the   observed   correlation  between  protests  and  fraud  may  be  due  to  the  fact  that  serious  electoral   fraud   provokes   post-­‐electoral   protests   (Hafner-­‐Burton,   Hyde,   and   Jablonski   2013).   In   fact,   the   correlation   between   post-­‐electoral   protests   (Hyde   and   Marinov   [2012]’s   NELDA  29)   and   the   one-­‐year   lagged   collective   action   variable   is   quite   high   (0.25)   and                                                                                                                   18  The   reversed   fractionalization   index   is   based   on   the   Herfindahl   formula   and   ! expressed   as   (1−   ! !!! 𝑃𝐷𝐸𝐺! )  where  𝑃𝐷𝐸𝐺!  is   the   share   of   a   politically   dominant   ethnic  group  in  total  population.  Thus,  the  ethnic  organizational  power  is  formulized   ! ! as  ( ! !!! 𝑃𝐷𝐸𝐺! ) ∗   (1 −   !!! 𝑃𝐷𝐸𝐺! ).     To   distinguish   politically   dominant   ethnic   groups  from  politically  excluded  ones,  I  rely  on  Wimmer,  Cederman  and  Min’s  (2010)   Ethnic  Power  Relations  Dataset.     53   statistically   significant,   suggesting   the   possibility   of   reversed   causality.   Therefore,   I   use  a  three-­‐year  moving  average  (one  year  lagged)  of  the  number  of  collective  actions   (riots,   demonstrations   and   strikes)   to   better   capture   the   history   of   opposition’s   organizational   strength.   Using   the   three-­‐year   moving   average,   the   correlation   between   the   collective   action   variable   and   post-­‐electoral   protests   becomes   much   lower  (0.09)  and  statistically  insignificant,  implying  that  the  variable  can  now  better   estimate  the  effect  of  collective  action  capabilities  on  fraud,  while  mitigating  the  risk   of  reversed  causality.                          In  addition,  I  use  a  host  of  control  variables,  based  on  previous  work  on  electoral   fraud,  authoritarian  politics  and  democratization.  Since  electoral  fraud  is  more  likely   in   repressive   authoritarian   regimes,   estimating   models   without   controlling   for   the   degree   of   pre-­‐existing   political   freedom   in   authoritarian   regimes   will   bias   my   estimates   for   the   three   hypotheses.   Therefore,   I   use   the   Freedom   House   Index.   This   variable   is   transformed   to   facilitate   interpretation   and,   thus,   ranges   from   2   (least   free)   to   14   (most   free),   capturing   the   extent   to   which   civil   liberties   and   political   freedom   are   guaranteed.       To   mitigate   the   risk   that   the   measure   picks   up   election   fraud   and   captures   the   overall   trends   in   political   openness,   the   variable   is   a   three-­‐ year  moving  average  lagged  one  year.              Domestic  institutional  covariates  are  also  important  explanations  of  electoral  fraud   and   thus   should   be   controlled   for.   Regime   types   are   important.   As   Geddes   (1999)   suggested,  military  dictators  are  more  likely  to  step  down  and  thus  may  allow  fair  and     54   free  elections  whereas  personalist  dictatorships  tend  to  cling  to  power  and  thus  may   rig   elections   more   seriously.   Using   Geddes,   Wright   and   Frantz   (2014),   I   introduce   dummy   variables   for   Military   and   Personalist   Dictatorships.   Some   research   suggests   that  Domestic  Election  Monitoring  is  effective  to  lower  the  level  of  fraud  (Ichino  and   Schuendeln   2012).   I   use   Kelley’s   (2012)   dummy   variable   identifying   whether   domestic   election   monitoring   exists   in   a   given   election.   Independent,   professional   electoral   management   bodies   (Election   Administrative   Capacities)   may   improve   the   quality   of   elections   (Hartlyn,   McCoy,   and   Mustillo.   2008).     In   elections   that   determine   who   will   hold   executive   power,   authoritarian   leaders   may   be   more   inclined   to   manipulate   elections   because   the   electoral   stakes   are   high   (Lehoucq   and   Molina   2002).   I   follow   Simpser   and   Donno   (2012)   and   include   a   dichotomous   variable   called   Main  Elections,  which  is  1  for  presidential  elections  in  presidential  (or  mixed)  systems   and  parliamentary  elections  in  parliamentary  systems.                In  addition  to  political  institutions,  international  reputation  is  also  a  crucial  factor   that  authoritarian  leaders  have  to  concern  at  elections.  The  international  community   may   decide   to   cut   foreign   aid   or   impose   economic   sanction   following   rigged   elections   (Bermeo  2011).  To  control  for  to  what  extent  dictators  will  care  about  international   reputation,   I   introduce   three   variables.   A   first   codes   whether   a   country   convenes   International   Election   Monitoring  (Hyde  2007;  Kelley  2012).  A  second  variable  codes   the   amount   of   Foreign  Aid   that   dictators   are   receiving.   I   use   percentage   of   foreign   aid   relative   to   GDP   (Ahmed   2012).   A   third   variable   is   Trade   Openness   (sum   of   imports     55   and  exports  relative  to  GDP)  is  included  to  accesses  how  economically  dependent  an   authoritarian  country  is  in  the  sphere  of  the  international  market  (Burch  2011).                To  account  for  socio-­‐economic  covariates  influencing  the  extent  of  electoral  fraud,  I   also   include   logged   GDP   per   capita   (one   year   lagged)   and   GDP   growth   (one   year   lagged)  to  control  for  the  level  of  and  change  in  economic  development  (Lehoucq  and   Molina  2002).  Rural  population  (one  year  lagged,  percent  of  total  population)  is  also   controlled   for   because   in   rural   societies   authoritarian   governments   find   it   easier   to   engage  in  electoral  fraud  (Burch  2011,  62).  Autocrats  may  be  more  likely  to  engage  in   electoral  fraud  depending  on  their  time  in  office.  Thus,  I  control  for  Leader’s  Tenure   Length.              As  the  dependent  variables  are  continuous,  I  use  random  effects  OLS  models  that   account   for   the   likelihood   that   the   effects   of   the   main   variables   may   differ   systematically   for   each   country.   Since   the   party-­‐based   regime   variable   and   the   ethnic   organizational   power   index   rarely   vary   over   time,   a   fixed   effects   model   cannot   be   employed   for   Hypothesis   2   and   thus   not   presented   here   as   main   analysis.   Instead,   I   show   its   results   in   robustness   check.19  Considering   the   likelihood   that   errors   are   correlated  within  each  country,  I  adopt  robust  standard  errors  clustered  by  country.  I   also  include  decade  interval  dummies  (1980s,  1990s  and  2000s)  to  control  for  time-­‐ specific  effects.                                                                                                                       19  Also   for   results   with   other   estimation   techniques,   see   the   robustness   check   section   and  Appendix  C2.     56   2.5.2  Results   Table  2-­‐1:  Determinants  of  Electoral  Fraud  in  Dictatorship   Oil2Gas&Value&per&capita&(1&year&lagged) Collective&Action&(Avg,&1&year&lagged) Party2Based&Regime Model&1 VCL 20.00937*** (0.003) 0.139** (0.070) 20.652 (0.671) Model&2 VC 20.00762** (0.003) 0.176** (0.073) 0.35 (0.671) 1.992*** (0.662) Model&3 VCL 20.00822*** (0.003) 0.129* (0.067) 20.293 (0.593) 20.0936*** (0.023) 1.535** (0.599) Model&4 VC 20.00671*** (0.003) 0.165** (0.072) 0.708 (0.604) 20.102*** (0.024) 1.893*** (0.625) 1.651*** (0.640) 22.045*** (0.78) 20.151 (0.768) 0.0326 (0.024) 20.482*** (0.079) 20.423*** (0.154) 0.124 (0.188) 0.099 (0.397) 20.0512** (0.021) 0.0283* (0.016) 20.008 (0.005) 20.2 (0.464) 0.26 (0.510) 20.0279 (0.018) 5.878 (4.179) 78 Yes 321 246.33*** 0.46 20.366 (0.78) 0.655 (0.760) 0.0102 (0.023) 20.219*** (0.073) 20.516*** (0.185) 0.138 (0.166) 20.28 (0.455) 20.0500** (0.023) 0.0177 (0.017) 20.00577 (0.004) 20.272 (0.482) 0.474 (0.480) 20.019 (0.020) 5.776 (4.532) 78 Yes 324 212.29*** 0.37 21.904*** (0.72) 20.00275 (0.709) 0.035 (0.023) 20.504*** (0.080) 20.452*** (0.147) 0.136 (0.188) 0.196 (0.402) 20.0583*** (0.022) 0.0275* (0.015) 20.00732 (0.005) 20.28 (0.456) 0.246 (0.501) 20.0300* (0.018) 5.476 (4.153) 78 Yes 321 529.3*** 0.48 20.245 (0.73) 0.78 (0.701) 0.012 (0.022) 20.241*** (0.073) 20.541*** (0.183) 0.150 (0.167) 20.185 (0.466) 20.0577** (0.023) 0.017 (0.016) 20.00519 (0.004) 20.339 (0.478) 0.459 (0.470) 20.0213 (0.019) 5.384 (4.551) 78 Yes 324 693.9*** 0.39 Oil*Party Ethnnic&Organizational&Power&(EOP) Oil*EOP Military&Regime Personalist&Regime Leader's&Tenure&Length Freedom&House&(Avg,&1&year&lagged) Election&Administrative&Capacity Main&Elections Logged&GDP&percapita&(1&year&lagged) Economic&Growth&(1&year&lagged) Rural&Population&(1&year&lagged) Trade&Openness&(1&year&lagged) Domestic&Election&Monitoring International&Election&Monitoring& Foreign&Aid&(1&year&lagged) Constant Number&of&Countries Decade&Dummies Number&of&Observations Wald&Chi2 R&squared&(overall) Mode&5 VCL 0.0148^ (0.016) 0.137* (0.070) 20.613 (0.657) Model&6 VC 0.0302* (0.016) 0.173** (0.073) 0.412 (0.653) 1.781*** (0.665) 20.115^ (0.076) 22.032*** (0.76) 20.135 (0.753) 0.031 (0.024) 20.489*** (0.079) 20.419*** (0.156) 0.130 (0.189) 0.116 (0.393) 20.0498** (0.021) 0.0276* (0.016) 20.00778 (0.005) 20.212 (0.466) 0.267 (0.510) 20.0284 (0.018) 5.775 (4.134) 78 Yes 321 254.4*** 0.46 2.199*** (0.675) 20.180** (0.077) 20.342 (0.765) 0.673 (0.736) 0.00777 (0.023) 20.230*** (0.072) 20.511*** (0.186) 0.148 (0.168) 20.262 (0.445) 20.0477** (0.023) 0.0168 (0.016) 20.00569 (0.004) 20.291 (0.482) 0.484 (0.477) 20.0204 (0.020) 5.679 (4.43) 78 Yes 324 229.95 0.38 Note:  Clustered  robust  standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;  *p<0.1.  ^   indicates  that  a  chi-­‐squared  joint  test  shows  that  the  interaction  effect  is  statistically   significant   at   the   1   percent   level.   VCL=   election   violence   +   election   cheating   +   legal   problems  (0-­‐15),  VC  =  election  violence  +  election  cheating  (0-­‐12).       57                In  Table  2.1,  Models  1  and  2  examine  Hypotheses  1  and  3.  The  variable  of  oil-­‐gas   value   per   capita   is   statistically   significant   and   negatively   correlated   with   electoral   fraud  in  both  models.  A  case  in  point  is  the  Suharto  regime  of  Indonesia.  In  the  late   1970s   and   mid   1980s,   Suharto   enjoyed   rich   oil   revenue   because   of   surges   in   international   oil   prices   (Smith   2007,   135-­‐137)20  and   thus   parliamentary   elections   held  in  1977  and  1982  were  both  relatively  clean  (the  election  fraud  level  [including   cheating,  violence  and  legal  problems]  was  2  and  4  respectively  in  my  data).  After  a   significant  drop  in  oil  price  in  the  late  1980s,  however,  the  Suharto  regime  began  to   seriously   rig   elections.   In   the   1997   parliamentary   elections,   when   oil-­‐gas   value   per   capita  was  just  98  dollars,  the  level  of  electoral  fraud  reached  15,  the  maximum  value   in  the  sample.                Models   1   and   2   also   show   that   the   opposition’s   collective   action   is   positively   correlated   with   electoral   fraud   in   a   statistically   significant   way,   supporting   Hypothesis  3.  The  more  riots,  demonstrations  and  strikes  authoritarian  leaders  face   during  three  years  before  an  election,  the  more  seriously  they  rig  elections.  Countries   like   Kenya   (1992),   Zimbabwe   (2000),   Jordan   (1997),   Tajikistan   (1995),   Bangladesh   (1986),  the  Philippines  (1984)  and  Indonesia  (1997)  are  typical  cases  where  strong   opposition   movements   forced   the   governments   to   resort   to   extensive   electoral   manipulation.                                                                                                                                                                                                                                                 20  Oil   revenue   in   Indonesia   in   the   early   1980s   (305   dollars)   was   about   ten   times   as   high  as  in  the  early  1970s.       58              The   party-­‐based   regime   dummy   by   itself   does   not   have   a   statistically   significant   impact   on   electoral   fraud   in   Models   1   and   2.   Ethnic   organizational   power   (EOP),   on   the   other   hand,   is   positively   correlated   with   electoral   fraud   in   both   models.   As   previous   studies   suggest   (Levitsky   and   Way   2010,   63),   this   suggests   that   extensive   organizational   bases   may   become   a   political   machine   to   systematically   fabricate   election  results.  It  does  so  especially  when  dictators  have  few  financial  resources  to   distribute  to  the  citizenry.       Figure  2-­‐2:  The  Different  Effects  of  Natural  Resource  Endowments  between   Dominant-­‐Party  and  Non-­‐Dominant-­‐Party  Regimes         Note:  The  left  and  right  graphs  are  based  on  estimation  results  of  Models  3  and  4,   respectively.                        In  Models  3  and  4,  I  examine  the  effect  of  oil-­‐gas  value  per  capita  conditional  on   the   presence   of  the   party-­‐based  regime.   The   evidence   supports   Hypothesis   2.   In   both   models,  the  oil-­‐gas  rent  variable  and  its  interaction  term  with  the  party-­‐based  regime   dummy   are   negative   and   statistically   significant.   Figures   2.2-­‐(a,b)   visually     59   demonstrate  the  difference  in  the  impact  of  natural  resources  on  electoral  fraud  for   party-­‐based   and   non-­‐party-­‐based   regimes.   In   non-­‐party-­‐based   regimes,   natural   resources   lower   electoral   fraud   very   little   (0.008).   If   the   country   is   party-­‐based   regime,   however,   the   impact   of   natural   resources   is   about   12   times   higher.   Under   dominant   party   regimes   that   include   Malaysia,   Mexico,   Tanzania   and   Gabon,   a   100-­‐ dollar  increase  in  oil  and  gas  revenue  lowers  level  of  electoral  fraud  by  0.1.       Figure  2-­‐3:  The  Marginal  Effect  of  Natural  Resource  Endowments  Conditional   upon  Ethnic  Organizational  Power       Note:  The  left  and  right  graphs  are  based  on  estimation  results  of  Models  5  and  6,   respectively.                  Models  5  and  6  provide  additional  evidence  supporting  for  Hypothesis  2  from  the   other   perspective   –   the   ethnic   organizational   power   (EOP).   Interaction   terms   of   the   EOP   variable   and   oil-­‐gas   value   per   capita   are   negative   and   statistically   significant.   Figures   2.3-­‐(a,   b)   depict   graphs   illustrating   how   the   coefficient   of   oil-­‐gas   per   capita   will   change   according   to   values   of   the   EOP.   Where   the   EOP   is   weak   (0   to   0.17)   like   Cameroon,   Zaire,   Tunisia,   and   Syria,   the   impact   of   natural   resource   endowments   is     60   not   statistically   distinguishable   from   0.   As   politically   dominant   ethnic   groups   become   more   cohesive   and   larger,   the   negative   impact   of   natural   resource   endowments   becomes   larger.   At   the   maximum   value   where   only   one   dominant   ethnic   group   occupies   98   percent   of   total   population   (Ethnic   Organizational   Power   =   0.98),   the   coefficient  of  oil-­‐gas  rent  per  capita  is  about  -­‐0.1,  which  is  12  times  larger  than  when   EOP  is  equal  to  0.2.         2.5.3  Robustness  Check  and  Additional  Data  Analysis            For   robustness   checks,   I   conduct   a   battery   of   sensitivity   analyses. 21  First,   I   alternatively   adopt   different   dichotomous   measures   of   political   regimes   by   using   Boix,  Miller  and  Rosato  (2012)  and  Polity  IV22  to  see  whether  the  results  are  sensitive   to   other   filtering   criteria   distinguishing   authoritarian   from   democratic   countries.   These   alternative   classifications   of   authoritarian   regimes   do   not   change   the   overall   results.   Second,   all   the   models   are   re-­‐estimated   with   different   measures   of   natural   resources  (1.  change  in  oil-­‐gas  value  per  capita23,  2.  three  years  moving  average  of  oil-­‐                                                                                                                 21  For  detailed  estimation  results,  see  Appendix  C2.       22  Following   the   convention,   I   define   a   country   as   authoritarian   if   its   Polity   IV   score   is   less  than  6.       23  Although   the   current   measure   focuses   on   the   level   of   oil-­‐gas   value   per   capita,   change  in  natural  resources  may  be  also  important,  considering  that  citizens’  voting   behavior   is   based   on   retrospective   evaluations   on   the   government’s   economic   performance   and   thus   sensitive   to   income   shocks.   Here   change   in   oil-­‐gas   value   per   capita  is  measured  by  taking  percent  change  in  the  variable  between  (t-­‐2)  and  (t-­‐1)   years  before  an  election.       61   gas   value   per   capita 24 ,   and   3.   Haber   and   Menaldo’s   [2011]   measure   of   natural   resource  wealth25),  and  the  results  are  robust  to  these  alternative  measurements  of   natural  resources.  Third,  the  extent  of  electoral  fraud  may  be  different  depending  on   the  country’s  regional  location  because  the  likelihood  of  democratic  transition  may  be   influenced   by   regional   diffusion   of   democratization.   This   may   make   measurement   errors   correlate   within   each   region,   which   possibly   induces   an   underestimation   of   standard   errors.   Thus,   instead   of   country-­‐clustered   robust   standard   errors,   I   use   robust  standard  errors  clustered  by  region,  yet  the  main  results  do  not  change.  Forth,   my   results   are   robust   to   the   simultaneous   inclusion   of   other   relevant   covariates   (memberships   of   democratic   international   organizations   [one   year   lagged], 26                                                                                                                   24  Since  the  dictators  may  be  able  to  store  natural  resources  for  some  period  of  time,   taking  average  for  several  years  may  be  a  better  indicator  to  measure  the  dictators’   available  financial  resources.         25  In   addition   to   oil   and   natural   gas,   Haber   and   Menaldo’s   (2011)   measure   includes   other   types   of   natural   resource   wealth   such   as   coals   and   metals.   Ross’   (2012)   measure   of   oil-­‐gas   value   per   capita   is   very   highly   correlated   with   Haber   and   Menaldo’s:  the  correlation  coefficient  is  0.989.             26  As   the   democratization   literature   argues,   strong   pressures   from   inter-­‐ governmental   organizations   induce   authoritarian   countries   to   democratiza   (e.g.   Pevehouse   2005).   I   first   identify   each   country’s   membership   of   international   organization   using   the   State   System   Membership   Dataset   by   the   Correlates   of   War   Project   (2011   version).   Then,   I   calculate   how   democratic   each   international   organization   is   by   taking   the   average   of   Cheibub,   Gandhi   and   Vreeland’s   (2009)   measure  of  democracy  for  all  member  states  other  than  the  county  under  study.       62   electoral   systems,27  foreign   direct   investment   [one   year   lagged],28  logged   population   size   [one   year   lagged], 29  military   spending   per   capita   [one   year   lagged], 30  and   duration   of   violent   conflicts   [one   year   lagged]31).   Lastly,   I   employ   other   estimation   methods   like   fixed   effects,   random   effects   with   regional   dummies,   and   random   intercept   models 32  to   see   to   what   extent   the   results   are   robust   to   alternative                                                                                                                   27  Burch   (2007)   contends   that   Single   Member   Districts   systems   are   more   likely   to   induce  serious  electoral  fraud.  Here  I  use  Effective  Electoral  Threshold  to  control  for   electoral  system  types.       28  I  use  inflow  of  foreign  direct  investment  relative  to  GDP.       29  Lehoucq  and  Molina  (2002)  and  Fukumoto  and  Horiuchi  (2011)  argue  that  election   fraud  tends  to  be  positively  associated  with  less  populated  areas.       30  Natural  resource  wealth  may  be  also  used  to  strengthen  security  apparatus  (Ross   2001,  332-­‐336).  To  make  sure  that  the  effect  of  natural  resource  endowments  largely   comes  from  economic  distribution,  I  control  for  military  spending  per  capita.           31  When   civil   conflict   is   ongoing,   the   governments   are   more   likely   to   violate   human   rights   and   thus   use   electoral   violence   (e.g.   Dunning   2011).   I   use   Armed   Conflict   Dataset  in  which  violent  conflict  is  defined  as  at  least  25  battle  deaths  per  year.           32  Since   the   party-­‐based   regime   and   EOP   dummies   rarely   change   over   time,   it   is   difficult   to   distinctively   estimate   the   within-­‐group   effects   from   the   between-­‐group   effects  by  using  fixed  effects  models  with  those  time-­‐invariant  variables.  On  the  other   hand,   fixed   effects   models   have   an   advantage   in   controlling   for   country-­‐specific   confounding   factors   that   do   not   change   over   time.   Therefore,   I   use   fixed   effects   estimation   to   test   Hypotheses   1   and   3.   For   Hypothesis   2,   instead   of   fixed   effects,   I   introduce   five   regional   dummies   to   control   for   regional-­‐specific   heterogeneous   effects.  Random  intercept  models  are  used  for  all  the  hypotheses.             63   estimation   strategies.   Other   than   one   exception,33  these   different   estimation   methods   do  not  affect  my  results.                                                                                                                                                       33  This   is   the   collective   action   variable   in   a   fixed   effect   model   where   VCL   is   the   dependent   variable.   In   this   model,   the   collective   action   variable   does   not   reach   to   the   10  percent  statistical  significance  (p-­‐value  =  0.17).         64   Figure  2-­‐4:  Additional  Analyses                   65     Note:   For   details   on   estimation   results,   see   Appendix   C2.   Ordered   logistic   regressions   are   estimated.   The   graphs   show   predicted   probabilities   that   any   level   of   electoral   violence  or  cheating  occurs.                    For   additional   analysis,   I   disaggregate   the   Election   Fraud   variable   into   election   violence   and   election   cheating   parts   and   run   ordered   logistic   regressions   with   clustered   robust   standard   errors   to   examine   whether   different   causal   mechanisms   exist   between   electoral   violence   and   electoral   cheating. 34  Oil-­‐gas   rent   per   capita   always  has  a  negative  association  with  both  electoral  violence  and  cheating  (Models  7   and   8).   Second,   the   collective   action   variable   is   better   at   explaining   variation   in   electoral   violence   rather   than   electoral   cheating   (Models   7   and   8).   Third,   an   interaction  between  dominant-­‐party  regime  and  natural  resource  endowments  has  a   statistically   significant   impact   on   reducing   both   electoral   violence   and   cheating   (Models   9   and   10).   Figures   2.4   (a-­‐d)   graphically   show   change   in   predicted   probabilities  of  any  level  of  electoral  violence  and  cheating,  suggesting  that,  when  an   authoritarian  country  is  party-­‐based  regime,  the  likelihood  of  electoral  violence  and   cheating  tends  to  shrink  more  rapidly  as  natural  resource  endowments  increase.  On   the   other   hand,   the   conditional   effect   of   the   oil-­‐gas   variable   upon   ethnic   organizational   power   explains   only   electoral   cheating   (Models   11   and   12.   See   also   Figure  2.4  [e,f]).  In  sum,  these  additional  analyses  suggest  that  all  other  things  being                                                                                                                   34  Lumping   pre-­‐electoral   and   election-­‐time   violence   together,   the   violence   variable   ranges   from   0   (no   problems)   to   6   (major   problems).   The   cheating   variable,   which   includes   two   electoral   cheating   variables   and   the   variable   of   legal   problems,   ranges   from  0  (no  problems)  to  9  (major  problems).           66   equal,  strong  oppositions  fuel  violence,  whereas  non-­‐party-­‐based  regimes  and  weak   politically   dominant   ethnic   groups   are   more   likely   to   commit   electoral   cheating   especially  when  they  lack  in  rich  financial  resources.                             2.6  Conclusion              This  chapter  has  presented  an  informational  theory  of  authoritarian  elections  and   cross-­‐nationally   explored   the   determinants   of   electoral   fraud   in   dictatorships.   Dictators  face  the  electoral  dilemma  between  the  certainty  and  credibility  of  election   results.  Focusing  on  the  balance  of  mobilization  power  between  the  dictator  and  the   elites,  I  have  argued  that  strong  dictators,  who  can  mobilize  a  large  amount  of  votes   via   public   spending,   are   more   willing   to   lower   the   degree   of   electoral   fraud.   The   findings  in  this  chapter  imply  that  the  mere  existence  of  free  and  fair  elections  does   not  necessarily  lead  to  further  democratization  in  authoritarian  countries.                International   development   assistance   organizations   send   election   observers   and/or   peacekeeping   agents   to   prevent   dictators   from   committing   serious   election   cheating   and   violence.   If   dictators   are   financially   weak,   however,   fair   elections   brought   by   international   election   monitoring   might   lead   to   post-­‐election   conflict   by   revealing  the  weakness  of  dictators,  which  in  turn  might  threaten  political  order  after   elections.   Conversely,   when   dictators   are   strong,   election   monitoring   and   subsequent   higher  quality  of  elections  might  legitimate  those  dictators  that  get  an  overwhelming     67   victory   by   mobilizing   their   rich   political   resources.   This   is   more   likely   to   prevent   conflict,   but   also   encourage   authoritarian   leaders   to   hold   on   to   power   through   “legitimized”   elections.   Based   on   this   understanding   of   the   trade-­‐off   between   democracy   and   political   order,   international   development   assistance   organizations   may   need   to   develop   strategies   to   transform   dictatorial   countries   into   democratic,   peaceful  ones.               68   CHAPTER  3     THE  CHOICE  OF  ELECTORAL  SYSTEMS  IN  DICTATORSHIPS     3.1  Introduction             Over  the  last  decade,  burgeoning  literature  on  authoritarian  politics  has  documented   how   authoritarian   elections   help   autocrats   hold   onto   power   (Gandhi   and   Lust-­‐Okar   2009).   Authoritarian   elections   enable   dictators   to   co-­‐opt   ruling   elites   (Boix   and   Svolik   2007;   Blaydes   2010),   party   members   (Magaloni   2006),   or   opposition   groups   within   society   (Gandhi   and   Przeworski   2007;   Gandhi   2008).   Specifically,   by   scoring   an   overwhelming  majority  at  elections,  authoritarian  leaders  can  show  the  invincibility  of   their  regimes  (Geddes  2006;  Magaloni  2006;  Simpser  2013).  In  addition,  election  results   may   also   inform   dictators   of   their   main   supporting   bases   and   opposition   strongholds   (Ames   1970;   Shi   1999;   Malesky   and   Schuler   2010;   Reuter   and   Robertson   2011).   Since   moderate  opposition  parties  participate  in  regime-­‐sponsored  elections,  whereas  radical   opposition   parties   boycott   the   elections,   elections   also   enable   a   “divide   and   conquer”   approach  of  the  opposition  camp  (Lust-­‐Okar  2004;  Beaulieu  2006).     Far  less  explored  in  the  current  scholarship,  however,  are  the  variations  of  electoral   systems   in   which   elections   in   authoritarian   regimes   take   place.   In   particular,   the   literature   on   electoral   system   choice   has   almost   exclusively   focused   on   democracies;   meanwhile  we  know  little  about  the  conditions  under  which  authoritarian  rulers  prefer   one  type  of  electoral  system  over  another.  This  analytical  imbalance  is  indeed  puzzling     69   given   our   abundant   knowledge   on   the   consequences   of   electoral   systems   in   the   literature   and   the   fact   that   authoritarian   leaders   do   often   embark   on   reforming   their   electoral   rules.   For   instance,   Putin’s   Russia   shifted   electoral   systems   from   a   mixed   system  to  a  pure  Proportional  Representation  system  (PR)  with  a  nationwide  district  in   2005.   Nazarvaev,   the   President   of   Kazakhstan,   also   took   the   same   direction   by   declaring   that  the  country  would  adopt  a  pure  PR  system  with  a  nationwide  electoral  district  prior   to   the   2007   parliamentary   elections.   By   contrast,   Belarus,   albeit   with   regional   similarities   to   Russia   and   Kazakhstan,   has   held   parliamentary   elections   with   Single   Member   Districts   (SMD)   systems   since   its   national   independence.   Furthermore,   electoral   authoritarian   countries   in   the   Middle   East   like   Egypt,   Tunisia,   and   Algeria   frequently   changed   their   electoral   institutions   and   district   size   during   the   1980s   and   1990s   (Lust-­‐Okar   and   Jamal   2002;   Pripstein   Posusney   2002).   As   the   cross-­‐national   data   I   examine   reveals,   there   is   wide   variation   in   the   Effective   Electoral   Threshold   among   electoral   authoritarian   regimes   across   time   and   space.   It   makes   us   wonder:   When   and   why  do  electoral  autocrats  decide  to  adopt  one  type  of  electoral  systems  over  others?                    This   chapter   is   further   motivated   by   another   intriguing   observation:   Just   like   in   advanced   democracies,   ruling   parties   in   electoral   authoritarianism   regimes   with   SMD   systems   typically   enjoy   significant   seat   premiums   due   to   the   seats-­‐votes   disproportionality.   In   addition,   as   authoritarian   leaders   delegate   local   elites   to   garner   support   in   their   localities   under   SMD   systems,   they   do   not   need   to   use   their   own   resources   to   mobilize   regime   supporters.   Together,   SMD   systems   seem   to   generate   a   pro-­‐regime   bias   in   electoral   authoritarianism   governments   and   should   be   the   natural     70   choice  for  self-­‐serving  dictators.  Yet,  PR  systems―a  seemingly  sub-­‐optimal  institutional   choice―are  still  used  in  some  electoral  authoritarian  regimes.  Why?   This  chapter  develops  a  theoretical  framework  to  solve  these  related  puzzles.  I  argue   that  the  choice  of  electoral  systems  poses  a  dilemma  to  authoritarian  rulers.  Specifically,   while   SMD   systems   provide   a   seat   premium   and   reduce   mobilization   costs   for   dictators,   SMD   systems   also   undermine   several   important   functionalities   of   elections   in   authoritarian  regimes.  For  instance,  under  SMD  systems,  the  parliamentary  dominance   built   on   the   seat   premium   makes   election   results   less   useful   and   less   credible   in   demonstrating  regime  strength.  Further,  SMD  systems  also  make  it  easier  for  opposition   parties   to   build   a   pre-­‐electoral   coalition   and   show   their   coherence   to   voters.   Importantly,  I  argue  that  different  dictators  develop  different  strategies  to  address  this   electoral   system   dilemma.   I   argue   that   “strong”   dictators,   that   is,   those   who   hold   rich   financial   resources   to   buy-­‐off   extensive   political   support   from   the   citizenry,   have   an   incentive   to   use   PR   systems.   Authoritarian   leaders   with   high   mobilization   power   need   not  depend  on  the  pro-­‐regime  bias  generated  by  SMD  systems.  In  addition,  PR  systems   make  opposition  parties  remain  divided,  which  persuades  people  that  the  opposition  is   not   a   viable   option   at   all   during   electoral   processes.   In   other   words,   authoritarian   leaders   with   rich   financial   resources   should   prefer   to   choose   a   more   proportional   representation  system  to  convey  a  clear  signal  of  regime  invincibility.       Using   newly   collected   cross-­‐national   datasets   on   electoral   systems   and   election   results  in  electoral  authoritarian  regimes  (95  countries,  1946-­‐2007),  I  find  that  there  is  a   robust,   negative   correlation   between   effective   electoral   threshold   and   dictators’   financial   strength   measured   by   their   natural   resource   wealth.   The   result   is   sustained     71   even   after   using   different   model   specifications,   alternative   estimating   strategies,   and   various   measurements   of   both   dependent   and   independent   variables.   Further,   I   explicitly  test  the  causal  mechanisms  and  show  the  pro-­‐regime  bias  and  the  coherency   of  opposition  parties  under  SMD  systems.     The   contribution   of   this   chapter   to   the   literature   is   threefold.   First,   I   try   to   explore   the   origins   of   political   institutions   in   dictatorship.   Taking   particular   emphasis   on   the   importance  of  the  dictator’s  power  of  popular  mobilization,  I  suggest  that  dictators  are   more   likely   to   adopt   less   biased   electoral   rules   to   credibly   show   their   regime   invincibility,  but  only  when  they  can  afford  to  do  so.  Second,  this  chapter  systematically   explores   one   technique   of   electoral   manipulation   in   dictatorships   that   has   avoided   close   scrutiny  in  the  extant  literature,  namely,  the  choice  of  electoral  systems.  In  addition  to   electoral   fraud   like   election   violence,   tampering   with   the   ballot   box,   media   bias,   and   severe   limits   on   electoral   laws,   as   well   as   pre-­‐electoral   economic   distribution   through   vote-­‐buying  and  fiscal  policy  manipulation,  the  chapter  suggests  that  dictators  can  bias   election  results  in  their  favor  by  manipulating  the  electoral  formula.  Lastly,  the  chapter   attempts  to  contribute  to  the  electoral  system  choice  literature.  Beyond  the  importance   of   opposition   threats   (Boix   1999),   partisan   bias   (Calvo   2009),   and   economic   interests   (Rogowski  1987;  Cusack,  Iversen,  and  Soskice  2007),  I  posit  a  new  theory  emphasizing  a   signaling  role  of  electoral  systems.     This  chapter  proceeds  as  follows.  In  the  next  section,   I  review  literature  on  electoral   system   choice,   suggesting   theoretical   gaps   that   should   be   filled   when   considering   dictators’   calculus   over   the   choice   of   electoral   systems.   The   third   section   provides   a   theory   of   electoral   system   design   in   electoral   autocracies   and   derives   observable     72   implications.   In   the   forth   section,   I   show   cross-­‐national   evidence   for   my   theoretical   expectations.  I  show  that  natural  resources  tend  to  lower  effective  electoral  threshold  in   electoral   authoritarianism.   Then,   to   confirm   causal   mechanisms   in   theory,   I   provide   additional  cross-­‐national  evidence:  (1)  natural  resource  wealth  tends  to  boost  up  regime   support,   suggesting   that   resource   abundance   is   a   good   proxy   of   the   dictator’s   mobilization  power;  (2)  SMD  systems  tend  to  provide  a  seat  premium  to  ruling  parties,   showing   the   existence   of   the   pro-­‐regime   bias   under   SMD   systems;   and   (3)   SMD   systems   tend   to   promote   a   pre-­‐electoral   opposition   coalition,   implying   that   SMD   systems   help   opposition  parties  send  a  good  signal  of  its  coherence.  Lastly,  conclusions  follow.           3.2  Literature  Review     There  are  broadly  two  perspectives  that  scholars  have  taken  to  study  the  choice  of   electoral   systems:   political   and   economic   explanations   (Leemann   and   Mares   2014).   Rokkan   (1970)   is   the   pioneering   work   that   proposes   two   “political”   hypotheses   explaining   the   adoption   of   proportional   representation   in   Europe   during   the   early   twentieth  century.35  Rokkan’s  first  hypothesis  argues  that  incumbents  adopt  PR  systems   to  avoid  a  devastative  electoral  defeat  in  the  face  of  socialist  mobilization.  Boix  (1999)   advanced   this   hypothesis   and   argues   that   ruling   parties   adopt   PR   systems   when   the                                                                                                                   35  For  criticism  of  Boix  (1999),  see  Cusack,  Iversen  and  Soskice  (2007)  and  Calvo  (2009).   Using  qualitative  sources  on  electoral  system  reforms  in  advanced  democracies,  Kreuzer   (2010)   argues   that   Boix’s   (1999)   explanation   is   more   valid   than   Cusack,   Iversen   and   Soskice’s  (2007).  Regarding  Cusack  and  his  colleagues’  response  to  Kreuzer  (2010),  see   Cusack,  Iversen  and  Soskice  (2010).         73   rightwing   parties   are   seriously   divided   between   the   conservative   and   liberal   camps   under   the   socialist   threat.   Put   differently,   parties   on   the   right   tend   to   maintain   majoritarian   electoral   systems   when   either   leftist   parties   are   considered   as   weak   or   one   of   the   old   parties   occupies   a   dominant   position   in   a   party   system.   Rokkan’s   second   hypothesis,   which   has   been   taken   up   recently   by   Calvo   (2009),   suggests   that   the   adoption   of   proportional   representation   is   also   driven   by   the   extent   to   which   the   established   parties   want   to   avoid   “partisan   bias”   induced   by   majoritarian   electoral   systems.  Parties  with  a  geographically  concentrated  distribution  of  votes  tend  to  enjoy   more   seats   than   those   having   geographically   dispersed   votes   under   SMD   exposed   to   severe  party  competition.  Therefore,  the  old  parties  with  dispersed  votes  prefer  to  shift   to  proportional  representation  to  moderate  the  partisan  bias.             Although  these  two  political  explanations  are  powerful  in  understanding  the  choice   of  electoral  systems  in  pre-­‐war  Europe,  it  is  difficult  to  directly  apply  these  insights  to   authoritarian   contexts.   Rokkan’s   first   hypothesis   views   strong   socialist   threats   as   a   driving   force   to   adopt   proportional   representation   systems,   yet   most   contemporary   authoritarian   countries   are   not   exposed   to   such   imminent   opposition   threats.   Indeed,   Boix   recognized   that   his   theoretical   expectation   is   only   applicable   to   democracies   satisfying   certain   presumptions   (Boix   1999,   622).   On   the   other   hand,   although   Rokkan’s   second   hypothesis   answers   why   incumbents   may   adopt   PR   systems   even   in   the   absence   of  strong  socialist  mobilization  (Calvo  2009,  268),  it  still  assumes  that  there  should  be   severe   party   competition   that   generates   the   partisan   bias   leading   to   government   turnover.   A   crucial   difference   between   democracies   and   autocracies   is   that   opposition   parties   in   authoritarian   states   are   too   weak   to   be   a   viable   alternative   at   elections.   In     74   electoral   authoritarian   states,   incumbent   parties   generally   retain   far   greater   resource   advantages   to   prevent   leadership   turnover   at   elections   (Greene   2007;   Levitsky   and   Way   2010).   Therefore,   opposition   parties,   although   allowed   to   participate   in   electoral   competition   in   these   countries,   suffer   severe   limits   on   their   organizational   capacities   and  monetary  resources  making  a  win  for  the  opposition  nearly  unlikely  (Lust-­‐Okar  and   Jamal   2002,   342-­‐345;   Schedler   2013).   However,   despite   the   fact   that   ruling   parties   do   not   need   to   worry   about   opposition   parties,   why   do   some   authoritarian   leaders   still   maintain  PR  systems  and  even  decide  to  change  electoral  systems  from  SMD  to  PR?   Apart  from  the  political  explanations,  other  scholars  have  tried  to  explain  the  origins   of  electoral  systems  by  focusing  on  the  distribution  of  economic  interests  in  a  country.   Focusing   on   domestic   politics,   Cusack,   Iversen,   and   Soskice   (2007)   argue   that   whether   rightist  parties  adopt  PR  systems  depends  on  the  extent  to  which  businesses  and  unions   forge   cooperative   relationships   via   skill   formation   and   well-­‐established   collective   bargaining   at   the   national   economy.   If   parties   on   the   right   are   embedded   in   the   cross-­‐ class,   consensus-­‐based   decision-­‐making   process,   then   they   can   enjoy   benefits   from   such   regulatory   politics   rather   than   incur   costs   induced   by   distributional   conflict   under   PR   systems.   Therefore,   they   argue   that   cross-­‐class   alliances   encourage   governments   to   choose   proportional   representation   systems.   Rogowski   (1987),   on   the   other   hand,   focuses  on  an  external  factor.  He  argues  that  trade-­‐dependent  countries  are  more  likely   to  adopt  PR  systems.  An  open  economy  encourages  governments  to  resist  protectionist   pressures,  maintain  high  efficiency,  and  keep  high  policy  stability  because  they  have  to   remain   competitive   in   the   international   market.   Under   PR   systems   that   tend   to   have   larger   electoral   districts   than   SMD   systems,   incumbents   can   contain   regional   and     75   sectorial  pressures  more  effectively,  making  governments  in  trade-­‐dependent  countries   preferring  PR  systems.                Unfortunately,   the   economic-­‐interests   explanations   do   not   provide   plausible   answers   about   why   electoral   authoritarian   countries   change   their   electoral   systems,   either.  First,  similar  with  the  political  explanations,  they  also  presuppose  that  countries   under  electoral  system  reforms  are  fully  democratic.  Cusack,  Iversen  and  Soskice  (2007)   and   Rogowski   (1987)   both   assume   distributional   conflict   among   different   economic   groups   becomes   severe   in   the   face   of   strong   party   competition.   Second,   based   on   the   “varieties  of  capitalism”  argument  (Hall  and  Soskice  2001),  Cusack,  Iversen  and  Soskice   (2007)   center   on   two   types   of   capitalisms―liberal   market   and   coordinated   market   economies―to  explain  the  choice  of  electoral  systems  in  pre-­‐war  Europe,  yet  neither  of   these   two   capitalisms   are   systematically   present   in   authoritarian   regimes.   Instead   of   capitalism’s   variants,   this   chapter   investigates   whether   and   to   what   extent   a   government’s   financial   resources   determine   electoral   system   design   in   electoral   authoritarian  regimes.             Beyond   the   context   of   advanced   democracies,   researchers   have   also   explored   electoral   system   reforms   in   new   democracies.   Many   point   to   the   important   role   of   political   uncertainty.   Specifically,   scholars   have   shown   that   electoral   system   change   in   transitioning  states  does  not  lead  reformers  to  gain  benefits  that  they  initially  expected   because   the   effects   of   electoral   reforms   are   highly   uncertain   in   those   countries   where     76   democratic   institutions   are   still   immature   (Przeworski   1991;   Remington   and   Smith   1996;  Moser  2001;  Andrews  and  Jackman  2005).36       In   authoritarian   regimes,   however,   electoral   institutions   have   high   certainty   about   how  they  work,  at  least  compared  to  transitioning  countries.  In  stark  contrast  with  the   findings   in   new   democracies,   a   large   body   of   literature   on   authoritarian   elections   compellingly   demonstrates   that   electoral   rules   help   dictators   consolidate   their   power   (Magaloni   2006;   Gandhi   and   Lust-­‐Okar   2009;   Blaydes   2010).   Discussing   electoral   systems   in   the   Middle   East,   Lust-­‐Okar   and   Jamal   (2002)   note   “both   sides   [incumbents   and   oppositions]   know   their   preferences   over   the   electoral   rules.   […]   Proportional   representation   and   larger   member   districts   tend   to   increase   the   number   of   effective   parties   and   the   possibility   of   minority   representation,   whereas   majoritarian   systems   and  single-­‐member  districts  tend  to  limit  the  participation  of  smaller  parties.  […]  Elites   hold   firm   preferences   over   electoral   laws   when   they   negotiate   with   each   other”   (345-­‐ 346).   My   empirical   analysis   confirms   the   proposition   that   electoral   system   design   is   a   crucial  factor  in  determining  the  allocation  of  seats  also  in  electoral  authoritarianism.                                                                                                                     36  On   the   other   hand,   Bielasiak   (2006)   illustrates   that   in   the   post-­‐communist   world,   democracy   and   semi-­‐authoritarian   states   tend   to   adopt   PR   systems   or   mixed   systems   while  more  authoritarian  states  are  prone  to  select  SMD  systems.  After  investigating  the   cases   of   Central   and   Eastern   Europe,   Ishiyama   (1997)   concludes   that   substantial   changes  in  electoral  systems  would  have  occurred  if  communist  parties  and  oppositional   forces  had  thought  of  their  organizations  as  seat-­‐maximizing  political  parties  rather  than   as   mass   movements   when   they   entered   the   phase   of   selecting   electoral   systems.   For   systematic   comparisons   regarding   the   impacts   of   electoral   systems   on   party   systems   between  advanced  and  new  democracies,  see  Moser  and  Scheiner  (2012).         77   Lastly,   it   has   been   argued   that   electoral   systems   are   highly   path-­‐dependent.   Especially  in  democracies,  a  dominant  view  is  that,  once  selected,  electoral  systems  are   surprisingly  stable  because  the  choice  of  electoral  system  is  strongly  influenced  by  the   stakes   of   the   preexisting   parties   (Cox   1998,   18). 37  Certainly,   electoral   systems   in   electoral   authoritarianism   also   have   such   a   path-­‐dependent   characteristic.   However,   there   have   been   significant   reforms   that   changed   electoral   systems   in   many   electoral   authoritarian  regimes  since  the  end  of  World  War  II.  This  is  contrary  to  the  experience   of   advanced   democracies   in   which   electoral   systems   tended   to   be   highly   stable   until   more  recently.  It  suggests  that  in  autocracies,  political  leaders  may  have  more  discretion   in   designing   pliable   electoral   systems.   In   subsequent   quantitative   analyses,   I   focus   on   this   strategic   nature   of   electoral   system   choice   in   autocracies   after   taking   into   account   the  path-­‐dependency  of  electoral  systems  in  a  methodologically  proper  way.                                                                                                                               37  For   example,   Tsebelis   (1990)   argues   that   extant   electoral   systems   may   shape   the   interests   of   legislators   within   each   party,   which   makes   it   difficult   to   change   electoral   systems,  even  if  an  alternative  electoral  system  is  rational  for  parties  as  a  whole.  Based   on   this   insight,   recent   literature   argues   that,   even   under   such   inertia,   ruling   parties   may   change   more   specific   and   micro-­‐level   electoral   rules   in   their   favor.   McElwain   (2008)   shows  that  incumbent  politicians  tended  to  enjoy  advantages  in  elections  by  restricting   lengths   of   the   electoral   campaign   period   in   Japan.   Diaz-­‐Cayeros   and   Magaloni   (2001)   assert  that  in  Mexico,  under  the  PRI  rule,  the  ruling  party  manipulated  electoral  systems   in   the   way   that   it   was   difficult   for   opposition   parties   to   coordinate   their   behavior   in   order  to  form  a  majority.         78   3.3  Electoral  System  Change  in  Electoral  Authoritarianism     Politicians   attempt   to   hold   onto   power   (Bueno   de   Mesquita   et   al.   2003).   This   is   particularity   true   for   authoritarian   leaders   who   may   lose   their   lives   after   losing   office   (Chiozza   and   Goemans   2011).   To   stay   in   power,   dictators   employ   a   variety   of   means,   such   as   violent   repression,   patronage   distribution,   and   the   construction   of   political   institutions   like   parties   and   legislatures.   When   authoritarian   rulers   use   elections   to   consolidate   their   rule,   previous   literature   suggests   that   they   try   to   score   an   overwhelming   majority,   not   a   slim   majority,   to   demonstrate   their   strength   (Geddes   2006;  Magaloni  2006;  Simpser  2013).  Winning  big,  dictators  convey  a  signal  to  potential   opponents  in  society  that  the  regimes  are  unshakable.     To   maintain   supermajorities,   there   are   at   least   three   techniques   that   dictators   can   use  at  the  ballot  box.  A  first,  well-­‐known  method  is  electoral  fraud  (Simpser  2013).  As   already   defined   in   the   previous   chapters,   electoral   fraud   is   a   series   of   illegal   measures   that   bias   election   results   in   favor   of   the   political   leader   (Lehoucq   2003).   Electoral   violence   (Strauss   and   Taylor   2012;   Hafner-­‐Burton,   Hyde,   and   Jablonski   2013),   tampering   with   the   ballot   box,   media   bias,   packing   election   management   bodies,   and   highly   restrictive   electoral   laws   (McElwin   2008;   Kelley   2012)   are   all   such   examples   of   blatant  electoral  manipulation  that  facilitate  obtaining  an  overwhelming  majority.     Second,  autocrats  may  engage  in  implementing  expansionary  economic  policies  prior   to  elections,  namely,  engineering  political  business  cycles  (e.g.  Shi  and  Svensson  2003).   Pre-­‐electoral  economic  distribution,  although  not  necessarily  breaking  any  laws,  allows     79   governments   to   derive   political   support   from   voters,   which   is   a   frequently   observed   phenomenon   in   electoral   authoritarian   regimes   (Magaloni   2006;   Pepinsky   2009;   Blaydes  2010;  Higashijima  2010;  Wright  2011).     The  last  and  least  researched  technique  in  the  literature  on  authoritarian  elections  is   the   design   of   electoral   systems.   As   previous   studies   demonstrate,   electoral   systems   significantly   influence   the   manners   in   which   votes   for   a   party   are   translated   into   parliamentary   seats   (Rae   1971;   Taagepera   and   Shugart   1989;   Lijphart   1994;   Cox   1998).   Similar   to   democratic   countries,   changing   voting   procedures   and   district   size,   authoritarian  leaders  are  able  to  increase  their  parliamentary  seat  shares,  which  would   not  otherwise  be  achievable  if  they  had  adopted  a  different  electoral  system  (Jamal  and   Lust-­‐Okar  2002;  Pripstein  Posusney  2002).     My   proposition   is   that   dictators   strategically   choose   different   electoral   systems   to   maximize  their  political  interests,  following  directly  from  various  studies  on  the  political   consequences   of   electoral   systems   in   democratic   settings.   Conventional   wisdom   holds   that   majoritarian   systems   usually   lead   to   high   disproportionality,   that   is,   “   [larger]   deviation[s]   of   parties’   seat   shares   from   their   vote   shares.”   (Lijphart   1994,   57)   Specifically,   due   to   their   “winner-­‐take-­‐all”   nature,   SMD   tends   to   mechanically   provide   more   seats   to   major   parties.   This   is   the   reason   why   a   drastic   swing   in   election   results   often   occurs   in   majoritarian   democracies.   Also   due   to   this   reason,   the   divided   political   right  feared  devastating  electoral  losses  to  emerging  social  democrats  in  pre-­‐war  Europe   (Rokkan  1970;  Boix  1999).         In   the   context   of   electoral   authoritarianism,   ruling   parties   are   the   larger   parties   which  in  turn  benefit  from  high  disproportionality  in  SMD  systems.  In  other  words,  SMD     80   systems   bias   election   results   in   favor   of   ruling   parties   in   authoritarian   regimes,   a   bias   that   I   term   here   “the   pro-­‐regime   bias.”   This   bias   is   first   generated   by   Duverger’s   well-­‐ known  mechanical  and  psychological  effects  of  majoritarian  systems  on  the  seats-­‐votes   elasticity   (Duverger   1954).   That   is,   expecting   that   a   large   portion   of   seats   will   not   be   guaranteed   to   opposition   parties,   voters   may   cast   their   ballot   for   the   ruling   party   or   abstain  from  voting  all  together.  The  pro-­‐regime  bias  is  well  illustrated  by  the  cases  of   Singapore   and   Malaysia,   two   well-­‐known   Asian   autocracies   usually   categorized   as   electoral   authoritarian   (Schedler   2013,   4-­‐5).   Both   countries   have   held   parliamentary   elections   since   independence   and   also   use   majoritarian   electoral   systems   with   single-­‐ member   districts.   On   average,   the   countries’   ruling   parties   (People’s   Action   Party   in   Singapore  and  Barisan  Nasional  in  Malaysia)  obtained  87  percent  of  the  total  seats  with   only  63  percent   of  the  total  votes  between  1959-­‐2008,  suggesting  that  they  have  been   receiving   very   large   seat   premiums:   24   percent   of   the   total   seats.38  Conversely,   PR   systems   prevent   major   parties   from   enjoying   such   seat   premiums   because   the   percentages   of   seats   tend   to   proportionally   coincide   with   those   of   votes.   Accordingly,   major   parties   need   to   collect   more   votes   to   win   an   overwhelming   legislative   majority   under  PR  systems.   A  second  mechanism  yielding  a  pro-­‐regime  bias  is  gerrymandering.  In  SMD  systems   where   governments   are   able   to   design   every   single   electoral   district,   dictators   can   engage  in  gerrymandering  to  favor  the  ruling  parties.  Ahmed  (2013)  finds  that,  in  mid-­‐ nineteenth   century   Europe   when   electoral   competition   was   minimal   and   socialist                                                                                                                   38  For   other   anecdotal   evidence   on   the   pro-­‐dictator   bias   from   some   Middle   Eastern   countries  (Yemen,  Palestine,  Tunisia,  and  Egypt),  see  Pripstein  Posusney  (2002).             81   threats  were  still  very  weak,  ruling  parties  tended  to  resort  to  redistricting  in  order  to   enjoy   the   seat   bias   to   maintain   electoral   dominance.   Conversely,   in   PR   systems,   both   electoral   magnitude   and   electoral   districts   are   generally   larger   and   districts   are   often   fixed  with  administrative  units.  Therefore,  in  the  PR  context,  it  is  more  difficult  for  ruling   parties  to  arbitrarily  redistrict,  at  least  compared  to  SMD  systems.     If   majoritarian   systems   help   dictators   win   big,   then   one   may   think   there   is   no   incentive  for  them  to  adopt  PR  systems.  However,  empirical  evidence  shows  that  many   electoral   autocracies   have   maintained   proportional   representation   systems   or   even   changed   their   systems   from   SMD   to   PR   (e.g.,     Russia   and   Kazakhstan).   Why   are   some   autocrats   willing   to   adopt   electoral   systems   that   are   not   necessarily   helpful   in   maintaining  an  overwhelming  majority  in  parliament?              I   argue   that   authoritarian   leaders   prefer   to   adopt   a   proportional   representation   system  with  an  eye  toward  demonstrating  regime  credibility  and  strength.  The  current   literature  on  authoritarian  politics  assumes  that  authoritarian  leaders  and  other  political   actors   are   more   likely   to   face   information   shortages   (Wintrobe   1998;   Egorov,   Gariev,   and   Sonin   2009).   In   dictatorships   where   political   transparency   is   seriously   circumscribed,   it   is   difficult   for   political   elites   to   know   to   what   extent   the   dictator   has   the  capability  of  governing  the  country  effectively.  Such  a  lack  of  information  may  result   in,   at   best,   miscommunication   between   the   dictator   and   potential   opponents,   and   at   worse,   political   conflicts   such   as   post-­‐electoral   coups   and   popular   protests   (Beaulieu   2014).   Therefore,   dictators   need   to   show   their   strength   in   the   most   credible   way   possible,  as  I  discussed  in  Chapter  2.  According  to  the  literature,  elections  may  work  as     82   such   a   mechanism   to   fill   the   information   gap   in   autocracies   (Geddes   2006;   Magaloni   2006;  Cox  2009;    Blaydes  2010;  Malesky  and  Schuler  2010;  Miller  2012)     Figure  3.1:  Hypothesis  and  Causal  Mechanisms  for  the  Origins  of  Electoral  Systems   in  Electoral  Authoritarianism         Causal  Mechanisms           Dictator’s  Strong   Mobilization  Power   (Rich  Financial   Resources)     PR  Systems     (More  Credible   Election  Results)          (I)  Less  “Pro-­‐Dictator  Bias”            (II)  Less  reliance  on  local  ruling  elites          (III)  Coordination  Problems  among                          Opposition       If  authoritarian  leaders  adopt  SMD  systems,  election  results  become  less  credible  to   inform   de  facto   mobilization   power   of   ruling   parties.   Under   SMD   systems,   other   political   actors   are   more   likely   to   think   that   the   election   results   and   subsequent   parliamentary   dominance   are   mainly   driven   by   the   pro-­‐regime   bias.   Even   if   the   regime   scores   a   landslide   victory,   the   signaling   function   of   elections   may   be   more   likely   to   be   undermined.     Further,   SMD   systems   allow   the   dictator   to   delegate   local   ruling   elites   to   garner   political  support  in  their  constituencies.  Since  electoral  districts  are  small  in  size  under   SMD   systems,   the   dictator   can   rely   on   powerful   local   elites   to   use   their   own   political   resources   to   mobilize   constituents.   On   the   flip   side,   such   political   delegation   to   local   ruling   elites   may   negatively   affect   the   credibility   of   regime   strength   because   subsequent   electoral  victory  is  likely  to  be  taken  as  a  reflection  of  ruling  elites’  mobilization  power,   rather  than  the  dictator’s.                 83   In   contrast,   PR   systems   are   more   likely   to   provide   the   dictators   with   institutional   circumstances   where   parliamentary   dominance   sends   a   credible   signal   of   regime   strength,   in   three   ways.   First,   since   the   pro-­‐dictator   bias   is   low   in   PR   systems,   authoritarian   leaders   need   to   garner   a   larger   number   of   votes   to   dominate   parliament   with  an  overwhelming  majority.  Second,  since  electoral  districts  are  generally  larger  in   size  and  the  electoral  battle  is  more  party-­‐based  under  the  PR  apparatus,  authoritarian   leaders   themselves   need   to   organize   a   wide   range   of   political   support   across   regions   without   relying   on   local   ruling   elites’   individual   resources   in   their   strongholds.   By   doing   so,  election  results  become  more  credible  to  inform  the  dictator’s  strength.   Third,   PR   systems   make   it   difficult   for   opposition   parties   to   unite   and   build   a   significant   pre-­‐electoral   coalition   (Strom,   Budge   and   Laver   1994,   315-­‐316;   Golder   2006).  Pre-­‐electoral  coalitions  are  “agreements  in  which  parties  publicly  announce  that   they   will   not   compete   independently   in   an   election.   Instead,   they   coordinate   their   campaigns  to  sponsor  joint  candidates  or  lists  or  to  not  run  against  each  other  in  some   constituencies”  (Gandhi  and  Reuter  2013,  138).  Since  opposition  parties  are  allowed  to   have   seats   with   smaller   vote   shares   under   PR   systems,   they   have   more   incentives   to   participate   in   elections.   Yet,   at   the   same   time,   they   are   less   likely   to   coordinate   their   electoral   campaigns   to   challenge   the   ruling   party.   Barbera   (2013)   demonstrates   that   PR   systems   tend   to   increase   the   number   of   opposition   parties   in   authoritarian   countries   with   multi-­‐party   elections.   This   coordination   problem   within   the   opposition   camp   persuades   voters   that   opposition   is   weak   and   regime   parties   are   strong   in   a   credible   manner.         84   PR  systems  increase  the  credibility  of  regime  invincibility  orchestrated  by  elections   so  long  as  ruling  parties  maintain  supermajorities.  Yet,  choosing  a  PR  system  poses  risk   to  authoritarian  leaders.  If  they  fail  to  obtain  high  political  support  at  the  ballot  box,  then   they  may  also  fail  to  obtain  adequate  seats  because  PR  systems  do  not  provide  the  seat   premiums  brought  by  the  pro-­‐regime  bias.  Rather,  PR  systems  allow  for  the  possibility   that   opposition   parties,   although   usually   divided,   gain   more   seats.   Although   the   opposition  rarely  claims  victory,  they  may  prevent  the  ruling  party  from  winning  big  in   the  legislature.  The  failure  to  maintain  overwhelming  support  may  in  turn  reveal  regime   weaknesses  rather  than  strengths,  complicating  regime  stability.     When   dictators   possess   strong   mobilization   power   to   derive   voluntary   support   from   citizens,   they   can   produce   an   overwhelming   majority   even   under   PR   systems.   And,   as   discussed   already,   economic   distribution   is   one   of   the   most   frequent   ways   to   collect   popular   support   in   electoral   authoritarianism   without   losing   the   credibility   of   election   results.   “Strong”   dictators,   who   have   abundant   monetary   resources   to   cultivate   voluntary   support   from   citizens,   should   be   more   willing   to   choose   a   PR   system   over   a   SMD   one   to   maximize   the   informational   benefits   of   elections.   PR   systems   help   such   strong   dictators   inform   regime   strength   by   conveying   a   costly   signal.   This   is   what   happened,   for   example,   in   Russia   and   Kazakhstan   during   the   mid-­‐2000s,   when   both   countries  transitioned  to  a  pure  PR  system  with  a  nationwide  electoral  district.  At  that   time,   two   autocrats   in   the   country,   Putin   and   Nazarbaev   enjoyed   very   high   popularity,     85   thanks   to   rich   natural   resource   wealth.39  While   increasing   mobilization   power,   both   decided   to   shift   to   PR   systems   and   the   ruling   parties,   United   Russia   and   Nur   Otan,   successfully   maintained   supermajorities   in   their   respective   legislatures.   Therefore,   I   propose   the   testable   hypothesis   on   electoral   system   choice   (See   also   Figure   3.1   for   summarized  causal  mechanisms).         Hypothesis:   The   greater   the   financial   strength   of   the   dictator,   the   more   likely   he   is   to   choose  PR  over  SMD.       3.4  Cross-­‐National  Statistical  Analysis     3.4.1  Sample:  Electoral  Authoritarianism   I  focus  on  electoral  authoritarian  regimes  for  the  period  of  1946-­‐2007.  According  to   Schedler   (2013,   2),   electoral   authoritarian   regimes   are   defined   as   political   regimes   where   authoritarian   leaders   “play   the   game   of   multiparty   elections,   […]   (and)   these   elections   are   broadly   inclusive   (they   are   held   under   universal   suffrage),   minimally   pluralistic   (opposition   parties   are   permitted   to   run),   minimally   competitive   (parties   and   candidates   outside   the   ruling   coalition,   while   denied   victory,   are   allowed   to   win   votes   and  seats),  and  minimally  open  (dissident  is  not  subject  to  massive,  but  often  to  selective   and  intermittent  repression),  […]  (yet)  not  minimally  democratic.”                                                                                                                       39  On   Putin’s   popularity   in   Russia,   see   Rose,   Mishler   and   Munro   (2011)   and   the   Economist  (2014).  On  the  same  issue  in  the  case  of  Kazakhstan,  see  Schatz  (2009)  and   Chapter  4.       86   I  use  the  following  two  data  sources  to  identify  electoral  authoritarian  regimes.  If  a   country   satisfies   necessary   conditions   in   either   one   of   the   two   datasets,   I   regard   the   country  as  an  electoral  authoritarian  regime.  The  first  data  source  is  National  Elections   in   Democracy   and   Autocracy   (NELDA).   Hyde   and   Marinov   (2012)   see   elections   as   minimally   competitive   if   there   is   ex   ante   uncertainty   over   election   results.   More   specifically,   they   regard   elections   as   minimally   competitive   if   (1)   multiple   parties   are   legal,   (2)   more   than   two   candidates   are   allowed   to   stand   in   electoral   districts,   and   (3)   opposition   is   allowed   to   participate   in   the   election.   Kinne   and   Marinov   (2013)   use   the   NELDA’s   criteria   to   identify   electoral   authoritarianism   from   the   sample   of   non-­‐ democracies   defined   by   Cheibub,   Gandhi   and   Vreeland’s   (2009)   dichotomous   measure   of   political   regimes.   Kinne   and   Marinov’s   (2013)   operationalization   and   NELDA   are   useful  because  they  provide  us  with  an  extensive  time  period  as  well  as  a  large  number   of  countries.  Yet,  unfortunately,  it  does  not  include  countries  where  political  parties  are   de  jure   illegal   but   there   are   relevant   political   groups   that   function   as   de  facto   political   parties   like   some   periods   in   Jordan,   Kuwait,   Swaziland,   and   Uganda.   Therefore,   I   compliment  NELDA  with  Svolik’s  (2012)  dataset  on  concentration  of  legislative  power  in   authoritarianism.  Relying  on  his  data  and  the  definition  of  Brownlee’s  (2009)  electoral   authoritarianism,   countries   are   regarded   as   electoral   authoritarian   if   multiple   political   actors   including   both   partisan   and   non-­‐partisan   opposition   groups   are   competing   in   a   legislative   election.   Svolik’s   data   enable   us   to   avoid   the   problem   of   nonpartisan   competition   because   Svolik’s   data   identify   electoral   competition   in   which   both   de  jure   and   de   facto   political   parties   participate.   However,   the   datasets   coverage   is   far   less     87   extensive  and  has  many  missing  values.  Therefore,  I  use  both  definitions  and  datasets  to   empirically  specify  electoral  authoritarian  countries.40       3.4.2  Dependent  Variable:  Effective  Electoral  Threshold     For   the   main   dependent   variable,   electoral   system   types,   I   use   the   Effective   Electoral   Threshold  (EET)  measure  originally  proposed  by  Lijphart  (1994,  2011)  and  Taagepera   (1998,  2007).  The  EET  measures  “the  proportion  of  votes  that,  for  each  electoral  system,   secures   parliamentary   representation   to   any   party   with   a   probability   of   at   least   50   percent”   (Boix   1999:   614).   It   is   well   known   that   the   EET   is   measured   by   using   the   following  formula:       𝐸𝐸𝑇 =   75%   (𝑀 + 1)       where   M   represents   average   district   magnitude   in   a   country-­‐year.   When   the   EET   is   lower   than   the   legal   threshold   that   sometimes   exists   in   PR   systems,   I   use   the   legal   threshold   as   an   Effective   Electoral   Threshold   in   the   country. 41  Using   various   data   sources   including   Nohlen   eds.   (1999,   2001,   2005,   2010),   Phil   Keefer’s   Database   of   Political  Institutions  (DPI),  the  Inter-­‐Parliamentary  Union  website,  Bormann  and  Golder   (2013),   Carey   and   Hix   (2011)   and   other   relevant   Internet   sources,   I   collect   data   on                                                                                                                   40  Both  measures  are  highly  correlated.  The  correlation  coefficient  is  0.62.       41  Even   when   I   do   not   recode   the   EET   variable   in   this   way,   basic   results   that   I   show   below  do  not  change.         88   district   size   and   legal   threshold   for   all   countries   in   the   world   between   1945-­‐2010.   To   measure  electoral  system  types,  the  effective  electoral  threshold  variable  is  adopted  by   Boix   (1999),   a   path-­‐breaking   work   on   the   choice   of   electoral   systems   in   advanced   democracies.42  Subsequent   studies   follow   this   tradition   and   rely   on   this   measure   to   explore  various  explanatory  factors  explaining  the  choice  of  electoral  systems  (Cusack,   Iversen  and  Soskice  2007;  Chang  et  al.  2010).       Figure  3.2:  Effective  Electoral  Threshold  in  Electoral  Authoritarian  Regimes       !"% Note:   Effective   Electoral   Threshold   (EET)   is   measured   by  𝐸𝐸𝑇 =   (!!!)  (Lijphart   1994;   Taagepera  1998)  where  M  stands  for  average  district  magnitude.  The  EET  measures  “the   proportion   of   votes   that,   for   each   electoral   system,   secures   parliamentary   representation  to  any  party  with  a  probability  of  at  least  50  percent”  (Boix  1999,  614).  If   the   legal   threshold   under   PR   systems   is   lower   than   the   EET,   then   I   use   the   legal   threshold  to  fill  the  EET.  Figure  2-­‐(a)  shows  the  level  of  average  EET  over  time.  Figure  2-­‐ (b)   shows   within-­‐country   time   series   change   in   the   EET   by   taking   average   of   the   first   difference.  The  vertical  dashed  lines  stand  for  standard  deviations.           Source:  Author’s  data                                                                                                                         42  Dichotomous   variables   of   electoral   system   types   (e.g.,   for   example,   SMD   and   PR   dummies)  may  be  also  available,  yet  as  Boix  (1999:  614)  suggested,  such  measures  are   often   too   rough   to   capture   significant   differences   within   both   systems   as   well   as   the   legal  threshold  often  adopted  under  PR  systems.     89   0 5 10 15 20 25 30 35 40 Effective Electoral Threshold 600 400 Frequency Whole Sample 200 Frequency 20 40 100 150 200 60 800 Middle East and North Africa 0 0 50 0 0 5 10 15 20 25 30 35 40 Effective Electoral Threshold 300 Frequency 100 0 0 5 10 15 20 25 30 35 40 Effective Electoral Threshold Asia Sub-Saharan Africa 200 60 Frequency 20 40 0 5 10 15 20 25 30 35 40 Effective Electoral Threshold Frequency 400 Latin America 0 Frequency Eastern Europe 0 20 40 60 80 100 Figure   3-­‐3:   Regional   Variations   in   the   Effective   Electoral   Threshold   in   Electoral   Authoritarianism   0 5 10 15 20 25 30 35 40 Effective Electoral Threshold 0 5 10 15 20 25 30 35 40 Effective Electoral Threshold   Note:  Western  Europe  and  North  America  are  not  in  the  sample  because  during  the  period  (1945-­‐2010),   there  are  no  electoral  authoritarian  regimes  in  the  regions.  Source:  Authors’  Data                In  Figures  3.2  and  3.3,  I  show  time-­‐series  variations  (both  level  and  change)  as  well  as   regional   variations   in   Effective   Electoral   Threshold   in   electoral   authoritarianism.43  As   you   can   see,   the   value   of   the   EET   has   been   declining   over   time.   In   addition,   there   are   wide  variations  in  the  EET  across  time  and  space.       3.4.3  Key  Independent  Variable:  Natural  Resource  Wealth   In  order  to  operationalize  the  main  independent  variable―mobilization  power  of  the   dictator―I   use   two   measures   of   natural   resource   wealth.   First,   I   use   Haber   and   Menaldo’s   (2011)   total   resource   income   per   capita.   This   variable   includes   natural   resources   that   can   produce   rents   captured   by   governments   such   as   oil,   gas,   coal,   and   metal   productions.   Second,   focusing   on   major   natural   resources—oil   and   gas,   Ross                                                                                                                   43  I  also  show  descriptive  statistics  of  all  the  variables  I  use  in  analysis  in  Appendix  C3.       90   (2012)   creates   a   variable   of   oil-­‐gas   value   per   capita,   which   is   calculated   multiplying   a   country’s  total  oil-­‐gas  production  by  the  current  oil-­‐gas  price  and  then  divided  by  total   population  (Ross  2012).  Both  measures  are  very  highly  correlated  (r  =  0.98).     These   two   variables   of   natural   resource   wealth   are   good   proxies   to   measure   dictators’   mobilization   power   from   a   theoretical   and   methodological   point   of   view,   especially   for   this   study.   First,   the   idea   that   natural   resource   wealth   strengthens   dictators’   distribution   capability   and   thus   makes   their   regimes   resilient   to   breakdown   is   a  well-­‐established  empirical  finding  in  the  study  of  authoritarian  durability  (Ross  2001;   Jensen  and  Wantchekon  2004;  Smith  2004;  Ulfelder  2007;  Desai,  Olofsgard,  and  Yousef   2009;   Morrison   2009;   Wright,   Frantz,   and   Geddes   2013). 44  Since   natural   resource   wealth,  particularly  non-­‐lootable  natural  resources  such  as  oil  and  gas,  has  been  mostly   dominated   by   state   or   state-­‐owned   companies   (Morrison   2009;   Andersen   and   Ross   2014),   it   contributes   to   magnifying   the   dictator’s   ability   to   use   patronage   distribution   in   deriving  political  support  from  constituencies.     Second,   natural   resource   wealth   is   also   a   good   indicator   in   a   methodological   sense   because   it   is   reasonable   to   think   that   the   natural   resources   variable   is   mostly   exogenously   determined   by   the   international   market   and   regional   potential   of   natural   resource  wealth,  independent  of  electoral  system  types.  The  nature  of  oil-­‐gas  value  per   capita   allows   me   to   avoid   endogeneity   problems   between   government   spending   and   electoral   system   choice.   In   fact,   several   studies   suggest   that   PR   systems   tend   to   have                                                                                                                   44  Haber   and   Menaldo   (2011)   argue   that   natural   resource   wealth   does   not   have   a   negative   impact   on   democratization.   Although   natural   resources   may   not   necessarily   discourage   a   country   to   democratize,   there   is   still   rich   evidence   that   natural   resources   allow   dictators   to   survive   longer   and   prevent   a   country   from   experiencing   authoritarian   breakdown  (Morrison  2009;  Wright,  Frantz  and  Geddes  2013).           91   higher   tax   rates   and   larger   redistribution   effects   in   democracies   (Austin-­‐Smith   2000;   Persson  and  Tabellini  2004;  Bawn  and  Rosenbluth  2006;  Iversen  and  Soskice  2006),  so   possibilities   of   reverse   causalities   should   be   taken   seriously.   Therefore,   more   direct   measures  such  as  general  fiscal  revenues  and  government  expenditure  are  very  hard  to   handle   in   observational   data.   In   this   respect,   natural   resource   wealth,   although   it   consists   only   a   part   (albeit   a   significant   part)   of   revenue   sources,   is   a   good   proxy   to   estimate  the  effect  of  fiscal  strength  on  the  choice  of  electoral  system,  while  mitigating   the  possibilities  of  reverse  causality.       3.4.4  Statistical  Method   The   unit   of   analysis   is   country-­‐year.   There   are   several   reasons   why   I   use   country-­‐ year  data  structure  rather  than  country-­‐election  year.  First,  my  data  collection  suggests   that   some   electoral   system   reforms   are   implemented   during   non-­‐election   years   (for   example,   Mexico,   Sri   Lanka,   and   Russia).   Thus,   country-­‐election   year   data   makes   the   analysis   less   accurate   in   the   sense   of   capturing   the   timing   of   electoral   system   change.   Second,  the  country-­‐year  data  format  enables  me  to  control  for  country-­‐fixed  effects  by   expanding   the   time-­‐series   dimension.   Since   it   is   not   hard   to   imagine   that   some   electoral   system  changes  are  possibly  driven  by  unobserved  country-­‐specific  effects  that  are  not   captured   by   natural   resource   wealth   and   other   standard   controls,   the   country-­‐year   data   structure  has  an  advantage  in  this  respect.     In  all  models,  I  add  a  lagged  dependent  variable  to  control  for  time  dependence,  or   path-­‐dependent   characteristics   of   electoral   systems   (Cox   1998).   To   deal   with   time-­‐ specific   effects,   I   include   half-­‐decade   dummies.   There   are   two   methods   to   test   the     92   hypothesis.   First,   I   employ   fixed-­‐effect   models   to   explain   within-­‐country   variation   in   the   EET  while  controlling  for  country-­‐specific  unobserved  heterogeneity.  Yet,  the  number  of   countries  (N)  is  still  larger  than  the  time-­‐series  (T)  (N   =   95,   T   =   65),  suggesting  that  the   Nickel  bias  might  be  a  serious  problem  if  I  include  a  lagged  dependent  variable  in  fixed   effect  models.  In  addition,  previous  studies  point  out  that  colonial  origins  (e.g.,  British,   French   and   Spanish   colonies)   are   crucial   in   determining   electoral   system   choice   in   developing   countries   (Blais   and   Massicote   1997)   but   fixed-­‐effects   models   do   not   allow   me   to   directly   take   into   account   the   effect   of   such   relevant   time-­‐invariant   variables.   Therefore,   I   also   estimate   system   GMM   models   (Arelano   and   Bover   1995;   Roodman   2007)   in   which   I   can   include   time   invariant   variables   such   as   colonial   heritage   and   regional  dummies  while  dealing  with  several  shortcomings  of  the  data,  such  as  the  short   time  span  in  the  sample,  fixed  individual  effects,  potential  heteroskedasticity,  and  auto-­‐ correlation  within  countries  (Roodman  2007).     I   expect   the   natural   resource   wealth   variables   to   be   negatively   correlated   with   the   Effective   Electoral   Threshold,   meaning   that   dictators   with   rich   financial   resources   are   more   likely   to   adopt   PR   systems.   Following   Boix   (1999),   I   add   logged   total   population   (one-­‐year   lagged;   World   Development   Indicators),   logged   territorial   size   (World   Development  Indicators),  trade  openness  (one-­‐year  lagged  sum  of  exports  and  imports   relative   to   GDP;   Rogowski   1987;   Penn   World   Table   7.1.),   and   Ethno-­‐Linguistic   Fractionalization  (Roeder  2002)  as  standard  controls.             93   3.4.5  Results   Table  3.1:  Determinants  of  Electoral  Systems  in  Electoral  Authoritarianism       Note:  Robust  standard  errors  are  reported  in  parentheses.       Table   3.1   displays   the   results.   Two   measures   of   natural   resource   wealth   are   both   consistently,  negatively  associated  with  the  EET  in  all  models  in  statistically  significant   ways   (at   the   .05   and   .01   levels).   The   results   suggest   that   strong   dictators   with   abundant   natural   resources   are   more   likely   to   adopt   PR   systems   by   lowering   the   Effective   Electoral   Threshold.   For   example,   Model   4   indicates   that   a   100   dollars   increase   in     94   natural   resource   income   per   capita   lowers   the   Effective   Electoral   Threshold   by   0.02.   Given   the   fact   that   average   change   in  the   Effective   Electoral   Threshold   ranges   between   -­‐ 0.46  and  0.24  with  1  to  4  standard  deviations  (Figure  3.2-­‐[b])  and  the  mean  of  natural   resource  wealth  is  708  USD  with  standard  deviations  of  3,563  USD,  the  impact  of  natural   resource  wealth  is  considerably  large.  These  results  provide  supporting  evidence  for  the   hypothesis―authoritarian  rulers  with  the  power  of  mobilization  are  more  likely  to  adopt   proportional  representation  systems.       3.4.6  Robustness  Check   I  run  additional  models  to  check  the  robustness  of  the  main  results.45  First,  one  may   think   that   authoritarian   leaders   feel   more   free   to   choose   PR   systems   over   SMD   when   they   severely   limit   political   competition   prior   to   elections   because   limited   party   competition   undermines   the   “tying-­‐the-­‐dictator’s-­‐hands”   character   of   PR   systems.   To   deal  with  this  possibility,  I  add  Polity  IV  index  (one-­‐year  lagged)  as  a  control;  controlling   for   political   competition   did   not   change   the   main   results.   Second,   I   adopt   clustered   standard  errors  by  region  for  fixed  effects  models  to  consider  the  likely  possibilities  that   measurement  errors  are  correlated  within  regions.  Clustering  by  region  does  not  affect   the  results  I  obtained  above.  Third,  I  use  an  alternative  measure  of  the  Effective  Electoral   Threshold,   in   which   I   do   not   adjust   its   score   by   the   legal   threshold   that   is   often   adopted   in  PR  systems,  and  the  main  findings  are  stable.  Finally,  according  to  Ross  and  Andersen   (2014:   4),   “[u]ntil   the   1960s,   most   of   the   rents   generated   by   oil   production   in   non-­‐ Western   countries   were   captured   by   a   handful   of   large,   vertically-­‐integrated                                                                                                                   45  For  details  on  the  results,  see  Appendix  C3.       95   international  oil  companies—sometimes  called  ‘the  Seven  Sisters.’  But  in  the  1970s,  the   industry   was   transformed   by   a   wave   of   nationalizations   and   contract   revisions   that   enabled   the   governments   of   host   countries   to   seize   control   of   these   rents.”   In   order   to   take  into  account  the  history  of  natural  resource  rents  as  a  “resource  curse,”  I  limit  the   sample   to   the   period   of   1970-­‐2007   and   run   system   GMM   models,46  but   natural   resource   endowments   remain   strongly,   negatively   correlated   with   the   Effective   Electoral   Threshold.     3.5  Testing  Causal  Mechanisms     The   above   analysis   found   the   likelihood   that   electoral   autocrats   with   resource   abundance   tend   to   adopt   PR   systems.   To   evidence   causal   mechanisms   and   several   assumptions   made   in   theory,   I   conduct   additional   cross-­‐national   analyses.   I   first   test   whether   natural   resource   wealth   does   contribute   to   mobilizing   regime   supporters   in   legislative  elections  under  electoral  authoritarian  regimes,  thereby  showing  that  natural   resource   wealth   is   a   good   proxy   to   measure   dictators’   ability   to   collect   political   support.   To   do   so,   I   estimate   the   impact   of   natural   resources   on   the   dictator’s   electoral   performance.   I   then   explore   whether   and   to   what   extent   SMD   systems   bring   the   pro-­‐ regime   bias.   Lastly,   I   show   cross-­‐national   evidence   suggesting   that   PR   systems   exacerbate   coordination   problems   within   the   opposition   camp   by   reducing   the   likelihood  of  building  a  pre-­‐electoral  coalition  among  opposition  parties.                                                                                                                         46  As   I   discussed   above,   fixed   effects   models   with   short   time   span   and   a   lagged   dependent   variable   induce   the   Nickel   bias.   Since   I   limit   the   sample   to   1970-­‐2007   with   88-­‐95  countries,  it  would  be  better  to  employ  system  GMM  models  for  estimation.           96   As   the   unit   of   analysis,   I   use   country-­‐election   year.   In   country-­‐election   year   data,   I   focus   on   parliamentary   elections,   which   electoral   systems   vary   significantly   across   electoral  authoritarian  regimes.  Since  the  time-­‐series  (election-­‐year;  it  is  4.2  on  average)   is  far  shorter  than  the  number  of  countries  (65  to  75  countries)  in  country-­‐election  year   data,   estimation   results   may   be   biased   if   fixed   effect   models   are   used   with   the   lagged   dependent   variable.   In   addition,   since   the   timing   when   elections   are   held   is   also   very   different   across   countries,   I   am   unable   to   use   other   methods   of   standard   panel   data   analysis   such   as   random   effects   and   system   GMM   models.   Therefore,   I   pool   the   data,   following   previous   studies   on   electoral   fraud   with   country-­‐election   year   data   (Kelley   2009).   To   control   for   regional   specific   effects,   I   add   regional   dummies.   To   deal   with   serial   correlation   and   time-­‐specific   effects,   I   add   a   one-­‐election   year   lagged   dependent   variable   and   half   decade   dummies   from   1955-­‐2010.   Robust   standard   errors   are   computed  to  cope  with  heteroskedasticity.  To  conduct  data  analysis,  I  originally  collect   data   such   as   vote   and   seat   shares   of   ruling   and   opposition   parties,   proportion   of   independent   politicians   in   parliament,   and   assembly   size   using   various   data   sources   such   as   Nohlen   eds.   (1999,   2001a,   2001b,   2005a,   2005b,   2010),   Database   of   Political   Institutions,   Arthur   Bank’s   Political   handbook   of   the   World   (1993-­‐2008,   various   volumes),  the  Inter-­‐Parliamentary  Union  website,  African  Election  Database,  and  other   area  studies  papers  and  Internet  sources.     3.5.1  Natural  Resource  Wealth  and  Authoritarian  Regime  Support     To  test  whether  natural  resource  wealth  contributes  to  boosting  regime  support  at   elections,  I  use  two  dependent  variables.  The  first  measure  is  total  percentage  of  votes     97   cast   for   ruling   parties.   This   is   a   straightforward   measure   because   I   can   directly   estimate   what   determines   ruling   party’s   popularity   at   the   ballot   box.   Yet,   authoritarian   leaders   may  care  more  about  to  what  extent  they  win  big  relative  to  opposition  parties  (Simpser   2013).   Thus,   I   use   the   second   measure,   margins   of   victory,   measured   as   the   gap   in   percentages  of  vote  shares  between  ruling  parties  and  opposition  parties.     Besides   financial   resources,   to   what   extent   authoritarian   rulers   can   garner   votes   is   determined   by   other   covariates.   To   consider   other   relevant   covariates,   I   introduce   the   following  control  variables.  The  stronger  politicians  who  are  not  affiliated  with  political   parties,   the   fewer   votes   ruling   parties   can   gain   at   elections.   To   measure   independents’   strength,   I   include   Vote   Shares   of   Independent   Politicians   in   the   previous   elections.   If   there   are   many   ethnic   groups   in   society,   it   may   be   more   difficult   for   ruling   parties   to   garner   political   support.   Ethno-­‐Linguistic   Fractionalization   constructed   by   Roeder   (2002)  is  included.     If   opposition   parties   refuse   to   join   elections,   then   ruling   parties   can   win   elections   more   easily   with   a   large   margin.   A   dummy   variable   of   whether   some   opposition   leaders   boycott   the   election   (Opposition’s   Boycott)   is   introduced   by   using   Hyde   and   Marinov’s   (2012)   NELDA   dataset   (NELDA   14).   Intuitively,   if   dictators   stuff   the   ballot   box,   ruling   parties   should   be   able   to   increase   their   vote   shares   and   win   the   elections   with   a   large   margin.  On  the  other  hand,  if  electoral  fraud  is  more  often  used  by  weak  dictators,  those   who  cannot  mobilize  regime  supporters,  such  positive  correlation  may  not  be  observed   between  the  variables.  A  dummy  variable  of  Electoral   Fraud  is  taken  from  NELDA  11.47                                                                                                                   47  “Before   elections,   were   there   significant   concerns   that   the   elections   would   not   be   free   and  fair?”       98   Previous   work   claims   that   pre-­‐electoral   violence,   which   is   mainly   exercised   by   ruling   parties   against   opposition   candidates   and   supporters,   is   conducive   to   ruling   party’s   electoral   victory   (Straus   and   Taylor   2012;   Hafner-­‐Burton,   Hyde,   and   Jablonski   2013).   Similar   to   electoral   fraud,   if   only   weak   dictators   use   election   violence,   its   effect   is   not   observed  in  the  form  of  ruling  party’s  vote  share  and  margins  of  victory.  I  use  the  NELDA   33  to  introduce  a  dummy  variable  of  Electoral   Violence.48  If  strong  Political   Competition   is   guaranteed,   it   is   harder   for   ruling   parties   to   win   elections   overwhelmingly.   For   this,   Polity  IV  score  (one-­‐year  lagged)  is  also  included  in  models.   When   a   legislative   election   is   held   in   parliamentary   system,   authoritarian   leaders   may   spend   more   efforts   to   win   the   election   than   a   legislative   election   in   presidential   system  because  election  results  directly  decide  who  holds  power.  A  dummy  variable  of   Executive-­‐Legislative   Relations   (0: presidentialism/semi-­‐presidentialism;   1:   parliamentarism)  is  introduced.     If  the  modernization  theory  is  correct,  ruling  parties  in  rich  countries  find  it  difficult   to   collect   votes   because   people   become   less   dependent   on   government   in   terms   of   economic   well-­‐being.   Logged   GDP   per   capita   (one-­‐year   lagged)   is   taken   from   Penn   World   Table   7.1   to   control   for   the   level   of   Economic   Development.   Better   economic   performance   should   make   dictators   and   their   parties   popular   among   citizens,   leading   to   better   electoral   performance.   GDP   growth   (one-­‐year   lagged)   is   measured   to   take   into   account  Economic  Growth  by  using  World  Development  Indicators.                                                                                                                         48  “Was   there   significant   violence   involving   civilian   deaths   immediately   before,   during   and  after  the  election?”  (NELDA  33)       99   Table  3.2:  Dictators’  Performance  at  the  Ballot  Box   DV:&Ruling&Party's&Electoral& Performance Natural&Resource&Variable DV&in&Previous&Elections Lagged&Natural&Resource&Wealth& Model&6 Share&of&Votes& (%) HM&(2011) 0.460*** (0.06) 0.005*** (0.002) Model&7 Margin&of& Victory&(%) HM&(2011) 0.410*** (0.07) 0.009*** (0.003) Vote&Share&of&Independents&&&&&&&&&&&&&&&&&&&&&&&&& (Previous&Elections) EthnoSLinguistic&Fractionalization Opposition&Boycott Electoral&Fraud Election&Violence Lagged&Polity&IV Parliamentarism Lagged&GDP&per&capita&(logged) Lagged&GDP&Growth Constant Regional&dummies HalfSdecade&dummies Number&of&Countries Observations RSsquared F&value 39.68*** (7.43) Yes Yes 75 291 0.348 9.90*** 27.23** (12.26) Yes Yes 75 285 0.261 6.24*** Model&8 Share&of& Votes&(%) HM&(2011) 0.339*** (0.08) 0.004** (0.002) Model&9 Model&10 Model&11 Margin&of& Share&of&Votes& Margin&of& Victory&(%) (%) Victory&(%) HM&(2011) Ross&(2012) Ross&(2012) 0.309*** 0.343*** 0.309*** (0.07) (0.08) (0.07) 0.007** 0.005** 0.007** (0.003) (0.002) (0.004) S0.243** S0.429*** S0.284** S0.454*** (0.12) (0.13) (0.11) (0.12) 1.365 0.821 2.376 3.104 (4.55) (8.82) (4.31) (8.51) 5.780* 15.66*** 5.190* 14.93*** (2.97) S1.73 (2.03) S8.645*** (2.75) S0.321 (0.22) 3.79 (2.37) S1.721 (1.38) 0.511** (0.21) 63.86*** (13.22) Yes Yes 65 246 0.442 8.39*** (5.13) S6.393* (3.83) S14.61*** (4.88) S0.727* (0.39) 4.73 (4.21) S3.591 (2.64) 1.183*** (0.40) 63.36*** (23.94) Yes Yes 65 244 0.415 8.06*** (3.02) S1.428 (1.96) S7.524*** (2.73) S0.284 (0.21) 2.74 (2.33) S0.795 (1.25) 0.547*** (0.20) 59.02*** (12.21) Yes Yes 67 255 0.447 8.45*** (5.32) S5.577 (3.74) S12.79*** (4.92) S0.698* (0.38) 3.09 (4.18) S1.439 (2.43) 1.261*** (0.39) 53.19** (22.40) Yes Yes 67 253 0.408 7.46*** Note:   Pooled   OLS   regressions   are   employed.   Robust   standard   errors   are   reported   in   parentheses.  The  margin  of  victory  is  measured  by  the  gap  between  ruling  parties’  and   opposition  parties’  vote  shares.  The  unit  of  analysis  is  country-­‐election  year.       Table   3.2   presents   the   statistical   results.   In   Models   6   and   7,   I   test   the   impact   of   natural   resource   wealth   on   regime   support   without   control   variables.   The   natural   resources  variable  has  positive  impacts  both  on  vote  shares  and  margins  of  victory  and     100     the   effects   are   statistically   significant   at   the   .01   level.   In   Models   8   through   11   where   I   include  the  controls  with  two  different  measures  for  natural  resource  wealth,  the  effects   remain   positive   and   statistically   significant   (at   the   .05   and   .10   levels).   Substantively,   a   100   dollars   increase   in   natural   resources   income   per   capita   tends   to   increase   ruling   parties’  share  of  votes  by  0.4  percent  and  their  margin  of  victory  by  0.7  percent  (based   on   Models   8   and   9).   The   results   suggest   that   natural   resource   wealth   is   positively   associated  with  popular  support  for  the  dictators.  These  findings  support  my  idea  that   natural  resource  wealth  is  a  good  surrogate  to  measure  dictators’  mobilization  power  at   the  ballot  box.       Looking   at   the   control   variables,   economic   growth,   opposition   boycott,   and   vote   share  of  independents  in  the  previous  election  all  show  expected  effects  in  statistically   significant   ways.   Meanwhile,   the   lagged   Polity   IV   score   has   a   negative,   statistically   significant   impact   only   on   the   margin   of   victory.   And   contrary   to   theoretical   expectations,   election   violence   is   negatively   correlated   with   both   ruling   party’s   vote   shares   and   the   margin   of   victory.   This   may   indicate   an   endogeneity   problem   between   election  violence  and  regime  strength:  weak  dictators,  those  who  are  not  popular  among   constituents,   may   be   more   inclined   to   resort   to   election   violence.   As   NELDA   33   does   not   distinguish   pre-­‐   and   post-­‐election   violence,   this   is   a   plausible   possibility.   Yet,   even   if   I   exclude  the  variable  of  election  violence,  the  overall  results  do  not  change  and  natural   resource  wealth  is  still  positively  associated  with  the  dependent  variables.                               101   3.5.2  Electoral  Systems  and  the  Pro-­‐Dictator  Bias     In  a  second  empirical  analysis,  I  investigate  whether  and  to  what  extent  SMD  systems   bring   the   seat   premiums   only   to   ruling   parties.   In   order   to   offer   evidence   of   the   pro-­‐ dictator  bias,  I  use  the  following  three  dependent  variables:         (i)  Disproportionality  Index.  I  compute  a  conventional  disproportionality  index,  which  is   widely   used   in   the   previous   literature.   Disproportionality   is   measured   by     !"#$%!"   !  !"#$%!" ! !    in   which   both   ruling   and   opposition   parties   are   included   to   compute   disproportionality.49  Employing   the   disproportionality   index,   I   can   test   whether   SMD   systems  bring  the  larger  number  of  seats  with  higher  seats-­‐votes  elasticity  regardless  of   whether  parties  are  governing  parties  or  not.       (ii)   Seats-­‐Votes  Gap  of  Opposition  Parties.   The   second   dependent   variable   is   seats-­‐votes   gap   of   opposition   parties.  The   Effective   Electoral   Threshold   should   not   have   a   positive   effect   on   the   seat   premiums   to   opposition   parties   because,   however   large   opposition   parties  might  be,  SMD  systems  bring  the  seat  premium  only  to  ruling  parties.           (iii)   Seats-­‐Votes   Gap   of   Ruling   Parties.   Third,   I   use   seats-­‐votes   gap   of   ruling   parties.   In   contrast  with  the  above-­‐stated   two   dependent   variables,   SMD   is   more   likely   to   bring   the   seat  bias  in  favor  of  ruling  parties.                                                                                                                     49  Disproportionality   is   calculated   by   using   seats-­‐votes   gaps   in   the   ruling   party   and   opposition  camps,  without  disaggregating  them  into  each  party’s  seats-­‐votes  gap.       102     For   control   variables,   my   models   are   largely   based   on   Lijphart   (1994),   who   introduces   variables   affecting   disproportionality   in   advanced   democracies   besides   my   variable   of   interest,   The   Effective   Electoral   Threshold   ―   logged   assembly   size   (the   natural   logarithm   of   total   number   of   seats   in   the   lower   house)   and   government   form   (presidentialism/semi-­‐presidentialism   vs   parliamentarism).   In   addition   to   these,   I   also   control   for   vote   shares   of   independents   in   the   previous   elections   to   partial   out   the   impact  of  independent  politicians’  strength.                                     103    Table  3.3:  The  Pro-­‐Dictator  Bias  under  Majoritarian  Electoral  Systems   Model&12 DV DV&in&Previous&Elections Effective&Electoral&Threshold&(EET) Model&13 Seats7Votes&Gap&of& &Disproportionality Opposition&Parties&(%) 0.311*** (0.08) 0.092 (0.06) 0.307*** (0.08) 70.079 (0.05) Model&14 Model&15 Seats7Votes&Gap&of& Ruling&Parties&(%) Seats7Votes&Gap&of& Ruling&Parties&(%) 0.223** (0.09) 0.09* (0.05) 0.188** (0.08) 70.417** (0.17) Vote&Share&of&Ruling&Parties 70.232*** (0.05) EET*&Ruling&Parties&Vote&Share 0.008*** (0.003) Vote&Share&of&Independents&& (Previous&Elections) Logged&Assembly&Size Parliamentarism Constant Regional&dummies Half7decade&dummies Number&of&Countries Observations R7squared F&value 70.016 0.0409 70.038 70.027 (0.05) 71.41 (1.02) 70.662 (1.72) 19.39** (8.86) Yes Yes 72 264 0.2587 4.54*** (0.05) 0.316 (1.01) 71.106 (1.38) 77.58 (8.92) Yes Yes 72 264 0.282 4.93*** (0.03) 70.16 (0.93) 0.15 (1.44) 9.75 (7.89) Yes Yes 72 264 0.273 5.22*** (0.04) 70.807 (0.90) 70.464 (1.42) 28.4 (8.44) Yes Yes 72 264 0.322 5.41*** Note:   Pooled   OLS   regressions   are   employed.   Robust   standard   errors   are   reported   in   parentheses.   Disproportionality   is   measured   by     !"#$%!"   !  !"#$%!" ! !    in   which   both   ruling   and  opposition  parties  are  included  to  compute  disproportionality.    The  unit  of  analysis   is  country-­‐election  year.       Table   3.3   shows   the   results.   First,   Model   12   tests   whether   SMD   systems   tend   to   be   associated   with   higher   disproportionality   in   72   electoral   authoritarian   regimes.   Although   the   direction   of   the   coefficient   is   positive,   the   EET   variable   does   not   reach   conventional   levels   of   statistical   significance   (p-­‐value   =   0.129).   Given   that   in   advanced     104     democracies,   majoritarian   electoral   systems   are   the   most   powerful   predictor   of   high   disproportionality   (Lijphart   1994;   Moser   and   Scheiner   2012),   the   result   suggests   that   electoral   systems   in   electoral   authoritarianism   do   not   necessarily   lead   to   high   disproportionality  to  both  ruling  and  opposition  parties.  Model  13  investigates  the  same   proposition  from  a  different  angle  by  setting  the  seats-­‐votes  gap  of  opposition  parties  as   a   dependent   variable.   Again,   the   effect   of   the   Effective   Electoral   Threshold   is   not   statistically  significant  (p-­‐value  =  0.149)  and  the  coefficient  is  negative,  suggesting  that   majoritarian   electoral   systems   may   tend   to   underrepresent   opposition   parties   in   parliament  although  the  impact  is  not  distinguishable  from  zero.     In   the   following   two   models,   I   test   whether   SMD   systems   bring   the   larger   seat   premiums  exclusively  to  ruling  parties.  In  Model  14,  the  Effective  Electoral  Threshold  is   positively   correlated   with   the   seats-­‐votes   gap   on   the   side   of   ruling   parties   at   the   .10   significance   level.   For   example,   when   the   Effective   Electoral   Threshold   is   37.5   percent   (SMD),   then   ruling   parties   are   more   likely   to   obtain   3.33   percent   more   seats,   when   compared   to   the   scenario   where   dictators   adopt   a   PR   system   and   the   Effective   Electoral   Threshold  is  just  5  percent.  This  result  supports  my  theoretical  expectation  that  the  pro-­‐ dictator  bias  comes  about  under  SMD  systems  in  electoral  authoritarian  regimes.                 105   Coefficient of Effective Electoral Threshold -0.4 -0.2 0 0.2 0.4 0.6 Figure   3.4:   The   Magnitude   of   the   Pro-­‐Dictator   Bias   Conditional   upon   Regime   Strength  (Vote  Shares  of  Ruling  Parties)     20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Vote Share of Ruling Parties (%) Coefficient of Effective Electoral Threshold 95% Confidence Intervals                        Model   15   estimates   an   interaction   model   to   show   how   the   effect   of   the   Effective   Electoral  Threshold  on  ruling  parties’  seats-­‐votes  elasticity  changes  depending  on  vote   shares  that  ruling  parties  enjoy  at  the  elections.  This  is  an  important  investigation  into   trying  to  know  the  existence  of  a  trade-­‐off  between  certainty  and  credibility  of  election   results,   which   I   focus   on   for   my   theory   of   electoral   system   choice   in   dictatorship:   If   strong  dictators  with  high  vote  shares  enjoy  larger  seat  premiums  under  SMD  systems,   then   the   signaling   function   of   election   results   may   be   complicated   because   of   the   stronger  pro-­‐dictator  bias.  The  interaction  term  is  positive  and  statistically  significant  at   the  .01  level,  suggesting  a  positive  feedback  character  of  the  pro-­‐dictator  bias:  the  seat   premiums   become   larger   if   ruling   parties   are   able   to   mobilize   a   large   number   of     106   supporters   at   elections.   For   instance,   if   a   dictator   adopts   a   SMD   system   (EET=37.5   percent)   and   the   ruling   parties   obtain   65   percent   of   total   votes   in   elections,   then   the   ruling  parties  receive  3.7  percent  more  seats  compared  to  the  scenario  when  she  adopts   a   PR   system   with   the   5   percent   EET   and   the   same   65   percent   vote.   In   contrast,   when   ruling   parties   get   more   vote   shares,   say,   85   percent   of   total   votes,   SMD   provides   8.45   percent  more  seats  than  a  PR  system  with  the  5  percent  of  EET.  This  result  suggests  a   possibility   that   strong   dictators,   who   can   mobilize   a   large   number   of   regime   supporters,   may  not  need  to  adopt  a  majoritarian  electoral  system.  Rather,  they  may  even  have  an   incentive  to  shift  an  electoral  system  more  to  a  PR  to  show  their  strength  because  their   vote   shares   are   sufficient   to   secure   an   overwhelming   majority   and   the   large   seat   premium  under  SMD  systems  may  obscure  de  facto  regime  strength.     Figure   3.4   also   provides   additional   evidence   that   SMD   systems   do   not   necessarily   backfire   on   weak   ruling   parties.   The   coefficient   of   the   EET   does   turn   negative   if   the   ruling   parties’   vote   share   is   less   than   about   50   percent,   yet   the   effect   is   not   distinguishable  from  0  at  .05  level.  The  negative  impact  becomes  statistically  significant   when   ruling   party’s   vote   share   is   less   than   25   percent,   which   is,   however,   rare   in   electoral   authoritarian   countries   (less   than   the   5th   percentile   in   the   sample).50  The   overall  results  again  support  the  theoretical  assumption  posited  in  my  theory:  Contrary   to   Rokkan’s   (1970)   and   Boix’s   (1999)   theoretical   expectations,   SMD   systems   do   not   necessarily   backfire   on   authoritarian   rulers   with   high   disproportionality   in   electoral   authoritarianism.   The   evidence   here   suggests   that   this   is   because   opposition   parties   are                                                                                                                   50  Such   countries   are   Ecuador   (2002),   Peru   (1990),   East   Germany   (1990),   Albania   (1992),   Cyprus   (1970),   Russia   (1995),   Belarus   (2000,   2004,   2008),   Liberia   (2005),   Kenya  (1992),  Iraq  (2010),  South  Korea  (1950),  and  Pakistan  (1985).         107   very  weak  and  even  if  ruling  parties  are  not  capable  of  maintaining  50  percent  of  total   votes,   authoritarian   rulers   can   contain   opposition   surges   in   elections   by   engaging   in   gerrymandering  and  other  indirect  manipulation  techniques  (Ahmed  2013).         3.5.3  PR  Systems  Inhibit  Pre-­‐Electoral  Opposition  Coalitions   Lastly,  I  investigate  whether  PR  systems  prevent  opposition  parties  from  uniting  to   challenge   the   authoritarian   ruler.   Here,   the   dependent   variable   is   whether   a   pre-­‐ electoral  opposition  coalition  is  formed.  I  use  Gandhi  and  Reuter’s  (2013,  143)  measure   of  opposition  coalition  making,  which  is  a  dichotomous  variable  coded  1  “if  there  was  a   significant  pre-­‐electoral  coalition  among  opposition  parties,  and  0  otherwise.”  And,  the   main  independent  variable  is  the  Effective  Electoral  Threshold.     For   model   specifications,   I   follow   Gandhi   and   Reuter   (2013),   the   most   comprehensive  analysis  on  pre-­‐electoral  coalition  making  in  non-­‐democracies.  Although   their  sample  is  all  non-­‐democracies  (1946-­‐2006),  my  sample  is  limited  only  to  electoral   authoritarian   regimes.   Based   on   Gandhi   and   Reuter’s   (2013)   models,   I   include   Ethno-­‐ Linguistic   Fractionalization,   number   of   opposition   parties,   parliamentarism,   age   of   the   largest   opposition   parties,   economic   growth   (one   year   lagged),   logged   total   population   (one  year  lagged),  election  violence,  ruling  party’s  share  of  seats  in  the  previous  elections,   natural  resource  wealth  (one  year  lagged),  and  five  regional  dummies.                 108   Table   3.4:   Electoral   Systems   and   Pre-­‐Electoral   Coalitions   in   Electoral   Authoritarian  Regimes   DV Effective&Electoral&Threshold &Natural&Resource&Wealth&(one&year&lagged,&100&dollars) Age&of&Largest&Opposition&Party Number&of&Opposition&Parties Ruling&Party's&Seat&Share&in&the&Previous&Election Parliamentarism Ethno.Linguistic&Fractionalization Economic&Growth&(one&year&lagged) Logged&Total&Population&(one&year&lagged) Electoral&Violence& Model&16 Pre.Electoral&Coalition 0.0521*** (0.019) .0.00114* (0.001) 0.276*** (0.076) 0.147*** (0.047) 0.0063 (0.009) .0.65 (0.647) 3.634*** (1.237) 0.0122 (0.036) .0.216 (0.162) .0.233 (0.509) Lagged&Dependent&Variable Constant .2.377 (2.697) Yes 71 225 .82.84 34.88*** Regional&dummies Number&of&Countries Observations Log&psedolikelihood& Wald&Chi&Squared Model&17 Pre.Electoral&Coalition 0.0512** (0.022) .0.00155 (0.001) 0.304*** (0.092) 0.117* (0.060) 0.0292* (0.015) .1.221 (0.953) 4.310** (2.029) .0.00536 (0.048) .0.0867 (0.258) 0.487 (0.615) 2.270*** (0.622) .6.990 (4.715) Yes 55 167 .53.97 44.94*** Note:  Pooled  logistic  regression  is  employed.  Robust  standard  errors  are  reported  in   parentheses.                   109     Probability of Pre-Electoral Opposition Coalition 0 .1 .2 .3 Figure  3-­‐5:  Effect  of  Electoral  Systems  on  Pre-­‐Electoral  Opposition  Coalitions   0 2.5 5 7.5 10 12.5 15 17.5 20 22.5 25 27.5 30 32.5 35 37.5 Effective Electoral Threshold   Note:  Dashed  lines  indicate  95  percent  confidence  intervals.  The  graph  is  based  on   Model  16.       Table   3.4   shows   the   results.   As   expected,   the   Effective   Electoral   Threshold   is   positively  correlated  with  the  probability  of  pre-­‐electoral  opposition  coalition  making  in   both   models,   suggesting   that   in   PR   systems   opposition   parties   are   less   likely   to   form   coalitions  before  elections.  Figure  3.5  graphically  illustrates  this  result  based  on  Model   16.   When   Effective   Electoral   Threshold   is   5   percent,   the   probability   of   opposition   coalition   making   is   just   4.4   percent.   When   a   country   adopts   a   SMD   system   (EET   is   equal   to   37.5   percent),   the   probability   of   coalition   increases   up   to   20.2   percent.     This   suggests   that   electoral   systems   are   also   important   to   explain   opposition   coordination   in   electoral   authoritarian  regimes.       110   On   the   other   hand,   consistent   with   Gandhi   and   Reuter   (2013),   the   number   of   opposition   groups   and   the   longevity   of   the   largest   opposition   party   both   have   positive   correlations   in   statistically   significant   ways   in   my   electoral   authoritarianism   sample.   Natural   resource   wealth   is   negatively   correlated   with   opposition   coalition   in   one   model,   implying   that   a   dictator’s   financial   resources   itself   tends   to   inhibit   the   opposition’s   coordination   at   elections,   yet   the   impact   is   uncertain   in   the   other   model.   Lastly   and   counterintuitively,   ethno-­‐linguistic   fractionalization   has   a   positive   correlation   with   the   dependent  variable.           3.6  Conclusion     This   chapter   has   explored   the   logic   of   electoral   system   choice   in   electoral   autocracies.  In  electoral  authoritarian  regimes  where  opposition  parties  are  very  weak,   dictators   and   their   parties   have   a   seat   bias   under   SMD   systems,   which   help   them   obtain   an   overwhelming   parliamentary   majority.   Such   a   pro-­‐dictator   bias,   however,   obscures   an  important  function  of  authoritarian  elections:  Conveying  a  credible  signal  of  regime   strength   via   elections.   On   the   flip   side,   strong   dictators,   who   are   capable   of   mobilizing   regime  supporters,  have  the  incentive  to  shift  electoral  systems  from  SMD  to  PR  because   PR  systems  provide  less  advantageous  institutional  settings  for  dictators,  through  which   they   can   use   election   results   as   a   device   to   send   a   costly   signal   by   winning   overwhelmingly.   Using   original   datasets   of   electoral   authoritarianism,   cross-­‐national   analyses   render   strong   empirical   evidence   endorsing   my   theoretical   expectations:   (1)   Natural   resource   endowments   are   negatively   correlated   with   the   Effective   Electoral     111   Threshold;   (2)   natural   resource   wealth,   a   good   proxy   to   measure   dictators’   power   of   distribution,   contributes   to   increasing   regime   support;   (3)   SMD   systems   tend   to   bring   the   larger   seat   premiums   only   to   ruling   parties;   and   (4)   SMD   systems   tend   to   promote   a   pre-­‐electoral  opposition  coalition.     The   analyses   presented   in   this   chapter   suggest   policy   implications   and   further   research  agendas  on  authoritarian  politics.  First,  this  research  shows  that  authoritarian   leaders  may  strategically  choose  electoral  systems  depending  on  their  strengths.  By  so   doing,   they   attempt   to   take   advantage   of   electoral   institutions   in   their   favor.   Without   closely   investigating   the   dictator’s   financial   conditions   and   distribution   power   in   an   authoritarian   country,   the   international   community   may   not   be   able   to   implement   effective  measures  to  reform  electoral  systems,  thereby  pushing  the  country  to  achieve   further  democratization.     Second,   the   present   study   proposes   a   further   need   to   research   indirect   manipulation   techniques   like   electoral   system   change,   gerrymandering   and   malapportionment   as   a   useful   set   of   tools   in   the   dictator’s   toolkit.   Because   direct,   blatant   electoral   fraud   often   hurts   authoritarian   leaders   by   sparking   popular   protests   and   other   dissents   (Tucker   2007;   Hafner-­‐Burton,   Hyde,   and   Jablonski   2013),   indirect   election   maneuvering   becomes  a  more  secure  strategy  for  authoritarian  rulers.  One  possible  research  agenda   may  be  to  explore  relationships  between  direct  and  indirect  manipulation  techniques  by   systematically   theorizing   when   authoritarian   rulers   are   tempted   to   use   blatant   measures  over  indirect  ones  and  vice  versa.         112   Third,   the   theory   proposed   here   makes   another   prediction   on   an   economic   consequence  of  electoral  systems  in  authoritarian  regimes:  as  dictators  need  to  mobilize   a   large   number   of   regime   supporters,   they   should   adopt   expansionary   fiscal   and   monetary   policies   to   maintain   electoral   dominance   under   PR   systems.   Studying   the   various   aspects   of   economic   outcomes   under   different   authoritarian   electoral   systems   would  be  another  promising  research  topic.     Finally,   this   chapter   also   suggests   the   possibility   that   natural   resources   may   be   significantly   affecting   institutional   designs   in   authoritarian   regimes.   Scholars   of   authoritarian   political   institutions   have   examined   how   political   institutions   change   the   prospect   of   the   dictator’s   survival.   However,   as   Pepinsky   (2014)   points   out,   every   political  institution  in  an  authoritarian  regime  should  be  endogenous  to  power  relations   and   the   distribution   of   economic   resources   in   the   country.   If   natural   resource   wealth   influences   the   origins   of   political   institutions   in   dictatorships,   then   we   will   have   to   elaborate   on   an   endogenous   theory   of   authoritarian   politics,   which   enables   us   to   take   into  account  both  causes  and  consequences  of  political  institutions  in  dictatorships,  and   test  the  theory  with  a  well-­‐planned  research  design.                       113          CHAPTER  4   POLITICAL  MANIPULATION  OR  FISCAL  MANEUVERING?       The  CASE  OF  KAZAKHSTAN       4.1  Introduction                In  the  previous  chapters,  I  presented  a  theory  of  authoritarian  elections  and  tested  its   observable   implications   by   conducting   cross-­‐national   analyses   of   electoral   fraud   and   electoral   system   choice   in   dictatorships.   This   chapter   provides   a   case   study   of   Kazakhstan  to  illustrate  causal  pathways  postulated  in  the  theory.                Kazakhstan   experienced   limited   political   liberalization   a   few   years   after   its   independence   following   the   collapse   of   the   Soviet   Union.   The   nascent   democratization   efforts  in  Kazakhstan  were,  however,  completely  reversed  in  1995.  Since  then,  President   Nursultan  Nazarbaev  has  successfully  maintained  his  rule,  and  the  country  has  been  an   electoral   authoritarian   regime.   The   intriguing   puzzle   here   is   that   the   Kazakh   government   has   permitted   opposition   parties   to   attend   elections   even   after   the   authoritarian  turn  in  1995.  Interestingly,  blatant  electoral  manipulation  in  the  forms  of   election  violence,  ballot  box  tampering,  and  the  manipulation  of  the  electoral  laws  have   become   gradually   less   severe   in   the   country.   This   stands   in   stark   contrast   with   Kyrgyzstan,   where   electoral   manipulation   has   been   escalated   under   the   Akaev   regime,   as   I   will   discuss   in   Chapter   6.   In   addition,   Nazarbaev   decided   to   adopt   a   Proportional   Representation   (PR)   system   in   2007,   which   did   not   bring   large   seat   premiums   to   his     114   party,   at   least   compared   to   the   previous   Single-­‐Member   District   (SMD)   systems.   Why   could  Nazarbaev  successfully  consolidate  his  rule  for  more  than  twenty  years,  despite  the   fact  that  he  became  less  dependent  on  political  manipulation  at  the  ballot  box?                      In   order   to   solve   this   puzzle,   I   conduct   a   case   study   of   Kazakhstan.   I   use   various   materials  including  original  quantitative  data  as  well  as  secondary  sources,  newspapers,   domestic   publications   including   government   documents,   and   semi-­‐structured   interviews  with  politicians,  bureaucrats,  NGO  activists  and  local  researchers  conducted   through  my  field  work  in  the  country  in  the  summer  of  2014.              The   case   study   proceeds   as   follows.   First,   I   briefly   illustrate   political   processes   through   which   the   country   experienced   limited   liberalization   and   turned   into   an   electoral   authoritarian   regime   by   1995.   Then,   I   describe   how   Nazarbaev   became   less   dependent  on  blatant  electoral  fraud  by  the  late  2000s  in  comparison  to  the  late  1990s.  I   also  describe  how  the  country  changed  its  electoral  system  from  SMD  to  PR  during  the   same  period,  although  the  SMD  system  had  brought  significant  seat  premiums  to  ruling   politicians.   In   the   third   section,   I   argue   that   these   “electoral   reforms”   were   possible   because   Nazarbaev   was   strong   enough   to   mobilize   large-­‐scale   voluntary   support   from   citizens:  He  increased  his  ability  to  distribute  economic  favors  to  at  least  some  portions   of   citizens   by   utilizing   rich   natural   resource   wealth   and   extensive   organizational   networks  of  the  government  as  well  as  the  dominant  party  Nur  Otan.  On  the  other  hand,   opposition  parties,  especially  since  the  middle  of  the  2000s,  have  become  too  weak  to  be   a   real   option   for   voters   at   the   ballot   box.   The   fourth   section   provides   systematic   quantitative  evidence  of  political  business  cycles  in  Kazakhstan,  demonstrating  that  pre-­‐ electoral   economic   distribution   as   manifestation   of   the   president’s   mobilization   power     115   tended   to   be   larger   in   the   second   half   of   the   2000s,   when   the   government   was   able   to   enjoy   the   rich   state   resources   and   strong   ruling   parties.   Lastly,   I   conclude   by   summarizing  main  arguments  and  findings  in  this  chapter.                       4.2  Limited  Liberalization         It   was   December   16,   1991   when   Kazakhstan   declared   its   independence.   The   collapse  of  the  Soviet  Union  and  national  independence  prompted  the  Kazakh  President,   Nursultan   Nazarbaev,   to   implement   political   and   economic   reforms.   President   Nazarbaev  took  measures  to  encourage  political  liberalization.  He  stated  in  a  newspaper   interview  in  October  1991  “I  see  Kazakhstan  as  a  democratic,  presidential  republic,  with   a   professional   parliament,   elected   on   a   multiparty   basis,   and   with   strong   executive   power  in  the  centre  and  in  the  region”(Cummings  2005,  24).       The  First  Constitution  of  Kazakhstan  that  was  adopted  on  January  28,  1993  was   indeed  more  liberal  than  those  of  other  Central  Asian  countries,  such  as  Turkmenistan   and   Uzbekistan.   Although   it   gave   the   president   decisive   control   over   the   local   governments   (Olcott   1995,   276;   Olcott   2002,   96-­‐101),   it   did   formally   divide   the   government   into   three   branches:   the   executive,   represented   by   the   large   presidential   staff;   the   legislative,   represented   by   the   unicameral   Kazakh   Supreme   Council,   whose   representatives  were  elected  on  a  territorial  basis;  and  the  judiciary.   Furthermore,  the   Supreme  Council  “enjoyed  a  wide  range  of  formal  rights  including  the  right  to  approve   the  budget,  amend  the  constitution  and  elect  the  constitutional  court”  (Anderson  1997,   307).   Under   the   First   Constitution,   by   utilizing   the   power   of   the   parliamentary     116   institution,   legislators   had   the   ability   to   challenge   the   president   over   the   issue   on   privatization.     Second,   Nazarbaev   allowed   the   formation   of   new   political   parties   and   organizations.   For   example,   Nevada-­‐Semipalatinsk   Movement,   Kazakhstan’s   largest   noncommunist   public   organization   during   the   glasnost   era,   transformed   itself   into   an   opposition   party   called   the   National   Congress   (NC)   in   October   1991.   The   new   party   was   headed  by  a  Kazakh  writer  and  political  activist,  Olzhas  Suleimenov  (RFE/RL  Research   Report,   January   3,   1992).   NC   received   much   financial   support   from   big   sponsors   and   private   businesses,   which   provided   the   party   with   an   independent   power   base   to   challenge   the   president   (Issacs   2011,   58).   Moreover,   the   Socialist   Party   (SP)   was   founded   in   August   1991   as   the   successor   to   the   Kazakhstan   Communist   Party   (KCP),   which  was  previously  dissolved  by  Nazarbaev.  After  dissolving  KCP,  Nazarbaev  quickly   lost   control   of   the   successive   party   (Olcott   2002:   93)   because   SP   consisted   of   many   deputies  in  the  Supreme  Soviet  anxious  about  radical  economic  reforms  (see,  e.g.  Issacs   2011:  57).    Given  its  powerful  organizational  base  and  its  large  membership,  SP  emerged   as  the  strongest  oppositional  group  at  that  time.  With  SP  having  turned  into  opposition,   Nazarbaev   himself   formed   a   pro-­‐presidential   party,   the   Union   of   People’s   Unity   of   Kazakhstan   (SNEK)   -­‐-­‐   the   precursor   of   the   current   dominant   party   Nur   Otan   -­‐-­‐   and   included   ruling   elites   that   would   later   play   important   roles   in   sustaining   the   authoritarian  rule  in  Kazakhstan  (Issacs  2011,  57).           117            Third,   various   kinds   of   social   movements   were   active   before   and   after   independence.51  For  instance,  the  aforementioned  Nevada-­‐Semipalatinsk  Movement  was   formed  in  1989  in  an  attempt  to  halt  Soviet  nuclear  testing  in  the  Northeastern  region  of   Semipalatinsk  (Uyama  2003,  49).  This  organization  had  encouraged  more  than  a  million   people  to  sign  petitions  demanding  to  ban  nuclear  tests  in  Kazakhstan’s  territory,  and  as   a   result   huge   crowds   participated   in   this   group’s   rallies   (Olcott   2002,   90).   Moreover,   the   Kazakh   nationalist   group   (and   also   party)   Alash   was   also   established   with   a   slogan   “Islam,   Turkism,   Democracy”,   and   a   political   agenda   supporting   the   exclusion   of   Russians   from   Kazakhstan.   Besides   Alash,   there   were  three  other   Kazakh   ethnic   groups,   namely,   the   movement   Azat,   the   party   Zeltoksan,   and   the   Republican   Party   of   Kazakhstan   (RPK),   which   had   a  popular   support   base   divided   between   Kyzylorda,   South   Kazakhstan,  and  Zhambul  oblasts  (Jones  Luong  2002,  144).  In  addition,  at  least  six  pro-­‐ Russian   groups   with   nationalist   and   federative   agendas   were   formed:   Edinstvo,   Civic   Contact,   Democratic   Progress,   Russkaia   Obshina,   Russian   Center,   and   Lad.   Their   main   goal  was  the  protection  of  Russian  culture.  Russkaia  Obshina  and  Lad,  especially,  carried   out   active   campaigns   for   bilingualism   and   dual   citizenship,   receiving   the   bulk   of   their   popular   support   from   the   North   Kazakhstan   oblast   and   East   Kazakhstan   oblast,   respectively   (Jones   Luong   2002,   144).   Lastly,   minority   ethnic   groups   such   as   Koreans   and  Germans  established  cultural  centers  during  Gorbachev’s  Perestroika  era  and  were   quite   active   and   in   constant   communication   with   the   governments   of   their   external   national  homelands  (Oka  2003,  473).                                                                                                                   51  For   more   detailed   discussion   on   the   social   movements   in   Kazakhstan,   see   Zhovtis   (1999),  and  Uyama  (2003).       118            Finally,   by   1994,   Kazakhstan   developed   relatively   free   and   vigorous   media   that   provided   the   legislators   with   opportunities   of   conveying   their   opinions   to   the   wider   public.  Although  the  two  largest  official  newspapers,  Kazakhstanskaia  Pravda  and  Vesti   Kazakhstana,  supported  the  government’s  positions,  they  also  reported  the  speeches  of   opposition   parliamentarians   (Olcott   2002,   104).   The   other   major   independent   newspapers  were  fully  reporting  opposition’s  political  campaigns.  The  largest  and  most   popular   of   these   was   Karavan,   which   would   engage   in   reporting   scandals.   The   other   major  independent  newspaper,  Panorama,  would  cover  parliamentary  maneuvering  as   well   as   social   and   economic   issues   (Olcott   2002,   104).   Furthermore,   Kazakhstan   had   a   few   independent   radio   stations   and   one   large   independent   television   company,   KTK,   which  would  freely  broadcast  these  issues  as  well.                By  1993,  Kazakhstan  came  to  be  perceived  in  the  West  as  one  of  the  more  democratic   states  in  Central  Asia  due  to  the  presence  of  substantial  political  competition,  relatively   free  press,  and  its  commitment  to  rapid  privatization  (RFE/RL  Research  Report,  January   7,  1994).  However,  radical  marketization  policies  introduced  by  Nazarbaev  provoked  a   confrontation   between   the   President   and   the   Parliament   which  led  to  the  backsliding  of   the  nascent  Kazakh  democracy.              Confronted   with   increasing   legislative   resistance,   the   government   encouraged   city,   raion   (region),   and   oblast   (state)   soviets   to   dissolve   themselves   in   November   1993.   Then,   Nazarbaev   ordered   the   Supreme   Council   to   “voluntarily”   dissolve   itself   in   December,  calling  for  elections  to  the  new  legislature  in  March  1994.              Nevertheless,   at   the   time   of   the   1994   parliamentary   election,   the   opposition   movements   in   Kazakhstan   were   still   vigorous.   Facing   strong   opposition   during   the     119   election  campaigns,  Nazarbaev  resorted  to  various  undemocratic  measures  to  preserve   his  hegemony.  For  example,  the  new  parliament  had  only  177  seats  (the  Supreme  Soviet   had   270   seats;   Olcott   1995,   263),   of   which   42   were   to   be   filled   from   a   “state   list”   by   which   the   president   selected   legislators.   Moreover,   electoral   districts   were   drawn   in   a   fashion   to   guarantee   Kazakh   pluralities   wherever   possible   (Olcott   1995,   278).   Nonetheless,  the  election  results  showed  that  opposition  was  still  considerably  strong;   the  distinct  opposition  parties,  NC,  SP,  Lad,  Azat  and  the  Communist  Party  of  Kazakhstan   (CPK),  occupied  43  seats  in  total,  which  were  roughly  equal  to  the  number  of  seats  of  the   pro-­‐presidential  Union  of  People’s  Party  (SNEK).              Soon   after   the   opening   of   the   Majlis,   an   opposition   bloc   named   Respublika   was   formed  largely  by  members  of  Azat,  Lad,  SP,  NC,  as  well  as  a  few  members  from  SNEK   and   from   the   state   list.   In   all,   the   opposition   group   controlled   at   least   40   percent   (69   seats)  of  the  parliament  and,  by  some  estimates,  a  slim  majority  (90  seats,  Bremmer  and   Welt   1996,   191).   Directing   its   criticism   toward   the   government,   the   members   of   Respublika  opposed  the  government’s  radical  economic  reforms.              The  strength  of  the  opposition  was  clear  when  members  of  Respublika  passed  a  vote   of   no-­‐confidence   against   Prime   Minister   Sergei   Tereshchenko,   who   was   a   champion   of   radical  economic  reforms,  with  a  majority  of  111  to  28  in  May  1994  (Olcott  2002,  103).   Since  the  Constitution  did  not  provide  for  a  vote  of  no-­‐confidence,  however,  Nazarbaev   declared   that   Tereshchenko   would   remain   as   prime   minister   until   privatization   was   completed.  This  event  drew  more  and  more  members  into  the  Respublika  camp,  and  led   to   the   formation   of   an   opposition   bloc   called   Otan-­‐Otechestvo,   which   called   for   Nazarbaev’s   resignation   (Olcott   2002,   103).   By   July   1994,   the   opposition   succeeded   in     120   overriding   Nazarbaev’s   veto   regarding   two   consumer   protection   bills.   Finally,   in   mid-­‐ October,   after   a   scandal   over   the   financial   improprieties   of   two   ministers   in   the   Tereshchenko   cabinet,   Nazarbaev   was   forced   to   accept   the   resignation   of   the   prime   minister  and  his  government  (Olcott  2002,  104).       4.3  Growing  Dominance  of  Ruling  Parties  Since  1995                Since   Nazarbaev   established   the   Second   Constitution   in   1995,   the   government   has   been  consolidating  its  authoritarian  rule.  This  section  will  briefly  describe  Kazakhstan’s   processes   to   a   stable   electoral   authoritarian   regime,   in   which  Nazarbaev   has  gradually   dominated  parliament  and  consolidated  authoritarian  rule.           On  March  1995,  Tatyana  Kvyatkovskaya,  a  journalist  and  a  failed  candidate  in  the   1994   election,   claimed   that   the   electoral   districts   had   been   disproportionately   drawn,   and   therefore,   legislators   represented   constituencies   of   vastly   differing   sizes.   She   also   charged  that  the  cross-­‐out  method  of  voting  enabled  the  vote  counters  to  accept  a  single   ballot  for  one  than  one  candidate  if  the  voter  had  marked  his  ballot  improperly.  Indeed,   in  several  districts  more  votes  were  recorded  than  the  number  of  voters  (Olcott  2002,   110;   Cummings   2005,   26).   Consequently,   the   Constitutional   Court   declared   the   whole   1994  parliamentary  elections  illegal  and  ordered  the  parliament  to  be  dissolved.              Paradoxically,  her  appeal  to  the  court  provided  Nazarbaev  with  an  opening  to  redraft   the   Constitution   and   avoid   presidential   elections.   While   the   parliament   had   been   dissolved,   he   held   two   nationwide   referenda   to   strengthen   his   rule.   In   April   1995,   the   first   referendum   was   on   Nazarbaev’s   terms   in   the   office.   At   that   time,   91.3   percent   of     121   voters   turned   out,   of   whom   95.8   percent   voiced   their   support   for   extending   his   presidential   term   until   2000   (Olcott   2002,   111).   In   a   second   referendum   held   at   the   end   of   August   of   the   same   year,   Nazarbaev   put   to   vote   an   entirely   new   Constitution,   the   result  of  which  further  bolstered  the  president’s  power.              The  Second  Constitution  arguably  expanded  presidential  power  at  the  expense  of  the   legislature   and   judiciary.   First,   the   unicameral   legislature   was   transformed   into   a   bicameral  system,  composed  of  the  Senate  and  the  Majlis  (the  Lower  House),  and  both  of   them   would   be   in   the   hands   of   Nazarbaev   and   his   inner   circle.   In   the   Senate,   seven   senators  would  directly  be  appointed  by  the  president,  and  the  rest  by  the  oblast  council   that   was   controlled   by   Nazarbaev.   The   new   Constitution   also   stipulated   that   Majlis   would   not   be   able   to   initiate   legislation   and   would   have   to   terminate   parliamentary   deliberation   on   a   bill   within   a   month,   seriously   undermining   the   legislative   power.   Second,   the   Constitutional   Court   was   abolished   and   replaced   with   a   Constitutional   Council,   where   the   president,   the   senate,   and   the   assembly   each   would   select   two   members   to   form   a   total   of   six   members.   All   possibility   for   the   Council   to   display   independence  was  nullified  by  a  simple  tool,  a  presidential  veto.52                                                                                                                               52  For  more  detailed  description  on  the  Second  Constitution,  for  example,  see  Bremmer   and  Welt  (1996:  193)  and  Cummings  (2000).       122   0 Proportion of Seats in Legislature (%) 20 40 60 80 100 Figure  4.1:  Increasing  Dominance  of  Regime  Parties  in  Kazakhstan  (1994-­‐2007)     1994 1995 1999 2004 2007 Election Year Ruling Parties Opposition Parties Sources:  Olcott  (2002)  and  Inter-­‐Parliamentary  Union  (http://www.ipu.org/parline-­‐ e/parlinesearch.asp)              Since  the  Second  Constitution  was  adopted,  ruling  parties  would  gradually  dominate   the   parliament   by   occupying   more   seats   all   the   way   up   to   2007   (See   Figure   4.1).53  When   parliamentary   elections   finally   took   place   on   December   9,   1995,   “candidates   were                                                                                                                   53  In   Figure   4.1,   I   show   time-­‐series   change   in   seat   shares,   rather   than   vote   shares   of   ruling  and  opposition  parties.  Detailed  data  on  vote  shares  are  not  available  for  the  1994   and  1995  elections,  so  for  the  purpose  of  comparison  between  the  elections,  the  data  of   seat  shares  are  more  useful.  One  may  think  that  seat  shares  are  easily  manipulated  by   electoral  systems.  As  I  argue,  this  is  also  the  case  for  Kazakhstan.  Yet,  even  under  SMD   systems   that   bring   seat   premiums   to   ruling   parties   (Chapter   3;   Pripstein   Posusney   2002),   ruling   parties   in   Kazakhstan   could   not   be   sufficiently   dominant   in   the   late   1990s,   suggesting  that  they  were  still  much  weaker  than  those  in  the  2000s.             123   arbitrarily   banned;   Russians   were   underrepresented   on   the   candidate   list;   and   Nazarbaev   supporters   dominated”   (Bremmer   and   Welt   1996,   193).   As   a   result,   many   opposition   parties   decided   to   boycott   the   election   due   to   the   accusations   of   electoral   fraud   (Bremmer   and   Welt   1996;   Oka   2003,   474).   The   proportion   of   seats   held   by   pro-­‐ presidential   parties   in   the   Lower   House   gradually   rose.   In   the   1995   legislative   elections,   the  People’s  Unity  Party  (PNEK)54  and  the  Democratic  Party  (DP)  won  52  percent  of  all   seats,  which  was  more  than  two  times  higher  than  their  seats  in  the  1994  election.              Between  the  1995  and  1999  elections,  ruling  parties  experienced  some  realignment.       The   Otan   party   was   established   to   support   Nazarbaev’s   electoral   campaigns   in   the   1999   presidential   elections   as   a   result   of   coalescing   between   pro-­‐presidential   parties   like   PNEK,   DP,   and   others.   Two   other   new   pro-­‐presidential   parties,   the   Civil   Party   of   Kazakhstan   and   the   Agrarian   Party   were   also   created   before   the   1999   parliamentary   elections.   Yet,   in   the   1999   election,   the   ruling   parties   (Otan,   Civil   Party,   and   Agrarian   Party)  occupied  only  the  same  total  of  seats  as  in  the  post-­‐1995  parliament  (49  percent   of  all  seats).              In  2003,  Dariga  Nazarbaeva,  the  president’s  daughter,  and  her  husband  Rakhat  Aliev   launched  a  new  pro-­‐presidential  party,  Asar.  After  the  Civil  Party  made  a  pre-­‐electoral   coalition   called   the   Aist   bloc   with   the   Agrarian   Party,   three   ruling   parties   joined   prior   to   the   2004   parliamentary   elections   (Otan,   Asar   and   the   Aist   bloc).   Although   the   ruling   parties   suffered   mutual   competition   within   and   between   the   parties   in   each   electoral   district,  they  dramatically  increased  their  seats  in  the  2004  elections  (75  percent).                                                                                                                     54  The  previous  name  of  this  party  is  SNEK.         124            In   September   2006,   Otan   merged   with   Asar.   Three   months   later,   the   Civil   Party   and   the   Agrarian   Party   announced   that   they   would   also   join   the   Otan   party,   leading   to   the   foundation   of   the   dominant   party,   Nur   Otan.   At   the   same   time,   Nazarbaev   proclaimed   that   he   would   serve   as   the   party   leader   of   Nur   Otan.   And,   in   the   August   2007   parliamentary  election,  the  newly  established  dominant  party  occupied  all  of  the  seats  in   the   Lower   House.   This   had   completed   the   processes   of   gradual   dominance   of   the   legislature  by  ruling  parties  in  the  country.                On  the  other  hand,  opposition  parties  had  gradually  lost  their  presence  in  parliament.   In  the  1994  elections,  they  obtained  as  many  seats  as  the  ruling  parties  (24  percent).  Yet,   they   had   lost   their   seats   from   24   percent   to   10   percent   in   the   1995   parliamentary   elections.  In  the  1999  legislative  elections,  two  main  opposition  parties,  the  Communist   Party   (led   by   the   former   presidential   candidate,   Serikbolsyn   Abdildin)   and   the   Republican   People’s   Party   (led   by   the   former   prime   minister,   Akezhan   Kazhegeldin)   obtained  only  3  and  1  seats,  respectively,  resulting  in  further  reduction  of  opposition’s   political   influence   in   the   parliament.   In   the   2004   parliamentary   elections,   the   Ak   Zhol   party,  which  was  created  as  a  result  of  the  strong  opposition  party  Democratic  Choice  of   Kazakhstan   movement   starting   in   2002,   obtained   a   single   seat   in   the   parliament   although  there  were  several  opposition  parties  participating  in  the  elections.  In  the  2007   parliamentary   elections,   which   was   the   country’s   first   elections   under   a   pure   proportional   representation   system   with   7   percent   electoral   threshold,   all   opposition   parties  failed  to  obtain  sufficient  number  of  votes  to  make  it  into  the  parliament.55                                                                                                                                 55  The   very   similar   change   in   electoral   systems   was   taken   also   by   Putin’s   Russia   and   Bakiev’s  Kyrgyzstan.       125   4.4  Electoral  Manipulation  in  Kazakhstan56                  As   discussed,   party   system   formation   in   Kazakhstan   has   been   the   process   by   which   ruling   parties   gradually   dominated   the   parliament   and   opposition   parties   incrementally   lost   their   political   influence.     The   electoral   dominance   of   the   ruling   parties,   however,   cannot   be   simply   attributed   to   the   fact   that   the   authoritarian   government   relied   exclusively   on   electoral   fraud   and   state   repression.   Nor   can   it   be   solely   explained   by   the   fact   that   the   president   just   maintained   electoral   systems   in   favor   of   ruling   coalitions.   Kazakhstan   is   an   electoral   authoritarian   regime,   and   its   elections   still   fall   short   of   international   standards   for   democratic   elections.   Although   the   government   sometimes   resorted  to  violent  measures  to  repress  opposition  groups,  the  regime  has  become  less   contingent   on   repressive,   fraudulent   tools   at   elections   to   win   big.   Moreover,   the   president   even   changed   the   country’s   electoral   system   from   Single-­‐Member   Districts   (SMD)   based   systems   (1994-­‐2006)   to   a   pure   Proportional   Representation   system   (2007),   which   resulted   in   bringing   a   smaller   seat   premium   to   the   president.   As   Schatz   (2009)   argues,   characteristics   of   the   Nazarbaev   regime   can   be   more   accurately   depicted   as   a   “soft   authoritarian   regime”   rather   than   a   highly   repressive,   closed   authoritarianism   like   neighboring   Uzbekistan   and   Turkmenistan.   Schedler   (2013,   4-­‐5)   also   refers   to   Kazakhstan  as  one  of  the  typical  electoral  authoritarian  regimes.  Here,  I  show  evidence                                                                                                                                                                                                                                                                                                                                                                 56  Here,   by   electoral   manipulation,   I   refer   to   1.   electoral   fraud   (election   violence,   electoral  cheating,  and  the  legal  problems  on  the  electoral  law)  and  2.  electoral  system   change.         126   that   the   government   became   less   dependent   on   fraud   and   manipulation   of   electoral   systems  at  least  in  comparison  to  its  reliance  on  such  measures  in  the  1990s.       4.4.1  Electoral  Fraud   Table  4.1:  Time-­‐Series  Change  in  Electoral  Fraud  in  Kazakhstan  (1995-­‐2007)     Note:   Each   election   is   accessed   by   Kelley’s   (2012)   Quality  of  Elections  Dataset   and   the   author’s   own   coding   for   the   2005   and   2007   elections.   Each   component   has   four   scales   (0:   no   problems,   1:   minor   problems   only,   2:   moderate   problems,   3.   serious   problems).   To  assess  overall  electoral  fraud,  I  aggregate  each  component.       Source:  Kelley  (2012)  and  the  Department  of  State’s  Country  Report              Hyde  and  Marinov’s  (2012)  NELDA  dataset  tells  us  that  all  the  elections  in  Kazakhstan   since  independence  were  “competitive”  in  the  sense  that  opposition  parties  and  multiple   candidates  were  allowed  to  participate  in  the  elections,  suggesting  that  the  country  is  an   electoral   authoritarian   regime.   Based   on   Kelley’s   (2012)   Quality   of   Elections   Dataset   (1995-­‐2004)  and  my  original  coding  for  the  2005  presidential  and  2007  parliamentary   elections,57  Table   4.1   shows   time-­‐series   changes   in   the   extent   of   electoral   fraud   in   parliamentary  and  presidential  elections  between  1995  and  2007.  The  level  of  electoral   fraud   in   the   1995   parliamentary   elections,   as   the   elections   after   limited   liberalization,   is                                                                                                                   57  I   follow   coding   schemes   that   Kelley   (2012)   refers   to   and   code   the   two   elections   by   using  the  US  Department  of  State’s  Country  Report.           127   relatively   lower   (6)   than   the   1999   elections   that   faced   the   most   serious   electoral   manipulation  (9).  Since  1999,  the  extent  of  electoral  fraud  became  gradually  less  serious   over  time  until  2007,  when  we  define  electoral  fraud  in  the  forms  of  electoral  violence,   election  cheating  and  the  legal  problems  regarding  the  electoral  law.                The   1999   elections   faced   the   most   extensive   electoral   manipulation   during   the   time   period   under   study.58  The   elections   were   minimally   competitive   in   the   sense   that   10   political   parties   and   547   candidates   were   registered   for   the   single-­‐mandate   constituencies.   The   OSCE   report,   however,   concludes   that   elections   were   exposed   to   severe   electoral   fraud   in   many   respects.   First,   electoral   cheating   was   widely   observed   during   election   campaigns. International   election   observers   found   numerous   practices   of  proxy  voting  among  the  regime  supporters  especially  in  rural  areas  (OSCE  2000,  16).   Official  election  activities  were  often  mingled  with  election  campaigns  carried  out  by  the   Otan   party   (OSCE   2000,   13).   Unfair   campaign   practices   by   ruling   parties   were   closely   associated   with   the   media   being   dominated   by   the   government.   Khabar,   a   large,   state-­‐ controlled   TV   station   operated   by   Dariga   Nazarbaeva,   played   an   important   role   in   biasing   the   tones   of   broadcasting   in   favor   of   the   ruling   parties.   Otan,   which   obtained   only  31  percent  of  seats  in  the  1999  parliamentary  election,  enjoyed  nearly  60  percent   of   the   media   coverage   (OSCE   2000,   14).   The   electoral   bias   was   also   brought   through   lower   level   election   commissions.   Election   commissions   had   become   more   effective   in   presiding  over  election  processes,  compared  with  the  past  elections.  But  their  members                                                                                                                   58  Below,  I  primarily  use  OSCE  final  reports  on  elections,  which  are  also  often  referred  by   the   US   Department   of   State’s   country   reports,   to   qualitatively   assess   changes   in   the   extent  of  electoral  fraud  in  Kazakhstan  (1995-­‐2007).  The  1994  legislative  elections  are   not  included  because  both  reports  are  not  available  for  the  elections.             128   were   overrepresented   by   the   ruling   parties.   70   percent   of   polling   stations   were   positively   assessed   for   voting   procedures   and   less   than   half   of   polling   stations   were   considered   as   fair   in   counting   procedures   (OSCE   2000,   17-­‐18).   Yet,   these   evaluations   were   much   worse   than   the   elections   during   the   2000s,   as   I   will   discuss   later.   Another   irregularity  in  the  elections  was  that  there  was  widespread  non-­‐violent  intimidation  of   the   opposition.   The   former   Prime   Minister   Akezhan   Kazhegeldin,   who   stated   his   intention  to  stand  for  the  presidential  elections  on  the  basis  of  his  high  popularity,  was   disqualified   as   a   presidential   candidate   and   was   detained   in   Moscow   on   the   charges   of   a   dubious   criminal   allegations   and   money   laundering   (Issacs   2011,   68).   As   a   result,   his   new   opposition   party,   Republican   People’s   Party,   was   forced   to   withdraw   from   the   party-­‐list   PR   part   of   the   election   and   ended   up   obtaining   only   one   seat   in   one   single-­‐ member  district  in  the  parliamentary  election.                    Second,   the   government   and   ruling   parties   were   highly   repressive   from   the   mid   to   late   1990s.   During   the   period,   opposition   parties   were   often   violently   intimidated   and   prevented   from   carrying   out   effective   electoral   campaigns.   For   example,   during   the   election   campaigns,   one   of   Kazhegeldin’s   secretaries   was   suddenly   beaten   up   by   unknown   assailants.   Kazhegeldin   himself   was   also   the   subject   of   an   assassination   attempt,  although  the  assassination  attempt  failed  (Furman  2006,  228).    According  to  a   political  scientist,       These   5   years   [1996-­‐2001]   were   really   difficult,   really   nightmare  for  the  opposition  in  Kazakhstan.  This  was  a  time  of   stagnation.   My   interviewees   all   claimed   that   if   they   wanted   to   organize   protests   in   the   1999   presidential   elections,   the   activists’   doors   of   flats   were   not   able   to   open.   This   is   how   the     129   regime   was   intimidating   the   opposition,   there   were   threats,   and   they   [regime   parties]   were   even   throwing   drugs   to   opposition   figures’  houses  in  order  to  be  able  to  prosecute  them.  This  menu   of   intimidation,   a   tool   to   crack   down   on   the   opposition,   was   so   huge   at   that   time   that   the   regime   was   extremely   repressive   in   the   end   of   the   1990s   (Author’s   interview   with   a   political   scientist  [#11]).              Likewise,  looking  back  on  the  1990s,  an  opposition  political  activist  states,         He  [Nazarbaev]  was  using  administrative  resources  to  threaten   political   movements   on   behave   of   the   government.   (…)   There   were   intimidations,   not   allowing   to   organize   party   meetings   with   voters.   At   that   time,   it   wasn’t   always   directly   against   opposition   parties,   but   sometimes   they   employed   repressive   measures.   I   remember   how   they   switched   off   electricity   in   a   building   where   we   were   gathering.   They   were   preventing   opposition   parties   and   groups   from   campaign   activities.   They   had  a  variety  of  methods  to  do  that.  (Author’s  interview  with  a   political  activist  [#20];  italic  by  the  author)                In   the   middle   of   the   2000s,   when   Kazakhstan   held   three   national   elections   (2004   parliamentary   elections,   2005   presidential   elections   and   2007   parliamentary   elections),   elections  became  less  fraudulent,  at  least  compared  to  those  of  the  late  1990s.  Looking   at  Table  4.1,  total  scores  of  electoral  fraud  in  the  2004,  2005  and  2007  elections  are  7,  7,   and   5,   respectively.   First   of   all,   the   parliament   had   adopted   the   amendments   of   the   Electoral   Law   on   March   16,   2004,   which   was   assessed   by   the   OSCE   as   a   considerable   progress,   even   though   further   improvements   had   to   be   realized   in   order   to   meet   the   international  criteria  on  democratic  elections  (OSCE  2004a,  5).  The  legal  framework  in   the   subsequent   2007   parliamentary   elections,   on   the   other   hand,   did   not   see   any     130   significant  progress  in  further  improvement  of  the  2004  Electoral  Law  (OSCE  2007,  7).59   Yet,   the   legal   framework   in   2007   was   no   worse   than   that   in   1999.     The   Quality   of   Elections  Data   reflects   the   improvements   in   the   Electoral   Law   by   evaluating   the   2004   and   2007   parliamentary   elections   as   2   (intermediate   fraud)   in   the   section   of   “legal   problems   on   electoral   law,”   compared   to   3   (major   fraud)   in   the   1999   parliamentary   elections  (see  also  OSCE  2004b,  1;  OSCE  2007,  7)  for  the  same  section.              Of   the   many   improvements   in   the   2004   Electoral   Law,   important   ones   to   prevent   ruling   parties   from   relentlessly   using   fraud   were   (1)   the   prohibition   of   undue   interference   in   the   work   of   the   election   commissions   by   the   authorities, 60  (2)   the   prohibition   of   the   presence   of   unauthorized   persons   in   polling   stations,   (3)   more   access   by   observers   to   the   entire   election   processes   and   the   receipt   of   relevant   election   documents,   (4)   posting   of   election   results   protocols   in   precinct   and   district   election   commissions   for   public   scrutiny,   (5)   stronger   efforts   to   provide   equal   conditions   for   election   contestants   during   the   election   campaign,   (6)   procedures   for   compilation   and   verification   of   the   accuracy   of   voter   lists,   and   (7)   expansion   of   the   list   of   prohibited   activities  that  could  interfere  with  the  election  process  (OSCE  2004a,  5-­‐6).    In  the  2007   parliamentary   elections,   there   was   also   some   improvement   in   the   Electoral   Law.   For   instance,  in  the  previous  elections,  the  Law  banned  public  meetings  between  the  end  of   electoral  campaigns  and  the  publication  of  final  results,  but  this  was  removed  from  the                                                                                                                   59  OSCE   reports   “despite   a   constructive   dialogue   with   the   authorities   since   2006,   recommendations   to   improve   the   legal   framework   made   by   the   OSCE/ODIHR   were,   in   general,  not  addressed  by  the  amendment  of  June  2007”     60  The  interference  of  local  authorities  to  electoral  commissions  in  each  oblast  was  one   of  the  most  serious  problems  in  the  1999  elections  (OSCE  2000,  2)       131   Electoral   Law   amended   in   December   2006.   Yet,   significant   parts   of   the   Law   were   not   improved   in   accord   with   the   international   standard   of   democratic   elections.   The   Electoral  Law  still  denied  the  “suffrage  right  of  a  citizen  ‘who  has  a  prior  conviction  not   cancelled  or  withdrawn,’  regardless  of  the  seriousness  of  the  crime”  (OSCE  2007,  7).  The   revised   Constitution   in   May   2007   added   a   provision   stating   that   eligible   electoral   candidates   must   have   been   permanently   resident   in   Kazakhstan   for   ten   years.   “Such   a   long  time  period  can  constitute  an  unreasonable  restriction  on  the  right  to  seek  public   office”  (OSCE  2007,  8).                      Improvements   on   paper   in   the   Electoral   Law   did   not   necessarily   mean   that   pro-­‐ presidential   parties   and   the   government   abided   by   the   Electoral   Law   without   any   infringements.   In   fact,   there   were   considerable   level   of   pre-­‐electoral   and   election-­‐day   cheating   practices   in   the   2004,   2005,   and   2007   elections.   Even   so,   the   Quality   of   Elections   Data   and   the   US   Department   of   State   country   reports   suggest   that   the   substantial   level   of   electoral   cheating   also   tended   to   become   lower   over   time   (total   cheating   scores   of   the   2004,   2005,   2007   elections   are   5,   4,   3,   respectively).   Media   was   still  significantly  biased  toward  the  ruling  parties  in  all  the  elections  and  far  from  equal   media  accesses  stipulated  in  the  Electoral  Law.  Yet,  in  all  three  elections,  there  were  no   reports  that  media  outlets  were  shut  down  or  journalists  were  prosecuted,  as  happened   in   the   past   elections   (OSCE   2004b,   15).   It   suggests   that   the   governments   had   become   less  inclined  to  resort  to  intimidation  against  the  opposition  media.  Moreover,  the  media   provided  all  party  leaders  with  more  chances  to  debate  political  issues  during  electoral   campaigns   (OSCE   2004b,   15).   In   this   respect,   the   OSCE’s   analysis   of   media   coverage   provides   evidence   suggesting   that   the   dominance   of   media   by   ruling   parties   tends   to   be     132   less   heavy   in   the   elections   in   comparison   to   the   previous   elections.   In   the   2004   parliamentary   elections,   the   major   state   channel,   Kazakhstan   TV   (Kazakhstan-­‐1),   devoted  64  percent  of  its  political  news  coverage  to  Otan  and  9  percent  to  the  rest  of  the   ruling  parties,  the  Aist  bloc  and  Asar.  Likewise,  the  other  well-­‐known  TV  station,  Khabar,   aired   44   percent   and   31   percent   of   political   news   favorable   for   Asar   and   Otan,   respectively   (OSCE   2004b,   14-­‐15).   In   the   2007   parliamentary   elections,   the   newly   established  dominant  party,  Nur  Otan  was  covered  by  only  20  percent  on  TV  Khabar  and   17  percent  on  Kazakhstan  TV  (OSCE  2007,  18).  Different  from  the  2004  elections,  these   media   outlets   broadcast   political   news   about   the   government   and   the   president   (30   percent  in  total),  which  generated  additional  bias  in  favor  of  the  dominant  party,  given   that   the   dominant   party   is   heavily   fused   with   the   government,   as   I   discuss   later.   Yet,   even  with  that  in  mind,  total  media  coverage  for  the  dominant  party  was  still  about  50   percent.   Considering   that   Nur   Otan   is   a   far   larger   party   in   size   and   has   stronger   organizational   bases   than   Otan   and   Asar,   the   difference   in   media   coverage   suggests   that   the  government  seems  to  have  refrained  from  relying  exclusively  on  media  bias  to  make   election  results  favorable  to  ruling  parties  even  though  the  media  is  still  biased.                All  the  three  elections  were  more  positively  assessed  by  the  OSCE  in  both  voting  and   counting  procedures,  at  least  compared  to  those  of  the  1999  elections.  The  proportion  of   polling   stations   where   voting   procedures   were   positively   assessed   was   87   (the   2004   elections),  92  (the  2005  elections),  and  94  percent  (the  2007  elections),  which  were  all   higher  than  the  70  percent  in  the  1999  elections  (OSCE  2000,  2004b,  2005,  2007;  Bader   2012,   53).   On   the   other   hand,   percentage   of   polling   stations   where   counting   procedures   were   positively   assessed   was   72   percent   (2004),   72   percent   (2005),   and   61   percent     133   (2007).   Although   the   2007   elections   were   a   bit   worse   than   the   rest   of   the   two   elections,   they  were  still  better  than  the  1999  elections  in  which  more  than  half  of  polling  stations   were  evaluated  as  negatively  in  terms  of  vote  counting  (OSCE  2000).                  Participation   and   registration   of   the   main   opposition   parties   were   highly   smooth   and   did   not   seem   to   face   serious   problems.   It   stands   in   stark   contrast   with   Akezhan   Kazhegeldin's  Republican  People’s  Party,  an  opposition  party  in  the  1999  elections  that   was   exposed   to   various   obstacles   in   electoral   participation.   It   was   true   that   during   election  campaigns  the  governments  prevented  the  main  opposition  from  carrying  out   effective  campaigns,  but  such  cases  were  very  rare.  Activists  from  the  Democratic  Choice   of  Kazakhstan  (DCK)  were  detained  by  police  only  three  times  in  the  Almaty  city  and  the   Pavlodar   oblast   and   local   authorities   inhibited   opposition   parties   from   holding   campaign   events   in   only   several   instances   (OSCE   2004,   11).   According   to   one   of   my   interviews   with   a   political   analyst,   the   Ak   Zhol   party   had   actually   obtained   25-­‐30   percent   of   total   votes,   but   the   official   election   result   was   just   12.8   percent,   indicating   electoral   falsification   by   the   government   (Author’s   interview   with   a   political   analyst   [#8]).  Yet,  in  contrast  to  the  Republican  People’s  Party,  a  strong  opposition  party  in  the   1999  elections,  opposition  parties  were  not  exposed  to  severe  restrictions  on  electoral   participation.   Main   opposition   parties   like   DCK   and   the   Communist   Party   in   the   2004   elections,  and  the  All  National  Social  Democratic  Party  and  Naghyz  Ak  Zhol  in  the  2007   elections,  were  all  registered  without  any  problems.    There  were  no  complainants  from   the   opposition   parties,   as   well   as   no   appeals   filed   concerning   the   registration   process   (OSCE  2004b,  10;  OSCE  2007,  14).       134            Finally,   extensive   electoral   violence   against   opposition   parties   was   no   longer   observed   by   the   2004   parliamentary   elections.61     Although   election-­‐time   violence   was   exercised   toward   presidential   candidate   Zharmakhan   Tuyakbay   during   the   campaigns   of  the  2005  presidential  elections,  this  was  the  last  case  of  such  violence.  Since  the  2007   elections   the   government   has   not   resorted   to   violent   measures   or   intimidated   the   opposition  parties  (Author’s  interview  with  an  opposition  activist  [#20]).         4.4.2  Electoral  System  Change:  From  SMD  to  PR   Table   4.2:   Electoral   System   Change   and   Effective   Electoral   Threshold   in   Kazakhstan        Note:  In  the  1994  elections,  24  percent  of  lawmakers  in  the  Lower  House  were  selected   by  the  president  via  a  “state  list.”                                                                                                                               61  This   does   not   mean   that   the   government   did   not   use   violent   repression   during   non-­‐ election   time.   The   regime   has   sometimes   used   violent   repression   and   intimidation   against   opposition   leaders   and   arrested   them   even   though   they   tended   to   de   detained   for   relatively   short   period   of   time.   Yet,   repression   was   not   severe   as   to   label   the   country   as  “closed  authoritarianism.”  Furthermore,  electoral  violence  has  been  clearly  reduced,   suggesting  that  the  government  may  be  reluctant  to  use  overt  repression  during  election   time  when  the  international  community  closely  looks  at  the  country’s  situations.             135     0 Effective Electoral Threshold (%) 10 20 30 40 Figure   4.2:   Effective   Electoral   Thresholds   in   Kazakh   Parliamentary   Elections   (1994-­‐2007)   1994 1995 1999 Year   2004 2007            Meanwhile,   the   Kazakh   government   has   changed   electoral   systems   in   the   Lower   House   elections   a   couple   of   times   since   independence,   shifting   electoral   systems   from   SMD   (high   Effective   Electoral   Threshold)   to   more   PR-­‐based   systems   (low   Effective   Electoral   Threshold;   Table   4.2   and   Figure   4.2).   As   discussed,   the   First   Constitution,   which   was   adopted   in   1993,   provided   a   177-­‐seat   parliament   in   which   135   lawmakers   were  elected  based  on  single-­‐member  districts  and  the  remaining  42  were  selected  by   the   president   without   electoral   competition   through   a   “state   list.”   In   September   1995,   the  president  issued  a  presidential  decree  on  the  Electoral  Law  in  which  the  number  of   seats   in   the   Lower   House   was   reduced   to   67   and   all   the   lawmakers   would   be   elected     136   under   SMD.   In   May   1999,   Nazarbaev   changed   electoral   rules   by   introducing   a   mixed-­‐ member   majoritarian   system   in   which   67   would   be   elected   in   single-­‐mandate   constituencies   while   the   rest   of   the   ten   members   would   be   determined   via   a   proportional  representation  system  in  a  nationwide  district   through   a   party   list   and   a   7   percent  electoral  threshold.  The  1999  and  2004  parliamentary  elections  would  be  held   under   this   new   electoral   system.     In   June   2007,   Nazarbaev   embarked   on   revising   the   Electoral   Law   and   initiated   a   big   change   in   Kazakhstan’s   electoral   system   subsequent   to   the   substantial   amendments   to   the   Constitution   in   the   previous   month.   In   this   amendment,   the   government   adopted   a   full   proportional   representation   system   by   a   party   list   with   the   same   7   percent   electoral   threshold.   In   sum,   effective   electoral   threshold  in  Kazakhstan  had  been  decreasing  since  1994  from  37.5  percent  (1994  and   1995;   SMD   with   a   state   list   for   the   1994   elections),   through   33.5   percent   (1999   and   2004;   mixed-­‐member   majoritarian   system),   to   7   percent   (2007;   a   pure   proportional   representation  system).                          In   the   1994,   1995,   and   1999   elections,   where   electoral   competition   with   opposition   parties   were   relatively   strong,   the   government   designed   elections   in   the   way   in   which   ruling  elites  would  be  able  to  keep  a  dominant  position  in  the  legislature.  Selecting  42   legislators  via  no  electoral  competition  (in  the  1994  parliamentary  elections)  and  more   serious   electoral   manipulation   (especially   in   the   1999   elections)   did   contribute   to   preventing   a   landslide   victory   for   the   opposition   camp   during   this   period   (Author’s   interview  with  a  political  analyst  [#4]).              More   importantly,   majoritarian   systems   significantly   protected   the   president   from   being   defeated   by   opposition   parties   in   Kazakhstan,   as   my   theory   and   cross-­‐national     137   empirical  analysis  in  Chapter  3  demonstrated  (see  also  Pripstein  Popsusney  2002).  First,   the   SMD-­‐based   systems   enabled   the   president   to   delegate   the   collection   of   political   support   to   popular   local   ruling   elites   in   small   electoral   districts,   who   ran   for   the   elections   as   independents   but   basically   supported   the   president.   Many   of   them   held   important  positions  in  local  governments  and  were  able  to  mobilize  regime  support  by   utilizing   their   “administrative   resources”   and   intensive   support   from   akims   (local   governors;   see,   e.g.   Issacs   2011,   88).   Second,   owning   to   the   seat   premium   under   SMD   systems,   the   ruling   elites   side   received   more   seats   than   vote   shares.   Although   vote   shares  for  each  party  and  candidate  in  the  1994  elections  are  not  formally  documented   and  published,  the  Central  Election  Commission  of  Kazakhstan  publishes  vote  shares  of   elected   candidates   in   each   electoral   district   in   the   1995   elections,   which   enables   us   to   estimate   to   what   extent   ruling   elites   enjoyed   the   seat   premiums   (Tsentralnaya   Izbiratelnaya   Komissiya   Respubliki   Kazakhstan   2010).   According   to   the   data,   average   vote   shares   of   elected   candidates   were   61   percent   in   the   1995   elections.   On   the   other   hand,  the  ruling  camp  including  pro-­‐regime  independents  occupied  about  80  percent  of   seats  in  parliament,  suggesting  that  the  government  was  enjoying  an  approximately  19   percent  seat  premium  under  the  SMD  system  in  the  elections.                Also,   in   the   1999   parliamentary   elections,   the   SMD   part   of   the   electoral   system   provided   a   large   seat   premium   to   ruling   elites.   Ruling   parties   and   independents   obtained  80.6  percent  of  seat  shares  with  61.7  percent  of  vote  shares.  On  the  other  hand,   the   main   opposition   party,   the   Communist   Party   of   Kazakhstan,   obtained   only   3.9   percent  of  seat  shares,  even  though  the  party  scored  17.7  percent  of  total  votes  (Nohlen,     138   Grotz   and   Hartmann   2001,   420-­‐423).   This   indicates   that   SMD   systems   undercounted   opposition  parties’  votes  when  translating  votes  into  seats  in  parliament.              In   the   2004   elections,   the   ruling   parties   and   pro-­‐regime   independents   became   stronger   and   increased   their   vote   shares   to   79   percent   under   the   SMD   system, 62   resulting  in  the  occupation  of  all  of  seats  determined  under  SMD.  In  the  2004  elections   as  well,  ruling  parties  had  received  21  percent  more  seat  shares  than  their  vote  shares,  a   considerably  large  seat  premium.                After   the   adoption   of   the   PR   system   prior   to   the   2007   legislative   elections,   the   gap   between  shares  of  votes  and  seats  decreased  significantly.  This  suggests  that  PR  systems   bring  smaller  seat  premiums  toward  the  governing  parties,  as  I  showed  cross-­‐nationally   in  Chapter  3.  Nur  Otan  scored  an  overwhelming  electoral  victory  with  88.41  percent  of   total   votes,   having   obtained   all   of   the   seats   determined   by   the   elections   (98   seats).63   Even   under   the   pure   proportional   representation   system,   the   dominant   party   enjoyed   a   favorable  bias  toward  its  seat  shares  (11.59  percent).  Yet,  the  bias  was  half  as  big  as  that   in   the   past   elections.   Furthermore,   as   evidenced   by   its   vote   shares,   the   government   significantly   increased   its   mobilization   power.   The   gap   in   this   new   electoral   system   derived   more   from   the   fact   that   the   other   opposition   parties,   be   they   radical   and   moderate,   could   not   clear   the   7   percent   electoral   threshold   that   had   been   in   effect   since   1999  and  were  cleared  by  some  opposition  parties  in  the  1999  and  2004  elections.                                                                                                                       62  The   calculation   is   based   on   district-­‐level   election   data   from   Nurmukhamedov   and   Chebotarev  (2005,  47-­‐49).  As  I  could  not  find  vote  shares  of  the  second  round  voting,  I   use  proportions  of  vote  shares  obtained  by  the  three  regime  parties  and  independents.       63  The   figure   was   taken   from   the   website   of   the   Central   Election   Commission   of   the   Republic  of  Kazakhstan.  (election.kz)       139   4.5  The  Dictator’s  Mobilization  Power              The   previous   sections   have   shown   that,   although   Nazarbaev   had   become   less   dependent  on  electoral  manipulations  -­‐-­‐  both  serious  electoral  fraud  and  the  adoption  of   single-­‐member  districts  systems  -­‐-­‐  between  1995  and  2007,  he  successfully  maintained   an  overwhelming  electoral  victory  and  kept  most  ruling  elites  in  line  with  the  regime.  As   numerous  extant  studies  on  politics  in  Kazakhstan  suggested,  there  is  little  doubt  that  he   continued   to   use   electoral   fraud   to   a   certain   extent.   Especially   in   the   late   1990s,   the   government  relied  heavily  on  the  disproportionality  feature  of  the  SMD  system,  as  well   as   extensive   electoral   cheating,   blatant   election   violence,   and   the   manipulation   of   the   electoral  law.  However,  given  that  the  level  of  electoral  fraud  gradually  declined  and  the   government   shifted   electoral   systems   from   a   pure   SMD   system   to   a   full   PR   system,   electoral   manipulation   alone   cannot   explain   the   ruling   parties’   electoral   dominance   in   Kazakhstan.   This   poses   important   puzzles   in   authoritarian   elections:   Why   did   the   president   become   less   reliant   on   electoral   fraud?   Why   did   he   give   up   the   seat   premiums   under  the  SMD  system  and  decide  to  adopt  the  PR  system  in  2007?64                                                                                                                       64  One   alternative   explanation   might   be   that   because   Kazakhstan   had   been   already   a   “closed   authoritarian   regime”   by   the   late   2000s   in   which   opposition   forces   had   been   already   severely   repressed   by   the   government,   government   no   longer   needed   to   rig   elections.   Given   the   fact   that   the   government   used   some   repressive   measures   in   the   2000s,   repressed   opposition   might   influence   the   government’s   calculus   over   electoral   manipulation  and  weak  opposition.  Yet,  as  many  researchers  point  out,  Kazakhstan  is  far   from  highly  repressive  authoritarian  regimes  like  Uzbekistan  and  Turkmenistan  where   oppositions  are  barred  and  completely  excluded  (e.g.  Schatz  2009;  Hale  2015).  Elections   provide   at   least   multiple   options   other   than   ruling   parties,   which   is   an   important   criterion   to   define   electoral   authoritarianism   (Hyde   and   Marinov   2012;   Kinne   and   Marinov   2013).   In   this   circumstance,   in   addition   to   repression,   I   emphasize   that   opposition’s   strategic   failure   to   coordinate   their   action   before   elections   contributes   to     140          I  argue  that  an  answer  can  be  found  in  the  fact  that  President  Nazarbaev  successfully   enhanced  his  mobilization  power  to  gain  “voluntary”  support  from  the  citizenry.  From   1999   until   2007,   the   growing   mobilization   power   of   ruling   parties   can   be   well-­‐observed   by  comparing  share  of  votes  in  the  PR  portion  of  the  electoral  systems  in  the  1999,  2004,   and   2007   legislative   elections.   As   the   PR   part   is   under   a   nation-­‐wide   electoral   district   without   any   gerrymandering,   vote   shares   are   relatively   comparable.   That   the   mobilization  power  of  ruling  parties  at  the  time  of  the  1999  parliamentary  elections  was   still   weak   is   well-­‐illustrated   by   the   fact   that   under   the   PR   portion,   vote   shares   of   the   regime  parties  were  smaller  (10  seats).  They  obtained  only  54.75  percent  of  total  votes   under  the  PR  system  (Ashimbaev  2008,  840):  Otan  scored  just  30.89  percent,  whereas   the  other  two  ruling  parties,  Agrarian  Party  and  Civil  Party  garnered  12.63  percent  and   11.23   percent,   respectively.   In   the   2004   parliamentary   elections,   the   three   regime   parties  (Otan,  Asar  and  Aist  Block)  obtained  79.06  percent  of  total  votes  under  the  PR   system  determining  10  seats,  suggesting  that  ruling  parties  significantly  enhanced  their   mobilization   power   by   about   25   percent   compared   to   the   1999   elections.   In   the   2007   legislative  elections,  Nur  Otan  achieved  an  overwhelming  victory  with  88.41  percent  of   total   votes,   suggesting   their   mobilization   power   had   increased   by   about   10   percent.   Given  my  findings  that  electoral  fraud  tended  to  be  less  serious,  growing  vote  shares  in   the  portion  of  PR  systems  indicate  that  the  president  had  attained  stronger,  nation-­‐wide   mobilization  power  between  1999  and  2007.                                                                                                                                                                                                                                                                                                                                                                                                         explaining   variations   in   electoral   manipulation   in   the   country.   Moreover,   other   than   opposition   strength,   there   are   other   important   factors   like   financial   resources   and   the   government’s   organizational   strengths   to   explain   electoral   manipulation.   I   also   look   at   the  government’s  strategies  to  buy-­‐off  people’s  support  through  such  “soft”  measures.                   141              How,   then,   did   Nazarbaev   succeed   in   improving   his   power   of   mobilization?   This   section   process-­‐traces   how   this   became   possible   by   focusing   on   the   president’s   mobilization  power  via  extensive  economic  distribution.  In  particular,  I  will  look  at  the   following   three   aspects   that   are   closely   related   to   the   president’s   mobilization   power:   (1)   natural   resource   wealth,   (2)   centralized   political   organizations,   and   (3)   opposition   strength.  I  argue  that  the  president  has  been  able  to  reduce  the  need  for  manipulating   elections   with   electoral   fraud   and   the   adoption   of   SMD   systems   because   he   could   rely   more   on   efficient   distribution   of   economic   favors   toward   (at   least)   some   portions   of   citizens.                                                                         142   4.5.1  Natural  Resource  Wealth  as  a  Source  of  Patronage     Figure  4.3:  Economic  Growth  and  Fiscal  Revenues  in  Kazakhstan  (1993-­‐2008)     12000   70   60   10000   50   8000   40   30   6000   20   4000   Growth   Revenue  (million  US$)   10   0   2000   -­‐10   0   -­‐20     Sources:  IMF  country  report  and  World  Development  Indicators.     Notes:   The   left   vertical   axis   stands   for   revenue   (million   US$).   The   right   vertical   axis   represents  for  economic  growth  (%).                       143   1000 2000 3000 (a) Oil-Gas Value per capita in Kazakhstan (1985-2008) 0 Oil-Gas Value per capita (2000 Constant USD) Figure  4.4:  Natural  Resource  Wealth  in  Kazakhstan  (1985-­‐2008)   1985 1990 1995 2000 2005 Year Oil Price (US Dollars) 0 10 20 30 40 50 60 70 80 (b) International Oil Price (Constant 2000 USD, 1985-2008) 1985 1990 1995 2000 2005 Year   Note:   Oil-­‐gas   value   per   capita   is   calculated   by   multiplying   the   country’s   total   oil-­‐gas   production  by  the  constant  2000  oil-­‐gas  price  and  dividing  it  by  total  population.   Source:  Ross  (2011).                Similar   to   the   other   post-­‐communist   countries,   Kazakhstan   had   suffered   serious   economic  decline  during  the  first  years  following  its  independence.  In  order  to  recover   from   recession,   the   Kazakh   government   adopted   a   major   economic   strategy   to   liberalize   its   trade   policy   and   become   part   of   the   international   market.   Particularly   after   1995,   when   most   firms   except   for   large   corporations   had   been   privatized,65  the   country   began   to   lower   tariffs   and   export   natural   resources   such   as   oil,   gas   and   minerals   at   higher   volumes.  Subsequent  to  1999,  when  the  international  oil  price  rapidly  increased,  natural                                                                                                                   65  On  the  processes  of  economic  reform  in  Kazakhstan,  see  Pomfret  (2006).       144   resource  sectors   substantively  boosted  the  economy,  which  allowed  the  country  to  keep   almost   two-­‐digit   economic   growth   until   2007   (Pomfret   2006,   7).   Figure   4.4a   shows   that   oil-­‐gas  value  per  capita  had  been  14  times  increased  between  1998  (207  USD)  and  2008   (2,975   USD).   This   drastic   increase   in   natural   resource   wealth   was   largely   driven   by   a   rapid  surge  in  international  oil  prices  during  the  same  period  (Figure  4.4b).                      The   economic   recovery   induced   by   exporting   natural   resources   greatly   enhanced   state   resources   available   to   the   president.   Figure   4.3   shows   that   the   government   had   increased   state   revenue   with   very   high   speed   between   1995   and   2008.   In   2008,   state   revenue  reached  10  billion  US  dollars,  which  is  five  times  as  the  total  revenue  in  1995.   Growing   financial   resources,   mostly   achieved   by   rich   natural   resource   wealth,   dramatically  enhanced  the  government’s  ability  to  extensively  distribute  public  goods  to   voters.                    In   order   to   take   advantage   of   natural   resources,   Nazarbaev   began   to   centralize   the   management   of   natural   resource   sectors,   thereby   facilitating   the   efficient   distribution   of   patronage   for   his   political   purposes.   Immediately   after   independence,   the   president   was   not  fully  in  charge  of  the  oil  sector.  The  energy  sector  in  the  country  was  the  prerogative   of   the   Ministry   of   Oil   and   Gas   as   well   as   the   state   oil   and   gas   company   Kazakhstanmunaigaz,  which  was  subsidiary  to  the  Ministry  (Hoffman  2000,  281).  Most   natural   resources   in   Kazakhstan   are   concentrated   on   Western   regions   and   local   elites   maintained  considerable  control  over  natural  resource  management  (Ostrowski  2010).              The   failure   to   seize   natural   resources,   as   well   as   their   relative   shortages   in   the   earlier   periods,   made   it   difficult   for   the   president   to   use   the   oil   money   for   the   purpose   of   sufficient   distribution.   Nazarbaev   was   under   pressure   from   the   internal   elites   in   both     145   the   Ministry   and   oil-­‐rich   Western   regions,   who   aspired   to   seize   full   control   of   the   industry   and   increase   their   political   influence   in   the   central   government.   In   particular,   ruling  elites,  who  went  up  their  career  ladders  within  the  oil  industry  during  the  Soviet   era   and   thus   were   closely   connected   to   indigenous   oil   enterprises   and   the   Ministry,   attempted  to  strengthen  their  grip  on  natural  resource  sectors.  The  case  in  point  is  the   appointment  and  dismissal  of  Ravil  Cherdabaev  as  the  Minister  of  Oil  and  Gas.  Born  in   the   oil-­‐rich   Atyrau   oblast   into   a   family   whose   members   had   worked   in   oil   enterprises   since  the  beginning  of  the  twentieth  century,  Cherdabaev  was  one  of  the  most  influential   “oil   men”   (Ostrowski   2010,   37).   Having   taken   over   the   Minister   of   Oil   and   Gas,   he   planned   to   create   a   vertically   integrated   oil   company,   which   would   be   monitored   by   the   Ministry   of   Oil   and   Gas.   “This   plan,   if   successfully   executed,   would   spell   a   significant   degree  of  control  over  the  republic’s  most  important  industry  and  allow  the  ministry  to   regain   the   authority”   (Ostrowski   2010,   37).   Fearing   the   growing   influence   of   Cherdabaev,  Nazarbaev  forced  Cherdabaev  to  step  down  in  October  1994.                      In   order   to   prevent   these   indigenous   “oil   men”   from   dominating   the   oil   industry,   Nazarbaev  accelerated  the  privatization  of  the  oil  sector  by  introducing  foreign  capital.   As   a   result,   “by   the   end   of   the   1990s,   Kazakhstan’s   oil   and   gas   industry   no   longer   reflected   its   indigenous   roots”   (Jones   Luong   and   Weinthal   2010,   259).   Strong   politicians   like   Kazhegeldin,   as   well   as   wealthy   businessmen,   emerged   in   strength   as   a   result   of   large  profits  yielded  by  the  privatization  of  companies.  These  businessmen  also  helped   to   establish   the   Democratic   Choice   of   Kazakhstan.   In   his   attempt   to   contain   strong   opponents  without  harming  the  privatization  process,  Nazarbaev  decided  to  take  charge   of   the   oil   sector   himself.   In   March   1997,   he   signed   a   decree   deciding   to   dissolve   the     146   Ministry  of  Oil  and  Gas  Industry  and  other  related  state  committees  and  established  in   their   place   the   KazakhOil   National   Oil   and   Gas   Company   (Ostrowski   2010,   47-­‐48).   By   transferring   many   of   the   Ministry’s   functions   to   KazakhOil,   which   presides   over   contracts   with   foreign   companies,   Nazarbaev   successfully   strengthened   his   control   over   the   country’s   oil   industry   (a   sector   that   provides   an   estimated   37   percent   of   state   revenues),   taking   it   firmly   within   his   own   presidential   apparatus   and   away   from   the   executive  powers  of  Prime  Minister  Kazhegeldin  (Hoffman  2000,  282).                In  making  KazakhOil  work  fully  as  a  political  machine  through  which  he  could  cement   political   support   for   himself,   Nazarbaev   began   to   construct   direct,   patron-­‐client   relationships  with  other  ruling  elites  and  his  family  members  within  this  newly  founded   national   oil   company.     Nurlan   Balgimbaev,   Nazarbaev’s   close   ally,   was   installed   as   the   president   of   KazakhOil,   while   Timur   Kulibaev,   Nazarbaev’s   son-­‐in-­‐law,   became   a   financial  director  and  vice  president  of  KazakhOil  (Ostrowski  2010,  49).  Nazarbaev  also   recruited   many   young   technocrats   to   the   oil   company   who   did   not   have   close   connections   with   the   oil-­‐rich   regions   and   thus   were   loyal   to   the   president   (Ostrowski   2010,  49-­‐50).  In  so  doing,  he  tried  to  centralize  patron-­‐client  relationships  built  around   the  president  and  oil  resources.                President   Nazarbaev   and   his   ruling   parties  were   able   to   spend   a   vast   amount   of   oil   money  for  electoral  purposes  by  utilizing  the  centralized  management  of  the  oil  industry   and   dramatically   increasing   natural   resource   wealth   after   2000.   Analyzing   the   1999   elections,   Hoffman   (2000)   points   out   the   likely   connection   between   oil   resources   and   electoral  processes  in  the  elections,  while  suggesting  the  importance  of  KazakhOil  as  an   informal  political  organization:       147     Kazakhoil’s   political   importance   stems   not   only   from   the   strategic  nature  of  petroleum  in  Kazakhstan;  the  organization   also   serves   an   important   budgetary   role   for   the   Presidency.   Politically   prioritized   projects,   such   as   the   construction   of   Kazakhstan’s   new   capital,   Astana,   and   the   financing   of   early   presidential   elections,   require   huge   amounts   of   capital,   yet   are   not   officially   funded   from   the   republican   budget.   It   is   widely   rumored  that  Kazakhoil  has  been  used  as  a  tool  for  redirecting   state   financing   to   such   off-­‐budget   items.   If   true,   this   would   help   to   explain   not   only   the   funding   sources   of   these   activities,   but   the   reluctance   by   top   oil   officials   to   open   Kazakhoil   to   privatization,   which   would   require   more   transparent   operations   and   bookkeeping.   It   may   also   serve   as   a   partial   explanation  for  why  Kazakhoil  through  October  1998  had  only   returned  a  total  of  $2.3  million  to  the  state  budget  —  fully  14   times  less  than  the  anticipated  amount.  (Hoffman  2000,  287)                In  the  mid-­‐1990s,  both  central  and  local  authorities  possessed  a  very  limited  ability   in   financing   pensions,   utilities,   health   care,   and   other   fiscal   policies.   This   is   because   serious   economic   decline   a   couple   of   years   after   the   collapse   of   the   Soviet   Union   significantly   deprived   the   government   of   its   fiscal   power   (Author’s   interview   with   a   political  activist  [#29]  who  had  been  working  in  a  local  government  during  the  1990s).   On   the   other   hand,   there   is   much   anecdotal   evidence   suggesting   that   the   government   and  ruling  parties  have  distributed  economic  favors  to  voters  in  the  2000s  when  natural   resources  dramatically  enriched  state  coffers  (Author’s  interview  with  an  officer  of  the   Soros-­‐Foundation   [#2]   and   a   political   analyst   [#9]).   Rakhat   Aliev,   who   was   a   son-­‐in-­‐law   of   Nazarbaev   yet   fell   from   power   and   defected   from   the   regime   in   2007,   criticized   the   president  and  alleged  that  Nur  Otan  used  10  billion  dollars  from  a  “secret  fund”  to  carry   out   electoral   campaigns   prior   to   the   2007   legislative   elections   (RFE/RL   Kazakhstan   2007.8.17).       148            Large-­‐scale   pre-­‐electoral   patronage   distribution   does   not   necessarily   mean   that   the   government   alleviates   serious   economic   inequality   among   citizens   by   engaging   in   extensive   public   goods   provisions   covering   every   citizen   in   an   equal   manner.   But,   it   would   be   hard   to   deny   that   there   is   widespread   popular   support   coming   from   the   provision   of   material   benefits   that   plays   a   crucial   role   in   the   electoral   victories   of   the   ruling   party   besides   electoral   fraud   and   manipulation   of   electoral   systems.   Sharipova   (2013)   finds   that   access   to   state   resources   such   as   public   jobs,   high-­‐quality   public   hospitals   and   state   housing   is   largely   determined   by   close   connections   with   powerful   and   wealthy   people.66  The   government   mainly   targets   those   who   are   most   likely   to   vote   for   ruling   parties   after   having   received   various   economic   benefits   from   the   government.   These   people   include   government   officials,   doctors   in   public   hospitals,   staff,   teachers   and   professors   in   public   schools,   and   old   people   whose   lives   are   dependent   upon   public   pensions.   The   government   encourages   heads   of   schools,   hospitals   and   universities   to   mobilize   votes   for   the   ruling   parties,   often   in   return   for   increases   in   salaries,   bonuses,   pension,   and   various   other   accommodations   (Author’s   interview   with   a   political   analyst   [#4]).  For  example,  one  opposition  activist,  who  had  been  working  as  a  deputy  in  a  local   legislature,  succinctly  describes  how  it  works:                                                                                                                             66  Sharipova   (2013)   used   a   survey   data   taken   in   2013   and   conducted   qualitative   case   studies  on  health  care  and  housing  policy  in  Kazakhstan,     149   I  myself  was  a  deputy  and  I  know  this  process.  […]  The  thing   is   administrative   resources.   Let's   say   we're   holding   an   election   in   a   village.   This   is   especially   true   in   small   villages,   home  to  1,000  people.  […]    Before  the  election,  the  mayor  of   the   village   collects   headmasters,   all   who   work   in   the   state   budget   section,   as   doctors,   or   in   small   and   medium-­‐sized   businesses   and   those   who   can   access   to   credit   commitments   and  tax.  And,  they  hold  meetings  and  say,  "we  need  to  ensure   people  will  be  participating  in  the  election  and   also  they  have   to   vote   for   us."   (Author’s   interview   with   a   political   activist   [#20])                                      Citizens   who   vote   for   the   governing   parties   receive   various   kinds   of   material   benefits.   The   Kazakh   government   often   employs   several   distribution   strategies   before   elections  in  order  to  garner  votes  from  citizens.67  First,  the  government  increases  public   spending   by   raising   salaries   for   public   officials   prior   to   elections   to   buy-­‐off   their   support.   For   example,   the   central   government   allocated   two   billion   dollars   in   order   to   raise   salaries   for   more   than   100,000   professionals   at   central   and   local   governments   before   the   2012   parliamentary   elections,   which   resulted   in   an   increase   in   wages   by   about   20-­‐30   percent   country-­‐wide   (Author’s   interview   with   an   economist   [#25];   Tengri   News  26,  December  2011).  Also,  when  the  legislative  elections  were  held  in  2007,  real   wages   in   the   country   had   annually   increased   by   30   percent,   which   is   about   twice   as   large  as  wage  increases  in  a  non-­‐election  year  (2006:  17.3  percent;  OECD  2011).              Second,   the   government   tends   to   raise   pensions   in   election   years   as   they   want   to   appeal  to  elderly  people.  For  example,  pensioners,  who  occupy  2  million  people  out  of   16   million   total   populations,   enjoyed   annual   increases   in   pensions   continuously   between  2009  and  2011.    In  non-­‐election  years  (2009  and  2010),  increases  in  pension                                                                                                                   67  I   provide   systematic   quantitative   evidence   of   these   political   business   cycles   in   Kazakhstan  in  the  following  section.     150   were   25   percent,   and   in   the   2011   election   year   (the   presidential   election   was   held   in   April  2011  and  the  parliamentary  election  was  held  in  January  2012)  they  raised  to  30   percent   (Author’s   interview   with   an   economist   [#25]).     “It   is   the   retired   people   that   support   Nur   Otan   because   they   think   this   is   the   party   doing   something   beneficial   for   them.   They   think   the   leader   of   Nur   Otan,   Nazarbaev,   is   the   only   person   who   can   keep   political  stability  in  Kazakhstan.  […]  Because  many  in  the  old  generation  as  well  as  the   middle-­‐aged  generation  still  believe  that  Nazarbaev  brings  stability,  they  also  think  Nur   Otan   is   a   real   powerful   instrument   to   keep   stability   and   development”   (Author’s   interview  with  a  political  analyst  [#4]).                Third,   as   many   people   become   more   dependent   on   state   resources   in   terms   of   salaries,   pensions,   and   loans,   it   is   difficult   for   them   to   not   vote   for   the   governing   parties   because  they  hope  to  continue  to  receive  these  material  benefits.  If  they  do  not  vote  for   the   governing   parties   at   elections   and   such   voting   behavior   is   detected   by   the   authorities,  they  may  be  deprived  of  income  sources  such  as  public  jobs,  tax  exemptions,   pensions  and  so  on  (Author’s  interview  with  a  political  activist  [#18]).  A  political  analyst,   who  had  been  working  in  a  public  university,  points  out:       Public   employees   –   teachers   in   schools,   professors   in   universities   and   doctors   in   hospitals   -­‐-­‐   are   mobilized   to   vote   for   ruling   parties.   During   election   campaigns,   these   institutions  not  only  asked  their  staff  to  vote  for  Nur  Otan,  but   sometimes  they  threaten  employees  by  saying  "if  you  do  not   vote  for  Nur  Otan,  then  there  would  be  some  measures,  some   implications   for   you,   even   being   fired   from   your   institution."   (Author’s  interview  with  a  political  analyst  [#4])         151   4.5.2  Centralization  of  Governing  Institutions     Table   4.3:   Percentage   of   Central   Government   Transfers   in   Total   Revenues   of   Regional  Governments         Source:  Sharipova  (2013,  113)                Kazakhstan   had   been   highly   decentralized   in   real   terms   until   the   end   of   the   1990s,   yet   since   the   early   2000s   the   country   has   become   more   centralized.   A   first   aspect   is   fiscal  centralization.  Although  the  Second  Constitution  stipulates  that  Kazakhstan  has  a   centralized  government,  de  facto  fiscal  decentralization  had  been  advanced  in  the  1990s.   For   the   purpose   of   accurately   responding   to   the   demands   of   local   populations,   decision-­‐ making   power   on   economic   and   social   policies   was   transferred   from   the   central   government  to  local  governments.    Moreover,  although  on  paper  the  central  government   has   the   sole   authority   to   collect   tax,   regional   governments   in   fact   seemed   to   have   had   considerable  discretionary  power  in  tax  collection  during  the  1990s  (Jones  Luong  2004).   Having   interviewed   with   state   officials   in   regional   governments   at   the   end   of   the   1990s,   Jones  Luong  (2004,  188-­‐189)  reports  that  directives  from  regional  governments  tended     152   to  be  more  prioritized  than  the  central  government’s,  when  the  two  conflict.  The  tide  of   fiscal   decentralization,   however,   had   been   reversed   from   2000   (Author’s   interview   with   an   economist   [#6]   and   a   political   analyst   [#9]),   and   the   central   government   tightened   fiscal   control   over   regional   governments   via   government   transfers   (Dave   2013).   Makhmutova  (2005,  287)  shows  that  the  central  government  tended  to  withdraw  more   money   from   the   Almaty   city,   the   oil-­‐rich   Mangistau   and   Atyrau   oblasts   to   poor,   pro-­‐ presidential  regions  such  as  the  South  Kazakhstan  oblast  by  2004.  Table  4.3  shows  time-­‐ series   change   in   central   government   transfers   to   regional   governments   between   2001   and   2009.   As   it   shows,   regional   governments   received   most   revenues   by   taxing   local   populations   at   the   time   of   2001.   Only   19.34   percent   of   total   revenues   derived   from   government   transfers.   The   amount   of   the   central   government   transfers   had   been   dramatically   raised   by   2009,   when   59.83   percent   of   regional   governments’   revenues   were   subsidized   by   the   central   government.   The   increasing   fiscal   dependence   on   the   central  government  encourages  regional  governments  to  be  more  loyal  to  the  president   (Author’s  interview  with  a  political  activist  [#29]).  Shapirova  (2013,  115)  acutely  points   to  this:  “The  main  problem  of  the  regions  is  their  dependence  on  the  center.  The  hands   and  legs  of  every  single  akim  (local  governors)  are  tied.  […]  If  an  akim  is  ‘good,’  then  he   receives   financial   resources;   if   he   is   not   good   and   if   he   does   not   manage   to   regulate   social   tension,   then   he   does   not   receive   money   from   the   center.   To   be   a   good   akim   means  to  be  loyal  and  provide  the  right  and  ‘appropriate’  indicators  of  socio-­‐economic   development.  The  functions  of  akims  then  are  to  squeeze  money  from  the  center  rather   than   to   develop   the   region.”   Although   regional   governments   have   discretionary   power   over   expenditures   such   as   education,   health-­‐care,   local   economic   policies,   the   center’s     153   strong  grip  on  revenues  limits  real  options  that  are  available  to  akims.       0.6 0.4 0.2 0 Herfindahl Fractionalization Index/Proportion of Independents Figure  4.5:  Coherence  of  Ruling  Parties  and  Proportion  of  Independents   1995 1999 2004 2007 Election Year Fractionalization of Ruling Parties Independents   Note:   Fractionalization   of   ruling   parties   is   computed   by   using   the   Herfindahl   index   ! ! !!! 𝑅𝑢𝑙𝑖𝑛𝑔  𝑃𝑎𝑟𝑡𝑦!  where  𝑅𝑢𝑙𝑖𝑛𝑔  𝑃𝑎𝑟𝑡𝑦!  is   a   share   of   a   ruling   party   in   total   seats   of   the   Lower  House.                A  second  aspect  is  de  facto  political  centralization.  Although  the  1995  constitution  had   already   given   strong   presidential   power   (as   discussed   before),   there   is   much   evidence   suggesting   that   Nazarbaev   in   the   1990s   was   not   as   strong   as   in   the   2000s.   First   of   all,   ruling  parties  had  tended  to  be  weaker  and  fractionalized  until  Nur  Otan  was  founded  in   December   2006.   Figure   4.5   shows   how   ruling   parties   had   become   less   fractionalized   between   1995   and   2007.   The   Party   of   People’s   Unity   of   Kazakhstan   (PNEK),   then   pro-­‐   154   presidential   party   that   Nazarbaev   seemingly   aspired   to   make   a   dominant   party,   could   not  obtain  the  dominant  position  in  parliament  as  there  were  a  number  of  ruling  parties   as  well  as  opposition  parties  in  the  parliament.  In  February  1999,  eight  months  before   the  legislative  elections,  Otan  was  established  as  a  result  of  the  merger  between  PNEK   and  several  pro-­‐presidential  parties,  although  there  were  still  three  other  ruling  parties   outside  this  merge  -­‐-­‐  Asar,  the  Agrarian  Party  and  the  Civic  Party.  In  fact,  these  political   parties  were  not  necessarily  satellite  parties  completely  subject  to  the  president  and  had   fierce   political   competition   with   each   other   (Dave   2004,   9).   For   instance,   in   the   2004   parliamentary   elections,   the   leader   of   the   Civic   Party,   Azat   Peruashev,   pointed   to   this   inter-­‐party   competition,   noting   that   “leaders   from   Otan   called   me   up   and   said   ‘deal   with   your  candidate  in  the  region  -­‐-­‐-­‐  he  is  trying  to  compete  with  the  Otan  candidate’”  (Issacs   2013,  17).  In  fact,  the  ruling  parties  failed  to  coordinate  their  candidates  and  put  more   than   two   candidates   in   a   district   in   38   out   of   67   electoral   districts   (56   percent   of   all   electoral   districts), 68  Asar,   which   was   directed   by   the   president’s   daughter   Dariga   Nazarbaeva  and  her  husband  Rakhat  Aliev,  was  not  only  subject  to  their  father  but  also   viewed  as  a  powerful  independent  force  arguing  for  democratic  reform  (Issacs  2013,  17).   As   I   detail   later,   however,   Nazarbaev   had   successfully   merged   all   the   ruling   parties   by   December  2006  and  founded  Nur  Otan,  the  dominant  party  in  the  country.                Second,   independent   politicians   had   been   thriving   throughout   the   1990s   until   the   mid-­‐2000s,  yet  they  disappeared  prior  to  the  2007  legislative  elections.  In  Kazakhstan,   most   independent   politicians   were   pro-­‐presidential   (Olcott   2002,   252),   so   it   would   be                                                                                                                   68  This  is  the  author’s  calculation  based  on  the  electoral  district  data  in  Nurmukhamedov   and  Chevotariev  (2005)       155   more   accurate   to   regard   them   as   members   of   the   ruling   coalition.   Yet,   independent   politicians   have   their   own   patron-­‐client   networks   in   their   strongholds   and   thus   retain   more  autonomous  power  than  lawmakers  affiliated  with  ruling  parties.  Figure  4.5  also   shows  how  the  proportion  of  independent  politicians  in  legislature  changed  from  1995   to   2007.   The   percentage   of   legislators   in   the   Lower   House   who   do   not   belong   to   any   political   party   tends   to   decrease   over   time.   Even   in   the   2004   parliamentary   elections   where   ruling   parties   became   stronger   than   in   the   past   elections,   18   self-­‐nominated   candidates  won  seats  and  14  of  them  ran  for  elections  from  the  regions  where  they  were   born  and/or  started  their  political  careers.69  It  indicates  that  self-­‐nominated  politicians   had   strong,   independent   support   base   built   upon   informal   connections   with   local   populations   via   kinship   or   clan   (Dave   2004).   By   the   time   of   the   2007   parliamentary   elections,   the   new   electoral   law   and   the   adoption   of   the   PR   system   encouraged   independent  politicians  to  be  affiliated  with  political  parties.  In  most  cases,  they  joined   the  dominant  party,  Nur  Otan.                      Third,  the  center-­‐periphery  relationship  has  also  changed  in  the  way  that  the  center   tightly   controls   regional   governments.   Since   independence,   the   president   has   reserved   the   right   to   appoint   and   dismiss   akims   of   oblasts   on   his   own   decision.   Yet,   a   couple   of   years  after  independence,  the  president  tended  to  appoint  elites  from  regions  as  akims   in   the   attempt   to   achieve   a   balance   of   power   between   the   political   elites   in   the   center   and   periphery.   Cummings   (2005,   106)   reports   that,   in   7   out   of   20   cases,   former   regional   executive   First   Secretaries   during   the   Soviet   period   became   akims   soon   after                                                                                                                   69  This   is   the   author’s   calculation   based   on   Tsentralnaya   Izbiratelnaya   Komissiya   Respubliki   Kazakhstan   (2010),   Ashimbaev   (2012),   and   Nyrmukhamedov   and   Chebotarev  (2005).         156   independence.  Jones  Luong  (2002,  287-­‐288)  documents  that  35  out  of  71  akims,  about   50  percent  of  total  akims,  were  regional  elites  who  held  a  political  position  in  the  same   region.   Schatz   (2004)   also   argues   that   during   the   1990s   the   president   appointed   regional   elites   as   well   as   central   ones,   drawing   much   attention   to   a   good   balance   among   three  Zhuz  or  clans.    Since  the  beginning  of  the  2000s,  however,  the  president  began  to   centralize  appointments  of  akims  by  using  several  strategies.  First,  more  members  of  the   national   elite,   who   went   through   their   career   in   the   central   government,   tended   to   be   parachuted   into   regions   as   akims   since   around   1997   than   ever   before   (Cummings   2005,   107),   while   regional   elites   were   appointed   to   the   positions   of   the   central   government   or   elected   as   legislators.   Such   “national-­‐regional   crossovers,”   combining   with   frequent   reshuffling   of   akim   appointments   every   two   or   three   years   (e.g.   Cummings   2005;   Junisbai  2010),  allowed  the  president  to  effectively  weaken  local  elites  who  had  strong   support  bases  in  their  home  regions.  Second,  it  seems  that  the  president  implements  not   only   crossovers   between   the   center   and   periphery   but   also   “parallel   appointments”   of   akims   between   regions   (Siegel   2014).   Employing   these   strategies,   the   president   successfully  tightened  his  control  over  regional  governments  in  order  to  use  them  as  a   political  machine  for  his  own  sake.         4.5.3  The  Dominant  Party,  Nur  Otan            Built   upon   the   fiscal   and   political   centralizations   of   the   government,   the   dominant   party  Nur  Otan  was  founded  in  December  2006  as  a  result  of  the  merger  between  Otan   and   three   other   pro-­‐presidential   parties:   Asar,   the   Civil   Party,   and   the   Agrarian   Party.   Even   at   its   inception,   the   party’s   organization   was   more   extensive   than   the   previously     157   existing  ruling  parties.  Currently,  it  officially  claims  that  there  are  about  850,000  party   members,70  a  large  number  for  a  country  of  17  million  people.  As  of  January  2015,  the   party   has   225   regional   branches   and   5,605   primary   party   organizations   across   all   the   oblasts  and  the  two  cities  Almaty  and  Astana.71                  Although   these   official   numbers   may   be   exaggerated   and   do   not   necessarily   reflect   real  numbers,  what  makes  the  party’s  organization  very  strong  is  that  it  is  highly  fused   with   the   government.   The   party   mobilizes   supporters   using   various   “administrative   resources.”  As  of  January  2015,  akims  in  most  regions  other  than  the  Almaty  oblast  and   the   Karaganda   oblast   serve   also   as   the   chairman   of   the   Nur   Otan’s   regional   branch   in   each   oblast.72  It   is   a   well-­‐known   fact   that   akims   become   a   main   actor   in   localities   and   mobilize   supporters   on   the   eve   of   elections   by   using   financial   and   other   kinds   of   administrative   resources   in   order   to   demonstrate   their   loyalty   to   the   party   and   the   president  (Author’s  interviews  with  political  analysts  [#8]  and  [#9]).  Mobilized  by  akims,   both   high   and   low   ranks   of   state   officials   in   oblasts   are   strongly   encouraged   to   work   for   the   party   during   election   campaigns,   while   it   is   often   implied   by   the   possibilities   of   salary   reduction   and/or   losing   their   jobs   if   they   do   not   comply   with   their   bosses’   orders   (Author’s   interview   with   a   political   analyst   [#4]).   As   I   discussed   above,   similar                                                                                                                   70  The   number   is   drawn   from   the   website   of   the   Central   Election   Committee.   (election.kz)           71  The  website  of  Nur  Otan.  (http://nurotan.kz/ru/regions)       72  The  website  of  Nur  Otan.  (http://nurotan.kz/ru/regions)         158   mobilization   structures   can   be   also   found   in   public   hospitals   and   schools,73  where   the   local   government   officials   are   entitled   to   appoint   deans   of   these   public   organizations,   who   are   then   encouraged   to   work   as   brokers   of   electoral   mobilization   in   their   work   places  (Del  Sordi  2012).  Although  all  these  strategies  did  exist  before  the  foundation  of   Nur   Otan,   the   efficiency   of   political   mobilization   has   increased   after   Nur   Otan   successfully  integrated  informal  networks  into  a  single-­‐pyramid  system  administered  by   the   president   at   the   top   (Author’s   interview   with   a   political   analyst   [#22]).   Using   the   hierarchical   mobilization   structures,   the   president   succeeds   in   using   the   financial   resources   effectively   and   mobilizing   a   large   number   of   supporters   nationwide   before   elections.   As   Issacs   (2013,   132)   rightly   points   out,   “Nur   Otan’s   preponderance,   which   was   achieved   primarily   through   its   relationship   with   the   president   and   other   state   executive   actors,   means   that   it   is   the   dominant   channel   between   society   and   the   state.   Nur   Otan,   however,   is   also   the   personal   political   vehicle   for   the   president   to   establish   greater   control   of   formal   political   institutions.   It   is   not   a   channel   to   articulate   societal   interests.”                 4.5.4  Weakening  of  Opposition            As   discussed   in   the   previous   sections,   opposition   movements   had   been   relatively   stronger   and   retained   considerable   mobilization   capabilities   at   elections   during   the                                                                                                                   73  Nur  Otan  also  organizes  a  youth  wing  called  Zhas  Otan  in  which  youngsters  carry  out   intensive  election  campaigns  mainly  in  public  universities  to  encourage  students  to  vote   for  Nur  Otan  (Del  Sordi  2012).  Many  of  them  become  party  officials  after  graduating  and   aspire  to  climb  up  a  career  ladder  to  party  cadres,  so  they  work  hard  for  the  party  in  this   youth   organization   (Author’s   interview   with   a   student   [#19]).   In   a   broad   sense,   they   are   also   an   important   political   actor,   who   mobilizes   potential   supporters   for   Nur   Otan   in   return  of  benefits.                           159   1990s.   In   particular,   before   1995,   opposition   activities   and   ethnic   movements   were   highly  powerful  and  the  power  distribution  between  opposition  and  ruling  parties  was   more   balanced.   This   was   one   of   the   most   important   factors   that   encouraged   the   president   to   resort   to   serious   electoral   fraud,   make   a   “state   list”   to   select   pro-­‐ presidential   legislators   in   the   1994   legislative   elections,   and   adopt   pure   SMD   systems   in   the  1994  and  1995  elections.              Even  after  1995,  opposition  parties  still  remained  popular  at  least  among  some  voters.   As   we   have   already   seen,   behind   his   popularity   and   rich   independent   resources   as   the   former   prime   minister,   Kazhegeldin   challenged   Nazarbaev   by   declaring   that   he   would   run   for   the   1999   presidential   elections   and   establishing   the   opposition   Republican   People’s   Party.   As   a   political   activist   who   had   defected   from   the   regime   together   with   Kazhegeldin   states,   his   severe   criticism   against   President   Nazarbaev   and   growing   nepotism   within   the   government   was   highly   compelling   and   appealing   for   a   large   number  of  citizens  because  he  had  been  served  as  the  second  position  until  recently  and   was   thought   to   know   the   internal   workings   of   the   government   very   well   (Author’s   interview   with   a   political   activist   [#24]).   Serikbolsyn   Abdildin,   the   chairman   of   the   Communist  Party  of  Kazakhstan,  also  gathered  high  political  support  especially  from  the   urban   poor   and   ran   for   the   1999   presidential   elections   as   a   strong   candidate   from   opposition.                In   the   run-­‐up   to   the   1999   elections,   Nazarbaev   responded   to   these   two   real   opponents   by   rigging   elections   seriously   and   repressing   and   harassing   the   opposition   leaders.   Kazhegeldin,   the   most   serious   contender   of   the   president,   was   barred   from   running   for   the   presidential   elections   due   to   accusations   of   money   laundering   in     160   Belgium  as  well  as  his  participation  in  an  unregistered  political  meeting  for  “Movement   for   Honest   Elections”   (Cummings   2005,   28).   Although   permitted   to   participate   in   electoral   battles,   Abdildin   and   his   Communist   Party   were   also   exposed   to   a   variety   of   electoral  manipulation  in  both  the  pre-­‐electoral  periods  as  well  as  the  election  days.  The   election   results   were   allegedly   seriously   falsified   against   Abdildin   and   the   Communist   Party  (Author’s  interview  with  an  opposition  politician  [#16];  Olcott  2010,  121).                In  November  2001,  a  significant  number  of  ruling  elites  defected  from  the  regime  and   declared   that   they   founded   the   Democratic   Choice   of   Kazakhstan.   This   opposition   movement  was  said  to  be  the  largest  and  most  serious  dissent  within  the  regime  since   independence,  and  that  the  party  had  rich,  independent  financial  resources  to  carry  out   effective   electoral   campaigns.   This   was   because   the   party   cadres   of   DCK   including   Galymzhan  Zhakiyanov,  Mukhtar  Ablyazov,  and  Nurzhan  Subkhanberdin  had  come  from   key   business   sectors,   held   important   government   positions   closely   related   to   businesspeople   financing   the   opposition   movements,   or   both   (see,   e.g.   Junisbai   and   Junisbai  2005;  Junisbai  2010;  Chebotariev  2009).              Due   to   political   infighting   within   the   party,   as   well   as   a   series   of   repression   and   harassment   exercised   by   the   government,   some   members   of   the   DCK   defected   and   established   the   Ak   Zhol   party   in   March   2002.   A   political   activist   who   had   participated   in   the  Democratic  Choice  of  Kazakhstan  before  joining  Ak  Zhol  stated:  “The  movement  was   so   spontaneous   that   leaders   in   the   movement   had   not   been   well-­‐prepared   for   how   to   lead   this   new   opposition   party.   We   launched   the   opposition   movement   to   demand   more   political   transparency,   but   our   views   on   how   we   could   achieve   this   goal   were   diverse,   which  led  some  people  to  prefer  to  a  moderate  course  and  defect  from  DCK.”  (Author’s     161   interview  with  a  political  analyst  [#8])    This  division  among  opposition  leaders  seriously   harmed   the   unity   and   strength   of   this   new   opposition   movement   before   the   2004   parliamentary  elections.  First  of  all,  financial  resources  of  the  opposition  were  dispersed,   making   it   difficult   for   them   to   carry   out   effective   electoral   campaigns   (Author’s   interviews   with   a   political   analyst   [#8]   and   an   opposition   activist   [#29]).   Second,   although  the  Communist  Party  made  a  pre-­‐electoral  opposition  coalition  with  DCK,  the   Ak   Zhol   party   did   not   coordinate   its   candidates   and   electoral   campaigns   with   the   two   parties,   which   helped   ruling   parties   win   seats   in   single-­‐member   districts.   In   the   2004   parliamentary  elections,  the  DCK-­‐CPK  bloc  and  Ak  Zhol  fielded  candidates  in  48  electoral   districts   out   of   67.   In   21   electoral   districts,   they   fielded   a   candidate   from   each   party.74   The  DCK-­‐CPK  bloc  and  Ak  Zhol  could  not  get  any  seats  in  parliament  and  obtained  only   3.44   percent   and   12.04   percent   of   total   votes   in   a   party   list,   respectively,   due   to   the   failure   to   coordination   among   opposition   parties   and   the   increasing   mobilization   power   of   the   ruling   parties   using   rich   oil   resources.   There   was   also   electoral   fraud   that   undercounted   the   votes   of   the   opposition   parties.   Yet,   as   we   have   seen,   the   scale   of   electoral  manipulation  was  a  bit  smaller  than  the  1999  elections.  It  is  said  that  Ak  Zhol   would   have   obtained   25-­‐30   percent   of   votes   if   the   elections   were   free   and   fair   (Author’s   Interview  with  a  political  activist  [#8]),  whereas  the  Communist  party  claimed  to  have   scored  9  percent  of  total  votes,  clearing  the  7  percent  electoral  threshold  (Dave  2004,  9).   However,  even  if  the  elections  had  been  completely  free  and  fair,  the  ruling  parties  and   pro-­‐presidential   candidates   would   still   have   scored   more   than   just   a   simple   majority,                                                                                                                   74  These   figures   are   based   on   electoral   results   showed   in   Nurmukhamedov   and   Chebotarev  (2005).         162   suggesting   that   the   regime   was   becoming   increasingly   stronger,   despite   the   still   fragmented   ruling   coalitions   that   were   due   to   several   governing   parties   and   many   independent  politicians.                              By  the  time  when  Kazakhstan  held  the  2005  presidential  and  2007  parliamentary   elections,   opposition   forces   had   already   been   suffering   further   divisions   and   realignments.   In   April   2005,   Ak   Zhol   split   because   of   its   leaders’   disagreement   over   whether   they   would   participate   in   a   pre-­‐electoral   opposition   coalition   (“For   a   Just   Kazakhstan”)  with  CPK  and  DCK.75  Defected  members  formed  Naghyz  Ak  Zhol.  In  2006   September,   National   Social   Democratic   Party   (NSDP)   was   also   formed   by   Jarmakhan   Yuyakbay,  the  former  Chairman  of  the  Lower  House  and  an  opposition  candidate  in  the   2005   presidential   elections.   Even   though   Naghyz   Ak   Zhol   decided   to   join   NSDP   two   months   before   the   parliamentary   elections   in   August   2007,   NSDP   gained   only   4.54   percent  of  votes,  failing  to  secure  any  seats  in  the  parliament.  Many  analysts  admit  that   by   the   time   of   the   2007   elections   the   opposition   camp   had   become   much   weaker   than   ever  before  due  to  factors  such  as  extremely  low  levels  of  political  support  in  the  rural   areas,  the  very  fragile  nature  of  their  nationwide  organizational  networks,  and  absence   of   financial   resources   to   achieve   effective   electoral   mobilization   (Author’s   interviews   with  opposition  activists  [#15,  #30]  and  political  analysts  [#3,  #22]).                                                                                                                                         75  Eurasia.net.   Ibragim   Aibekov.   “Kazkahstan’s   Leading   Opposition   Party   Faces   Split”   (http://www.eurasianet.org/departments/insight/articles/eav021605.shtml)       163   4.6  Political  Business  Cycles  in  Kazakhstan              The  previous  section  argued  that  the  president  was  able  to  become  less  dependent  on   extensive   electoral   manipulation   to   bring   an   overwhelming   victory   because   he   successfully  increased  mobilization  capabilities  to  cultivate  votes  from  the  citizenry  vis-­‐ à-­‐vis  opposition  parties.    The  qualitative  process  tracing  also  suggests  that  the  president   came   to   distribute   more   economic   favors   as   the   country   was   enjoying   abundant   state   resources  and  centralized  political  organizations.              This   section   tests   the   existence   and   size   of   pre-­‐electoral   economic   distribution,   namely,   political   business   cycles,   by   using   newly-­‐collected   monthly   economic   data   of   Kazakhstan  (1995-­‐2008).  I  use  the  following  three  indices  as  my  dependent  variables,  all   of  which  capture  economic  manipulation  before  elections  from  different  angles.  The  first   measure  is  a  monthly-­‐specified  Consumer  Price  Index  (CPI).  Setting  the  CPI  at  December   1994   as   100,   I   calculate   the   CPI   between   January   1995   and   December   2008.   As   many   studies   on   political   business   cycles   have   argued,   inflation   after   elections   can   be   interpreted   as   strong   evidence   that   governments   adopt   expansionary   fiscal   and   monetary   policies   before   elections.   Second,   the   unemployment   rate,   which   is   also   monthly   variant   data,   is   a   valuable   measure   to   check   whether   opportunistic   budget   cycles  can  be  observed.  Because  the  National  Bank  of  Kazakhstan  does  not  publicly  show   the   unemployment   rate   in   monthly   units,   I   calculate   the   unemployment   rate   by   dividing   the   number   of   unemployed   workers   (which   is   monthly   data   from   Statisticheskii   Byulleten)   by   annual   total   number   in   the   working   population   (from   Regiony   Kazakhstana).  Finally,  electoral  budget  cycles  are  observed  to  see  if  real  wage  increases     164   before   elections.   As   I   have   shown   with   some   anecdotal   evidence,   the   government   increases   wages   of   public   employees   before   elections.   Statistical   analysis   using   the   variable   of   real   wage   increases   enables   me   to   present   more   systematic   evidence   on   pre-­‐ electoral   fiscal   maneuvering   than   anecdotal   data.   Since   the   National   Bank   of   Kazakhstan   reports   the   quarterly   average   nominal   wage   among   workers,   I   transform   it   into   average   real   wage   dividing   nominal   wage   by   CPI.   Data   sources   for   all   indices   are   from   Statisticheskii  Byulleten,  published  monthly  by  the  National  Bank  of  Kazakhstan.       100 Consumer Price Index (1994.12=100) 200 300 400 500 Figure  4.6:  Time  Series  Change  in  Consumer  Price  Index     1995.1 1997.1 1999.1 Source: Statisticheskii Byulleten 2001.1 2003.1 2005.1 2007.1 2009.1 Month-Year (Jan 1995 - Dec 2008)       165            Figure  4.6  plots  time-­‐series  changes  in  the  consumer  price  index.76  Visual  inspection   reveals   that   while   each   index   has   certain   general   patterns   (e.g.   consumer   price   index   constantly  keeps  increasing  over  time)  they  have  some  seasonal  fluctuations  as  well.  The   Figure  allows  us  to  roughly  grasp  the  tendencies  the  index  experiences  a  certain  amount   of   fluctuation   around   the   election   times.   For   instance,   inflation   appears   to   increase   precisely  after  elections.  After  the  elections  in  2004,  2005  and  particularly  in  2007,  the   country   was   exposed   to   higher   inflation.   Moreover,   in   order   to   confirm   whether   there   are  significant  differences  between  (pre-­‐  and  post-­‐)  electoral  and  non-­‐electoral  periods   for  the  three  variables,  I  conduct  simple  t-­‐tests  and  find  that  four  out  of  six  differences   are   in   the   predicted   directions. 77  Controlling   for   seasonal   effects   and   coping   with   problems  accompanied  with  time-­‐series  data,  I  also  conduct  OLS  time-­‐series  regressions   with  the  three  dependent  variables.78                                                                                                                                 76  Although  I  omitted  graphs  on  unemployment  rate  and  real  wage  increases  because  of   a  space  constraint,  these  are  available  upon  request.       77  (1)   Inflation   tends   to   increase   around   elections.   (2)   Prior   to   elections,   the   unemployment   rate   has   been   mitigated,   while   it   tends   to   increase   after   elections.   (3)   Real  wages  are  more  likely  to  grow  before  elections,  while  they  will  shrink  thereafter.       78  For  details  on  statistical  methods,  see  Appendix  C4.       166   Table  4.4:  Political  Business  Cycles  in  Kazakhstan,  1995-­‐2008     Note:  ***p  <  .01,  **p  <  .05,  *p<.10.    Standard  errors  are  reported  in  parentheses.  Durbin-­‐ Watson  test  is  performed  after  correcting  serial  autocorrelation  through  AR(1)  process.                Models  1-­‐3  (Table  4.4)  estimate  the  impact  of  (pre  and  post)  electoral  periods  on  the   three  dependent  variables.  Model  1,  in  which  the  variable  being  regressed  is  CPI,  shows   that  in  the  first  six  months  after  elections  the  inflation  rate  increases  by  1.31  percent  at   the   5   percent   significance   level.   In   Model   3,   we   can   confirm   that   real   wages   are   more   likely  to  increase  before  elections.  Even  after  controlling  for  seasonal  effects,  the  raise  in   salaries   is   statistically   significant   (534   Tenges   per   quarter   before   elections   at   the   5   percent  level).              On   the   other   hand,   Model   2   shows   that   elections   do   not   have   an   impact   on   unemployment  rates  with  any  statistical  certainty.  Although  unemployment  rates  before   elections   tend   to   decrease,   this   is   not   statistically   significant.   This   result   suggests   that   the   political   business   cycles   are   not   observed   in   real   economic   outcomes   but   as   the   manipulation   of   policy   instruments   under   autocracy   as   in   democracies.   By   manipulating   policy   instruments   before   elections,   the   government   might   try   to   show   their   "competence"   of   economic   management   to   their   constituencies   (Rogoff   and   Silbert   1988;  Drazen  2000,  228-­‐246).     167   Table  4-­‐5:  Electoral  Cycles,  Organizational  Strength,  and  Financial  Resources     Note:   Note:   ***p   <   .01,   **p   <   .05,   *p<.10.     ^p   :   jointly   statistically   significant   at   the   10   percent   level.   For   details,   see   Figure   4.7.   Standard   errors   are   reported   in   parentheses.   The   Durbin-­‐Watson   test   is   performed   after   correcting   serial   autocorrelation   through   AR   (1)  process.                The   statistical   results   presented   so   far   demonstrate   that   political   business   cycles   do   exist   in   Kazakhstan.   What   are   relationships   among   organizational   strengths,   financial   resources   and   the   size   of   electoral   business   cycles?   According   to   Models   5-­‐7,79  the   inflation   rates   tend   to   increase   around   elections   under   highly   centralized   governing   institutions   or   rich   financial   revenues.   For   Model   5,   I   introduce   the   independent   variables80  -­‐-­‐  the  number  of  pre-­‐  and  post-­‐elections  periods  experienced  since  1995  -­‐-­‐  in   order  to  investigate  whether  the  size  of  electoral  budget  cycles  has  become  larger  as  the                                                                                                                   79  The  dependent  variable  in  these  models  is  the  first  difference  of  CPI.     80  There   are   6   elections   in   total   between   January   1995   and   December   2008.   For   the   variable,   I   coded   the   electoral   periods   as   follows:   the   December   1995   parliamentary   election  =  1,  January  1999  presidential  election  =2,  October  1999  parliamentary  election   =   3,   September   2004   parliamentary   election   =   4,   December   2005   presidential   election   =   5,  and  August  2007  parliamentary  election  =  6.                 168   country  experienced  more  elections.  Given  the  fact  that  President  Nazarbaev  gradually   and   consistently   centralized   the   government   while   accumulating   financial   resources   since  1995,  the  size  of  electoral  business  cycles  should  have  a  positive  association  with   the   number   of   elections   that   the   country   experienced.   The   coefficient   is   positive   and   statistically   significant   below   a   1   percent   probability   of   error,   which   means   that   while   after  the  1995  election  the  inflation  rate  increased  by  just  0.425  percent,  post-­‐electoral   inflation   escalated   to   2.55   percent   at   the   6th   election   in   August   2007.   Compared   to   Model   1,   adjusted   R-­‐squared   of   Model   4   has   improved   by   about   4   percent,   suggesting   that  Model  4  is  better  at  explaining  the  variation  in  inflation  rates  than  Model  1.                                   169   Figure   4.7:   Political   Business   Cycles,   Organizational   Strength,   and   Financial   Resources       Figure 6b: Marginal Effect of Elections on Inflation Conditional upon Fiscal Revenues 1 0 Consumer Price Index (%) 0 -2 -4 -1 -2 Consumer Price Index (%) 2 2 4 3 Figure 6a: Marginal Effect of Elections on Inflation Conditional upon Proportion of Ruling Parties 0 .2 .4 .6 .8 1 Difference in Seats Proportion between Incumbents and Oppositions 21.5 22 22.5 23 Fiscal Revenues (log) Note:   The   straight   line   stands   for   inflation   rates   6   months   after   elections.   The   dotted   lines  represent  90  percent  confidence  intervals.                          I  expect  Models  5  and  6  to  test  more  directly  the  interactive  effects  of  organizational   strengths/state   resources   and   elections   on   CPI.   Here,   I   operationalize   organizational   strengths  by  calculating  the  difference  between  shares  of  parliamentary  seats  occupied   by   ruling   parties   and   opposition   parties.   Given   the   fact   that   the   centralization   of   government   institutions   coincides   with   the   strengthening   of   ruling   parties   and   weakening   of   oppositions   in   the   country’s   context,   taking   the   gap   between   the   seat   proportions   of   ruling   and   opposition   parties   works   as   a   good   proxy   to   measure   organizational   strength   in   Kazakhstan.   To   operationalize   government   resources,   I   use     170     logged   state   revenue   (Annual   data   from  IMF   country   reports).   Model   5   examines   how   the   impact   of   elections   will   change   according   to   the   level   of   the   government’s   organizational   strengths.   Figure   4.7a   visually   tells   us   that   inflation   after   elections   becomes   more   extensive   as   ruling   parties   occupy   more   seats   in   the   parliament   (confidence   intervals   are   the   90   percent   level).   Notably,   when   the   difference   of   the   proportions   between   the   parties   is   more   than   0.6,   the   government   is   more   likely   to   manipulate   the   economy   around   elections.   Model   7   tests   the   marginal   effect   of   post-­‐ electoral  periods  on  CPI  conditional  upon  the  logged  state  revenue.  The  coefficient  of  the   post-­‐election   dummy   changes   from   -­‐0.5   (when   logged   state   revenue   takes   the   minimum)   to   1.8   (when   logged   state   revenue   is   maximum)   and   turns   statistically   significant   when   the   logged   revenue   becomes   more   than   about   22.0   (Figure   4.7b).   Finally,   I   estimate   how   real   wage   increases   will   intensify   as   the   country   experiences   more   elections   in   Model   8.   The   coefficient   of   the   pre-­‐election   variable   is   positive   and   statistically  significant  at  the  10  percent  level,  suggesting  that  distributive  policy  before   elections  becomes  more  extensive  as  the  autocratic  regime  has  become  more  centralized   and   held   richer   financial   resources.   This   result   further   supports   my   argument   that   strong  organizational  strength  and  increasing  fiscal  revenues  are  two  important  factors   increasing  the  magnitude  of  pre-­‐electoral  economic  manipulation.81                                                                                                                             81  For  robustness  checks,  see  Appendix  C4.       171   4.7  Conclusion              This   chapter   has   empirically   examined   the   causal   mechanisms   posited   in   theory   by   conducting   an   in-­‐depth   case   study   of   Kazakhstan.   The   case   of   Kazakhstan   provides   us   with  an  intriguing  puzzle  on  electoral  manipulation  and  authoritarian  stability.  Although   President   Nazarbaev   seems   to   have   become   less   dependent   on   electoral   fraud   and   electoral   system   manipulation,   he   successfully   consolidated   his   rule   and   won   big   at   elections,   leading   to   Nur   Otan’s   perfect   dominance   of   the   parliament   in   2007.   I   have   explored   this   puzzle   of   authoritarian   elections   by   focusing   on   the   distribution   of   mobilization   power   between   the   president   and   political   elites.   My   in-­‐depth   case   study   has   shown   a   series   of   qualitative   and   quantitative   evidence   demonstrating   that   Nazarbaev   enhanced   economic   distribution   before   elections,   which   helped   him   score   an   overwhelming  majority  without  employing  extensive  electoral  fraud.  Further,  I  have  also   shown  that  Nazarbaev  succeeded  in  streamlining  pre-­‐electoral  economic  distribution  by   constructing  centralized,  hierarchical  political  organizations  such  as  the  dominant  party,   political  and  financial  centralizations  of  the  government,  and  a  top-­‐down  style  national   oil  company.  On  the  other  hand,  opposition  parties,  which  were  powerful  in  the  1990s,   suffered   financial   difficulties   and   internal   divisions.   This   reduced   the   need   for   the   president   to   employ   blatant   electoral   fraud   and   keep   to   adopt   single-­‐member   districts   system  that  would  continue  to  bring  a  seat  premium  to  his  ruling  coalition.  The  case  of   Kazakhstan   suggests   that   dictators   have   incentives   to   not   rely   completely   on   electoral   manipulation,  as  long  as  they  can  garner  voluntary  political  support  from  the  citizenry   through  extensive  distribution  of  money  and  goods.         172   CHAPTER  5     PROTESTS  AND  LEADERSHIP  TURNOVER     AFTER  AUTHORITARIAN  ELECTIONS     5.1  Introduction              Since   the   end   of   the   Cold   War,   growing   pressure   from   the   international   community   has   made   it   difficult   for   authoritarian   leaders   to   avoid   holding   periodical   elections.   Coinciding   with   the   proliferation   of   autocracies   with   elections   (Diamond   2002;   Schedler   2006;   Levitsky   and   Way   2010),   scholars   of   authoritarian   politics   began   to   draw   their   attention   to   elections’   role   in   authoritarian   regimes,   asserting   that   authoritarian   leaders   may   use   elections   as   a   tool   to   consolidate   their   rule   (e.g.   Magaloni   2006;   Gandhi   and   Lust-­‐Okar   2009;   Blaydes   2011).   However,   the   more   puzzling   fact   about   autocratic   elections   is   that   elections   do   not   always   benefit   autocrats   to   the   extent   some   research   suggests.  Rather,  elections  often  induce  more  political  conflicts  like  popular  protests  that   undermine   authoritarian   stability.   For   example,   the   Color   Revolutions   in   post-­‐Soviet   countries   (Georgia,   Ukraine,   and   Kyrgyzstan)   during   the   mid-­‐2000s   all   occurred   immediately   after   elections   (Tucker   2007;   Kuntz   and   Thompson   2009;   Bunce   and   Wolchik  2010).  Likewise  after  the  Côte  d’lvoire  2000  elections,  massive  protests  erupted   in   favor   of   opposition   parties,   which   subsequently   ousted   the   incumbent   president   Robert  Guéï.  Protests  allow  the  opposition  to  send  a  clear  signal  of  public  dissent  to  the   international   community.   Thus,   even   if   subdued   by   dictatorial   governments,   the     173   eruption   of   serious   protests   may   threaten   authoritarian   stability   in   the   long   run,   with   international  actors  tightening  economic  sanctions  and  adopting  coercive  diplomacy.              Another  form  of  unexpected  post-­‐electoral  political  change  is  leadership  turnover  as  a   result   of   elections   producing   surprising   results   for   the   incumbent   leader.   For   instance,   unpopular   incumbent   Prime   Minister   Bandaranaike   lost   the   Sri   Lankan   1977   elections   by  the  largest  margin  in  the  country’s  history  to  increasingly  popular  opposition  parties.   This  resulted  in  not  only  the  opposition’s  electoral  victory  but  also  the  strengthening  of   Tamil   opposition   people’s   sentiment   toward   separatism   (Samaraweera   1977,   1201),   which   helped   contribute   to   the   Sri   Lankan   Civil   War.   Similarly,   in   December   1991,   the   Algerian   President,   Chadli   Bendjedid   called   the   first   multi-­‐party   election   in   Algeria’s   history.  This  election  also  unexpectedly  brought  a  sweeping  victory  to  the  radical  Islamic   Salvation   Front   opposition   party,   triggering   a   military   coup   and   a   civil   war   (Bouandel   1993).   Separately,   in   the   1989   Polish   elections,   the   opposition   Solidarity   Movement   obtained  an  overwhelming  majority  both  in  the  lower  and  upper  houses,  paving  the  way   for  a  democratic  transition.  “No  one  in  the  political  elite  anticipated  the  replacement  of  a   Communist   government   by   a   Solidarity   government.   (…)   The   purpose   of   (…)   election   procedures   was   to   permit   Solidarity   to   enter   Parliament   but   to   preserve   the   continuation  of  Communist  rule”  (Olson  1993,  417).              Cross-­‐national   data82  on   78   authoritarian   countries   (1977-­‐2004)   shows   that   a   small   but   significant   minority   of   authoritarian   leaders   faces   either   political   turnover   or   popular   protests   after   elections:   14   percent   of   authoritarian   elections   experienced                                                                                                                   82  Data   range   in   time   from   1977   to   2004   and   is   compiled   using   Hyde   and   Marinov’s   (2012)   National   Elections   in   Democracy   and   Autocracy   (NELDA)   and   Kelley’s   (2012)   Quality  of  Elections  (QOE).     174   leadership  turnover,  whereas  popular  protests  occurred  in  19  percent  of  them.  Political   leaders   in   dictatorships   like   Indonesia   (1997),   Cameroon   (1993),   Azerbaijan   (2000,   2003),   and   Mexico   (1988,   1994)   experienced   post-­‐election   popular   protests,   while   Uruguay  (1984),  Bolivia  (1980),  Chile  (1988),  Haiti  (1995,  2000),  Sri  Lanka  (1977),  and   Liberia   (1997)   saw   their   elections   lead   to   political   turnover.   These   variations   in   post-­‐ electoral   outcomes   in   authoritarian   states   leave   us   with   several   puzzles:   why   do   authoritarian   elections,   which   are   expected   to   help   autocrats   to   stay   in   power,   often   backfire?   Specifically,   why   do   autocrats   face   two   different   types   of   threats–   popular   protests   and   political   turnover—after   elections   and   how   can   we   understand   the   sources   of  these  two  distinct  political  conflicts  in  dictatorship?            In   untangling   the   puzzling   relationships   among   protests,   turnover   and   authoritarian   elections,   this   chapter   suggests   an   answer:   authoritarian   leaders   likely   face   either   protests  or  overthrow  when  they  fail  to  successfully  manipulate  elections  in  light  of  their   mobilization  power.  To  do  so,  I  first  describe  the  dilemma  that  political  leaders  face  at   the  ballot  box.  Recent  literature  on  authoritarian  politics  suggests  that  autocrats  try  to   take  advantage  of  elections  to  show  their  invincibility  as  well  as  obtain  information  on   key   actors’   strengths   (e.g.   Magaloni   2006;   Geddes   2006;   Cox   2009;   Blaydes   2010;   Simpser  2013).  To  achieve  these  ends,  elections  need  to  be  free  and  fair  enough  to  make   the   results   believable.   On   the   one   hand,   if   election   results   are   seen   to   be   completely   predetermined,  then  authoritarian  leaders  cannot  enjoy  the  informational  benefits  of  a   more  competitive  election.  On  the  other  hand,  if  elections  are  too  free  and  fair,  it  is  more   likely   that   autocrats   fail   to   win   overwhelmingly.   In   other   words,   autocrats   face   a   serious   trade-­‐off   between   the   credibility   of   election   results   and   the   certainty   of   winning   big.     175   Under   the   constraint   of   this   dilemma,   authoritarian   leaders   need   to   carefully   make   a   decision  about  how  much  they  should  manipulate  elections.              To   what   extent   authoritarian   leaders   open   up   the   electoral   field   given   the   constraints   of   this   electoral   dilemma   will   be   determined   by   their   ability   to   mobilize   voluntary   popular  support  from  citizens,  or  autocrats’  mobilization  power,  as  discussed  in  Chapter   2.     When   the   political   leader   is   able   to   buy   a   large   portion   of   popular   support   through   extensive  pre-­‐electoral  economic  distribution,  he  will  be  able  to  win  big  without  relying   much   on   the   tools   of   electoral   fraud   such   as   election   violence,   electoral   cheating,   and   manipulation   of   electoral   law.   Thus,   if   the   autocrat   is   financially   strong,   the   more   credibly  he  is  able  to  signal  his  strength  by  producing  an  overwhelming  majority  at  the   polls   without   making   the   electoral   field   extremely   favorable   to   herself.   By   contrast,   when   the   authoritarian   leader   lacks   of   such   financial   resources,   he   is   unable   to   organize   large-­‐scale   mobilization   of   popular   support.   In   this   case,   fair   elections   are   more   likely   to   produce   surprising   results,   so   that   the   authoritarian   leader   has   a   strong   interest   in   biasing  election  results  by  engaging  in  electoral  manipulation.                  If   authoritarian   leaders   are   able   to   overcome   this   electoral   dilemma   by   optimally   setting  the  level  of  electoral  fraud  according  to  their  mobilization  power,  then  elections   contribute   to   authoritarian   stability   via   the   signaling   and   information-­‐gathering   functions.   The   more   complicated   fact,   however,   is   that   autocrats   may   have   difficulties   in   setting  the  appropriate  level  of  electoral  fraud  in  light  of  their  power.  When  this  is  the   case,   autocrats   fail   to   solve   the   electoral   dilemma,   and   they   are   more   likely   to   face   political   conflict   after   elections   –   popular   protests   or   political   turnover.   More   specifically,   I   argue   that   there   are   two   distinct   pathways   through   which   authoritarian     176   elections   induce   political   conflict.     First,   when   autocrats   underuse   electoral   fraud   relative  to  their  power,  election  results  are  more  likely  to  credibly  reveal  the  weakness.   This   brings   about   leadership   change   as   a   result   of   post-­‐electoral   coups   within   ruling   coalitions   or   via   opposition   parties’   electoral   victory.   Second,   when   autocrats   overuse   electoral  fraud  relative  to  their  power  of  mobilization,  elections  deteriorate  the  quality   of   electoral   information   and   hence   cannot   work   as   a   credible   tool   to   show   regime   strength.  This  encourages  post-­‐electoral  protest  movements.              In   order   to   test   these   empirical   implications,   I   conduct   a   cross-­‐national   statistical   analysis  including  78  authoritarian  countries  between  1977-­‐2004.  I  employ  a  two-­‐stage   estimation  to  test  my  theoretical  expectations.  I  first  run  a  model  in  which  I  predict  the   level   of   electoral   fraud   using   a   series   of   regressors   measuring   mobilization   power   of   authoritarian   leaders   and   other   controls   that   are   found   to   be   important   to   explain   electoral  fraud  in  the  previous  literature.  Then,  using  predicted  values  in  the  first-­‐stage   model,   I   measure   differences   between   the   predicted   level   of   electoral   fraud   that   the   autocrat  is  expected  to  employ  in  light  of  his  strength  and  the  real  level  of  electoral  fraud   that  he  actually  exercised  in  the  election.  In  doing  so,  it  is  possible  to  empirically  observe   how  well  the  autocrat  dealt  with  the  electoral  dilemma.  Then,  in  the  second-­‐stage  model,   I   estimate   the   likelihoods   of   leadership   turnover   and   popular   protests   using   probit   regressions   with   this   fraud   gap   variable.   My   empirical   analysis   shows   that   the   more   negative  the  gap  variable  is,  the  more  likely  elections  are  to  bring  leadership  turnover,   suggesting   that   the   autocrat’s   underuse   of   electoral   fraud   is   more   likely   to   lead   to   leadership   change.   By   contrast,   when   the   gap   variable   takes   positive   values   signifying     177   that  the  autocrat  overuses  electoral  fraud,  the  elections  are  more  likely  to  be  followed  by   popular  protests.       5.2  Literature  Review              The   extant   literature   of   authoritarian   politics   contends   that   formal   institutions   play   crucial  roles  in  consolidating  authoritarian  rule.  Since  dominant  parties  institutionalize   their   patronage   system   and   enable   the   autocrat   to   make   credible   commitment   to   the   internal   elite,   party   regimes   are   more   likely   to   survive   than   military   and   personalist   regimes   (Geddes   1999;   Magaloni   2008).   Multi-­‐party   legislatures   also   increase   autocrats’   survival   rate   because   it   provides   a   forum   through   which   autocrats   can   make   policy   concession   to   a   large   portion   of   society   (Gandhi   and   Przeworski   2007;   Gandhi   2008),   credibly  share  patronage  with  the  elite  (Malesky  and  Schuler  2010;  Blaydes  2011;  Boix   and  Svolik  2011),  and  divide  and  rule  opposition  parties  (Lust-­‐Okar  2004).  For  similar   reasons,   multi-­‐party   legislatures   also   make   political   order   stable   by   preventing   civil   war   and  labor  protests  (Vreeland  and  Gandhi  2004;  Kim  and  Gandhi  2010).              Among   these   institutions,   elections   have   been   seen   as   one   of   the   most   important   political   tools   that   autocrats   can   use   to   stay   in   power   (Gandhi   and   Lust-­‐Okar   2009;   Blaydes  2011).  First,  autocratic  elections  are  viewed  as  an  institution  for  authoritarian   leaders   to   acquire   information   on   competence   of   ruling   and   opposition   elites.   Semi-­‐ competitive   elections   provide   information   on   the   popularity   of   local   officials   and   candidates  in  their  electoral  districts  (Ames  1970;  Shi  1999;  Magaloni  2006).  The  total   number   of   votes   that   candidates   gain   in   their   districts   works   as   an   opportunity   for     178   autocrats   to   judge   who   among   the   elites   is   powerful   as   well   as   who   is   loyal   to   the   dictator   (Blaydes   2011).   Election   results   also   render   information   on   the   geographical   distribution  of  popular  support  for  opposition  parties  (Magaloni  2006;  Cox  2009;  Miller   2012).  Second,  elections  work  as  an  efficient  method  to  communicate  with  the  elites  by   conveying  information  on  regime  strengths.  By  holding  elections  and  winning  them  with   a   large-­‐margin,   autocrats   can   credibly   demonstrate   to   potential   opponents   that   the   regime   is   so   unshakable   that   any   rebellious   attempt   against   the   current   ruler   will   fail   (Simpser  2013;  Magaloni  2006;  Geddes  2006).              The   current   literature   of   authoritarian   politics   tends   to   focus   on   how   elections   help   autocrats   stay   in   power.   Other   strands   of   research,   on   the   other   hand,   have   suggested   that   elections   in   hybrid   regimes   and   authoritarian   regimes   often   contribute   to   democratization.   Lindberg   (2006,   2009)   argue   that   repetitive   elections   in   multi-­‐party   contexts   contribute   to   further   democratization   and   improve   the   quality   of   democracy   in   Africa.   Employing   a   comprehensive   cross-­‐national   dataset   covering   193   countries   between   1919-­‐2004,   Teorrel   and   Hadenius   (2009)   find   both   current   and   cumulative   effects   of   holding   elections   on   democratization,   which   resonates   with   Lindberg’s   finding   in   the   context   of   Africa.   In   a   similar   vein,   Roessler   and   Howard   (2009)   and   Brownlee   (2009)   assert   that   competitive   authoritarian   regimes   are   more   likely   to   democratize   than   both   hegemonic   and   closed   authoritarian   regimes.   In   this   context,   Huntington   (1991:   174)   notes   that   “the   lessons   of   the   third   wave   [of   democratization]   is   that   elections  are  not  only  the  life  of  democracy;  they  are  also  the  death  of  dictatorship.”            In   a   similar   vein,   researchers   also   maintain   that   fraudulent   elections   provide   an   opportunity  for  opposition  parties  and  anti-­‐regime  supporters  to  protest  (Tucker  2007;     179   Kalandadze   and   Orenstein   2009).   Although   most   protests   are   repressively   subdued,   some   post-­‐electoral   manifestations   of   public   dissent   include   large   scale,   anti-­‐ government   demonstrations.   In   some   cases,   these   demonstrations   lead   to   the   breakdown  of  authoritarian  regimes,  which  has  occurred  in  the  Philippines  (1986),  the   post-­‐Soviet   countries   (the   Color   Revolutions,   Tucker   2007;   Thompson   and   Kuntz   2009),   and  Côte  d’Ivoire  (2000).              In   reconciling   these   different   findings   about   authoritarian   elections,   scholars   have   begun   to   illuminate   the   background   conditions   determining   authoritarian   elections’   effects  on  democratization,  leadership  change  and  protest  movements.  Conducting  both   a   cross-­‐national   quantitative   analysis   of   31   competitive   authoritarian   countries   and   a   Kenyan   case   study,   Howard   and   Roessler   (2006)   find   that   election   results   tend   to   become  more  open  when  opposition  parties  succeed  in  forming  coalitions  and  launching   pre-­‐election   anti-­‐government   protests.   Donno   (2013)   also   asserts   that   competitive   authoritarian   states   are   more   likely   to   democratize   either   when   domestic   oppositions   form   coalitions   or   when   pre-­‐electoral   political   and   economic   conditionality   is   imposed   from   international   actors.   In   a   study   on   electoral   violence   in   developing   countries,   Hafner-­‐Burton,   Hyde,   and   Jablonski   (2013)   provide   cross-­‐national   evidence   that   serious   pre-­‐electoral   violence   is   positively   associated   with   the   probability   of   post-­‐electoral   protests.  Similarly,  Kuhn  (2012)  shows  that  electoral  fraud  increases  the  propensity  of   popular  protests  after  elections  (only  in  fairly  close  elections,  however).  And  Bunce  and   Wolchik   (2010)   emphasize   the   importance   of   the   opposition’s   electoral   campaign   strategies.   They   argue   that   in   hybrid   regimes   where   opposition   parties   can   carry   out     180   sophisticated,   energetic   electoral   campaigns   elections   are   more   likely   to   trigger   both   political  protests  and  leadership  turnover.              Echoing   these   previous   studies,   this   chapter   posits   conditional   hypotheses   about   authoritarian   elections’   effects   on   turnover   and   protests.   This   research,   however,   contributes  to  the  literature  in  two  different  and  original  ways.  First,  taking  into  account   the   costs   and   benefits   of   authoritarian   elections,   this   chapter   theoretically   and   empirically   endogenizes   the   authoritarian   leader’s   electoral   manipulation   calculations.   Assuming   that   autocrats   will   strategically   decide   on   a   level   of   electoral   manipulation   that   maximizes   the   informational   benefits,   I   argue   that   autocrats   will   likely   face   post-­‐ electoral  conflicts  such  as  leadership  turnover  and  protests  when  they  miscalculate  on   the   extent   of   electoral   fraud.   Second,   I   explain   both   leadership   turnover   and   popular   protests   in   a   unified   theoretical   framework.   Briefly,   I   argue   that   popular   protests   and   leadership  turnover  both  result  from  different  types  of  mistakes  that  autocrats  make  at   the  ballot  box.       5.3  Dictator’s  Calculus  over  Electoral  Manipulation              According  to  the  recent  literature  on  authoritarian  politics,  political  leaders  and  their   potential  opponents  are  more  likely  to  lack  reliable  informational  sources  to  know  each   other’s   strength   and   intention   (Wintrobe   1998;   Egorov,   Griev,   and   Sonin   2009).   Since   political   rights   and   civil   liberties   are   not   institutionalized   in   authoritarian   regimes,   people   are   difficult   to   know   to   what   extent   the   political   leader   is   able   and   popular   through   reliable   media   outlets.   In   such   circumstances,   potential   opponents   among     181   people   are   less   likely   to   accurately   estimate   the   strength   of   the   autocrat.   Such   misinformation  may  increase  the  likelihood  that  a  conflict  accidentally  occurs  between   an   autocrat   and   potential   opponents.   Strengthening   the   military   is   a   frequently   used   strategy   by   which   an   autocrat   can   credibly   demonstrate   his   power.   Yet,   history   suggests   that  a  heavy  reliance  on  the  sword  risks  an  autocrat’s  tenure  by  giving  the  military  too   much   power   (Svolik   2012).   Therefore,   strengthening   the   security   apparatus   is   not   a   perfect  solution.                On  the  other  hand,  the  political  leader  also  faces  difficulties  in  knowing  what  people   think   in   authoritarian   regimes,   because   people   have   an   incentive   to   conceal   their   preferences  fearing  tortures  and  repression  by  the  government  (Kuran  1991;  Wintrobe   1998).  This  is  problematic  because  if  he  is  not  familiar  with  the  distribution  of  popular   support,   it   is   more   difficult   to   govern   the   country   efficiently.   Strengthening   domestic   surveillance  may  be  an  available  option  to  the  autocrat.  Yet  such  methods  do  not  always   garner  high  quality  information,  because  in  such  situations  people  will  falsify  their  true   preferences   in   the   public   fearing   possible   sanctions   by   the   government,   as   previous   studies  acutely  pointed  out  (Wintrobe  1998).                                    Recent   studies   of   authoritarian   politics   see   elections   as   an   important   institution   to   overcome   this   information   shortage.   According   to   the   current   literature,   authoritarian   elections   enable   political   leaders   to   (1)   demonstrate   their   strength   via   large-­‐scale   electoral   mobilization   to   potential   opponents   (e.g.   Magaloni   2006;   Geddes   2006;   Simpser   2013)   and   (2)   acquire   information   on   the   distribution   of   popular   support   of   both   the   opposition   and   incumbent   politicians—both   of   whom   may   turn   against   the   political   leader   (e.g.   Magaloni   2006;   Cox   2009;   Blaydes   2010).   When   autocrats   call     182   elections,   however,   they   face   a   serious   trade-­‐off   between   the   certainty   of   gaining   an   overwhelming  victory  and  the  credibility  of  election  results,  as  I  discussed  in  Chapter  2.   Authoritarian  leaders  can  stay  in  power  by  winning  through  ballot  stuffing,  repression,   intimidation,   and   the   manipulation   of   election   rules   and   institutions.   Resorting   to   serious   electoral   manipulation,   authoritarian   leaders   can   effectively   deter   opposition   parties   from   winning.   Yet,   at   the   same   time,   excessive   electoral   manipulation   makes   elections  meaningless  or  even  harmful  to  their  authoritarian  rule  for  two  reasons.  First,   extremely  pro-­‐regime  election  results  make  it  difficult  for  autocrats  to  convey  a  credible   signal  of  their  regime’s  strength  to  potential  opponents  because  the  more  manipulated   elections   are,   the   less   election   results   reflect   the   autocrat’s   real   popularity.   In   such   predetermined   plebiscite   elections,   citizens   tend   to   be   indifferent   or   cynical   about   the   electoral  process  and  its  results  as  in  the  Soviet  Union  (e.g.  White  1988:  13;  Tedin  1994).   Therefore,   the   signaling   effect   of   elections   will   be   significantly   reduced   in   heavily   manipulated   elections.   The   second   problem   is   that   if   elections   are   just   a   façade,   autocrats   can   no   longer   obtain   accurate   information   about   popularity   of   potential   opponents  among  ruling  elites  and  opposition  leaders.  Obviously,  electoral  manipulation   biases   election   results   in   favor   of   the   autocrat,   so   that   election   results   will   suffer   non-­‐ negligible   noises   on   electoral   information.   Deteriorated   electoral   information   makes   it   very  difficult  for  autocrats  to  maintain  their  authoritarian  rule  efficiently.  This  is  because   autocrats  need  to  govern  the  country  without  reliable  information  that  would  have  been   obtained   if   the   political   system   had   been   more   transparent   (Wintrobe   1998;   Egorov,   Griev,  and  Sonin  2009).         183            Therefore,  although  which  electoral  benefits  autocrats  want  to  exploit  the  most  might   differ   across   countries,   they   all   have   incentives   to   open   the   electoral   field   and   introduce   some  degree  of  competition  via  electoral  reforms.  This  is  what  happened,  for  example,  in   the   Soviet   Union   during   the   perestroika   era   (White   1988),   in   village-­‐level   elections   in   Communist  China  (Shi  1999),  and  in  Mexico  during  the  PRI’s  heyday  (Eisenstadt  2004,   32-­‐44).   That   being   said,   it   does   not   necessarily   mean   that   relatively   free   and   fair   elections   are   always   good   for   autocrats.   Given   the   strategic   interactions   between   the   autocrat   and   potential   opponents   under   the   electoral   dilemma,   if   autocrats   make   elections   too   transparent,   then   they   are   more   exposed   to   risks   and   may   fail   to   obtain   electoral   victory   with   a   large   margin,   thereby   revealing   their   weaknesses.   This   may   give   an   opportunity   for   potential   opponents   to   challenge   the   political   leader.   On   the   other   hand,   excessive   electoral   manipulation   deteriorates   the   information   problems   that   I   mentioned   above,   which   makes   efficient   communication   between   the   autocrat   and   opponents  more  difficult.  Under  the  constraint  of  the  electoral  dilemma,  autocrats  need   to   decide   the   level   of   electoral   manipulation   while   considering   likely   responses   from   potential  opponents,  in  a  way  that  autocrats  can  balance  the  credibility  and  the  certainty   of  election  results.       5.4  Backfiring  at  the  Ballot  Box              When  an  autocrat  wins  an  election  by  an  overwhelming  margin,  the  total  number  of   votes   that   he   obtains   consists   of   “clean”   and   “dirty”   parts.   The   “clean”   part   is   the   total   number   of   real   votes   from   his   supporters.   These   citizens   vote   for   the   dictator   after     184   positively   evaluating   his   economic   and   policy   performance.   In   particular,   previous   studies   suggest   that   authoritarian   leaders’   popular   support   depends   on   the   breadth   of   their  distribution  of  economic  favors  to  the  citizenry  (e.g.  Magaloni  2006;  Greene  2007,   2009).   For   example,   non-­‐tax   revenues   like   natural   resource   wealth   or   foreign   aid   significantly   enrich   state   coffers   and   thus   enable   increased   public   spending   without   having  to  levy  taxes  on  their  citizens  (e.g.,  Ross  2001;  Desai  et  al.  2009;  Morrison  2009;   Wight,   Frantz   and   Geddes   2013).   Even   if   state   revenue   is   raised   by   taxation,   autocrats   can  buttress  public  support  by  selectively  collecting  taxes  from  opposition  loyalists  and   using   it   to   benefit   regime   supporters,   as   is   the   case   in   many   authoritarian   regimes   (Levitsky   and   Way   2010,   10-­‐11,   Chapters   5-­‐7).   Gaining   voluntary   support   is   therefore   costly  because  governments  must  invest  large  amounts  of  financial  resources  to  satisfy   their   citizens.   Therefore,   an   election   victory   upheld   by   costly   mobilization   of   citizens’   support  makes  election  results  credible  to  know  the  autocrat’s  strength  and  popularity.   In   this   study,   I   refer   to   citizens’   voluntary   support   for   the   political   leader   through   economic  and  policy  performance  as  the  leader’s  “mobilization  power.”                                      The  second,  “dirty”  part  is  the  total  number  of  votes  resulting  from  various  kinds  of   electoral   manipulation.   As   already   defined   before,   electoral   fraud   is   a   series   of   illegal   measures   that   bias   election   results   in   favor   of   the   political   leader   (Lehoucq   2003),   including  election  violence,  election  cheating,  and  undemocratic  restrictions  on  electoral   law.   Election   violence   is   physical   intimidation   exercised   largely   by   incumbent   parties   during   elections   (Straus   and   Taylor   2012;   Hafner-­‐Burton,   Hyde,   and   Jablonski   2013).   Using   electoral   violence   against   opposition   leaders   and   anti-­‐regime   supporters,   autocrats   can   undermine   oppositions’   effective   campaigns   and   decrease   opposition     185   supporters’   turnout.   Cheating   also   allows   autocrats   to   affect   the   electoral   result   with   nonviolent   but   still   illegal   measures   such   as   undermining   of   oppositions’   freedom   to   campaigns,  media  bias,  ballot  stuffing,  vote-­‐buying,  and  nonviolent  intimidation  (Kelley   2012).  Restrictions  on  electoral  laws  refer  to  a  series  of  regulations  that  prevent  citizens   and   electoral   candidates   from   effectively   participating   in   elections,   including   limits   on   voting  rights  based  on  certain  social  characteristics  such  as  gender  and  ethnicity,  flaws   in   the   complaints   procedures,   high   thresholds   for   new   parties   to   get   registered   and   gain   seats,   constraints   on   the   right   to   run   for   office   such   as   language   and   educational   requirements,   and   so   on   (Kelley   2012).   All   three   fraud   techniques,   though   different,   contribute  to  an  electoral  victory  with  a  margin  that  could  not  be  achieved  without  these   techniques.              Making   full   sense   of   his   mobilization   power,   if   the   autocrat   can   tactfully   match   the   level  of  electoral  fraud  with  his  strength,  he  can  exploit  as  much  informational  benefit  as   possible   while   maintaining   an   overwhelming   majority.   When   this   is   the   case,   elections   contribute  to  authoritarian  stability.  In  fact,  as  Chapters  2  and  4  showed,  authoritarian   regimes  with  substantial  financial  resources  and  a  weak  opposition  tend  to  have  lower   levels   of   electoral   fraud,   suggesting   that   authoritarian   leaders   strategically   manipulate   elections   based   on   their   ability   to   cultivate   voluntary   popular   support.   When   the   autocrat   fails   to   adequately   deal   with   the   electoral   dilemma,   the   elections   are   more   likely  to  backfire.  More  specifically,  autocrats  fail  to  deal  with  the  electoral  dilemma  in   two  ways.              First,   stability   may   not   be   achieved   when   authoritarian   leaders   are   overconfident   about   their   popularity,   hold   multi-­‐party   elections,   and   then   lose   a   supermajority   (or     186   even   an   electoral   victory).   Researchers   have   provided   substantial   anecdotal   evidence   and   noted   that   autocrats’   overconfidence   unexpectedly   paves   the   way   for   democratization   and   leadership   change—e.g.   in   Brazil   (1974),   Pinochet’s   Chile   (1988),   Marcos’   Philippines   (1986),   Myanmar   (1990)   and   Algeria   (1992)   (Huntington   1991:   174-­‐178;  Diamond  2008:  53-­‐54).  In  Poland,  for  example,  the  authoritarian  government   held   multi-­‐party   elections   in   1989   without   using   serious   electoral   fraud.   The   incumbent   government  did  not  doubt  its  popularity,  and  the  opposition  Solidarity  party  also  did  not   expect   its   eventual   electoral   triumph   (Olson   1993,   425).   Nevertheless,   after   the   vote   count,   Solidarity   scored   a   sweeping   electoral   victory,   which   resulted   in   Poland’s   transition   to   democracy.   Algeria’s   1991   election   exhibited   similar   characteristics   to   Poland’s;  however,  elections  there  did  not  result  in  democratization.  Algeria’s  president   did  decide  to  hold  multi-­‐party  elections  with  a  free  and  fair  electoral  process.  In  the  first   round  of  elections,  the  opposition  Islamic  Salvation  Front  emerged  victorious  with  87.7   percent   of   the   total   seats   decided   (Bouandel   1993,   13).   Fearing   the   rise   of   the   radical   Islamists,  the  army  annulled  the  election  results  and  removed  the  president  from  power   in   a   military   coup.   This   military   intervention   then   led   to   the   civil   war   between   the   government   and   Islamist   rebel   groups.   These   Polish   and   Algerian   cases   suggest   that   when   an   autocrat   does   not   increase   electoral   fraud   up   to   the   level   that   his   de   facto   weakness   demands,   election   results   can   credibly   reveal   his   regime’s   true   weakness   to   potential  opponents  and  lead  to  a  leadership  change.  Revealed  weakness  in  an  election   is  most  likely  to  result  in  leadership  turnover  via  electoral  victory  of  opposition  parties   and  hence  democratization  like  the  cases  of  Poland  and  Chile.  Or,  like  Côte  d’Ivoire  and     187   Algeria,  such  dictator’s  weakness  may  encourage  ruling  coalitions  to  change  their  leader   via  a  military  coup  or  civil  war.       Hypothesis  1:  When  an  autocrat  underuses  electoral  fraud  relative  to  his  power,  political   turnover  is  more  likely  to  occur  after  an  election.              Second,   autocrats   may   likely   face   another   type   of   political   conflict—popular   protests—after   they   use   excessive   electoral   fraud.   When   the   autocrat   excessively   rigs   elections,   potential   opponents   are   more   likely   to   think   that   election   results   will   be   largely   driven   by   political   manipulation,   rather   than   voluntary   popular   support.   Therefore,   the   signals   conveyed   by   the   elections   to   potential   opponents   are   more   mixed   when   it   comes   to   knowing   an   autocrat’s   true   popularity   and   strength.   In   particular,   previous   studies   suggest   that   “sticks”   (blatant   electoral   fraud)   without   sufficient   accompanying  “carrots”  (economic  favors)  encourages  potential  dissidents  to  speculate   that   the   regime   is   now   too   weak   to   hold   up   its   anti-­‐regime   collective   action.   As   Bunce   and   Wolchik   (2010,   38)   put   it,   “while   signals   in   the   admittedly   murky   political   environment  of  mixed  regimes  are  always  hard  to  read,  repression  can  also  be  read  as   an   indication   that   political   leaders   have   become   increasingly   nervous   about   their   hold   on   power.”   In   fact,   various   studies   show   that   both   harsh   repression   and   excessive   election  cheating  without  much  patronage  distribution  fuel  the  escalation  of  protests  in   authoritarian   regimes.   Bratton   and   van   de   Walle   (1997)   and   Wood   (2000)   argue   that   African   autocracies   (when   failing   to   provide   goods   to   citizens)   have   faced   anti-­‐regime   popular   mobilization   after   adopting   harsh   state   repression.   Investigating   the     188   experiences   of   South   Asian   countries   during   the   Cold   War   era,   Goodwin   (2001)   also   asserts   that   political   revolution   is   more   likely   to   occur   in   the   countries   where   the   government   has   relied   on   indiscriminate   violence   against   anti-­‐government   forces.   Color   Revolutions   in   post-­‐Soviet   countries   were   all   preceded   by   rigged   elections   (Tucker   2007).  In  the  "Tulip  Revolution"  in  Kyrgyzstan,  violence  perpetrated  by  state  police  and   electoral   fraud   exercised   by   incumbents   fueled   opposition   forces’   grievances,   which   activated  anti-­‐regime  mobilization  against  the  Akaev  regime  (Jones  2007).  Therefore,  I   hypothesize   that   after   being   exposed   to   excessive   electoral   fraud,   anti-­‐government   popular  protests  are  more  likely  to  be  observed.       Hypothesis   2:   When   an   autocrat   overuses   electoral   fraud   relative   to   his   power,   post-­‐ election  popular  protests  are  more  likely  to  occur.       5.5  Cross-­‐National  Statistical  Analysis     5.5.1  Data  and  Modeling  Strategies   In   order   to   empirically   test   the   two   hypotheses,   I   conduct   a   cross-­‐national   statistical   analysis.   The   unit   of   analysis   is   country-­‐election   year   in   an   authoritarian   country   between   1977-­‐2004.   I   limit   my   sample   to   authoritarian   countries   using   a   binary   classification   of   political   regime   by   Geddes,   Wright   and   Frantz   (2014),   a   frequently   used   dataset  to  identify  authoritarian  regimes  in  the  literature.              To   measure   the   gap   between   the   degree   of   electoral   fraud   and   the   autocrat’s   mobilization   power,   I   adopt   a   two-­‐stage   model.   In   the   first-­‐stage,   I   use   Ordinary   Least     189   Squares  (OLS)  to  regress  a  series  of  predictors  on  electoral  fraud,  which  is  continuously   measured   with   values   ranging   between   0   (no   fraud)   and   15   (most   fraudulent)   from   Kelley’s  (2012)  Quality  of  Elections  (QOE)  Dataset.  The  electoral  fraud  variable  includes   five  subcomponents  that  bias  election  results  in  favor  of  the  incumbent:  (1)  pre-­‐electoral   election   violence   (0-­‐3),   (2)   election-­‐day   electoral   violence   (0-­‐3),   (3)   pre-­‐electoral   election   cheating   (0-­‐3),   (4)   election-­‐day   election   cheating   (0-­‐3)   and   (5)   restrictions   on   electoral   participation   and   electoral   law   (0-­‐3).   0   indicates   no   fraud,   whereas   3   represents  for  serious  fraud.  To  measure  the  dictator’s  power  of  mobilization,  the  first   model   is   based   on   Chapter   2,   which   is   briefly   described   below   again.   In   addition   to   a   series   of   variables   measuring   mobilization   power,   I   also   include   other   variables   that   are   seen   as   relevant   in   the   study   of   electoral   manipulation.   I   then   calculate   the   gap   between   predicted  values  in  this  first-­‐stage  model  and  real  values  of  electoral  fraud.  In  the  second   stage,   I   regress   this   gap   variable   (predicted   values   –   real   values)   and   other   relevant   controls   on   the   two   dependent   variables   –   leadership   turnover   and   popular   protests.   Capturing  this  gap  enables  us  to  see  how  well  dictators  match  the  level  of  electoral  fraud   with  his  power  of  mobilization.  In  other  words,  adopting  this  two-­‐stage  estimation,  we   can   empirically   assess   how   successfully   the   electoral   dilemma   is   resolved.   If   the   gap   variable   takes   more   positive   values,   then   it   suggests   that   the   dictator   manipulates   elections   more   blatantly   than   he   needs.   My   theoretical   expectation   is   that   excessively   manipulated   elections   should   be   associated   with   a   higher   probability   of   popular   protests,  while  having  a  lower  likelihood  of  political  turnover.  When  the  variable  takes   more   negative   values,   elections   are   excessively   transparent   in   light   of   the   autocrat’s     190   strength.   Therefore,   I   expect   that   political   turnover   is   more   likely   to   follow   such   elections.       5.5.2  First  Model  Specification:  A  Mobilization  Model    As   I   explained   above,   the   dependent   variable   of   the   first   model   is   electoral   fraud.   To   measure   the   main   explanatory   factor,   the   autocrat’s   power   to   mobilize   popular   support,   I   focus   on   the   factors   that   I   introduced   in   Chapter   2.   The   first   is   to   what   extent   autocrats   possess  the  ability  to  efficiently  distribute  economic  favors  to  a  wide  range  of  citizens.   To   do   so,   they   need   to   have   (1)   substantial   financial   resources   and   (2)   strong   political   organizations   to   discipline   ruling   elites   (and   hence   streamline   economic   distribution).   As   discussed   before,   the   importance   of   financial   resources   to   buy   popular   support   has   been   established   by   previous   studies.   Without   abundant   public   resources,   autocrats   cannot   buy   off   popular   support   through   public   goods   provision   to   a   sufficient   extent   (see,   e.g.,   Ross   2001;   Morrison   2009).   To   make   economic   distribution   to   the   citizenry   efficient,   disciplinary   organizations   are   also   necessary   because   such   organizations   can   deter  ruling  elites  from  engaging  in  exploiting  state  resources.  To  measure  the  financial   resources  that  autocrats  control,  I  use  Ross’  (2011)  oil-­‐gas  value  per  capita  in  constant   2000   dollars.   This   variable   is   calculated   by   multiplying   a   country’s   total   oil   and   gas   production   by   the   current   oil   and   gas   price   and   then   dividing   this   amount   by   the   total   population.   The   oil-­‐gas   value   per   capita   variable   is   interacted   with   (1)   a   dominant-­‐party   regime  dummy  (Geddes,  Wright  and  Franz  [2014])  and  (2)  the  size  and  cohesiveness  of     191   politically   dominant   ethnic   groups83  (from   Cederman,   Min,   and   Wimmer’s   [2009]   Ethnic   Power   Relations   Dataset).   Making   long   lasting   power-­‐sharing   possible   between   the   autocrat   and   ruling   elites   (Magaloni   2008;   Svolik   2012)   and   thus   preventing   ruling   elites’   myopic   appropriation   of   state   resources,   dominant-­‐party   regimes   contribute   to   reducing   the   need   for   election   fraud   by   increasing   the   efficiency   of   economic   distribution.  Coherent,  large  politically  dominant  ethnic  groups84  make  it  easier  for  the   autocrat   to   monitor   ruling   elites’   behavior   (Fearon   and   Laitin   1995)   while   facilitating   public   goods   provision   to   a   large   portion   of   citizens   (e.g.   Alesina,   Baqir   and   Easterly   1999;   Habyarimana   et   al.   2008).   Thus,   such   dominant   ethnic   groups   help   dictators   streamline   economic   distribution.   I   expect   the   negative   impact   of   natural   resource   wealth  on  electoral  fraud  will  be  magnified  when  authoritarian  regimes  have  dominant   parties  and/or  less  fractionalized,  large  dominant  ethnic  groups.              A   second   way   to   measure   an   autocrat’s   mobilization   power   is   the   extent   to   which   clear  opposition  exists.  In  authoritarian  regimes,  challenging  an  autocrat  is  an  extremely   costly   political   behavior   because   in   most   cases   anti-­‐regime   protests   are   brutally   repressed   (Davenport   2007).   On   the   flip   side,   once   initiated,   this   costly   action   would   result  in  credibly  showing  the  authoritarian  leader  that  a  considerable  number  of  people   are   unsatisfied   with   the   regime   and   strong   opposition   does   exist   at   both   national   and   local   levels   (e.g.   Zimbabwe’s   Movement   for   Democratic   Change   during   the   2000s)   (Kuran   1991;   Kricheli,   Livne   and   Magaloni   2011;   Weiss   2012).   To   measure   anti-­‐                                                                                                                 83  This   variable   is   measured   multiplying   fractionalization   index   of   politically   dominant   ethnic  groups  by  the  proportion  of  the  dominant  groups  relative  to  total  population.         84  Here   politically   dominant   ethnic   groups   refer   to   ethnic   groups   that   have   access   to   political  posts  at  the  executive  level  in  the  country.       192   government  collective  action,  I  follow  the  previous  literature  like  Bueno  de  Mesquita  and   Smith   (2010)   and   Howard   and   Roessler   (2006:   372)   and   use   indicators   counting   the   number   of   demonstrations,   riots,   and   strikes   from   Arthur   Banks’   (2010)   Cross-­‐National-­‐ Time-­‐Series  Data  Archive.   I   compute   three   years   moving   averages   of   the   number   of   riots,   strikes  and  demonstrations  (one  year  lagged)  to  operationalize  opposition’s  strength.              Besides   these   variables   measuring   the   autocrat’s   power   of   mobilization,   I   also   add   control  variables:  regime  types  (personalist,  military  and  monarchy),  political  rights  and   civil  liberties  (measured  by  Freedom  House  Index,  using  three  years  moving  average  of   the  index  one  year  before  the  election),  election  administrative  capacity  (using  Kelley’s   [2012]   QOE),   logged   GDP   per   capita   (using   World   Development   Indicators   [WDI]   and   Maddison  2011,  one  year  lagged),  GDP  growth  (WDI,  one  year  lagged),  trade  openness   (measured  by  sum  of  exports  and  imports  relative  to  GDP,  using  Penn  World  Tables.  One   year  lagged),  rural  population  (WDI,  one  year  lagged),  types  of  elections  (if  an  election  is   parliamentary  in  parliamentarism  or  presidential  in  presidentialism,  then  1  is  assigned,   otherwise  0),   leader’s   tenure   length,  the   presence   of   domestic   and   international   election   monitoring  (Kelley  2012),  foreign  aid  (percent  of  GDP,  Ahmed  2012.  One  year  lagged),   and  decade  dummies.                   193   Table  5.1:  The  First-­‐Stage  Model  Predicting  Electoral  Fraud     Note:  Decade  dummies  are  included.  Clustered  robust  standard  errors  in   parentheses.    ***p  <  .01,  **p  <  .05,  *p<.10.           194              In  Table  5.1,  I  show  statistical  results  of  the  first  stage  model.  As  I  expected,  the  oil-­‐ gas   value   is   negatively   associated   with   electoral   fraud   when   autocrats   have   dominant   parties  or  more  coherent,  larger  dominant  ethnic  groups.85  The  collective  action  variable   is   positively   correlated   with   the   level   of   electoral   fraud.   These   results   suggest   that   stronger   dictators   with   mobilization   power   tend   to   refrain   from   using   a   series   of   manipulation   techniques.   R-­‐squared   is   0.446,   suggesting   that   the   first   model   explains   more   than   40   percent   of   total   variations   in   electoral   fraud.   Based   on   this   result,   I   compute  predicted  values  of  electoral  fraud,  which  is  showed  in  Appendix  C5.                                                                                                                                               85  The   interaction   between   the   oil   variable   and   the   ethnic-­‐organizational   power   index   is   jointly  statistically  significant  at  the  1  percent  level.         195   5.5.3  Second  Model  Specification:  Turnover  and  Protests   0 .05 Density .1 .15 .2 Figure  5.1:  Gap  in  Electoral  Fraud  under  Dictatorship   -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 Gap in Electoral Fraud (Predicted Values - Real Values) 6 7              To   measure   the   dependent   variable   for   Hypothesis   1,   leadership   turnover,   I   use   a   variable   capturing   broadly   defined   post-­‐electoral   leadership   turnover   from   Hyde   and   Marinov  (2012).    This  variable  is  coded  as  1  if  the  incumbent  leader  is  replaced  after  the   election,  0  otherwise  (NELDA39).  It  includes  all  types  of  post-­‐election  leadership  change   including  turnover  brought  by  hereditary  successions  and  nomination  of  the  next  leader   by  the  current  ruler  before  an  election.  As  these  types  of  leadership  change  do  not  relate   to  an  incumbent’s  electoral  performance,  I  remove  them  from  the  sample  by  referring  to   another  variable  (NELDA23).  This  variable  captures  if  a  successor  assumes  power  after   elections.  I  also  found  nine  additional  cases  where  leadership  change  occurred  because   of   pre-­‐electoral   successions   within   ruling   parties   —   rather   than   turnover   as     196   consequences   of   election   results.   I   excluded   these   cases   and   rerun   the   model   to   check   the  robustness  of  empirical  results.86    Further,  in  both  models,  I  do  not  include  cases  that   experienced  political  turnover  as  a  result  of  large-­‐scale  popular  protests  because  these   political   turnover   cases   are   not   driven   by   election   results   but   by   protests   (the   2003   Georgian  election  and  the  2000  election  in  Côte  d’lvoire).              My   second   dependent   variable,   popular   protests,   is   measured   using   the   NELDA   dataset.   NELDA   includes   a   variable   indicating   whether   there   were   riots   and   protests   after   the   election   (NELDA29).   If   either   riots   or   protests   occur   after   the   election,   then   the   variable  is  coded  as  1.  As  a  robustness  check,  another  variable  including  only  riots  and   protests  over  electoral  fraud  is  also  used  (NELDA30).                I   calculate   an   election   fraud   gap   by   using   predicted   values   from   the   first   model   and   then  taking  the  difference  between  predicted  and  real  values  of  fraud  (predicted  values   –   real   values).   Figure   5.1   shows   the   distribution   of   the   gap.   Using   the   gap   variable   as   the   main   independent   variable,   I   estimate   probit   models   to   empirically   test   my   theoretical   expectations.  My  empirical  tests  consist  of  two  parts  —  a  protest  model  and  a  turnover   model.  Regarding  controls,  I  include  the  same  set  of  control  variables  for  the  protest  and   turnover  models.87  I  add  political  rights  and  civil  liberties  (one  year  lagged,  measured  by   Freedom  House  Index’s  three  years  moving  average),  GDP  per  capita  (one  year  lagged,                                                                                                                   86  The   nine   cases   are   the   Tanzanian   1995   elections   (both   parliamentary   and   presidential),  the  Algerian  1999  elections  (presidential),  the  Mozambican  2004  elections   (both   parliamentary   and   presidential),   the   Namibian   2004   elections   (both   parliamentary   and   presidential),   and   the   Zambian   2001   elections   (both   parliamentary   and  presidential).               87  Even  if  I  try  every  different  combination  of  controls  in  both  models,  main  results  do   not  change.         197   WDI   and   Maddison   2011),   economic   growth   (one   year   lagged,   WDI),   trade   openness   (measured   by   Penn   World   Tables,   one   year   lagged),   rural   population   (WDI,   one   year   lagged),   types   of   elections,   presence   of   domestic   and   international   election   monitoring   (Kelley   2012),   election   administrative   capacity   (Kelley   2012),   election   boycotts   (Hyde   and  Marinov  2012,  NELDA  14),  military  spending  per  capita  (one  year  lagged,  Correlates   of   War   Project),   leader’s   age   (Goemans,   Gleditsch,   and   Chiozza   2009),   leader’s   tenure   length   (Goemans,   Gleditsch,   and   Chiozza   2009),   logged   population   (one   year   lagged,   WDI)   and   violent   conflict   incidence   (one   year   lagged,   from   PRIO’s   Armed   Conflict   Dataset,   Harbom   and   Wallensteen   2009).   I   also   control   for   regional   and   time   specific   heterogeneities   by   employing   regional   and   decade   dummies.   To   deal   with   possible   temporal   dependence,   duration   of   peace   years   is   also   included   in   all   models   (Beck,   Katz,   and  Tucker  1998).                                           198   5.5.4  Results   Table  5.2:  Probit  Analysis  of  Post-­‐Electoral  Turnover  and  Protests       Note:  Decade  dummies  and  regional  dummies  are  all  included  in  the  models.   Robust  standard  errors  are  in  parentheses.    ***p  <  .01,  **p  <  .05,  *p<.10.                 199     Figure  5.2:  Predicted  Probabilities  of  Turnover  and  Protests   (b) Efect of Fraud Gap on the Probability of Protests Probability of Protests (%) 40 60 40 0 0 20 20 Probability of Turnover (%) 60 80 80 (a) Effect of Fraud Gap on the Probability of Turnover -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 Gap in Electoral Fraud Gap in Electoral Fraud   Note:  Shaded  areas  are  the  90  percent  confidence  interval.  The  graphs   (a)  and  (b)  are  based  on  Models  1  and  3,  respectively.                      Table   5.2   reports   results   of   the   probit   analysis.   In   Model   1   where   the   dependent   variable  is  political  turnover,  the  fraud  gap  is  statistical  significantly  negative  at  the  .05   level.   This   suggests   that   if   autocrats   fail   to   increase   the   level   of   electoral   fraud   despite   their  need  to  do  so,  they  are  more  likely  to  experience  political  turnover  after  elections.   Model   2   limits   the   sample   by   excluding   the   nine   cases   where   political   turnover   was   driven  by  pre-­‐electoral  leadership  succession,  and  the  fraud  gap  variable  has  the  same   negative   and   statistically   significant   effect   on   the   likelihood   of   post-­‐electoral   turnover.   Based   on   Model   1,   Figure   5.2-­‐(a)   graphically   illustrates   how   a   predicted   probability   of     200   turnover  changes  as  the  fraud  gap  variable  increases.  When  the  gap  variable  takes  the   value   of   more   than   0,   the   predicted   probability   is   still   small.   Yet,   when   the   variable   becomes   more   negative   (between   -­‐1   and   -­‐7),   the   probability   of   turnover   exponentially   increases   in   a   statistically   significant   way.   When   the   variable   is   3,   the   probability   of   turnover   is   no   more   than   10   percent,   whereas   the   probability   increases   to   42   percent   when  the  gap  variable  is  -­‐6.  These  results  support  Hypothesis  1.  Closely  looking  at  the   data,   countries   such   as   Sri   Lanka   (1977),   Bolivia   (1980),   Honduras   (1981),   Guatemala   (1982),  Uruguay  (1984),  Zambia  (1991),  Azerbaijan  (1992),  Haiti  (1995,  2000),  Liberia   (1997),   and   Niger   (1999)   underused   electoral   fraud   in   their   elections,   resulting   in   political  turnover.                                                                                                                                                                                                                                                                                                                                                        Then,   Models   3   and   4   estimate   the   fraud   gap’s   impact   on   the   likelihood   of   popular   protests.   In   Model   3,   the   fraud   gap   has   a   positive   coefficient,   which   is   statistically   significant  at  the  .01  level,  meaning  that  when  elections  are  more  exposed  to  excessive   electoral  manipulation  relative  to  dictators’  mobilization  power,  they  are  more  likely  to   face  protesters  in  the  aftermath  of  elections.    In  Model  4  where  I  focus  only  on  protests   clearly  over  the  government’s  electoral  fraud,  a  similar,  positive  and  significant  effect  of   fraud   gap   is   confirmed.   Using   estimation   results   of   Model   3,   Figure   5.2-­‐[b]   shows   how   the   probability   of   protests   changes   with   the   values   of   the   fraud   gap.   When   the   gap   variable  is  negative  between  -­‐7  and  -­‐3,  its  impact  is  not  distinguishable  from  0.  Yet,  as   the   variable   gets   more   positive   and   bigger,   the   impact   of   the   fraud   gap   also   tends   to   increase.   For   instance,   when   the   gap   variable   is   0,   the   probability   of   protests   is   no   more   than   8   percent,   whereas   when   the   gap   variable   is   7,   the   probability   rises   up   to   57   percent.   Some   examples   in   which   overused   electoral   manipulation   induced   post-­‐   201   electoral  protests  include  Haiti  (1984),  Senegal  (1988),  Kenya  (1992,  1997),  Mauritania   (1992),   Cameroon   (1992),   Togo   (1994),   Indonesia   (1997),   Algeria   (1999),   and   Côte   d’Ivoire  (2000).         5.5.5  Robustness  Check            To  make  sure  the  extent  to  which  the  results  are  robust,  I  conduct  the  following  four   robustness  checks.88  First,  I  use  two  alternative  datasets,  Boix,  Miller,  and  Rosato  (2012)   and   Polity   IV,   to   identify   authoritarian   countries   and   reran   the   models   using   new   samples   of   authoritarian   countries.89  Second,   one   may   think   that   the   results   in   the   second-­‐stage   model   might   be   unstable   depending   on   model   specifications   of   the   first   model.  To  minimize  this  concern,  I  alternatively  use  the  electoral  fraud  variable  per  se  as   a  main  independent  variable  and  I  regress  it  on  the  dependent  variables  with  the  same   sets   of   controls.90  Finally,   I   try   every   possible   combination   of   control   variables   in   the   second   models   to   see   if   results   may   change   according   to   model   specifications   in   the                                                                                                                   88  The  estimation  results  are  available  in  Appendix  C5.       89  Boix,   Miller,   and   Rosato   (2012)   are   an   alternative,   binary   measure   of   political   regimes   (democracy   vs.   non-­‐democracies)   covering   1800-­‐2007.   Regarding   Polity   IV   (which   ranges   from   -­‐10   and   10),   I   use   a   conventional   threshold   of   Polity2   score   =   6   to   empirically   identify   non-­‐democracies.   If   a   country’s   Polity2   score   is   less   than   6,   then   the   country  is  seen  as  an  authoritarian  country.  As  Polity2  score  is  covering  countries  that   are  exposed  to  civil  war,  we  are  able  to  avoid  possible  bias  in  estimation  that  we  might   have   by   using   Cheibub,   Gandhi   and   Vreeland’s   (2010)   binary   measure,   which   excludes   countries  under  civil  war.               90  From  the  theoretical  point  of  view,  using  the  electoral  fraud  variable  per  se  does  not   necessarily   represent   the   idea   of   the   “fraud   gap.”   However,   a   correlation   between   the   fraud  gap  and  the  electoral  fraud  variable  is  very  high,  0.72.         202   second   stage   estimation.91  As   a   result   of   these   robustness   checks,   I   find   that   all   results   are  vertically  the  same  as  the  ones  that  I  reported  above.       5.6  Conclusion                This   chapter   explores   the   conditions   under   which   elections   contribute   to   political   conflict   in   authoritarianism—specifically   leadership   turnover   and   popular   protests.   Pointing  to  the  fact  that  authoritarian  leaders  face  a  trade-­‐off  between  the  certainty  of   winning   an   overwhelming   majority   and   the   credibility   of   election   results,   I   argue   that   when  autocrats  fail  to  match  their  electoral  fraud  to  their  de  facto  power  balance  with   political   elites,   elections   are   more   likely   to   be   followed   by   political   conflict.   A   cross-­‐ national   statistical   analysis   of   78   authoritarian   countries   from   1977   to   2004   rendered   empirical   support   for   my   theoretical   predictions.   The   theory   and   empirical   analysis   of   this  chapter  suggest  that  elections  are  a  double-­‐edged  sword  for  authoritarian  leaders:   Elections   may   provide   a   good   chance   for   autocrats   to   improve   information   shortage   inherent   in   authoritarian   regimes,   yet   the   failure   of   choosing   an   appropriate   level   of   fraud   backfires   on   authoritarian   leaders   themselves.   Recognizing   this   election   paradox   and   preventing   autocrats   from   flexibly   manipulating   elections   via   international   pressures   and   foreign   policies,   the   international   community   and   domestic   opposition   may   be   able   to   transform   authoritarian   elections   into   the   window   of   opportunity   to   achieve  democratization.                                                                                                                             91  Due   to   space   limit,   I   do   not   show   all   the   results   for   this   part   of   robustness   checks.   The   results  are  available  upon  request.         203   CHAPTER  6     The  FAILURE  OF  STRATEGIC  FRAUD  AND  POPULAR  PROTESTS:     THE  CASE  OF  KYRGYZSTAN     6.1  Introduction                        The   previous   chapter   empirically   examined   the   conditions   under   which   post-­‐ electoral  protests  and  leadership  change  are  more  likely  to  occur  through  cross-­‐national   statistical   analysis.   Centering   on   the   electoral   dilemma   and   power   distribution   in   dictatorships,  I  have  argued  that  excessive  electoral  fraud  is  more  likely  to  spark  popular   protests,   whereas   leadership   turnover   is   more   likely   to   occur   when   weak   dictators   mistakenly   open   the   electoral   field.     With   the   theory   in   mind,   this   chapter   provides   a   case  study  of  Kyrgyzstan  (1990-­‐2005)  to  illustrate  how  excessive  electoral  fraud  leads   to  post-­‐electoral  popular  protests.                There   are   two   reasons   why   I   choose   Kyrgyzstan   as   the   object   of   my   case   study   for   post-­‐electoral   political   conflicts.   First,   employing   the   method   of   difference,   two   similar   countries,   Kazakhstan   (Chapter   4)   and   Kyrgyzstan,   enable   me   to   explore   when   authoritarian   elections   may   (or   may   not)   backfire   on   dictators.   At   the   time   of   independence,  Kyrgyzstan  shared  many  commonalities  with  Kazakhstan,  such  as  ethnic   diversity,   implementation   of   radical   economic   reforms,   economic   crisis,   presidential   systems,   authoritarian   regime   types   (civilian   dictatorship),   center-­‐periphery   relationships,   and   a   trajectory   of   political   regimes   (nascent   democratization   after     204   national   independence   was   soon   followed   by   the   emergence   of   electoral   authoritarianism).   President   Askar   Akaev   of   Kyrgyzstan,   however,   faced   a   large-­‐scale   popular   protest   after   the   2005   parliamentary   elections,   the   so-­‐called   “Tulip  Revolution.”   On   the   other   hand,   as   I   demonstrated   through   the   in-­‐depth   case   study   of   Kazakhstan   (Chapter  4),  President  Nursultan  Nazarbaev  successfully  consolidated  his  authoritarian   rule   in   the   same   period   of   time.   Comparing   the   two   countries   under   the   most-­‐similar   systems  design  (Przeworski  and  Teune  1970),  the  two  case  studies  allow  me  to  focus  on   the   impact   of   the   variable   of   interests   on   post-­‐electoral   protests   while   controlling   for   many  other  factors.                Second,  in  addition  to  the  comparison  with  Kazakhstan,  I  also  conduct  a  within-­‐case   analysis   of   Kyrgyzstan   to   illustrate   to   what   extent   my   theory   can   explain   time-­‐series   variations   in   electoral   fraud   and   post-­‐electoral   protests   within   Kyrgyzstan.   Although   comparative   analysis   of   Kazakhstan   and   Kyrgyzstan   permits   me   to   match   many   important   covariates   influencing   electoral   fraud   and   post-­‐electoral   conflicts   because   both  countries  have  much  in  common,  there  are  still  important  differences  between  the   two  countries  that  could  potentially  influence  authoritarian  politics  and  regime  change   such   as   population   size,   country   size,   absolute   volume   of   natural   resource   wealth,   and   sensitivity  to  international  influences.  Comparing  different  elections  within  Kyrgyzstan   over   time,   I   am   able   to   control   for   these   additional   confounding   factors   that   are   very   different  from  those  of  Kazakhstan.              This  chapter  is  organized  as  follows.  In  next  section,  while  comparing  with  the  similar   experience   of   Kazakhstan   that   I   detailed   in   Chapter   4,   I   briefly   illustrate   processes   in   which  Kyrgyzstan  transitioned  from  an  electoral  democracy  to  an  electoral  authoritarian     205   regime  by  the  mid-­‐1990s.  Then,  in  stark  contrast  with  Kazakhstan,  I  show  that  President   Akaev’s   mobilization   power   had   tended   to   decrease   over   time,   preparing   background   conditions   for   the   Tulip   Revolution   in   March   2005.   Focusing   on   the   change   in   the   distribution  of  power  between  the  president  and  other  political  elites,  I  explain  the  level   of   electoral   fraud   in   the   1995   and   2000   elections   as   well   as   the   eruption   of   massive   popular   protests   after   the   2005   parliamentary   elections.   I   argue   that   President   Akaev   faced  massive  protests  in  2005  because  he  excessively  manipulated  the  2005  elections   expecting   that   he   would   not   collect   extensive   support   from   voters.   Such   excessive   manipulation   and   overwhelming   election   victory   at   the   elections   made   opposition   believe   that   the   regime   would   be   weaker,   given   the   high   unpopularity   of   the   Akaev   regime.  Lastly,  conclusions  follow.         6.2  From  Electoral  Democracy  to  Electoral  Authoritarianism              Much  like  President  Nazarbaev  of  Kazakhstan,  President  Askar  Akaev  also  embarked   on   political   liberalization   after   winning   the   presidential   election   of   the   Republic   of   Kyrgyz   in   1991.   Democratization   initiated   by   Akaev   was   deeper   than   that   in   Kazakhstan   and   admired   by   the   Western   media   broadcasting   that   Kyrgyzstan   was   an   “island   of   democracy”   in   authoritarian   Central   Asia.   Indeed,   during   the   first   couple   of   years   after   independence,   the   country   could   be   more   accurately   depicted   as   an   “electoral   democracy”   satisfying   the   procedural   definition   of   democracy.   First,   the   country   had   already  held  free  and  fair  presidential  elections  in  October  1991,  through  which  Akaev   was   elected.   Although   parliament   had   banned   the   Communist   Party   from   running   a     206   candidate,   no   other   candidates   were   prohibited   and   Akaev   and   his   backers   did   not   intimidate   opposition   parties.   Positively   assessing   Akaev’s   consensus-­‐based,   multi-­‐ ethnic  approaches  that  he  had  been  taking  after  the  ethnic  riots  occurred  in  Osh  in  June   1990   (Spector   2004,   8),92  both   democratic   activists   and   the   opposition   Democratic   Movement   of   Kyrgyzstan   strongly   supported   Akaev   and   did   not   offer   an   alternative   candidate  (Collins  2006,  179).  Although  there  was  a  lack  of  contest,  the  elections  were   widely   recognized   as   free   and   fair,   both   within   Kyrgyzstan   and   abroad   (Collins   2006,   179).              Second,   the   separation   of   power   between   the   executive   and   the   legislature   was   guaranteed  in  the  Constitution.  Since  independence,  there  had  been  a  long  process  and   much   debate   among   various   political   actors   over   the   Constitution.   After   numerous   discussions   within   the   government   as   well   as   on   the   public,   the   Constitution   was   established  in  May  1993.93    Although  Akaev  sought  to  endow  the  president  with  a  strong   power   to   implement   political   and   economic   reforms   under   a   presidential   system,   he   simultaneously   pursued   a   constitutional   framework   in   which   the   presidential   power   would   be   checked   by   the   legislature   and   judiciary.   Some   political   figures   in   the   presidential   office   such   as   Felix   Kulov   wanted   a   strong   president   and   a   weaker   legislature,   but   Akaev   himself   rejected   the   idea   to   adopt   such   a   “super-­‐presidential”   system  (Collins  2006,  182-­‐183).                                                                                                                           92  Before  the  1991  presidential  election,  Akaev  had  been  serving  as  the  president  of  the   Kyrgyz  Soviet  Socialist  Republic  for  one  year.       93  Regarding   detailed   political   processes   on   the   1993   Constitution,   see,   for   instance,   Anderson  (1999,  25-­‐27)  and  Collins  (2006,  179-­‐181;  182-­‐184).       207            Third,  political  parties  and  social  movements  grew  before  and  after  the  declaration  of   independence.   In   February   1991,   Akaev   signed   a   law   on   social   organizations,   which   created   a   framework   for   the   activities   of   associations,   interest   groups,   and   political   parties.   And,   in   practice,   the   law   paved   the   way   for   people   to   create   various   voluntary   associations.   By   February   1993,   the   Ministry   of   Justice   had   registered   258   social   organizations,   including   15   political   movements   or   parties,   31   professional   bodies,   21   national-­‐cultural  centers  or  organizations,  41  sporting  bodies,  11  children’s  and  young   people’s  and  5  women’s  organizations  and  the  number  had  been  further  increased  up  to   nearby  1,000  by  the  summer  of  1997  (Anderson  1999,  31).  Twelve  political  parties  were   registered   for   the   1995   presidential   and   parliamentary   elections,   including   the   Social   Democratic  Party  and  the  Ata  Meken  party,  opposition  parties,  both  of  which  stemmed   from   the   strongest   opposition   movement   at   that   time,   the   Democratic   Movement   of   Kyrgyzstan  (Collins  2006,  184).                    Lastly,  relatively  free  media  had  emerged  during  the  same  period.  Already  soon  after   October   1990,   when   Akaev   was   approved   as   the   president   of   the   Kyrgyz   Socialist   Republic  in  parliament,  the  official  media  such  as  Sovetskaya  Kirgiziya  (later  renamed  as   Slovo  Kyrgyzstana)   began   to   develop   an   independent   tone   and   informatively   explore   a   variety   of   political   issues,   followed   by   numerous   newly   established   newspapers   and   other   media   outlets     (Anderson   1999,   29;   Collins   2006,   186-­‐188).   One   of   the   most   critical  newspapers  was  the  parliamentary  paper  Svobodny  Gory,  which  came  to  severely   criticize  the  president.  Another  weekly  newspaper  Res  Publika  also  reported  corruption   scandals  (Anderson  1999,  30).       208            For   about   two   years   after   independence,   the   parliament   and   president   had   forged   cooperative   relationships   due   to   Akaev’s   tactful   co-­‐optation   techniques   toward   legislators   and   Medetkan   Sherimkulov’s   (the   speaker   of   the   parliament)   effectively   restraining   the   fractured   opposition   in   parliament   (Huskey   1997,   256-­‐257).   Yet,   as   economic   crisis   deepened,   different   views   on   radical   economic   reforms   between   the   parliament   and   the   president   contributed   to   heightening   tension   between   the   two.   Similar   to   Kazakhstan,   members   of   the   assembly   started   to   challenge   President   Akaev,   criticizing   alleged   practices   of   political   corruption   over   privatization   and   foreign   trade   among  government  officials.94  The  communists  in  the  parliament  went  on  the  offensive   against  the  president  because  most  of  them  were  supposed  to  lose  their  seats  due  to  the   new  Constitution  stipulating  that  total  seats  in  the  two  new  chambers  would  be  reduced   from   450   to   300   in   the   next   elections   (Spector   2004,   19;   Collins   2006,   227).   Midst   the   serious   confrontation   between   the   parliament   and   the   presidential   palace,   Prime   Minister   Tursunbek   Chyngyshev   was   forced   to   resign   in   January   1994   after   the   investigation  over  the  selling  of  mining  rights  at  the  Kumtor  goldmine  to  a  Canadian  firm,   Cameco   (Huskey   1997,   257).   This   is   also   very   similar   to   the   case   of   Prime   Minister   Tereschchenko  of  Kazakhstan,  who  presided  over  privatization  processes  from  October   1991  until  October  1994,  yet  ended  up  resigning  after  having  faced  the  similar  scandals   over   privatizations.   Chyngyshev’s   resignation   impressed   people   that   opposition   in   the   parliament  had  been  enhancing  political  pressures  on  the  president.                                                                                                                       94  For  detailed  analysis  on  economic  reform  such  as  privatization  and  foreign  trade,  see   Pomfret  (2006,  Chapter  5).       209            Facing   increasing   pressure   from   the   parliament,   Akaev   began   to   take   authoritarian   measures   by   late   1994,   turning   the   country   into   an   electoral   authoritarian   regime   (Huskey   1997;   Anderson   1999;   Spector   2004;   Collins   2006).   In   order   to   overcome   the   impasse   between   the   assembly   and   the   executive,   Akaev   strengthened   presidential   power   through   undemocratic   ways.   In   September   1994,   some   members   in   parliament   refused   to   attend   opening   session   of   the   fall   legislature.   Taking   advantage   of   this   opportunity,   President   Akaev   dismissed   the   parliament,   calling   for   new   elections,   because   the   1993   Constitution   stipulated   that   when   the   parliament   fails   to   satisfy   a   quorum   the   president   is   entitled   to   dismiss   it   (Collins   2006,   227).   At   the   same   time,   during   the   parliamentary   boycott,   Akaev   convoked   a   referendum   to   call   for   a   new,   smaller,   bicameral   parliament   having   only   105   seats   in   total   (the   Upper   House:   35   deputies;  the  Lower  House:  70  deputies)  as  well  as  the  electoral  law,  reducing  the  power   of   parliament   and   strengthening   presidential   power   (Collins   2006,   227-­‐228).   Further,   convoking   another   referendum   in   1996,   Akaev   made   amendments   on   the   1993   Constitution  and  strengthened  the  formal  power  of  the  president.  This  violated  the  1993   Constitution   because   constitutional   changes   via   referenda   were   prohibited   on   the   Constitution  (Human  Rights  Watch  World  Report  1997,  227).            Media  and  social  movements  also  were  beginning  to  be  exposed  to  intimidation  from   the   government   from   the   mid   1990s.   Harassment   toward   the   media   and   opposition   became   more   serious   over   time   during   the   Akaev   regime.   In   June   1994,   the   government   began   to   restrict   the   media   freedom   by   closing   down   two   newspapers,   including   the   parliamentary   paper   Svobodnye  Gory,   which   was   highly   critical   of   the   president.   In   the   spring   of   1995,   the   president   launched   a   criminal   prosecution   for   defamation   against     210   Zamira   Sydykova   and   Tamara   Slashcheva,   editors   of   the   independent   newspaper,   Res   Publica   (Huskey   1997,   258).   Besides   these   direct   measures,   the   government   put   pressure  on  state-­‐run  papers  to  engage  in  self-­‐censorship  and  on  other  media  to  replace   their   editors   from   1995   to   1997   (Anderson   1999,   57).   The   number   of   incidents   of   intimidation   against   opposition   leaders   and   religious   activities   grew,   seriously   threatening   pluralism   in   the   country   (Anderson   1999,   56-­‐59).   By   the   late   1990s,   the   optimistic  assessment  of  the  West  toward  the  prospect  of  democracy  in  Kyrgyzstan  had   rapidly   faded;   an   electoral   authoritarian   regime   had   emerged,   much   like   that   of   Kazakhstan.                                 6.3  The  Weakening  of  Akaev’s  Mobilization  Power            There   are,   however,   important   differences   between   Akaev   and   Nazarbaev   in   temporal   changes   in   their   mobilization   power.   And,   I   argue   that   this   difference   led   to   the   contrasting   impact   of   authoritarian   elections   on   political   order   –   massive   protests   and   authoritarian   breakdown   after   the   2005   elections   in   Kyrgyzstan   and   post-­‐electoral   consolidation   of   the   Nazarbaev   regime   in   the   2000s.   As   discussed   in   Chapter   4,   Nazarbaev   succeeded   in   enhancing   his   mobilization   power   vis-­‐à-­‐vis   other   political   elites   primarily  with  the  help  of  growing  natural  resource  wealth  in  the  late  1990s  to  the  late   2000s,   reducing   the   need   for   him   to   use   extensive   electoral   manipulation.   Conversely,   Akaev  tended  to  deteriorate  his  mobilization  power  over  time,  which  encouraged  him  to   gradually   resort   to   blatant   electoral   fraud   in   elections.   Facing   declining   power,   he   consequently   used   excessive   fraud   in   the   2005   elections   and   won   the   elections     211   overwhelmingly,   which   triggered   popular   protests   mobilized   by   opposition   elites   who   believed   the   regime   was   already   weak.   While   comparing   with   Kazakhstan,   I   examine   Akaev’s  decreasing  power  of  mobilization  by  looking  at  the  three  factors  in  my  theory  of   authoritarian  elections:  (1)  Financial  resources,  (2)  weakening  of  organizational  bases,   and  (3)  the  emergence  of  strong  opposition.             6.3.1  Financial  Resources:  Gold  and  Aid            Although   he   initially   possessed   financial   resources   stemmed   from   gold   and   foreign   aid,   the   amount   of   such   resources   to   which   Akaev   was   accessible   to   distribute   had   declined   from   the   mid-­‐1990s   all   the   way   up   to   the   mid-­‐2000s.   Different   from   Kazakhstan,   Kyrgyzstan   does   not   possess   rich   natural   resources   such   as   oil   and   gas.95   Only   relevant   source   of   natural   resource   wealth   in   the   country   is   the   Kumtor   gold   mine,   which  accounted  for  nearly  50  percent  of  industrial  output  in  Kyrgyzstan  between  1996-­‐ 2000   (Pomfret   2006,   80).   In   2002,   when   a   landslide   occurred   and   shut   down   the   gold   mine,   GDP   growth   dropped   to   zero,   indicating   that   the   country’s   whole   economy   was   highly   dependent   on   this   gold   mine   (Pomfret   2006,   80).   Yet,   compared   to   Kazakhstan,   the   gold   wealth   did   not   necessarily   contribute   to   improving   Akaev’s   mobilization   power   especially   after   the   2000s   for   two   reasons.   First,   mineral   income   per   capita   in   the   country   was   still   too   small   to   dramatically   improve   people’s   living.   Gold   income   per   capita   accounts   for   only   1   percent   of   GDP   per   capita,   which   is   far   smaller   than   Kazakhstan’s   natural   resource   wealth   occupying   15   percent   of   GDP   per   capita   on                                                                                                                   95  According   to   Ross   (2011),   oil-­‐gas   value   per   capita   is   only   3.9   USD   on   average   between   1992-­‐2005.         212   average  between  1995-­‐2005.96  Second,  especially  from  the  early  2000s,  it  is  said  that  the   gold   wealth   were   put   into   high   officials’   pockets   for   their   personal   use.   In   2004,   a   financial   restructuring   of   the   joint   venture   for   the   Kumtor   mine   created   a   new   company   Canterra  listed  on  the  Tronto  Stock  Exchange.  The  company  was  politically  controversial   because   high-­‐ranking   government   officials   were   accused   of   enjoying   vast   amount   of   rents  from  the  company  (Pomfret  2006,  80).              As  Kyrgyzstan  did  not  possess  abundant  natural  resources,  Akaev  attempted  to  fill  the   gap   by   turning   to   foreign   aid   and   stabilization   loans   from   international   organizations   and  the  Western  countries.  One  of  his  motivations  behind  rapid  political  and  economic   reform   was   to   appeal   to   the   international   community   to   receive   financial   assistance   (Tordoff   1995,   496).   Akaev   “took   the   view   that   only   by   adopting   such   a   pro-­‐reform   position   in   advance   of   many   neighboring   states   could   his   country   hope   to   attract   investment   and   economic   support   from   the   outside   world”   (Anderson   1999,   75-­‐76).   Responding   to   the   radical   reforms,   international   organizations   agreed   to   provide   extensive  financial  support  to  Akaev;  the  IMF  provided  over  60  million  dollars  to  back   up  the  introduction  of  the  Kyrgyz  som  in  May  1993;  the  World  Bank  offered  a  number  of   substantial   credits   to   support   the   reform   programs;   the   European   Bank   for   Reconstruction  and  Development  (EBRD)  and  the  Asian  Development  Bank  (ADB)  also   provided   large   amounts   of   financial   assistance   to   smooth   privatization   processes   (Anderson   1999,   76).   In   sum,   foreign   aid   from   the   Western   countries   had   increased                                                                                                                   96  I  made  this  comparison  on  people’s  dependence  on  natural  resource  wealth  by  using   Haber  and  Menaldo  (2011).       213   dramatically  from  less  than  25  million  US  dollars  in  1992  to  more  than  275  million  US   dollars  in  1995  (McGlinchey  2011,  89;  World  Development  Indicators).                    Yet,   as   Akaev   turned   into   an   authoritarian   leader   and   strengthened   authoritarian   rule,  financial  assistance  from  the  international  community  had  been  cut,  leading  Akaev   to  suffer  a  lack  of  financial  resources  from  the  early  2000s.  Foreign  aid,  which  had  been   maintained   between   230   and   300   million   USD   between   1995-­‐1999,   decreased   to   less   than  200  million  USD  between  2000-­‐2003.  In  addition,  serious,  long-­‐standing  economic   downturn  after  the  rapid  economic  reform  made  it  difficult  to  improve  tax  revenues.              Essentially,  financial  resources  stemmed  from  foreign  assistance  are  more  difficult  to   use   directly   for   political   purposes   than   natural   resource   wealth   because   international   actors  pay  close  attention  to  how  the  central  government  uses  foreign  aid.  As  a  legislator   in  the  parliament,  who  had  been  working  in  the  agency  of  state  budget  during  the  Akaev   regime,  stated:  “Akaev  and  his  government  could  not  use  foreign  assistance  for  electoral   purposes   through   fiscal   policies   due   to   international   organizations’   monitoring   budget   of   the   country”   (Author’s   interview   with   a   deputy   [#33]).   Yet,   as   I   discuss   later,   Akaev   tended  to  share  the  use  of  foreign  aid  with  local  ruling  elites,  which  helped  him  improve   his   mobilization   power   in   locality   during   the   mid-­‐1990s.   This   in   turn   enabled   him   to   hold  relatively  fair  and  free  elections.              In  the  wake  of  the  Afghanistan  War  in  2001,  the  Kyrgyz  government  decided  to  supply   the   air   base   at   Manas   to   the   United   States,   which   resulted   in   generating   unexpected   windfalls   to   the   Akaev   regime.   It   is   estimated   that   U.S.   air   base   at   Manas   provided   on   average   revenues   of   40   million   USD   for   their   fuel   subcontracts,   2   million   USD   for   rent   of   the   U.S.   air   base,   as   well   as   7   thousand   dollars   every   time   a   U.S.   military   took   off   from     214   Manas   to   the   Akaev   government   between   2003   and   2005   (McGlinchey   2011,   98).   All   these   payments,   however,   never   passed   through   official   Kyrgyz   accounts.   The   Akaev   family   was   alleged   to   have   put   the   money   into   their   pockets   because   family   members   controlled   companies   operating   the   Manas   International   Airport   (McGlinchey   2011,   98).   Therefore,   fiscal   revenues   from   U.S.   air   base   did   not   contribute   to   the   president’s   ability   to  distribute  economic  favors  to  other  actors.                             6.3.2  Decentralization  and  Lack  of  Organizational  Bases                    In   stark   contrast   with   President   Nazarbaev   of   Kazakhstan   who   gradually   attained   the   centralization   of   his   government,   President   Akaev   could   not   build   up   centralized   political   organizations   to   streamline   the   distribution   of   economic   resources.   Economic   reform   and   privatization   promoted   de   facto   economic   decentralization   in   the   country.   Apart   from   Nazarbaev   who   always   carefully   dealt   with   the   risk   of   fiscal   decentralization   and  controlled  the  processes  of  privatization  for  fear  of  opposition  forces,  Akaev  allowed   economic  decentralization.  According  to  Jones  Luong  (2002,  115),  Akaev  was  “convinced   that  a  significant  degree  of  influence  over  the  economy  must  be  decentralized  in  order   for   the   transition   to   the   market   to   succeed.   He   was   hesitant   to   design   an   economic   reform  agenda  from  the  center  without  careful  consideration  of  local  conditions,  which   he   believed   regional   leaders   were   in   the   best   position   to   determine.”   His   decision   to   delegate   economic   decision-­‐making   to   local   governments   resulted   in   generating   independent   businessmen   and   strong   local   elites   who   hold   economic   resources   independent   of   the   central   government.   The   privatization   of   land   did   not   empower   ordinary  farmers  but  benefit  collective  farm  directors,  most  of  whom  were  strongmen  in     215   locality   because   the   latter   could   control   over   privatized   land   by   utilizing   personal   connections  with  regional  governments  (Bloch  and  Rasmussen  1998,  125;  Radnitz  2010,   62-­‐63).  Local  businessmen,  who  launched  their  companies  amid  of  the  privatization  and   received  large  profits,  often  ran  for  legislative  elections  from  their  strongholds  by  using   their   own   local   networks   and   independent   resources   and   gained   parliamentary   seats   (Radnitz  2010;  Sjoberg  2011).              Political  decentralization  went  hand-­‐in-­‐hand  with  economic  decentralization.    “At  this   time  [November  1990],  and  especially  after  his  popular  elections  as  the  first  president  of   independent  Kyrgyzstan  the  following  October,  Akaev  had  the  opportunity  to  decrease   or  at  least  impose  greater  limits  on  regional  akims’  growing  authority.  Yet,  he  did  little  to   halt   the   ‘spontaneous’   devolution   of   power,   and   instead,   supported   policies   that   directly   contributed   to   this   process”   (Jones   Luong   2002,   108).   In   March   1992,   he   supported   amendments   to   the   Law   on   Local   Self-­‐Governance   and   Local   Administration   in   the   Republic   of   Kyrgyzstan,   which   strengthened   the   role   of   akims   in   decision-­‐making   and   the  implementation  of  policies  at  the  local  level  (Jones  Luong  2002,  108-­‐109).  Akims  had   strong  powers  in  their  regions.  They  supervise  departments  in  their  oblast  governments,   serving  as  the  personal  and  political  representatives  of  the  head  of  governments.  Taking   advantages   of   their   local   authority,   they   also   have   an   extensive   local   network   “held   together  bonds  of  friendship  and  loyalty,  [while  having]  exerc[ing]  extensive  patronage   (Tordoff  1995,  500).  They  decide  their  own  budgets,  relying  less  on  transfers  from  the   central   government,   especially   in   autonomous   oblasts.   For   example,   the   Southern   oblasts  such  as  Osh  and  Jalal-­‐Abad,  opposition’s  strongholds  critical  of  Akaev,  tended  to   not   receive   government   transfers   in   1993   (0   percent   and   16   percent,   respectively.     216   Tordoff  1995,  502)  and  the  high  autonomy  of  regional  governments  was  not  deprived  by   the  president  and  rather  tended  to  have  got  strengthened  until  the  collapse  of  the  Akaev   regime   in   2005   (Radnitz   2010).   Between   1992-­‐1995,   7   akims   out   of   8   came   from   the   same  oblasts  (Jones  Luong  2002,  290),  indicating  that  decentralization  had  been  much   more  advanced  than  that  of  Kazakhstan.  It  was  very  difficult  for  Akaev  to  parachute  his   own  favorite,  as  an  akim  into  regions  where  the  latter  did  not  have  local  networks,  thus   was   unable   to   obtain   support   from   local   populations.   Siegel   (2014)   documents   that   only   12  out  of  78  akims  between  1991-­‐2014  was  the  ones  who  “slided”  from  an  akim  of  one   region  to  that  of  another,  suggesting  that  the  center  needed  to  consider  strong  regional   power  in  the  cadre  rotation  of  akims.              It   is   not   hard   to   imagine   that,   in   a   country   where   regional   elites   were   very   strong   vis-­‐ à-­‐vis   the   president,   its   party   system   was   not   well   institutionalized.   The   under-­‐ development  of  national  party  systems  makes  the  president’s  organizational  base  very   weak  because  he  cannot  use  the  party  organization  to  discipline  ruling  elites.  The  under-­‐ institutionalization   of   national   party   systems   is   manifested   by   the   following   two   factors:   (1)   frequent   realignments   of   political   parties   and   (2)   proliferation   of   independent   politicians  who  hold  political  and  financial  resources  independent  of  political  parties  (cf.   Hale  2006).  Soon  after  independence,  numerous  political  parties  emerged  in  the  country,   yet  most  of  them  were  initiated  by  powerful  regional  leaders  and/or  strongmen  in  clans   and   thus   attempted   to   represent   only   local   interests   in   parliament   (Jones   Luong   2002,   112-­‐114;   Collins   2006,   231-­‐240):   “Parties   did   not   serve   the   function   that   they   do   in   Western   democracies,   aggregating   the   interests   of   society   and   translating   those   preferences   into   public   policy.   Instead,   parties   were   vehicles   for   ambitious   elites   to   gain     217   or  retain  a  seat  in  parliament.  They  were  regionally  concentrated  and  weakly  rooted  in   society,   and   had   little   organizational   capacity”   (Radnitz   2010,   71).     In   the   1995   legislative  elections,  there  were  11  political  parties,  which  obtained  at  least  one  seat  in   the  parliament  (Nohlen  et  al.  2001,  447),  indicating  that  the  national  party  system  was   extremely   fragmented.   In   addition,   their   support   bases   were   concentrated   on   certain   regions.   For   example,   Erkin   Kyrgyzstan,   an   opposition   party   led   by   Topchuibek   Turgunaliev,  had  its  main  supporters  in  the  Osh  oblast,  whereas  Ata  Meken,  which  was   formed   when   it   split   from   Erkin   Kyrgyzstan,   had   the   Jalal-­‐Abad   oblast   as   its   main   support   base.   Also   in   the   2000   legislative   elections,   in   which   a   party   list   with   a   nationwide  district  was  introduced  for  15  seats  out  of  60  to  make  the  party  system  more   nationalized,  9  political  parties  gained  at  least  one  seat  and  their  regional  concentrations   of   supporters   were   still   salient   (Nohlen   et   al.   2001;   Abazov   2003,   548;   Collins   2006,   240).   The   Kyrgyz   party   system   did   not   achieve   nationalization   but   rather   remained   highly   fractionalized,   contrasted   with   Kazakhstan   where   especially   pro-­‐presidential   parties  became  less  fractionalized  during  the  same  period  of  time.              Independent   politicians   also   proliferated   in   elections   and   the   proportion   of   independents  did  not  decrease  during  the  Akaev  regime.  Because  the  country  was  highly   decentralized,   independent   politicians   tended   to   have   close   relationships   with   autonomous   akims   and   local   “notables”   and   thus   they   were   not   necessarily   pro-­‐ presidential   (Collins   2006,   237-­‐238),   which   was   different   from   Kazakhstan   where   independents  were  mostly  supporting  the  president,  as  discussed  in  Chapter  4.    In  the   1995   legislative   elections,   63.8   percent   of   elected   legislators   were   independents.   The   proportion   of   independents   in   the   2000   parliamentary   elections   increased   to   69.5     218   percent  (Nohlen,  Grotz  and  Hartmann  eds.  1999,  447),  strongly  suggesting  that  the  party   system   remained   to   not   be   institutionalized   and   was   more   driven   by   self-­‐nominated   politicians.   This   also   makes   a   large   difference   with   the   case   of   Kazakhstan.   Although   Kazakhstan   had   a   large   number   of   independents   until   the   early   2000s,   Nazarbaev   succeeded  in  gradually  incorporating  these  independents  into  ruling  parties.              More   important   is   that   Akaev   could   not   consolidate   his   dominant   party   in   such   a   fractionalized   party   system   and   regionally   dispersed   clan   networks.   In   the   1995   legislative   elections,   the   pro-­‐presidential   party   Social   Democratic   Party   got   only   13.3   percent   of   total   seats.   Also   in   the   2000   legislative   elections,   the   two   pro-­‐presidential   parties,  the  Union  of  Democratic  Forces  and  My  Country,  obtained  15.2  percent  of  total   seats,   which   was   far   from   a   majority   (Nohlen,   Grotz   and   Hartmann   eds.   1999,   447;   Abazov   2003,   551).   In   the   party   list   PR   portion   of   the   2000   elections,   the   pro-­‐ presidential   parties   scored   only   25.1   percent   (Nohlen,   Grotz   and   Hartmann   eds.   1999,   445),   suggesting   very   weak   mobilization   power   of   Akaev,   given   that   this   was   achieved   only   after   his   seriously   having   rigged   the   elections.     Two   years   before   the   2005   parliamentary  elections,  Akaev  created  a  new  pro-­‐presidential  party,  Alga  Kyrgyzstan  by   merging   the   existing   pro-­‐presidential   parties,   seemingly   mimicking   dominant   parties   like  Russia’s  United  Russia  and  Kazakhstan’s  Otan.  Yet,  given  that  the  Alga  party  failed  to   coordinate   with   the   other   pro-­‐presidential   party   and   candidates   in   the   2005   elections,   the  party  did  not  incorporate  ruling  elites  very  well  (Uyama  2006,  49).               219   6.3.3  Emerging  Opposition                Lastly,  once  it  turned  out  that  Akaev  could  not  maintain  financial  resources  to  share   with   other   political   actors   and   cement   strong   political   supporters   via   powerful   organizational  bases,  strong  opposition  leaders  began  to  challenge  him.  After  the  1995   elections,   political   elites   who   had   local   networks   in   the   Southern   regions,   began   to   dissent   against   the   president.   Two   lawmakers,   Omurbek   Tekebaev   and   Dooronbek   Sadyrbaev,  played  an  important  role  to  form  opposition  in  the  south,  being  followed  by   Usen   Sydykov,   a   deputy   chairman   of   the   Council   of   Ministers   (Radnitz   2010,   74).   As   Akaev  had  lost  much  of  his  popularity,  opposition  had  gotten  stronger  by  the  time  of  the   2000  elections,  while  also  involving  prominent  Northern  elites.    In  particular,  powerful   ruling   elites,   who   previously   supported   Akaev   and   thus   occupied   important   positions   in   the   government,   defected   from   the   regime   and   formed   opposition   parties   while   taking   advantage  of  their  dense,  patronage  networks  with  local  populations.  Felix  Kulov,   who   came   from   the   north   and   had   served   as   various   important   posts   in   the   government   including  the  Mayer  of  Bishkek  between  1998-­‐1999,  determined  to  form  the  opposition   party   Ar-­‐Namys   in   1999   and   declared   to   participate   in   the   2000   legislative   elections.   Daniyar   Usenov,   another   famous   politician   from   the   north,   also   turned   into   opposition   and  proclaimed  to  run  for  the  2000  presidential  elections  as  an  opposition  candidate.                Up  until  the  2000  elections,  these  opposition  forces  had  been  rather  divided  without   making  any  efforts  in  coordinating  their  opposition  movements  (Lewis  2008,  125-­‐126).   But,  a  riot  in  2002  in  Aksy,  a  village  which  is  located  in  the  South  of  the  country,  became   “a  focal  point  that  cemented  alliances  between  new  and  old  oppositionists  and  pushed   fence-­‐sitters   into   the   opposition   camp”   (Radnitz   2010,   74).   Azimbek   Beknazarov,   the     220   parliamentary  deputy  born  in  and  elected  from  Aksy,  was  arrested  on  dubious  changes   relating  to  an  affray  several  years  ago.  Yet,  in  reality,  Akaev  ordered  his  arrest  because   nationalist   Beknazarov   planned   to   impeach   Akaev   in   parliament   on   the   ground   that   Akaev  agreed  with  the  Chinese  government  to  transfer  some  remote  territory  in  Eastern   Kyrgyzstan  to  China  (see,  e.g.  Lewis  2008,  127).  In  order  to  contest  the  abuse  of  power   by  the  president,  Beknazarov  mobilized  his  local  supporters  and  they  started  to  throw   stones   at   the   police,   which   then   opened   fire   into   them,   killing   five   unarmed   demonstrators.  Subsequent  anti-­‐government  protests  cemented  a  network  of  opposition   politicians   by   encouraging   a   wide   range   of   participation   regardless   of   regions.   Kurmanbek  Bakiev,  who  was  at  that  time  serving  as  the  Prime  Minister  and  later  became   the  president  of  Kyrgyzstan  after  the  collapse  of  the  Akaev  regime  in  2005,  resigned  to   take   a   responsibility   for   this   Aksy   incident,   and   then   defected   from   the   regime   and   joined  the  opposition  camp.    Prior  to  the  2005  parliamentary  elections,  he  united  nine   opposition   parties   into   the   pre-­‐electoral   opposition   bloc   People’s   Movement   of   Kyrgyzstan,  consolidating  the  opposition  (Kulov  2010,  342)                  In   sum,   opposition   forces   had   been   much   stronger   before   the   2005   parliamentary   elections.   Strong   opposition   also   makes   a   clear   contrast   with   the   case   of   Kazakhstan   where  the  opposition  camp  became  weak  by  the  middle  of  the  2000s.                                                                                       221   6.4  Electoral  Fraud  and  the  Rise  of  Massive  Protests:     Comparing  Elections  in  the  Akaev  Regime              By  the  time  Kyrgyzstan  held  the  parliamentary  elections  in  February  2005,  Akaev  had   already   lost   much   of   his   mobilization   power   and   popularity.   His   available   financial   resources   had   been   declining.   For   his   remaining   scarce   resources,   he   did   not   share   with   other   ruling   elites   and   rather   he   relied   more   on   his   family   members   by   appointing   them   to  important  government  positions,  resulting  in  the  family  monopolizing  state  resources   (Specter   2004;   Collins   2006;   McGlinchey   2011).   This   discouraged   local   ruling   elites   to   continue   to   ally   with   Akaev   like   they   did   during   the   1990s.   In   addition,   opposition   figures,   most   of   which   were   based   on   former   ruling   elites,   were   challenging   the   president.   In   order   to   explain   electoral   fraud   in   Kyrgyz   elections   since   1995   and   the   Tulip   revolution   in   the   aftermath   of   the   2005   elections,   I   analyze   the   1995,   2000   and   2005  elections  in  light  of  Akaev’s  mobilization  power.                         6.4.1  The  1995  Elections:  Relatively  Fair  Elections  and  the  Durable  Akaev  Regime              In   the   1995   elections   (both   parliamentary   and   presidential),   although   there   were   some   electoral   fraud   and   irregularities   observed,   OSCE   positively   assessed   the   democratic   characters   of   the   elections   (OSCE   1995,   Huskey   1997,   261).   According   to   Collins  (2006,  224-­‐225),  “despite  some  minor  violations  –  far  more  circumscribed  than   had  been  expected,  given  Akaev’s  uncertain  chances  –  the  elections  were  given  a  pass  by   the  OSCE  and  other  international  observers.”  For  example,  OSCE  reported  that  in  some   electoral   districts   multiple   voting,   vote-­‐buying   and   non-­‐violent   intimidation   were     222   confirmed   and   Akaev   admitted   to   electoral   malpractices   (OSCE   1995).   Also,   in   the   presidential  race,  it  was  said  that  the  presidential  candidate  Masaliev,  who  was  the  first   secretary   of   the   Soviet   Socialist   Republic   of   Kyrgyzstan   and   the   strongest   opposition   candidate   at   that   time,   could   not   score   a   slim   majority   in   the   Osh   oblast   where   he   held   a   strong  support  base.  This  was  probably  because  Akaev  extensively  stuffed  the  ballot  box   in  the  region  by  appointing  akims  in  Osh  and  Jalal-­‐Abad  loyal  to  him  (Author’s  interview   with   a   political   activist   [#32];   Collins   2006,   236).   There   were   many   practices   of   illegal   vote-­‐buying   during   election   campaigns   as   well   in   rural   area,   as   Huskey   (1997,   261)   reported   that   “in   some   districts   campaign   vodka   flowed   like   a   river.”   Yet,   the   government   neither   eliminated   strong   candidates   by   deregistering   them   in   advance   before  those  elections  nor  seriously  biased  media  in  favor  of  the  president  and  their  pro-­‐ presidential   candidates.   Although   there   was   some   intimidation   to   the   media,   there   were   also   still   active,   independent   media,   and   Akaev   did   not   try   to   hold   powerful   media   outlets   broadcasting   political   views   in   favor   of   the   president.   Many   of   electoral   irregularities  were  probably  due  to  the  low  quality  of  bureaucracy  and  an  inexperienced   electoral  management  body  in  charge  of  operating  the  elections.              One  of  the  most  plausible  reasons  why  Akaev  won  the  presidential  and  parliamentary   elections  with  relatively  free  and  fair  elections  was  that  he  could  successfully  mobilize   local   networks   by   making   cooperative   relationships   with   akims   and   other   local   elites   (Collins   2006).   At   that   time,   sharing   with   ruling   elites   financial   resources   largely   obtained   from   international   assistance   and   gold,   he   managed   to   maintain   relatively   stable   ruling   coalitions   to   grip   on   power.   As   McGlinchey   (2011)   succinctly   points   out,   “paradoxically,   the   diffuse   nature   of   economic   and   political   reform   aid   –   the   fact   that     223   reform  aid  is  distributed  not  directly  to  the  president  but  to  ministers,  regional  and  local   governments,  members  of  parliament,  even  NGO  activists  -­‐-­‐  forces  leaders  like  Akaev  to   pursue  the  very  wealth  redistribution  policies  that  are  most  likely  to  sustain  autocratic   rule”   (McGlinchey   2011,   94).   In   exchange   for   financial   resources   drawn   from   the   president,   members   of   the   ruling   coalition   then   mobilized   their   supporters   in   their   electoral  districts  by  utilizing  their  own  local  patronage  networks  and  resources,  which   was  very  effective  in  bringing  electoral  victories  to  Akaev  in  the  1995  parliamentary  and   presidential   elections   (Huskey   1997,   258-­‐259;   Collins   2006,   231-­‐240).   Although   the   alliance   between   Akaev   and   ruling   elites   was   not   established   well   enough   to   be   sustainable   in   the   long   run,   the   former   could   derive   support   from   the   latter   by   distributing  patronage  that  he  held  even  amid  of  serious  economic  crisis.  In  addition,  my   analysis  of  political  business  cycles  in  Kyrgyzstan  confirmed  that  although  the  effect  is   not   distinguishable   in   a   statistically   significant   way,   post-­‐election   inflation   rates   are   about   4   percent,   which   is   even   higher   than   the   later   elections   in   2000   and   2005,   suggesting   that   Akaev’s   efforts   of   distribution   might   have   been   larger   in   the   later   elections.97         6.4.2  The  2000  Elections:  Electoral  Manipulation  and  the  Absence  of  Large-­‐Scale  Protests                            However,   as   financial   assistance   decreased   and   opposition   became   more   active,   Akaev  turned  more  inclined  to  use  serious  electoral  manipulation.  OSCE  concluded  that   the   2000   legislative   and   presidential   elections   both   were   more   seriously   manipulated   than   the   1995   elections   and   thus   fell   short   of   international   standards   of   democratic                                                                                                                   97  Details  on  the  analysis  are  showed  in  Appendix  C6.       224   elections  (OSCE  2000b;  OSCE  2001).  The  Quality  of  Elections  Data  constructed  by  Kelley   (2012)   codes   that   the   2000   elections   were   more   fraudulent   than   the   1995   elections:   “The  pre-­‐election  period  was  marred  by  a  high  degree  of  interference  in  the  process  by   state   officials,   a   lack   of   independence   of   the   courts,   resulting   in   a   selective   use   of   sanctions  against  candidates,  and  a  bias  in  the  state  media”  (OSCE  2000b,  1).  Two  major   opposition   parties   –   Ar   Namys   and   Bei   Bechara   -­‐-­‐   both   of   which   were   formed   by   powerful   opposition   politician   defected   from   the   government,   Felix   Kulov   and   Daniyar   Usenov,   were   barred   from   registering   for   the   election   on   the   ground   that   the   parties   did   not  qualify  under  the  new  regulation  requiring  parties  to  have  been  registered  for  a  year   before  the  election.  Kulov  himself  ran  for  the  parliamentary  elections  and  obtained  the   largest  vote  shares  in  the  first  round  of  vote,  yet  on  the  second  round  of  vote  his  votes   decreased   for   unknown   and   unexplainable   reasons   and   he   lost   the   elections   (Uyama   2006,  47).  After  the  elections,  he  was  indicted  for  embezzlement  and  jailed  until  the  end   of   the   Akaev   regime   in   2005.   Usenov   and   Topchubek   Turgunaliev,   other   famous   opposition   leaders,   were   also   imprisoned   after   the   elections.   State-­‐owned   newspapers   and   TV   news   broadcasted   political   news   in   favor   of   leading   pro-­‐presidential   parties,   such  as  the  Union  of  Democratic  Forces  and  the  Democratic  Party  of  Women,  as  well  as   the  president  (OSCE  2000b,  12;  OSCE  2001,  9).              Although  the  2000  elections  were  more  seriously  rigged  than  those  in  1995,  they  did   not   provoke   nation-­‐wide   mobilization.   Although   there   were   occasional   protests   in   support  of  defeated  opposition  candidates,  they  did  not  develop  into  larger  ones  (Kulov   2010,   117).   Why?   First   of   all,   opposition   forces   were   not   united   at   that   time,   yet   some   prominent  opposition  candidates  could  gain  some  seats,  which  may  lead  them  think  that,     225   although   elections   were   surely   rigged,   electoral   fraud   was   not   extensive   to   completely   crowd   out   the   opposition   parties   (Kulov   2010,   120).   In   fact,   although   the   most   powerful   opposition   leaders   like   Kulov   and   Usenov   were   repressed   and   exposed   to   serious   electoral   fraud,   some   others   such   as   Adahan   Madumarov,   Azimbek   Beknazarov,   Ismail   Isakov,   and   Omurbek   Tekebaev   could   secure   seats   in   the   2000   elections   by   mobilizing   their  local  networks.  Different  recognition  over  the  seriousness  of  electoral  fraud  among   opposition   leaders   may   have   made   it   difficult   for   them   to   unite   together   after   the   elections.   Second,   although   financial   resources   that   Akaev   could   use   had   been   decreasing,  he  still  did  not  monopolize  state  resources  at  the  time  of  2000  to  squander   them   for   his   family   like   he   did   after   the   2000   elections   (McGlinchey   2011,   88-­‐86).   Although   pre-­‐electoral   distribution   is   not   observed   in   my   analysis   of   political   business   cycles   (post-­‐electoral   inflation   rates   are   0.029   percent;   see   Appendix   C6),   he   could   manage   to   keep   most   of   them   to   stay   in   the   regime,   sharing   scant   resources   with   members  of  the  ruling  coalition.       6.4.3  The  2005  Elections:  Excessive  Fraud,  Popular  Protests  and  the  Collapse  of  the  Akaev   Regime              The   2005   parliamentary   elections   sparked   popular   protests   in   some   electoral   districts  in  the  first  round  of  voting  on  February  27th,  which  escalated  into  national-­‐level   mobilization  in  the  second  round  of  voting  on  March  13th.  After  the  second  round,  the   central   election   commission   declared   that   pro-­‐presidential   candidates   obtained   an     226   overwhelming  majority  of  seats.98  Protesters,  many  of  whom  came  from  local  places  and   were  mobilized  by  local  elites,  set  out  to  the  capital  Bishkek  and  held  meeting  to  demand   Akaev’s  resignation  and  called  for  another  election  due  to  serious  electoral  manipulation.     By  March  24th,  the  number  of  protesters  had  reached  15,000-­‐20,000,  clashing  with  pro-­‐ presidential  supporters  as  well  as  the  police  (Marat  2008,  7).  Finally,  protesters  stormed   the  presidential  palace  only  to  discover  that  Akaev  had  already  flown  from  there  a  few   hours  ago.              OSCE   reported   that   the   parliamentary   elections   were   relatively   competitive   in   the   election-­‐day   (OSCE   2005)   in   that   multiple   candidates   were   contesting   in   a   large   number   of  districts,  but  at  the  same  time  they  also  documented  that  prior  to  the  elections,  Akaev   resorted   to   extensive   electoral   fraud   to   seriously   bias   election   results   in   favor   of   pro-­‐ presidential  candidates  as  well  as  the  ruling  party  Alga  Kyrgyzstan.  This  extensive  pre-­‐ electoral   manipulation   made   people   suspect   that   election   results   did   not   reflect   real   popularity   of   the   regime.   First   and   foremost,   de-­‐registration   of   powerful   opposition   candidates   convincingly   impressed   people   that   Akaev   may   not   be   able   to   win   the   elections   overwhelmingly   without   excluding   these   opposition   leaders   in   advance.   Rosa   Otunbaeva,  a  former  diplomat  having  resided  in  Moscow  and  a  famous  opposition  leader,   was  barred  from  running  for  the  election  because  she  did  not  meet  the  permanent,  in-­‐ country   residency   requirement   of   five   years   prior   to   candidate   nomination.   Besides   her,   11  other  candidates  were  rejected  from  registration  (OSCE  2005).  Second,  illegal  vote-­‐ buying   by   pro-­‐presidential   candidates   was   widespread.   During   election   campaigns,   as                                                                                                                   98For   detailed   description   on   the   Tulip   revolution,   see,   for   example,   Uyama   (2006),   Marat  (2008),  Cummings  (2010),  and  Radnitz  (2010,  Chapter  6).         227   they  could  not  rely  on  the  government’s  expansionary  fiscal  policy,  candidates  gave  out   small  gifts  like  food,  clothes,  soup,  tea,  and  vodka  to  a  large  number  of  voters  (Radnitz   2010,   132-­‐133).   According   to   an   international   observer,   “pre-­‐election   vote   buying   [was]   the  worst  she  had  seen  in  fifteen  years  of  Kyrgyz  elections”(Radnitz  2010,  133).  Third,   media   was   exposed   to   more   extensive   control   by   the   government.   “Most   media   monitored   by   the   OSCE/ODIHR   failed   to   provide   impartial   and   fair   coverage   of   the   campaign.   Almost   all   media   paid   extensive   attention   to   the   authorities,   mainly   to   the   President”  (OSCE  2005,  16).                      Due   to   pre-­‐electoral   manipulation,   election   results   in   the   first   round   of   voting   were   highly   favorable   to   pro-­‐presidential   candidates.   Out   of   725   seats   in   legislature,   32   electoral  districts  decided  winners.  Of  these,  surprisingly  only  2  electoral  districts  were   won   by   the   opposition,   and   the   rest   were   taken   by   pro-­‐government   or   independent   politicians   (Radnitz   2010,   136).   Even   among   pro-­‐presidential   candidates,   they   often   competed  with  each  other,  signaling  that  ruling  parties  and  the  president  did  not  control   the  ruling  camp  well  (Uyama  2006,  49).    Bermet  Akaeba,  the  president’s  daughter  who   ran   for   the   elections   in   Bishkek,   only   obtained   45   percent   of   votes   despite   enormous   pressure  on  students  to  vote  for  her  and  restrictions  on  the  campaigns  of  her  opponents   (Lewis  2008,  138).              For  many  opposition  supporters,  their  devastating  loss  of  was  very  surprising,  given   that   the   united   opposition   actively   campaigned   amid   of   serious   electoral   manipulation   by  the  government.  In  addition,  people  generally  perceived  that  the  Akaev  regime  was   highly   unpopular   due   to   growing   nepotism   and   corruption   within   the   government,   as   described   in   the   previous   sections.   Motivated   by   this   large   gap   between   unpopularity   of     228   the   Akaev   regime   and   the   overwhelming   victory   of   pro-­‐presidential   candidates   in   elections,  opposition  leaders,  most  of  whom  were  losing  the  elections,  found  it  easier  to   mobilize  their  supporters  to  denounce  this  much  electoral  fraud  by  the  government.  As   Henry  Hale  rightfully  points  out,  “when  the  first  round  of  voting  indicated  that  Akaev’s   supporters   were   winning   far   more   seats   than   the   regime’s   popularity   level   made   credible,   when   major   opposition   figures   like   Roza   Otunbaeva   were   disqualified,   and   when  Akaev’s  relatives  […]  appeared  to  be  headed  to  parliamentary  seats  […],  losers  of   the  formal  counts  rallied  their  forces  and  quickly  joined  efforts,  with  southern  network   leading  the  way”  (Hale  2015,  196).    After  electoral  defeats  of  opposition  candidates,  they   started  to  mobilize  their  supporters  to  protest  against  electoral  fraud  committed  by  the   government.   Once   large   protests   occurred   in   some   electoral   districts,   protests   over   electoral   fraud   spread   to   other   districts   and   regions,   which   reduced   the   costs   for   people   to  join  the  public  dissents  and  thus  escalated  into  national-­‐level  mobilization.99                                     6.5  Conclusion              This   chapter   has   explored   the   case   of   Kyrgyzstan.   In   the   1990s,   the   Akaev   regime   tended   to   share   financial   resources   derived   from   international   assistance   and   mineral   resources   with   members   of   ruling   coalitions,   which   made   him   to   outpace   opposition   leaders   and   mobilize   a   large   number   of   supporters   even   under   a   decentralized   circumstance   without   resorting   to   serious   electoral   fraud.   Yet,   as   financial   resources                                                                                                                   99  The   dynamics   of   protests   in   the   Tulip   revolution   is   well   depicted   by   Marat   (2006),   Cummings  (2010),  and  Radnitz  (2010).       229   became  scarce  and  Akaev  relied  more  on  his  family  and  cronies  to  maintain  his  regime,   he   became   more   inclined   to   use   serious   electoral   manipulation.   In   the   2000   elections,   using   extensive   electoral   fraud   was   still   effective   to   manage   to   sustain   the   regime   because  the  opposition  was  still  weaker  than  later.  Yet,  as  opposition  united  before  the   2005   parliamentary   elections,   Akaev   resorted   to   far   more   extensive   pre-­‐electoral   manipulation,   which   contributed   to   generating   a   big   gap   between   his   perceived   popularity   and   pro-­‐presidential   candidates’   electoral   performance   at   the   first   round   of   voting.   This   sparked   massive   opposition   protests   soon   after   the   elections.   The   case   of   Kyrgyzstan   illustrates   well   my   argument   that   excessive   electoral   fraud   in   light   of   the   dictator’s   mobilization   power   encourages   people   to   take   to   the   streets,   which   then   results  in  threatening  the  stability  of  authoritarian  regimes.                                                   230   CHAPTER  7     CONCLUSION            This  dissertation  has  explored  the  causes  and  consequences  of  authoritarian  elections.   Authoritarian   leaders   need   to   deal   with   the   dilemma   at   the   ballot   box:   In   order   to   maintain  their  rule,  they  need  to  win  big  in  elections,  yet  manipulation  of  election  results   loses   the   informational   benefits   of   authoritarian   elections   –   credibly   showing   regime   strength  and  knowing  the  distribution  of  political  support  from  the  citizenry.  Under  the   constraint  of  the  electoral  dilemma,  authoritarian  leaders  design  authoritarian  elections.   Designed  elections  then  have  important  implications  on  post-­‐electoral  political  order  in   autocracies.              In   this   dissertation,   I   have   argued   that   domestic   power   relationships   between   the   dictator   and   political   elites   determine   how   much   the   dictator   designs   authoritarian   elections   in   his   favor.   Strong   dictators,   who   can   mobilize   regime   supporters   by   using   rich  financial  resources  in  efficient  ways,  have  an  incentive  to  not  manipulate  elections.   By   refraining   from   serious   manipulation   of   election   results,   authoritarian   leaders   can   take  advantage  of  authoritarian  elections  to  fill  the  gap  of  information  in  authoritarian   regimes.    On  the  other  hand,  “weak”  dictators,  who  lack  of  financial  resources,  binding   political   organizations   or   face   strong   oppositions,   are   likely   to   rely   more   on   electoral   manipulation   because   revealing   their   de   facto   weakness   at   the   ballot   box   leads   authoritarian  elections  to  have  destabilizing  effects  on  political  order.       231            With  this  information  theory  of  authoritarian  elections  in  mind,  Chapter  2  examined   the   determinants   of   electoral   fraud   cross-­‐nationally.   Employing   cross-­‐national   data   covering  78  authoritarian  countries  (1977-­‐2004),  I  found  that  dictators,  who  have  rich   natural   resource   wealth,   strong   political   organizations,   and   weak   opposition,   tend   to   “liberalize”   elections   by   relying   less   on   election   violence,   election   cheating,   and   the   manipulation  of  electoral  law.              Chapter   3   turned   to   the   logic   of   the   electoral   system   change   in   electoral   authoritarianism.   Choosing   SMD   systems,   dictators   can   enjoy   a   larger   seat   bias   with   fewer   votes,   which   enable   dictators   to   win   big   with   less   mobilization   power.   Yet,   SMD   systems   make   election   results   a   mixed   signal   to   convey   regime   strength   as   well   as   facilitate  opposition  parties’  pre-­‐electoral  election  coalition  making,  which  undermines   the  dictator’s  demonstration  of  invincibility  at  the  ballot  box.  PR  systems,  on  the  other   hand,   require   dictators   to   obtain   a   larger   number   of   votes   to   win   elections   overwhelmingly.  In  so  doing,  they  can  send  a  costly  signal  of  their  regime  strength  and   also   can   prevent   oppositions   from   making   a   pre-­‐electoral   coalition   at   the   expense   of   the   probability   of   winning   big.   Using   an   original   cross-­‐national   data   of   electoral   system   changes  in  electoral  authoritarianism  (1946-­‐2007),  I  found  that  dictators  with  abundant   natural  resource  wealth  tend  to  choose  a  more  PR-­‐based  system.              Chapter   4   then   provided   an   in-­‐depth   case   study   of   Kazakhstan   in   which   President   Nazarbaev   consolidated   his   rule   despite   the   fact   that   he   became   less   dependent   on   blatant  electoral  fraud  and  electoral  system  manipulation.  My  case  study  demonstrated   that   during   the   1990s   when   Nazarbaev   was   still   too   weak   to   mobilize   a   wide   range   of   political   support   from   citizens   he   was   apt   to   manipulate   election   results   with   election     232   fraud   and   keeping   SMD   systems.   As   he   got   powerful   to   mobilize   supporters   via   the   efficient   distribution   of   natural   resource   wealth,   however,   he   became   more   inclined   to   use  pre-­‐electoral  economic  distribution  rather  than  electoral  fraud  and  decided  to  shift   the  country’s  electoral  system  to  more  PR-­‐based  ones.                        In   Chapter   5,   I   investigated   the   determinants   of   post-­‐electoral   political   conflicts   in   dictatorships   –   leadership   turnover   and   popular   protests.   The   theory   provides   implications   on   political   order   after   authoritarian   elections.   If   the   dictator   fails   to   strategically   manipulate   elections   and   cannot   deal   with   the   electoral   dilemma,   then   authoritarian   elections   backfire   on   him.   More   specifically,   there   were   two   causal   pathways   through   which   elections   threaten   authoritarian   stability.   If   the   dictator   employ  excessive  electoral  fraud,  then  he  is  more  likely  to  face  popular  protests  because   political  elites  cannot  are  unable  to  make  sense  of  de  facto  strength  of  the  regime.  On  the   contrary,   if   the   autocrat   fails   to   use   the   sufficient   level   of   fraud,   then   election   results   reveal  the  weakness  of  the  dictator,  leading  to  leadership  turnover  via  a  post-­‐electoral   coup  or  opposition’s  landslide  victory  in  elections.  My  cross-­‐national  statistical  analysis   covering   78   authoritarian   countries   (1977-­‐2004)   has   rendered   strong   supporting   evidence  on  these  theoretical  expectations.              Chapter   6   has   investigated   the   case   of   Kyrgyzstan   in   which   President   Akaev   faced   massive   protests   after   the   2005   elections,   what   is   the   so-­‐called   “Tulip   Revolution.”   During  the  1990s,  he  managed  to  maintain  his  regime  and  hold  relatively  free  and  fair   elections   by   using   financial   resources   and   maintaining   regional   alliances   with   local   elites.  From  the  early  2000s,  he  became  more  inclined  to  use  serious  electoral  fraud  as   he  increasingly  suffered  from  a  lack  of  financial  resources  and  faced  strong  oppositions.     233   Despite   the   fact   that   people   were   speculating   that   the   regime   became   weak,   the   2005   parliamentary   elections   with   serious   electoral   fraud   announced   an   overwhelming   electoral  victory  of  the  president  and  his  ruling  parties.  This  sparked  popular  protests  by   opposition   leaders   and   supporters   who   believed   that   the   regime   in   fact   was   weak,   resulting  in  the  collapse  of  the  Akaev  regime.              The   theory   and   the   empirical   findings   in   the   dissertation   lead   us   to   at   least   the   following   two   policy   implications   on   international   assistance   for   democratization   and   election   monitoring.   First,   the   dissertation   suggests   that   policy   makers   may   need   to   provide   international   assistance   for   elections   while   carefully   considering   conditions   of   domestic   politics   -­‐-­‐   the   power   balance   between   the   dictator   and   political   elites   -­‐-­‐   in   authoritarian   regimes.   If   the   dictator   possesses   abundant   financial   resources   and   strong   disciplinary   organizations,   then   international   election   monitoring   and   subsequent   relatively   free   and   fair   elections   may   help   the   dictator   use   authoritarian   elections   to   consolidate   his   regime   by   improving   the   quality   of   information   that   elections   bring.   Conversely,   if   strong   opposition   exists   and   the   dictator   suffers   from   a   shortage   of   resources  to  distribute,  international  pressures  to  hold  free  and  fair  elections  may  pave   the  way  for  destabilizing  the  country’s  political  order  after  elections  reveal  the  dictator’s   weakness.   To   what   extent   international   election   monitoring   becomes   effective   to   promote   peaceful   democratization   at   least   partly   depends   on   domestic   power   distribution   in   an   authoritarian   country.   Strong   international   pressures   without   taking   into   account   domestic   politics   in   authoritarian   countries   may   bring   “unintended   consequences”   to   policy   makers,   such   as   the   consolidation   of   autocracy   through     234   relatively   free   and   fair   elections   or   the   demise   of   political   order   in   the   wake   of   a   relatively  free  election.                Second,   the   dissertation   also   suggests   that   mere   existence   of   free   and   fair   elections   and   proportional   representation   systems   do   not   necessarily   contribute   to   further   democratization  in  authoritarian  countries.  The  findings  in  the  dissertation  tell  us  that   dictators   may   strategically   manipulate   elections   and   electoral   systems   under   the   constraints   of   the   electoral   dilemma.   It   implies   that   dictators,   who   monopolize   rich   resources   and   successfully   centralize   political   organizations,   may   dare   to   hold   relatively   free   and   fair   elections   and   adopt   a   proportional   representation   system   to   obtain   the   informational   benefits   while   disguising   their   regimes   as   democratic   to   appeal   to   the   international  community  as  well  as  the  domestic  audience.  In  order  for  elections  to  be   competitive   enough   to   be   called   democratic   elections,   elections   need   to   have   a   real   potential   of   achieving   peaceful   government   alternation.   To   do   so,   international   organizations   may   need   to   also   strengthen   their   support   for   non-­‐electoral   aspects   of   democratization,   such   as   cultivating   the   rule   of   law,   preventing   the   government   from   monopolizing   financial   resources,   undermining   the   dictator’s   centralized   mobilization   structures,   helping   opposition   forces   unite.   Combining   these   measures   with   international   election   monitoring,   elections   are   more   likely   to   bring   government   alternation.   While   ensuring   that   such   political   turnover   is   achieved   peacefully,   international   assistance   may   become   more   effective   to   democratize   authoritarian   countries.                        That   being   said,   there   are   a   number   of   topics   for   which   further   research   will   be   needed.   First,   future   research   should   make   clear   under   what   conditions   authoritarian     235   rulers   might   miscalculate   on   the   use   of   electoral   manipulation.   Although   there   are   numerous   examples   in   which   authoritarian   elections   turn   out   to   be   “stunning”   ones,   systematic   theoretical   explorations   on   why   some   authoritarian   leaders   face   such   elections  have  not  yet  been  done.  This  dissertation  dealt  with  such  cases  as  “off-­‐paths”   and  just  treated  them  as  empirical  implications  that  should  be  observed  if  the  dictator   failed   to   “appropriately”   manipulate   elections.   Yet,   theorizing   the   miscalculation   may   be   relevant   when   it   comes   to   making   a   better   prediction   on   a   country’s   odds   of   democratization  and/or  civil  conflicts.                  Second,  although  the  dissertation  has  presented  cross-­‐national  evidence  of  “political   manipulation”   such   as   electoral   fraud   and   electoral   system   change,   it   does   not   cross-­‐ national   investigate   “economic   manipulation,”   namely,   political   business   cycles.   If   my   theory  is  correct,  we  should  be  able  to  expect  that  dictators  with  holding  relatively  free   and   fair   elections   should   rely   more   on   pre-­‐electoral   economic   distribution:   A   trade-­‐off   between  political  and  economic  manipulations.              Third,  although  I  have  argued  that  less-­‐manipulated  elections  should  be  able  to  play   the  role  of  signaling  the  dictator’s  strength  as  long  as  he  can  win  big,  other  researchers   contend   that,   the   more   heavily   elections   are   manipulated,   the   more   successfully   dictators   can   deter   challengers   via   elections   (Simpser   2013).   Although   my   analysis   found   supporting   evidence   on   my   theoretical   perspective,   more   direct   empirical   evidence   may   need   by   conducting   survey   data   analysis.   Setting   people’s   perception   about   the   government’s   legitimacy   as   a   dependent   variable,   we   would   be   able   to   investigate   how   the   effect   of   election   results   on   political   legitimacy   may   change     236   depending   on   the   level   of   electoral   fraud   and   electoral   systems.   In   so   doing,   we   could   provide  additional  evidence  on  the  signaling  effect  of  authoritarian  elections.              Lastly,   the   dissertation   did   not   provide   a   case   of   leadership   turnover   to   illustrate   causal  mechanisms  through  which  excessively  free  and  fair  elections  lead  to  leadership   turnover.   Cases   such   as   Poland   (1989)   and   Algeria   (1991)   could   be   useful   to   provide   case-­‐study  evidence  on  the  part  of  leadership  turnover.                                         237   APPENDIX     Appendix  Part  I:  Cross-­‐National  Statistical  Analyses     Chapter  2     Table  C2-­‐1:  Descriptive  Statistics  of  Chapter  2                                     238   Table  C2-­‐2:  List  of  Authoritarian  Countries  (1977-­‐2004)         Note:  “Par”  stands  for  parliamentary  elections,  while  “Pres”  represents  presidential   elections.         239     Table  C2-­‐3:  Additional  Analyses   Note:  Models  7-­‐12  are  estimated  using  ordered  logistic  regression.  Clustered  robust   standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;  *p<0.1.  ^  indicates  that  a  chi-­‐ squared  joint  test  shows  that  the  interaction  effect  is  statistically  significant  at  the  1   percent  level.         240     Table  C2-­‐4:  Alternative  Definition  of  Political  Regimes  –  Boix,  Miller  and  Rosato   (2012)   Note:  Country-­‐clustered  robust  standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;   *p<0.1.  ^  indicates  that  a  chi-­‐squared  joint  test  shows  that  the  interaction  effect  is   statistically  significant  at  the  1  percent  level.                     241       Table  C2-­‐5:  Alternative  Definition  of  Political  Regimes  –  Polity  IV   Note:  A  countries  is  seen  as  authoritarian  if  its  Polity  IV  score  is  less  than  6.  Country-­‐ clustered  robust  standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;  *p<0.1.  ^   indicates  that  a  chi-­‐squared  joint  test  shows  that  the  interaction  effect  is  statistically   significant  at  the  1  percent  level.                     242       Table  C2-­‐6:  Alternative  Measure  of  Natural  Resources  –  Change  in  Oil-­‐Gas  Value   per  capita   Note:  Country-­‐clustered  robust  standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;   *p<0.1.  ^  indicates  that  a  chi-­‐squared  joint  test  shows  that  the  interaction  effect  is   statistically  significant  at  the  1  percent  level.           243       Table  C2-­‐7:  Alternative  Measure  of  Natural  Resources  –  3  Years  Moving  Average  of   Oil-­‐Gas  Value  per  capita   Note:  Country-­‐clustered  robust  standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;   *p<0.1.  ^  indicates  that  a  chi-­‐squared  joint  test  shows  that  the  interaction  effect  is   statistically  significant  at  the  1  percent  level.           244       Table  C2-­‐8:  Alternative  Measure  of  Natural  Resource  Wealth  –  Haber  and  Menaldo   (2011)   Note:  Country-­‐clustered  robust  standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;   *p<0.1.  ^  indicates  that  a  chi-­‐squared  joint  test  shows  that  the  interaction  effect  is   statistically  significant  at  the  1  percent  level.             245       Table  C2-­‐9:  Region-­‐Clustered  Robust  Standard  Errors   Note:  Region-­‐clustered  robust  standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;   *p<0.1.  ^  indicates  that  a  chi-­‐squared  joint  test  shows  that  the  interaction  effect  is   statistically  significant  at  the  1  percent  level.     246       Table  C2-­‐10:  Additional  Controls   Note:  Country-­‐clustered  robust  standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;   *p<0.1.  ^  indicates  that  a  chi-­‐squared  joint  test  shows  that  the  interaction  effect  is   statistically  significant  at  the  1  percent  level.     247       Table  C2-­‐11:  Alternative  Methods  -­‐  Fixed  Effects  and  Regional  Specific  Effects   Note:  Country-­‐clustered  robust  standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;   *p<0.1.  ^  indicates  that  a  chi-­‐squared  joint  test  shows  that  the  interaction  effect  is   statistically  significant  at  the  1  percent  level.      W  indicates  Wald-­‐chi  squared  test.  F   indicates  F  test.                   248       Table  C2-­‐12:  Alternative  Methods  –  Random  Intercept  Models     Note:  Country-­‐clustered  robust  standard  errors  in  parentheses.  ***p<0.01;  **p<0.05;   *p<0.1.  ^  indicates  that  a  chi-­‐squared  joint  test  shows  that  the  interaction  effect  is   statistically  significant  at  the  1  percent  level.           249     C2:  Data  Sources  (Chapter  5’s  data  sources  are  identical  with  the  one  listed  here)     Banks,  Arthur.  2009.  The  Cross-­‐National  Time-­‐Series  Data  Archive.   Cederman,  Lars-­‐Erik,  Brian  Min  and  Andreas  Wimmer.  2009.  Ethnic  Power  Relations   Dataset.    Available  at  http://www.epr.ucla.edu/.  (accessed  on  May  24,  2013)   The  Correlates  of  War  Project  (State  System  Membership  Dataset,  National  Capabilities   Dataset)   http://www.correlatesofwar.org/COW2%20Data/SystemMembership/2011/System20 11.html     Freedom  House  Index.  Available  at  https://freedomhouse.org/  (accessed  on  January  4th,   2015)   Geddes,  Barbara,  Joseph  Wright  and  Erica  Frantz.  2010.  Autocratic  Regime  Data.     Available  at  http://dictators.la.psu.edu/.  (accessed  on  May  24,  2013)   Gleditsch,  Nils  Petter,  Peter  Wallensteen,  Mikael  Eriksson,  Margareta  Sollenberg,  and   Håvard  Strand  (2002)  Armed  Conflict  1946-­‐2001:  A  New  Dataset.  Journal  of  Peace   Research  39(5).   Goemans,  Hein,  Kristian  Skrede  Gleditsch,  Giacomo  Chiozza.  2009.  “Introducing   Archigos:  A  Data  Set  of  Political  Leaders.”  Journal  of  Peace  Research  46-­‐2:  269-­‐283.   Archigos  version  2.9.    http://www.rochester.edu/college/faculty/hgoemans/data.htm     Kelley,  Judith.  2012.  Quality  of  Elections  Data.  Available  at     http://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/31461/version/1.  (accessed   on  May  24,  2013)   Maddison,  Angus.  2011.  The  Maddison  Project  Database.  Available  at       250   http://www.ggdc.net/maddison/maddison-­‐project/home.htm.  (accesses  on  May  24,     2013)   Marshall,  Monty  and  Keith  Jagger,  Polity  IV  Project.   Heston,  Alan,  Robert  Summers  and  Bettina  Aten,  Penn  World  Table  Version  7.1,  Center   for  International  Comparisons  of  Production,  Income  and  Prices  at  the  University  of   Pennsylvania,  Nov  2012.   Ross,  Michael.  2011.  Oil  and  Gas  Production  and  Values,  1932-­‐2009.  Available  at   http://dvn.iq.harvard.edu/dvn/dv/mlross.  (accessed  on  May  24,  2013)   The  European  Bank  for  Reconstruction  and  Development,  Transitional  Report.   World  Bank,  World  Development  Indicators.                                                       251       Chapter  3     Table  C3-­‐1:  Descriptive  Statistics  of  Chapter  3                               252       Table  C3-­‐2:  List  of  Electoral  Authoritarian  Countries  (1946-­‐2007)                   253       Table  C3-­‐3:  Including  Polity  IV  Scores  as  a  Control  Variable               254       Table  C3-­‐4:  Region-­‐Clustered  Robust  Standard  Errors                               255       Table  C3-­‐5:  Using  a  Different  Measure  of  the  Effective  Electoral  Threshold   Note:  The  dependent  variable  is  the  Effective  Electoral  Threshold  which  does  not   consider  legal  threshold  under  proportional  representation  systems.                           256         Table  C3-­‐6:  Limiting  Sample  into  1970-­‐2007     Note:  The  sample  is  limited  to  1970-­‐2007,  given  the  fact  that  most  developing  countries   had  finished  nationalizing  oil  companies  by  the  1970s.               257       C3:  Data  Sources     African  Elections  Database.  http://africanelections.tripod.com/     Banks,   Arthur   and   Thomas   Muller   eds.   Political   Handbook   of   the   World   (1993-­‐2008,   various  volumes).  CSA  Publications.     Inter-­‐Parliamentary  Union.  http://www.ipu.org/parline-­‐e/parlinesearch.asp   Keefer,  Philip.  Database  of  Political  Institutions.     Nohlen,  Dieter,  Michael  Krennerich,  and  Bernard  Thibaut  eds.  Elections  in  Africa:  A  Data   Handbook.  Oxford  University  Press.     Nohlen,   Dieter,   Florian   Grotz,   and   Christof   Hartmann   eds.   2001a.   Elections  in  Asia  and   the   Pacific:   A   Data   Handbook,   Volume   I:   The   Middle   East,   Central   Asia   and   South   Asia.   Oxford  University  Press.     Nohlen,   Dieter,   Florian   Grotz,   and   Christof   Hartmann   eds.   2001b.   Elections  in  Asia  and   the  Pacific:  A  Data  Handbook,  Volume  II:  South  East  Asia,  East  Asia  and  the  South  Pacific.   Oxford  University  Press.     Nohlen,   Dieter   ed.   2005a.   Elections   in   the   Americas,   Volume   I:   North   America,   Central   America,  and  the  Caribbean.  Oxford  University  Press.     Nohlen,   Dieter   ed.   2005b.   Elections   in   the   Americas,   Volume   II:   South   America.   Oxford   University  Press.     Nohlen,   Dieter   and   Philip   Stoever   eds.   2010.   Elections   in   Europe:   A   Data   Handbook.   Nomos.       Roeder,  Philip.  Ethno-­‐Linguistic  Fractionalization  Indices  for  1961  and  1985.     http://pages.ucsd.edu/~proeder/data.htm       258     Chapter  5   Table  C5-­‐1:  Descriptive  Statistics  of  Chapter  5                   259     0 .05 Density .1 .15 .2 Figure  C5-­‐1:  Histogram  of  Predicted  Values  of  Electoral  Fraud   -3 -2 -1 0 1 2 3 4 5 6 7 Predicted Values of Electoral Fraud 8 9 10                         260     Table  C5-­‐2:  Alternative  Measures  of  Political  Regimes       Note:   Decade   dummies   and   regional   dummies   are   all   included   in   the   models.   Robust   standard   errors   are   in   parentheses.     ***p   <   .01,   **p   <   .05,   *p<.10.     ^p:   Although   the   gap   variable   does   not   reach   to   the   10   percent   statistical   significance,   graphs   for   the   marginal   effects   show   that   it   has   statistically   significant  effects  in  a  wide  range  of  values  in  the  fraud  gap  variable.           261     Table  C5-­‐4:  Using  the  Electoral  Fraud  Variable  per  se       Note:  Decade  dummies  and  regional  dummies  are  all  included  in  the  models.  Robust   standard  errors  are  in  parentheses.    ***p  <  .01,  **p  <  .05,  *p<.10.         262   Appendix  Part  II:  Case  Studies     Chapter  4   C4:  List  of  Interviewees  in  Kazakhstan  and  Kyrgyzstan   #1  Officer  of  the  Soros  Foundation-­‐Kazakhstan,  May  5th,  2014,  Almaty,  Kazakhstan   #2  Officer  of  the  Soros  Foundation-­‐Kazakhstan,  May  5th,  2014,  Almaty,  Kazakhstan   #3  Political  Analyst,  May  6th,  2014,  Almaty,  Kazakhstan   #4  Political  Analyst,  May  7th,  2014,  Almaty,  Kazakhstan     #5  Officer  of  the  Kazakhstan  International  Bureau  for  Human  Rights  and  Rule  of  Law,   May  12nd,  2014,  Almaty,  Kazakhstan     #6  Officer  of  the  Kazakhstan  International  Bureau  for  Human  Rights  and  Rule  of  Law,   May  12nd,  2014,  Almaty,  Kazakhstan     #7  Officer  of  the  Kazakhstan  International  Bureau  for  Human  Rights  and  Rule  of  Law,   May  12nd,  2014,  Almaty,  Kazakhstan     #8  Political  Analyst,  May  14th,  2014,  Almaty,  Kazakhstan     #9  Political  Analyst,  May  14th,  2014,  Almaty,  Kazakhstan     #10  Political  Scientist,  May  15th,  2014,  Almaty,  Kazakhstan     #11  Political  Scientist,  May  15th,  2014,  Almaty,  Kazakhstan     #12  Political  Scientist,  May  16th,  2014,  Almaty,  Kazakhstan     #13  Political  Activist,  May  26th,  Almaty,  Kazakhstan     #14  Political  Activist,  May  27th,  Almaty,  Kazakhstan       #15  Political  Activist,  May  27th,  Almaty,  Kazakhstan     #16  Political  Activist,  May  28th,  Almaty,  Kazakhstan     #17  Political  Analyst,  May  28th,  Almaty,  Kazakhstan     #18  Officer  of  the  Kazakhstan  International  Bureau  for  Human  Rights  and  Rule  of  Law,   May  29th,  2014,  Almaty,  Kazakhstan     263     #19  Student  of  KIMEP,  May  29th,  2014,  Almaty,  Kazakhstan.       #20  Political  Activist,  May  29th,  2014,  Almaty,  Kazakhstan     #21  Political  Scientist,  May  30th,  2014,  Almaty,  Kazakhstan     #22  Political  Analyst,  June  3rd,  2014,  Almaty,  Kazakhstan     #23  Journalist,  June  6th,  2014,  Almaty,  Kazakhstan     #24  Political  Activist,  June  6th,  2014,  Almaty,  Kazakhstan     #25  Economist,  June  7th,  2014,  Almaty,  Kazakhstan     #26  Ak  Zhol  Official,  June  9th,  2014,  Astana,  Kazakhstan     #27  Nur  Otan  Official,  June  10th,  2014,  Astana,  Kazakhstan     #28  Ak  Zhol  Official,  June  12nd,  2014,  Astana,  Kazakhstan       #29  Political  Activist,  June  16th,  2014,  Astana,  Kazakhstan     #30  Political  Activist,  June  18th,  2014,  Astana,  Kazakhstan     #31  Political  Scientist,  July  2nd,  2014,  Bishkek,  Kyrgyzstan       #32  Political  Activist,  July  4th,  2014,  Bishkek,  Kyrgyzstan     #33  Deputy of Ata Zhurt, July 7th, Bishkek, Kyrgyzstan   #34  Senior  Government  Official,  July  9th,  2014,  Bishkek,  Kyrgyzstan                             264     C4-­‐2:    Methodological  Details  and  Robustness  Checks:  the  Political  Business  Cycles   Analysis  of  Kazakhstan     1.  Note  on  Methodology              In   order   to   employ   time-­‐series   data   like   the   one   I   used   in   the   analysis   of   political   business   cycles   in   Kazakhstan,   researchers   have   to   deal   with   the   following   two   problems:  (1)  non-­‐stationarity  of  data  and  (2)  autocorrelation.  Since  a  Dickey-­‐Fuller  test   revealed   that   all   dependent   variables   are   non-­‐stationary,   I   took   the   first   difference   to   transform   them   into   stationary   data.100  In   addition,   because   both   Breush-­‐Godfrey   and   Durbin-­‐Watson   tests   suggested   that   CPI   and   unemployment   rate   all   suffered   from   serious   autocorrelation,101  I   coped   with   it   through   AR   (1)   process   through   the   Prais-­‐ Winsten   method.   Monthly   dummies   (reference   category   is   December)102  are   included   in   all  models  on  account  of  controlling  for  seasonal  effects.  I  set  six  months  (for  real  wage,   3   quarters)   before/after   elections   as   electoral   periods   (1);   otherwise   the   months   were   coded  as  non-­‐electoral  (0).103                                                                                                                             100  In   the   first   difference   data   of   all   four   induces,   the   null   hypothesis,   "the   variable   includes   a   unit   root",   was   rejected   at   0.1   per   cent   level,   which   means   that   the   first   difference  data  is  stationary.       101  As  in  real  wage  data  the  null  hypothesis  that  "there  is  no  serious  correlation"  was  not   rejected,  a  statistical  model  for  real  wage  is  examined  without  using  the  Prais-­‐Winsten   method.         102  For  real  wage,  I  used  quarterly  dummies  (reference  category  is  the  fourth  quarterly).         103  Blaydes   (2011,   Chapter   5)   also   sets   six   months   as   election   periods   to   make   the   election  dummies.       265   2.  Robustness  Checks            In   order   to   check   whether   the   results   are   robust,   I   conduct   the   following   four   robustness   checks.   First,   following   previous   studies   (Fearon   and   Laitin   2003),   I   alternatively   use   the   logged   GDP   per   capita   (Annual   data   from   World   Development   Indicators)   as   another   proxy   to   operationalize   state   resources   instead   of   state   revenue.104  Even  if  I  use  GDP  per  capita,  the  overall  results  do  not  change  and  I  obtain   almost  the  same  results  in  all  of  the  models.  Second,  taking  into  account  the  context  of   Kazakh   politics   in   which   almost   all   independents   are   pro-­‐presidential,   I   include   independents  into  the  proportion  of  ruling  parties  and  rerun  the  models,  and  the  same   results   are   confirmed.   Third,   instead   of   using   AR   (1)   process   to   eliminate   autocorrelation,   I   introduce   a   lagged   dependent   variable   and   rerun   the   same   models.   The   results   remain   robust.   These   additional   analyses   demonstrate   that   the   statistical   results   remain   stable   even   if   I   adopted   alternative   ways   of   operationalization   and   estimation.                                                                                                                                                 104  The  correlation  between  logged  revenue  and  logged  GDP  per  capita  is  0.95.       266   600 400 200 Dec 2010 Jan 2010 Jan 2009 Jan 2008 Jan 2007 Jan 2006 Jan 2005 Jan 2004 Jan 2003 Jan 2002 Jan 2001 Jan 2000 Jan 1999 Jan 1998 Jan 1997 Jan 1996 Jan 1995 0 Consumer Price Index (%) (1994 December=100) 800 Chapter  6     Figure  C6-­‐1:  Time-­‐Series  Change  in  Inflation  Rates  in  Kyrgyzstan   Month-Year   Note:  red—Akaev  regime,  blue—Bakiev  regime,  green—Interim  government,  solid   line—parliamentary  elections,  dash  line—presidential  elections,  dot  line—referendum.                This  section  examines  political  business  cycles  in  Kyrgyzstan  under  the  Akaev  regime.   For  the  purpose  of  comparison,  I  also  include  two  regimes  in  the  analysis,  which  were   established   after   Akaev   –   the   Bakiev   regime   (2005-­‐2010)   and   the   interim   government   led   by   Roza   Otunbaeva   (2010-­‐2011).   To   be   consistent   with   the   analysis   of   political   business  cycles  in  Kazakhstan,  I  use  monthly-­‐specified  Consumer  Price  Index  (CPI)  as  a   dependent   variable.   The   data   comes   from   the   Bulletin   of   the   National   Bank   of   Kyrgyz   Republic.  Setting  the  CPI  at  December  1994  as  100,  I  calculate  the  CPI  between  January   1995   and   December   2010.   As   many   studies   of   the   PBCs   have   argued,   inflation   after     267   elections  can  be  interpreted  as  evidence  that  governments  adopted  expansionary  fiscal   and  monetary  policies  before  elections.              Figure  C6-­‐1  plots  time  series  changes  in  CPI.  Vertical  lines  represent  the  months  when   elections   were   held   (red—Akaev   regime,   blue—Bakiev   regime,   green—Interim   government,   solid   line—parliamentary   elections,   dash   line—presidential   elections,   dot   line—referendum).     Table  C6-­‐1:  Political  Business  Cycles  in  Kyrgyzstan  (1995-­‐2010)     Note:  ***p  <  .01,  **p  <  .05,  *p<.10.    Standard  errors  are  reported  in  parentheses.  Durbin-­‐ Watson  test  is  performed  after  correcting  serial  autocorrelation  through  AR(1)  process.         268            Using   the   same   time-­‐series   OLS   regressions   and   model   specifications   with   the   analysis  of  political  business  cycles  in  Kazakhstan,  Table  C6-­‐1  shows  the  results.  Model  1   confirms  that  Kyrgyzstan  also  tends  to  experience  a  post-­‐electoral  surge  in  inflation  by   4.794   percent,   suggesting   that   the   government   engineers   political   business   cycles   in   general  during  the  period  of  1995-­‐2010.    Interestingly,  dividing  elections  into  the  ones   under  three  different  regimes,  Model  2  shows  that,  while  elections  under  the  Bakiev  and   the   interim   government   have   strong   statistically   significant   impacts   on   inflation   rates,   elections  under  the  Akaev  are  not  exposed  to  pre-­‐electoral  economic  manipulation  in  a   statistically   significant   way.   This   suggests   that   the   Akaev   regime   was   not   financially   strong  enough  to  adopt  large-­‐scale  expansionary  economic  policies  before  elections,  as   Nazarbaev  of  Kazakhstan  does.  More  closely  looking  at  the  three  parliamentary  elections   under   the   Akaev   regime,   there   is   a   declining   pattern   on   the   post-­‐electoral   increases   in   inflation  rates,  although  the  effects  fail  to  satisfy  the  10  percent  significance  level  (Model   3).   In   the   1995   parliamentary   elections,   inflation   rates   after   the   election   were   3.9   percent.  But  it  became  almost  0  percent  in  the  2000  elections  and  turned  negative  in  the   2005  parliamentary  elections,  which  triggered  the  Tulip  revolution.                     269                                     BIBLIOGRAPHY                                                           270   BIBLIOGRAPHY       Abazov,   Rafis.   2003.   “The   Parliamentary   Elections   in   Kyrgyzstan,   February   2000.”   Electoral  Studies  22:  545-­‐552.       Acemoglu,   Daron   and   James   Robinson.   2006.   Economic   Origins   of   Dictatorship   and   Democracy.  New  York:  Cambridge  University  Press.       Ahmed,   Amel.   2013.   Democracy  and  the  Politics  of  Electoral  System  Choice:  Engineering   Electoral  Dominance.  New  York:  Cambridge  University  Press.       Akhmedov,   Akhmed   and   Ekaterina   Zhuravskaya.   2004.   “Opportunistic   Political   Cycles:   Test  in  a  Young  Democracy  Setting.”  Quarterly  Journal  of  Economics,  119-­‐4:1301-­‐1338.       Alesina,   Alberto,   Reza   Baqir   and   William   Easterly.   1999.   “Public   Goods   and   Ethnic   Division.”  Quarterly  Journal  of  Economics  114:  1243-­‐1284.           Ames,   Barry.   1970.   “Bases   of   Support   for   Mexico’s   Dominant   Party.”   American  Political   Science  Review  64-­‐1:  153-­‐167.       Anderson,   John.   1999.   Kyrgyzstan:   Central   Asia’s   Island   of   Democracy?   Harwood   Academic  Publishers.       Andersen,   Jørgen   Juel   and   Michael   Ross.   2014.   “The   Big   Oil   Change:   A   Closer   Look   at   the   Haber-­‐Menaldo  Analysis.”  Forthcoming  in  Comparative  Political  Studies.       Andrews,   Josephine   and   Robert   Jackman.   2005.   “Strategic   Fools:   Electoral   Rule   Choice   under  Extreme  Uncertainty.”  Electoral  Studies,  24:  65-­‐84.           Arellano,  Manuel  and  Olympia  Bover.  1995.  “Another  Look  at  the  Instrumental  Variables   Estimation  of  Error  Components  Models.”  Journal  of  Econometrics  68:  29-­‐51.           271   Arendt,   Hannah.   1951.   The   Origins   of   Totalitarianism.   Cleveland   and   New   York:   The   World  Publishing  Company.           Ashimbaev,  Daniyar  and  Vitalii  Khlyupin.  2008.  Kazakhstan:  Istoirya  Vlasti  [Kazakhstan:   History  of  Power].  Almaty:  CREDOS.         Austin-­‐Smith,  David.  2000.  “Redistributing  Income  under  Proportional  Representation.”   The  Journal  of  Political  Economy  108:  1235-­‐1269.       Bader,   Max.   2012.   “Trends   and   Patterns   in   Electoral   Malpractice   in   Post-­‐Soviet   Eurasia.”   Journal  of  Eurasian  Studies  3:  49-­‐57.       Barbera,   Pablo.   2013.   “When   Duverger   Becomes   Autocratic:   Electoral   Systems   and   Opposition  Fragmentation  in  Non-­‐Democratic  Regimes.”  Working  Paper.       Bawn,  Kathleen  and  Frances  Rosenbluth.  2006.  “Short  versus  Long  Coalitions:  Electoral   Accountability   and   the   Size   of   the   Public   Sector.”   American   Journal   of   Political   Science   50(2):  251-­‐265.       Beaulieu,   Emily.   2006.   Protesting  the  Contest:  Election  Boycotts  around  the  World  1990-­‐ 2002.  Ph.D.  Dissertation  Submitted  to  the  Department  of  Political  Science  at  University   of  California,  San  Diego.       Beaulieu,   Emily.   2014.   Electoral   Protest   and   Democracy   in   the   Developing   World.   New   York:  Cambridge  University  Press.       Beblawi,  Hazem  and  Giacomo  Luciani  eds.  1987.  The  Rentier  State,  Croom  Helm.       Beck,  Nathaniel  and  Richard  Katz.  1995.  “What  to  do  (and  not  to  do)  with  Time-­‐Series   Cross-­‐Section  Data,.  American  Political  Science  Review  89-­‐3:  634-­‐647.       Beck,  Nathaniel,  Jonathan  Katz,  and  Richard  Tucker.  1998.  “Taking  Time  Seriously.”   American  Journal  of  Political  Science  42-­‐4:  1260-­‐1288.         272   Bermeo,   Sarah   Blodgett.   2011.   “Foreign   Aid   and   Regime   Change:   A   Role   for   Donor   Intent.”  World  Development  39-­‐11:  2021-­‐2031.       Bielasiak,   Jack.   2006.   “Regime   Diversity,   Electoral   Systems   in   Post-­‐Communism.”  Journal   of  Communist  Studies  and  Transition  Politics,  22-­‐4:  407-­‐430.       Birch,  Sarah.  2007.  “Electoral  Systems  and  Electoral  Misconduct.”  Comparative  Political   Studies  40-­‐12:  1533-­‐1556.       Birch,  Sarah.  2011.  Electoral  Malpractice.  Oxford:  Oxford  University  Press.       Blais,   Andre   and   Louis   Massicotte.   1997.   “Electoral   Formulas:   A   Microscopic   Perspective.”  European  Journal  of  Political  Research  32:  107-­‐129.       Blaydes,   Lisa.   2011.   Elections   and   Distributive   Politics   in   Mubarak's   Egypt.   Cambridge   University  Press.     Bouandel,   Youcef.   1993.   “Algerian   National   Popular   Assembly   Election   of   December   1991.”  Representation  32:  10-­‐14.       Bratton,   Michael   and   Eldred   Masunungure.   2008.   “Zimbabwe’s   Long   Agony.”   Journal  of   Democracy  19-­‐4:  41-­‐55.       Bratton,   Michael   and   Nicholas   van   de   Walle.   1997.   Democratic   Experiments   in   Africa:   Regime  Transitions  in  Comparative  Perspective.  New  York:  Cambridge  University  Press.       Blaydes,   Lisa,   2011,   Elections   and   Distributive   Politics   in   Mubarak's   Egypt.   New   York:   Cambridge  University  Press.     Bloch,   Peter   and   Kathryn   Rasmussen.   1998.   “Land   Reform   in   Kyrgyzstan.”   In   Stephen   Wegren   ed.   Land   Reform   in   the   Former   Soviet   Union   and   Eastern   Europe,   New   York:   Routledge,  pp.  111-­‐135.           273   Boix,   Carles.   1999.   “Setting   the   Rule   of   the   Game:   The   Choice   of   Electoral   Systems   in   Advanced  Democracies.”  American  Political  Science  Review  93-­‐3:  609-­‐624.       Boix,   Carles.   2003.   Democracy   and   Redistribution.   New   York:   Cambridge   University   Press.       Boix,   Carles   and   Milan   Svolik.   2007.   “Non-­‐Tyrannical   Autocracies.”   Paper   presented   at   the  UCLA  Comparative  Politics  Seminar.         Boix,   Carles   and   Milan   Svolik.   2013.   “The   Foundations   of   Limited   Authoritarian   Government:   Institutions,   Commitment,   and   Power-­‐Sharing   in   Dictatorships.”   forthcoming  in  the  Journal  of  Politics.       Boix,  Carles,  Michael  Miller  and  Sebastian  Rosato.  2013.  “A  Complete  Data  Set  of  Political   Regimes,  1800-­‐2007.”  Comparative  Political  Studies  46-­‐12:  1523-­‐1554.       Bremmer,   Ian   and   Cory   Welt.   1996.     “The   Trouble   with   Democracy   in   Kazakhstan.”   Central  Asian  Survey  15-­‐2:  179-­‐199.     Brender,  Adi  and  Allan  Drazen.  2006.  “Political  Budget  Cycles  in  New  versus  Established   Democracies.”  Journal  of  Monetary  Economics  52-­‐7:  1271-­‐1295.     Brownlee,   Jason.   2006.   Authoritarianism   in   an   Age   of   Democratization.   New   York:   Cambridge  University  Press.         Brownlee,   Jason.   2009.   "Portents   of   Pluralism:   How   Hybrid   Regimes   Affect   Democratic   Transitions."  American  Journal  of  Political  Science  53-­‐3:  515-­‐532.       Bueno   de   Mesquita,   Bruce,   Alastair   Smith,   Randolph   Siverson,   and   James   Morrow.   2003.   The  Logic  of  Political  Survival.  Cambridge,  Massachusetts:  MIT  Press.       Bueno   de   Mesquita,   Bruce   and   Alastair   Smith.   2010.   “Leader   Survival,   Revolutions   and   the  Nature  of  Government  Finance.”  American  Journal  of  Political  Science  54-­‐4:  936-­‐950.         274   Bunce,   Valerie   and   Sharon   Wolchik.   2010.   Defeating   Authoritarian   Leaders   in   Post-­‐ Communist  Countries.  Cambridge  University  Press.       Calvo,   Ernesto.   2009.   “The   Competitive   Road   to   Proportional   Representation:   Partisan   Bias   and   Electoral   Regime   Change   under   Increasing   Party   Competition.”   World  Politics   61-­‐2:  254-­‐295.     Carey,   John   and   Simon   Hix.   2011.   “The   Electoral   Sweet   Spot:   Low-­‐Magnitude   Proportional  Electoral  Systems.”  American  Journal  of  Political  Science  55-­‐2:  383-­‐397.       Cederman,   Lars-­‐Erik,   Kristian   Skrede   Gleditsch,   and   Simon   Hug.   2012.   “Elections   and   Ethnic  Civil  War.”  Comparative  Political  Studies  46-­‐3:  387-­‐417.       Cederman,  Lars-­‐Erik,  Min  Brian,  and  Wimmer  Andreas.  2009.  “Ethnic  Politics  and  Armed   Conflict:  A  Configurational  Analysis.”  American  Sociological  Review  74-­‐2:  316-­‐337.       Chandra,  Kanchan.  2004.  Why   Ethnic   Parties   Succeed:   Patronage   and   Ethnic   Head   Counts   in  India.  New  York:  Cambridge  University  Press.       Chang,  Eric  and  Miriam  Golden.  2010.  “Sources  of  Corruption  in  Authoritarian  Regimes.”   Social  Science  Quarterly  91-­‐1:  1-­‐20.       Chang,   Eric,   Mark   Andreas   Kayser,   Drew   Linzer,   and   Ronald   Rogowski.   2010.   Electoral   Systems  and  the  Balance  of  Consumer-­‐Producer  Power.  New  York:  Cambridge  University   Press.       Cheibub,   Antonio,   Jeniffer   Gandhi   and   James   Vreeland.   2009.   “Democracy   and   Dictatorship  Revisited.”  Public  Choice  143-­‐2:  67-­‐101.       Chebotarev,   Andrei.   2009.   Politicheskaya   Transformantsiya   Respubliki   Kazakhstan   v   Kontekste   Smeny   Upravlentseskoi   Elity   [Political   Transformation   of   the   Republic   of   Kazakhstan   in   the   Contexts   of   Changing   Ruling   Elites].   Almaty:   Tsentr   Aktualnykh   Issledovanii  “Alyternativa.”             Chiozza,  Giacomo   and   H.E.   Goemans.   2011.   Leaders  and  International  Conflict.  New  York:     275   Cambridge  University  Press.       Collier,   David   ed.   1979.   The   New   Authoritarianism   in   Latin   America.   New   Jersey:   Princeton  University  Press.         Collier,   Paul,   and   Anke   Hoeffler,   2004,   “Greed   and   Grievance   in   Civil   War.”   Oxford   Economic  Papers  56:563–595.     Collins,   Kathleen.   2006.   Clan   Politics   and   Regime   Transition   in   Central   Asia.   New   York:   Cambridge  University  Press.       Cox,   Gary.   1998.   Making   Votes   Count:   Strategic   Coordination   in   the   World’s   Electoral   Systems.  Cambridge  University  Press.       Cox,  Gary.  2009.  "Authoritarian  Elections  and  Leadership  Succession,  1975-­‐2004."     Unpublished  Manuscript.       Cummings,  Sally.  2005.  Kazakhstan:  Power  and  the  Elite.  London,  New  York:  I.B.  Tauris.       Cummings,   Sally.   2000.   Kazakhstan:   Centre-­‐Periphery   Relations,   the   Royal   Institute   of   International  Affairs.     Cummings,  Sally  ed.  2010.  Domestic  and  International  Perspectives  on  Kyrgyzstan’s  ‘Tulip   Revolution’:  Motives,  Mobilization  and  Meanings.  Routledge.       Cusack,  Thomas,  Torben  Iversen  and  David  Soskice.  2007.  “Economic  Interests  and  the   Origins  of  Electoral  Systems.”  American  Political  Science  Review  101-­‐3:  373-­‐391.       Cusack,   Thomas,   Torben   Iversen,   and   David   Soskice.   2010.   “Coevolution   of   Capitalism   and   Political   Representation:   The   Choice   of   Electoral   Systems.”   American   Political   Science  Review  104-­‐2:  393-­‐403.       Dahl,  Robert.  1972.  Polyarchy:  Participation  and  Opposition.  New  Haven:  Yale  University   Press.             276     Dave,   Bhavna.   2004.   “Kazakhstan’s   2004   Parliamentary   Elections:   Managing   Loyality   and  Support  for  the  Regime.”  Problems  of  Post-­‐Communism  52-­‐1:  3-­‐14.       Dave,   Bhavna.   2013.   “Nations   in   Transit:   Kazakhstan   2013.”   Freedom   House.   (https://freedomhouse.org/report/nations-­‐transit/2013/kazakhstan#.VOo098a-­‐uQE)       Davenport,   Christian.   2007.   State   Repression   and   the   Domestic   Democratic   Peace,   Cambridge  University  Press.       Del   Sordi,   Adele.   2012.   The   Party   of   Power   of   Kazakhstan:   The   Origins,   Resources   and   Regime-­‐Supporting  Functions  of  Nur-­‐Otan.  PhD  Thesis  Submitted  to  the  IMT  Institute  for   Advanced  Studies,  Lucca,  Italy.       Desai,   Raj,   Anders   Olofsgard,   and   Tarik   Yousef.   2009.   “The   Logic   of   Authoritarian   Bargains.”  Economics  and  Politics  21:  93-­‐125.       Diamond,  Larry.  2002.  “Elections  without  Democracy:  Thinking  about  Hybrid  Regimes.”   Journal  of  Democracy  13-­‐2:  21-­‐35.       Diaz-­‐Cayeros,   Alberto   and   Beatritz   Magaloni.   2001.   “Party   Dominance   and   the   Logic   of   Electoral   Design   in   Mexico’s   Transition   to   Democracy.”   Journal  of  Theoretical  Politics   13-­‐ 3:  271-­‐297.       Dietrich,   Simone   and   Joseph   Wright.   2014.   “Foreign   Aid   Allocation   Tactics   and   Democratic  Change  in  Africa.”  Forthcoming  in  the  Journal  of  Politics.       Donno,  Daniela.  2013.  Defending   Democratic   Norms:   International   Actors   and   the   Politics   of  Electoral  Misconduct.  New  York:  Oxford  University  Press.       Drazen,  Allan.  2000.  Political  Economy  in  Macroeconomics,  Princeton  University  Press.     Dunning,   Thad.   2004.   “Conditioning   the   Effects   of   Aid:   Cold   War   Politics,   Donor   Credibility,  and  Democracy  in  Africa.”  International  Organization  58:  409-­‐423.         277     Dunning,  Thad.  2008.  Crude  Democracy:  Natural  Resource  Wealth  and  Political  Regimes.   New  York:  Cambridge  University  Press.       Dunning,   Thad.   2011.   “Fighting   and   Voting:   Violent   Conflict   and   Electoral   Politics.”   Journal  of  Conflict  Resolution  55-­‐3:  327-­‐339.       Duverger,   Mourice.   1954.   Political  Parties:  Their  Organizations  and  Activity  in  the  Modern   State.  London:  Methuen  &  CO.  LTD.       The  Economist.  2014.  “Putin’s  Russia:  Sochi  or  Bust.”  January  30.       Egorov,   Georgy,   Sergei   Guriev   and   Konstantin   Sonin.   “Why   Resource-­‐poor   Dictators   Allow  Freer  Media:  A  Theory  and  Evidence  from  Panel  Data.”  American   Political   Science   Review  103-­‐4:  645-­‐668.           Eisenstadt,   Todd.   2004.   Courting   Democracy   in   Mexico:   Party   Strategies   and   Electoral   Institutions.  New  York:  Cambridge  University  Press.       Ezrow,   Natasha   and   Erica   Frantz.   2011.   Dictators   and   Dictatorships:   Understanding   Authoritarian  Regimes  and  Their  Leaders.  Continuum.       Fearon,   James.   1999.   “Why   Ethnic   Politics   and   ‘Pork’   Tend   to   Go   Together?’   Working   Paper.       Fearon,   James   and   David   Laitin.   1995.   “Explaining   Interethnic   Cooperation.”   American   Political  Science  Review  90-­‐4:  715-­‐735.       Fearon,   James   and   David   Laitin,   2003,   “Ethnicity,   Insurgency   and   Civil   War.”   American   Political  Science  Review  97-­‐1:  75-­‐90.       Franck   Raphael   and   Ilia   Rainer.   2012.   “Does   the   Leader’s   Ethnicity   Matter?   Ethnic   Favoritism,   Education,   and   Health   in   Sub-­‐Saharan   Africa.”   American   Political   Science   Review  106-­‐2:  294-­‐325.       278     Friedgut,   Theodore   H.   1979.   Political   Participation   in   the   USSR.   New   Jersey:   Princeton   University  Press.       Friedrich,   Carl   J.   and   Zbigniew   Brzezinski.   1965.   Totalitarian  Dictatorship  and  Autocracy.   Cambridge,  Massachusetts:  Harvard  University  Press.       Fukumoto,   Kentaro   and   Yusaku   Horiuchi.   2011.   “Making   Outsiders’   Votes   Count:   Detecting   Electoral   Fraud   through   a   Natural   Experiment.”   American   Political   Science   Review  105-­‐3:  586-­‐603.       Furman,  Dmitrii.  2006.  “The  Regime  in  Kazakhstan.”  In  Boris  Rumer  ed.  Central  Asia  at   the  End  of  the  Transition,  M.E.  Sharpe,  pp.  195-­‐266.         Gandhi,   Jennifer.   2008.   Political   Institutions   under   Dictatorship.   New   York:   Cambridge   University  Press.     Gandhi,  Jennifer  and  Lust-­‐Okar,  Ellen.  2009.  “Elections  under  Authoritarianism.”  Annual   Review  of  Political  Science  12:  403-­‐422.       Gandhi,   Jennifer   and   Adam   Przeworski.   2007.   “Authoritarian   Institutions   and   the   Survival  of  Autocrats.”  Comparative  Political  Studies,  40-­‐11:  1279-­‐1301.       Gandhi,  Jennifer  and  James  Vreeland.  2004.  "Political  Institutions  and  Civil  War:   Unpacking  Anocracy."  Unpublished  Manuscript.       Geddes,  Barbara.  1999.  “What  Do  We  Know  about  Democratization  after  Twenty  Years?”   Annual  Review  of  Political  Science  2:  115-­‐144.       Geddes,  Barbara.  2006.  “Why  Parties  and  Elections  in  Authoritarian  Regimes.”  Working   Paper.       Geddes,   Barbara,   Joseph   Wright   and   Erica   Frantz.   2014.   “New   Data   on   Autocratic   Regimes.”  Forthcoming  in  Perspective  on  Politics.       279     Gehlbach,   Scott   and   Philip   Keefer.   2011.   “Investment   without   Democracy:   Ruling-­‐Party   Institutionalization   and   Credible   Commitment   in   Autocracies.”   Journal   of   Comparative   Economics  38:  123-­‐139.         Goemans,   Hein,   Kristian   Skrede   Gleditsch,   and   Giacomo   Chiozza.   2009.   “Introducing   Archigos:  A  Data  Set  of  Political  Leaders.”  Journal  of  Peace  Research  46-­‐2:  269-­‐283.       Golder,   Sona.   2006.   The   Logic   of   Pre-­‐Electoral   Coalition   Formation.   The   Ohio   State   University  Press.       Goodwin,  Jeff.  2001.  No   Other   Way   Out:   States   and   Revolutionary   Movements,   1945-­‐1991.   Cambridge:  Cambridge  University  Press.       Greene,   Kenneth.   2007.   Why   Dominant   Parties   Lose:   Mexico's   Democratization   in   Comparative  Perspective.  New  York:  Cambridge  University  Press.     Greene,   Kenneth.   2009.   “The   Political   Economy   of   Authoritarian   Single-­‐Party   Dominance.”  Comparative  Political  Studies  43-­‐7:  807-­‐834.         Haber,  Stephen.  2006.  “Authoritarian  Government.”  In  The  Oxford  Handbook  of  Political   Economy,  eds.,  Barry  Weingast  and  Donald  Wittman.  Oxford  University  Press,  693-­‐707.       Haber,   Stephen   and   Victor   Menaldo.   2011.   “Do   Natural   Resources   Fuel   Authoritarianism?   A   Reappraisal   of   the   Resource   Curse.”   American   Political   Science   Review  105-­‐1:  1-­‐26.         Habyarimana,  James,  Macartan  Humphreys,  Daniel  Posner  and  Jeremy  Weinstein.  2007.   “Why   Does   Ethnic   Diversity   Undermine   Public   Goods   Provision?”   American   Political   Science  Review  101-­‐4:  709-­‐725.       Hadenius,   Axel   and   Jan   Teorrel.   2007.   “Pathways   from   Authoritarianism.”   Journal   of   Democracy  18-­‐1:  143-­‐156.         280   Hafner-­‐Burton,   Emilie,   Susan   Hyde   and   Ryan   Jablonski.   2012.   “When   Do   Governments   Resort  to  Election  Violence?”  Forthcoming  in  British  Journal  of  Political  Science.       Haggard,   Stephan   and   Robert   Kaufman.   1995.   The   Political   Economy   of   Democratic   Transitions.  New  Jersey:  Princeton  University  Press.         Hale,   Henry.   2006.   Why   Not   Parties   in   Russia?   Democracy,   Federalism   and   the   State,   Cambridge  University  Press.   Hale,   Henry.   2015.   Patronal   Politics:   Eurasian   Regime   Dynamics   in   Comparative   Perspective.  New  York:  Cambridge  University  Press.     Hall,   Peter   and   David   Soskice   eds.   2001.   Varieties   of   Capitalism:   The   Institutional   Foundations  of  Comparative  Advantage.  Oxford  University  Press.         Hartlyn,   Johathan,   Jennifer   McCoy,   and   Thomas   Mustillo.   2008.   “Electoral   Governance   Matters:   Explaining   the   Quality   of   Elections   in   Contemporary   Latin   America.”   Comparative  Political  Studies  41:  73-­‐98.       Hermet,  Guy,  Richard  Rose,  and  Alain  Rouquie  eds.  1978.  Elections   without   Choice.  John   Wiley  and  Sons.     Higashijima,  Masaaki.  2010.  “Electoral  Cycles,  Party  Competition  and  State  Resources  in   Authoritarianism:   Evidence   from   Kazakhstan.”   Paper   Presented   at   the   2010   Annual   Meeting  of  the  Midwest  Political  Science  Association.           Hoffman,   David.   2000.   Oil   and   State-­‐Building   in   Post-­‐Soviet   Azerbaijan   and   Kazakhstan.   PhD   Dissertation   Submitted   to   the   Department   of   Political   Science   at   the   University   of   California,  Berkeley.         Howard,   Marc   and   Philip   Roessler.   2006.   “Liberalizing   Electoral   Outcomes   in   Comparative   Authoritarian   Regimes.”   American   Journal   of   Political   Science   50-­‐2:   365-­‐ 381.       Human  Rights  Watch.  1997.  “Human  Rights  Watch  World  Report  1997.”         281     Humphreys,   Macartan,   2005,   “Natural   Resources,   Conflict   and   Conflict   Resolution:   Uncovering  the  Mechanisms.”  Journal  of  Conflict  Resolution  49:  508-­‐537.       Huntington,   Samuel.   1968.   Political   Order   in   Changing   Societies.   New   Haven:   Yale   University  Press.       Huntington,   Samuel.   1991.   The   Third   Wave:   Democratization   in   the   Late   Twentieth   Century,  University  of  Oklahoma  Press.       Huntington,   Samuel   and   Clement   Moore   eds.   1970.   Authoritarian   Politics   in   Modern   Society:  The  Dynamics  of  Established  One-­‐Party  Systems.  Basic  Books.       Huskey,   Eugene.   1997.   “Kyrgyzstan:   The   Fate   of   Political   Liberalization.”   In   Karen   Dawisha   and   Bruce   Parrott   eds.,   Conflict,   Cleavage,   and   Change   in   Central   Asia   and   the   Caucasus.  Cambridge  University  Press.       Hyde,   Susan.   2007.   “The   Observer   Effect   in   International   Politics:   Evidence   from   a   Natural  Experiment.”  World  Politics  60:  37-­‐63.       Hyde,  Susan.  2011.  The   Pseudo   Democrats   Dilemma:   Why   Election   Monitoring   Became   an   International  Norm?  New  York:  Cambridge  University  Press.       Hyde,   Susan   and   Nichorai   Marinov.   2012.   “Which   Elections   Can   be   Lost?”   Political   Analysis  20-­‐2:  191-­‐210.       Hyde,   Susan   and   Angela   O’Mahoney.   2011.   “International   Scrutiny   and   Pre-­‐Electoral   Fiscal  Manipulation.”  Journal  of  Politics  72-­‐3:  1-­‐14.       Ichino,   Nahomi   and   Matthias   Schuendeln.   2012.   “Deterring   or   Displacing   Electoral   Irregularities?   Spillover   Effects   of   Observers   in   a   Randomized   Field   Experiment   in   Ghana.”  The  Journal  of  Politics  74-­‐1:  292-­‐307.       International  Monetary  Fund.  IMF  Country  Reports  for  Kazakhstan.  Various  issues.       282     Ishiyama,   John.   1997.   “Transitional   Electoral   Systems   in   Post-­‐Communist   Eastern   Europe.”  Political  Science  Quarterly  112-­‐1:  95-­‐115.       Issacs,  Rico.  2011.  Party  System  Formation  in  Kazakhstan:  Between  Formal  and  Informal   Politics.  Routledge.       Issacs,  Rico.  2013.  “Nur-­‐Otan,  Informal  Networks  and  the  Countering  of  Elite  Instability   in  Kazakhstan:  Bringing  the  ‘Formal’  Back  In.”  Europe-­‐Asia  Studies  65-­‐6:  1055-­‐1079.       Iversen,   Torben   and   David   Soskice.   2006.   “Electoral   Systems   and   the   Politics   of   Coalitions:  Why  Some  Democracies  Redistribute  More  Than  Others.”  American  Political   Science  Review  100-­‐2:  165-­‐181.         Jensen,   Nathan,   Edmund   Malesky,   and   Stephen   Weymouth.   2013.   “Unbundling   the   Relationship   between   Authoritarian   Legislatures   and   Political   Risk.”   Forthcoming   in   British  Journal  of  Political  Science.         Jensen,  Nathan  and  Leonard  Wantchekon.  2004.  “Resource  Wealth  and  Political  Regimes   in  Africa.”  Comparative  Political  Studies  37-­‐7:  816-­‐841.       Jones,   Kevin.   2007.   The   Dynamics   of   Political   Protests:   A   Case   Study   of   Kyrgyz   Republic.   Ph.D  dissertation,  University  of  Maryland.     Jones   Luong,   Pauline.   2002.   Institutional   Change   and   Political   Continuity   in   Post-­‐Soviet   Central  Asia.  New  York:  Cambridge  University  Press.       Jones   Luong,   Pauline.   2004.   “Economic   ‘Decentralization’   in   Kazakhstan:   Causes   and   Consequences.”   In   Pauline   Jones   Luong   ed.   The   Transformation   of   Central   Asia:   States   and   Societies   from   Soviet   Rule   to   Independence.   Ithaca   and   London:   Cornell   University   Press,  pp.  182-­‐210.       Jones   Luong,   Pauline   and   Erika   Weinthal.   2011.   Oil  Is  Not  a  Curse:  Ownership  Structure   and  Institutions  in  Soviet  Successor  States.  New  York:  Cambridge  University  Press.       283   Junisbai,  Barbara.  2010.  “A  Tale  of  Two  Kazakhstans:  Sources  of  Political  Cleavage  and   Conflict  in  the  Post-­‐Soviet  Period.”  Europe-­‐Asia  Studies  62-­‐2:  235-­‐269.       Junisbai,  Azamat  and  Barbara  Junisbai.  2005.  “The  Democratic  Choice  of  Kazakhstan:  A   Case   Study   in   Economic   Liberalization,   Intra-­‐Elite   Cleavage   and   Political   Opposition.”   Demokratizatsiya:  The  Journal  of  Post-­‐Soviet  Democratization  13:  385-­‐421.         Kalandadze,   Katya   and   Mitchell   Orenstein,   “Electoral   Protests   and   Democratization   beyond  the  Color  Revolutions.”  Comparative  Political  Studies  42-­‐11:  1403-­‐1425.       Kelley,   Judith.   2009.   “D-­‐Minus   Elections:   The   Politics   and   Norms   of   International   Election  Observation.”  International  Organization  63:  765-­‐787.       Kelley,   Judith.   2012.   Monitoring   Democracy:   When   International   Election   Observation   Works,  and  Why  It  Often  Fails.  New  Jersey:  Princeton  University  Press.       Kim,  Wonik  and  Jennifer  Gandhi.  2010.  “Coopting  Workers  under  Dictatorship.”  Journal   of  Politics  72-­‐3:  646-­‐658.         King,  Gary,  Jeniffer  Pan,  and  Margaret  Roberts.  2013.  “How  Censorship  in  China  Allows   Government   Criticism   but   Silences   Collective   Expression.”   American   Political   Science   Review  107-­‐2:  1-­‐18.       Kinne,   Brandon   and   Nikolay   Marinov.   2013.   “Electoral   Authoritarianism   and   Credible   Signaling  in  International  Crisis.”  forthcoming  in  Journal  of  Conflict  Resolution.       Kreuzer,  Marcus.  2010.  “Historical  Knowledge  and  Quantitative  Analysis:  The  Case  of  the   Origins   of   Proportional   Representation.”   American  Political  Science  Review   104-­‐2:   369-­‐ 392.         Krichelli,  Ruth,  Yair  Livne  and  Beatriz  Magaloni.  2011.  “Taking  to  the  Streets:  Theory  and   Evidence  on  Protests  under  Authoritarianism.”  Working  Paper.       Kuhn,  Patrick.  2012.  “Elections,  Information,  Fraud  and  Post-­‐Electoral  Protest.”  Working   Paper.     284       Kulov,   Emir.   2008.   “March   2005:   Parliamentary   Elections   as   a   Catalyst   of   Protests.”   Central  Asian  Survey  27-­‐3,4:  337–47     Kuntz,   Philipp   and   Mark   Thompson.   2009.   "More   Than   Just   the   Final   Straw:   Stolen   Elections  as  Revolutionary  Triggers."  Comparative  Politics  41-­‐3:  253-­‐273.     Kuran,  Timur.  1991.  “Now  Out  of  Never:  The  Element  of  Surprise  in  the  East  European   Revolution  of  1989.”  World  Politics  44-­‐1:  7-­‐48.           LeBas,   Adrienne.   2006.   “Polarization   as   Craft:   Explaining   Party   Formation   and   State   Violence  in  Zimbabwe,”  Comparative  Politics  38-­‐4:  419-­‐438.       Leemann,   Lucas   and   Isabela   Mares.   2014.   “The   Adoption   of   Proportional   Representation.”  The  Journal  of  Politics.  76-­‐2:  461-­‐478.       Lehoucq,  Fabrice  and  Ivan  Molina.  2002.  Stuffing  the  Ballot  Box:  Fraud,  Electoral  Reform,   and  Democratization  in  Costa  Rica.  New  York:  Cambridge  University  Press.       Lehoucq,   Fabrice.   2003.   “Electoral   Fraud:   Causes,   Types,   and   Consequences.”   Annual   Review  of  Political  Science  6:  233-­‐256.       Levitsky,   Steven   and   Lucan   Way.   2002.   “Elections   without   Democracy:   The   Rise   of   Competitive  Authoritarianism.”  Journal  of  Democracy  13-­‐2:  51-­‐65.         Levitsky,   Steven,   and   Lucan   Way.   2010.   Competitive   Authoritarianism:   Hybrid   Regimes   after  the  Cold  War.  New  York:  Cambridge  University  Press.       Lewis,   David.   2008.   The   Temptations   of   Tyranny   in   Central   Asia.   London:   Hurst   &   Company.       Lijphart,   Arend.   1994.   Electoral   Systems   and   Party   Systems:   A   Study   of   Twenty-­‐Seven   Democracies,  1945-­‐1990.  Oxford:  Oxford  University  Press.         285   Lijphart,   Arend.   2011.   Patterns   of   Democracy,   2nd   edition.   New   Haven:   Yale   University   Press.       Lindberg,  Staffan.  2006.  Democracy  and  Elections  in  Africa,  Baltimore:  The  Johns  Hopkins   University  Press.       Linz,   Juan.   1975.   “Totalitarian   and   Authoritarian   Regimes.”   In   Fred   Greenstein   and   Nelson   Polsby   eds.   Handbook  of  Political  Science,   Vol.   3.   Reading,   M.A:   Addison   Wesley   Publishing  Company,  pp.  191-­‐357.       Linz,   Juan   and   Alfred   Stepan.   1996.   Problems  of  Democratic  Transition  and  Consolidation:   Southern  Europe,  South  America,  and  Post-­‐Communist  Europe.   Johns   Hopkins   University   Press.       Lust-­‐Okar,   Ellen.   2004.   Structuring   Conflict   in   the   Arab   World:   Incumbents,   Opponents,   and  Institutions.  New  York:  Cambridge  University  Press.       Lust-­‐Okar,   Ellen.   2008.   “Competitive   Clientelism   in   Jordanian   Elections.”   In   Ellen   Lust-­‐ Okar  and  Saboua  Zehouni  eds.,  Political  Participation  in  the  Middle  Eats,  Lynne  Rienner   Publishers,  pp.75-­‐94.       Lust-­‐Okar,   Ellen   and   Amaney   Ahmad   Jamal.   2002.   “Rulers   and   Rules:   Reassessing   the   Influence   of   Regime   Type   and   Electoral   Law   Formation.”   Comparative   Political   Studies   35-­‐3:  337-­‐366.     Mahdavy,   Hussein.   1970.   “The   Patterns   and   Problems   of   Economic   Development   in   Rentier  States:  The  Case  of  Iran.”  In  Studies  in  Economic  History  of  the  Middle  East,  M.A.   Cook.  ed.,  428-­‐467.  London:  Oxford  University  Press.       Magaloni,   Beatriz.   2006.   Voting  for  Autocracy:  Hegemonic  Party  Survival  and  Its  Demise  in   Mexico.  New  York:  Cambridge  University  Press.     Magaloni,   Beatritz.   2008.   “Credible   Power-­‐Sharing   and   the   Longevity   of   Authoritarian   Rule.”  Comparative  Political  Studies  41:  715-­‐41.         286   Magaloni,   Beatritz.   2010.   “The   Game   of   Electoral   Fraud   and   the   Ousting   of   Authoritarian   Rule.”  American  Journal  of  Political  Science  54-­‐3:  751-­‐765.       Magaloni,   Beatritz   and   Ruth   Kricheli.   2010.   “Political   Order   and   One-­‐Party   Rule.”   Annual   Review  of  Political  Science  13:  123-­‐143.       Mahdavi,   Paasha.   2014.   “Explaining   the   Oil   Advantage:   Effects   of   Natural   Resource   Wealth  on  Incumbent  Reelection  in  Iran.”  Forthcoming  in  World  Politics.       Makhmutova,   Meruert.   2005.   “Local   Government   Organization   and   Finance:   Kazakhstan.”   In   Anwar   Shah   ed.   Local   Governance   in   Developing   Countries.   The   World   Bank.       Malesky,   Edmund   and   Paul   Schuler.   2010.   “Nodding   or   Needling:   Analyzing   Delegate   Responsiveness   in   a   Authoritarian   Parliament.”   American  Political  Science  Review   103-­‐3:   482-­‐502.       Marat,   Erica.   2006.   The   Tulip   Revolution:   Kyrgyzstan   One   Year   After,   March   15,   2005   –   March  24,  2006.  The  Jamestown  Foundation.       McElwain   Mori,   Kenneth.   2008.   “Manipulating   Electoral   Rules   to   Manufacture   Single-­‐ Party  Dominance.”  American  Journal  of  Political  Science  52-­‐1:  32-­‐47.     McGlinchey,   Eric.   2011.   Chaos,   Violence,   and   Dynasty:   Politics   and   Islam   in   Central   Asia.   University  of  Pittsburgh  Press.           Miller,   Michael.   2012.   “Elections,   Information,   and   Policy   Responsiveness   in   Autocratic   Regimes.”  Forthcoming  in  Comparative  Political  Studies.       Montinola,   Gabriella,   Yingyi   Qian   and   Barry   Weingast.   “Federalism,   Chinese   Style:   The   Political  Basis  for  Economic  Success.”  World  Politics  48-­‐1:  50-­‐81.       Moore,   Barrington   Jr.   1966.   Social   Origins   of   Dictatorship   and   Democracy:   Lord   and   Peasant  in  the  Making  of  the  Modern  World.  Boston:  Beacon  Press.         287   Morrison,   Kevin.   2009.   “Oil,   Nontax   Revenue,   and   the   Redistributional   Foundations   of   Regime  Stability.”  International  Organization  63:  107-­‐138.       Moser,   Robert.   2001.   Unexpected   Outcomes:   Electoral   Systems,   Political   Parties,   and   Representation  in  Russia.  University  of  Pittsburgh  Press.     Moser,  Robert  and  Ethan  Scheiner.  2012.  Electoral  Systems  and  Political  Context:  How  the   Effects   of   Rules   Vary   across   New   and   Established   Democracies.   New   York:   Cambridge   University  Press.       Mutlu,   Gulay.   2012.   “Transition   to   Multi-­‐Party   System   in   Kazakhstan?”   The   Journal   of   Turkish  Weekly.  January  12nd,  2012.       The  National  Bank  of  Kazakhstan.  Statisticheskii  Byulleten.  Various  issues  (1995-­‐2008).       Nohlen,   Dieter,   Florian   Grotz   and   Christof   Hartmann.   2001.   Elections   in   Asia   and   the   Pacific:   A   Data   Handbook,   Volume   I:   The   Middle   East,   Central   Asia,   and   South   Asia.  Oxford   University  Press.         Nordhaus,  William.  1975.  “The  Political  Business  Cycle.”  Review   of   Economic   Studies  42-­‐ 2:  169-­‐190.       Norris,   Pippa.   2014.   Why   Electoral   Integrity   Matters.   New   York:   Cambridge   University   Press.       Nurmukhamedov,  Burikhan  and  Andrei  Chebotarev.  2005.  Itogi  Parlamentskikh  Vyborov   2004  Goda:  Statistika  i  Analiz  [Results  of  the  2004  Parliamentary  Elections:  Statistics  and   Analysis].  Almaty:  Institute  of  National  Studies.           O’Donnell,   Guillermo.   1979   [1973].   Modernization   and   Bureaucratic-­‐Authoritarianism.   Institute  of  International  Studies,  University  of  California,  Berkeley.         O’Donnell,   Guillermo   and   Philipe   Schmitter.   1986.   Transitions   from   Authoritarian   Rule:   Tentative  Conclusions  about  Uncertain  Democracies.  Johns  Hopkins  University  Press.         288   OECD.   2011.   Competitiveness   and   Private   Sector   Development:   Kazakhstan   2010.   OECD   Publishing.       Oka,   Natsuko.   2003.   “Kazafusutan   ni   okeru   Minzoku   Undo   no   Yokusanka”   [Conforming   Ethnic  Movements  in  Kazakhstan]  In  Takeuchi  Shinichi  ed.  Kokka,  Bouryoku,  Seiji  [State,   Violence,   and   Politics]   Institute   of   Developing   Economies,   Japan   External   Trade   Organization.       Olcott,  Martha  Brill.  1995.  The  Kazakhs,  the  2nd  Edition.  Stanford  University  Press.       Olcott,   Martha   Brill.   2002   [2010:   the   2nd   edition].   Kazakhstan:   Unfulfilled   Promise?   Carnegie  Endowment  for  International  Peace.         Olcott,   Martha.   2012.   “Kazakhstan’s   Political   (R)evolution.”   Carnegie   Endowment   for   International  Peace.       Olson,   David.   1993.   “Compartmentalized   Competition:   The   Managed   Transitional   Election  System  of  Poland.”  The  Journal  of  Politics  55-­‐2:  415-­‐441.       OSCE.   1995.   “The   Parliamentary   Election   in   Kyrgyzstan,   Fenruary   5,   1995”   Report   Prepared   by   the   Staff   of   the   Commission   on   Security   and   Cooperation   in   Europe.   Bishkek,  Kyrgyzstan.       OSCE.  2000a.  “Republic  of  Kazakhstan  Parliamentary  Elections  10  and  24  October  1999:   Final  Report.”  Warsaw:  OSCE  Office  for  Democratic  Institutions  and  Human  Rights.         OSCE.   2000b.   “Kyrgyzstan,   Parliamentary   Elections,   20   February   and   12   March   2000:   Final  Report.”  Warsaw:  OSCE  Office  for  Democratic  Institutions  and  Human  Rights.           OSCE.   2001:   “Kyrgyzstan   Republic   Presidential   Elections,   29   October   2000:   Final   Report.”  Warsaw:  OSCE  Office  for  Democratic  Institutions  and  Human  Rights.         OSCE.   2004a.   “Republic   of   Kazakhstan   Parliamentary   Elections   19   September   2004,   OSCE/ODIHR  Needs  Assessment  Mission  Report.”  Warsaw:  OSCE  Office  for  Democratic   Institutions  and  Human  Rights.       289     OSCE.   2004b.   “Republic   of   Kazakhstan   Parliamentary   Elections   19   September   and   3   October   2004   OSCE/ODIHR   Election   Observation   Mission   Report:   Final   Report”   Warsaw:  OSCE  Office  for  Democratic  Institutions  and  Human  Rights.       OSCE.  2005.  “The  Kyrgyz  Republic  Parliamentary  Elections,  27  February  and  13  March   2005,   OSCE/ODIHR   Election   Observation   Mission   Final   Report.”   Warsaw:   OSCE   Office   for  Democratic  Institutions  and  Human  Rights.       OSCE.   2007.   Republic   of   Kazakhstan   Parliamentary   Elections,   18   August   2007,   OSCE/ODIHR   Election   Observation   Mission   Final   Report.”   Warsaw:   OSCE   Office   for   Democratic  Institutions  and  Human  Rights.       Ostrowski,  Wojclech.  2011.  Politics  and  Oil  in  Kazakhstan.  Routledge.       Paler,   Laura.   2013.   “Keeping   the   Public   Purse:   An   Experiment   in   Windfalls,   Taxes,   and   the   Incentives   to   Restrain   Government.”   American  Political  Science  Review   107-­‐4:   706-­‐ 725.       Pepinsky,   Thomas.   2007.   “Autocracy,   Elections   and   Fiscal   Policy:   Evidence   from   Malaysia.”  Studies  in  Comparative  International  Development  42:  136-­‐163.       Pepinsky,   Thomas.   2014.   “The   Institutional   Turn   in   Competitive   Authoritarianism.”   British  Journal  of  Political  Science  44-­‐3:  631-­‐653.       Persson,   Torsten   and   Guido   Tabellini.   2004.   The   Economic   Effects   of   Constitutions.   Cambridge:  MIT  Press.       Pevehouse,   Jon.   2005.   Democracy   from   Above:   Regional   Organizations   and   Democratization.  Cambridge,  UK:  Cambridge  University  Press.       Pomfret,  Richard.  2006.  The  Central  Asian  Economies  Since  Independence.  Princeton  and   Oxford:  Princeton  University  Press.         290   Pripstein   Posusney,   Marsha.   2002.   “Multi-­‐Party   Elections   in   the   Arab   World:   Institutional   Engineering   and   Oppositional   Strategies.”   Studies   in   Comparative   International  Development  36-­‐4:  34-­‐62.       Przeworski,   Adam.   1991.   Democracy  and  the  Market.   New   York:   Cambridge   University   Press.       Przeworski,   Adam,   Michael   Alvarez,   Antonio   Cheibub,   and   Fernando   Limongi.   2000.   Democracy  and  Development.  New  York:  Cambridge  University  Press.       Przeworski,   Adam   and   Henry   Teune.   1970.   The   Logic   of   Comparative   Social   Inquiry.   Krieger  Pub.  Co.       Radio  Free  Europe.  RFE/  RL  Research  Report  (1992-­‐1994).         Radio  Free  Europe.    RFE/  RL  Online  News,  (http://www.rfel.org/).       Radnitz,  Scott.  2010.  Weapons  of  the  Wealthy:  Predatory  Regimes  and  Elite-­‐Led  Protests  in   Central  Asia.  Ithaca  and  London:  Cornell  University  Press.       Rae,  Douglass.  1971.  The  Political  Consequences  of  Electoral  Law.  Yale  University  Press.       Remington,  Thomas  and  Steven  Smith.  1996.  “Political  Goals,  Institutional  Context,  and   the   Choice   of   an   Electoral   System:   The   Russian   Parliamentary   Election   Law.”   American   Journal  of  Political  Science,  40-­‐4:  1253-­‐1279.       Reuter,   Ora   John   and   Graeme   Robertson.   2011.   “Sub-­‐National   Appointments   in   Authoritarian  Regimes:  Evidence  from  Russian  Gubernatorial  Appointment.”  The  Journal   of  Politics  74-­‐4:  1023-­‐1037.       Rogoff,   Kenneth   and   Anne   Silbert.   1988.   "Elections   and   Macroeconomic   Policy   Cycles."   Review  of  Economic  Studies  55:  1-­‐16.       Rogowski,   Ronald.   1987.   “Trade   and   the   Variety   of   Democratic   Institutions.”     291   International  Organization  41:  203-­‐223.       Roessler,   Philip.   2011.   “The   Enemy   Within:   Personal   Rule,   Coups,   and   Civil   War   in   Africa.”  World  Politics  63-­‐2:  300-­‐346.       Roessler,   Philip   and   Marc   Haward.   2009.   “Post-­‐Cold   War   Political   Regimes:   When   Do   Elections   Matter?”   in   Staffen   Lindberg   ed.   Democratization   by   Elections,   the   Johns   Hopkins  University  Press,  pp.  101-­‐127.       Rokkan,   Stein.   1970[2009].   Citizens,   Elections,   Parties:   Approaches   to   the   Comparative   Study  of  the  Processes  of  Development.  Colchester:  The  ECPR  Press.       Roodman,   David.   2007.   “How   to   Do   xtaband2:   An   Introduction   to   Difference   and   System   GMM  in  Stata.”  Center  for  Global  Development  WP  103.       Rose,   Richard,   William   Mishler,   and   Neil   Munro.   2011.   Popular   Support   for   an   Undemocratic   Regime:   The   Changing   Views   of   Russians.  New  York:  Cambridge  University   Press.         Ross,  Michael,  2001.  “Does  Oil  Hinder  Democracy?”  World  Politics  53:  325-­‐61.       Ross,  Michael.  2012.  The  Oil  Curse,  New  Jersey:  Princeton  University  Press.       Rueschemeyer,   Dietrich,   John   D.   Stephens   and   Evelyne   Huber   Stephens.   Capitalist   Development  and  Democracy.  Chicago:  University  of  Chicago  Press.       Sachs,   Jeffrey   and   Andrew   Warner.   1995.   “Natural   Resource   Abundance   and   Economic   Growth.”   Development   Discussion   Paper   517a.   Harvard   Institute   for   International   Development.       Samaraweera,   Vijaya.   1977.   “Sri   Lanka’s   1977   General   Election:   The   Resurgence   of   the   UNP.”  Asian  Survey  17-­‐12:  1195-­‐1206.       Schatz,   Edward.   2004.   Modern   Clan   Politics:   The   Power   of   “Blood”   in   Kazakhstan   and     292   Beyond.  Seattle:  University  of  Washington  Press.       Schatz,   Edward.   2009.   “The   Soft   Authoritarian   Tool   Kit:   Agenda-­‐Setting   Power   in   Kazakhstan  and  Kyrgyzstan.”  Comparative  Politics,  41-­‐2:  203-­‐222.       Schedler,   Andreas.   2002.   “Elections   without   Democracy:   The   Menu   of   Manipulation.”   Journal  of  Democracy  13-­‐2:  36-­‐50.       Schedler,   Andreas   ed.   2006.   Electoral   Authoritarianism:   The   Dynamics   of   Unfree   Competition.  Boulder:  Lynne  Rienner.       Schedler,  Andreas.  2013.  The  Politics  of  Uncertainty:  Sustaining  and  Subverting  Electoral   Authoritarianism.  Oxford:  Oxford  University  Press.       Sharipova,   Dina.   2013.   State-­‐Building   and   Clientelism   in   Post-­‐Soviet   Kazakhstan:   Continuity   and   Transformation   of   Informal   Institutions.   PhD   Dissertation   Submitted   to   the  Department  of  Political  Science  at  Indiana  University.       Shi,   Min   and   Jakob   Svensson.   2006.   “Political   Budget   Cycles:   Do   they   Differ   across   Countries  and  Why?”  Journal  of  Public  Economics  90:  1367-­‐1389.       Shi,  Tianjian.  1999.  “Voting  and  Nonvoting  in  China:  Voting  Behavior  in  Plebiscitary  and   Limited-­‐Choice  Elections.”  Journal  of  Politics  61-­‐4:  1115-­‐1139.       Siegel,   David.   2014.   “State   Formation   in   Post-­‐Soviet   Central   Asia:   The   Politics   of   Territorial   Control   in   Kazakhstan   and   Kyrgyzstan.”   Paper   Delivered   at   the   ASEEES-­‐CESS   Conference  held  in  Astata,  Kazakhstan  in  June  2014.         Simpser,   Alberto   and   Daniela   Donno.   2012.   “Can   International   Elections   Monitoring   Harm  Governance?”  Journal  of  Politics  74-­‐2:  501-­‐513.       Simpser,   Alberto.   2013.   Why   Governments   and   Parties   Manipulate   Elections:   Theory,   Practice  and  Implications.  New  York:  Cambridge  University  Press.           293   Sjöberg,  Fredrik.  2011.  Competitive  Elections  in  Authoritarian  States:  Weak  State,  Strong   Elites,  and  Fractional  Societies  in  Central  Asia  and  Beyond.    PhD  Dissertation  Submitted  to   the  Department  of  Government  at  Uppsala  University.       Smith,  Benjamin.  2004.  “Oil  Wealth  and  Regime  Survival  in  the  Developing  World,  1960-­‐ 1999.”  American  Journal  of  Political  Science  48-­‐2:  232-­‐246.       Smith,   Benjamin.   2007.   Hand   Times   in   the   Lands   of   Plenty:   Oil   Politics   in   Iran   and   Indonesia,  Cornell  University  Press.       Snyder,   Jack.   2000.   From   Voting   to   Violence:   Democratization   and   nationalist   Conflict.   New  York:  W.W.  Norton  &  Company.       Spector,   Regine.   2004.   “The   Transformation   of   Askar   Akaev,   President   of   Kyrgyzstan.”   Berkeley  Program  in  Soviet  and  Post-­‐Soviet  Studies  Working  Paper  Series.         Statistical  Agency  of  Kazakhstan.  Regiony  Kazakhstana.  Various  issues  (1995-­‐2008)     Stokes,  Susan,  Thad  Dunning,  Marcelo  Nazareno  and  Valeria  Brusco,  Brokers,  Voters  and   Clientelism:  The  Puzzle  of  Distributive  Politics,  Cambridge  University  Press.       Straus,  Scott  and  Charlie  Taylor.  2012.  “Democratization  and  Electoral  Violence  in  Sub-­‐ Saharan   Africa,   1990-­‐2008.”   in   Voting  in  Fear:  Electoral  Violence  in  Sub-­‐Saharan  Africa,   ed.  Dorina  Bekoe.  Washington  DC:  United  States  Institute  of  Peace,  15-­‐38.       Strom,  Kaare,  Ian  Budge  and  Michael  Laver.  1994.  “Constraints  on  Cabinet  Formation  in   Parliamentary  Democracies.”  American  Journal  of  Political  Science.  38-­‐2:  303-­‐335.       Svolik,  Milan.  2012.  The  Politics  of  Authoritarian  Rule.  New  York:  Cambridge  University   Press.       Taagepera,  Rein.  1998.  “Effective  Magnitude  and  Effective  Threshold.”  Electoral  Studies   17:  393-­‐404.           294   Taagepera,   Rein.   2007.   Predicting   Party   Sizes:   The   Logic   of   Simple   Electoral   Systems.   Oxford:  Oxford  University  Press.       Taagepera,   Rein   and   Matthew   Shugart.   1989.   Seats   and   Votes:   The   Effects   and   Determinants  of  Electoral  Systems.  New  Haven  and  London:  Yale  University  Press.         Tedin,   Kent.   1994.   “Popular   Support   for   Competitive   Elections   in   the   Soviet   Union.”   Comparative  Political  Studies  27:  241-­‐271.       Tengri  News.  Available  at  http://en.tengrinews.kz/.       Teorrel,  Jan  and  Alex  Hadenius.  2009.  “Elections  as  Levers  of  Democratization:  A  Global   Inquiry,”  in  Staffan  Lindberg  ed.  Democratization  by  Elections:  A  New  Mode  of  Transition.   Baltimore:  The  Johns  Hopkins  University  Press,  pp.  77-­‐100.       Tordoff,   William.   1995.   “Local   Government   in   Kyrgyzstan.”   Public   Administration   and   Development  15:  495-­‐505.       Treisman,   Daniel.   1999.   After   the   Deluge:   Regional   Crises   and   Political   Consolidation   in   Russia.  Ann  Arbor:  University  of  Michigan  Press.       Tsebelis,  George.  1990.  Nested  Games.  University  of  California  Press.     Tsentralnaya   Izbiratelnaya   Komissiya   Respubliki   Kazakhstan.   2010.   Informatsionnyi   Sbornik   o   Vyborakh   i   Respyblikanskikh   Referendymakh,   Sostoyavshisya   v   Respyblike   Kazakhstan   v   Period   s   25   Mapta   1990   Goda   po   4   Oktyabrya   2008   Goda   [Information   Collection   about   Elections   and   Republican   Referenda,   Conducted   in   the   Republic   of   Kazakhstan   from   March   1990   until   4   October   2008].   Almaty:   Central   Election   Commission  of  the  Republic  of  Kazakhstan.           Tucker,  Joshua  A.  2007.  “Enough!  Electoral  Fraud,  Collective  Action  Problems,  and  Post-­‐ Communist  Colored  Revolutions.”  Perspective  on  Politics  5-­‐3:  535-­‐551.     Tufte,  Edward.  1978.  Political  Control  of  the  Economy.  New  Jersey:    Princeton  University   Press.     295     Tullock,  Gordon.  1987.  Autocracy.  Kluwer  Academic  Publishers.       Ulfelder,   Jay.   2007.   “Natural-­‐Resource   Wealth   and   the   Survival   of   Autocracy.”   Comparative  Political  Studies  40-­‐8:  995-­‐1018.       Uyama,   Tomohiko.   2003.   “Why   Are   Social   Protest   Movements   Weak   in   Central   Asia:   Relations  between  the  State  and  People  in  the  Era  of  Nation-­‐  Building  and  Globalization”,   in   Sakai,   Keiko   eds.,   Social   Protests   and   Nation-­‐Building   in   the   Middle   East   and   Central   Asia,  IDE  Development  Perspective  Series  No.1,  Institute  of  Developing  Economies,  Japan   External  Trade  Organization.     Uyama,  Tomohiko.  2006.  “Kyruguzusutan  (Kirugisu)  no  Kakumei:  Erito  no  Rigoushusan   to   Shakainettowaku   no   Douin”   [The   Revolution   in   Kyrgyzstan:   Changing   Alliances   among   Elites   and   the   Mobilization   of   Social   Networks]   In   Tomohiko   Uyama,   Hirotake   Maeda   and   Shinkichi   Fujimori.   “Minshukakakumei   toha   Nan   dattanoka:   Gurujia,   Ukuraina,   Kuruguzusutan”   [What   was   the   Revolutions   for   Democratization?   Georgia,   Ukraine,  and  Kyrgyzstan]  Sapporo:  Slavic  Eurasian  Studies  Working  Paper  Series  No.  16.           Weiss,   Jessica   Chen.   2012.   “Authoritarian   Signaling,   Mass   Audiences,   and   Nationalist   Protest  in  China.”  International  Organization  67-­‐1:  1-­‐35.       White,  Stephen.  1988.  “Reforming  the  Electoral  System.”  Journal  of  Communist  Studies,  4-­‐ 4:  1-­‐17.       Wilson,   Matthew,   C.   2013.   “A   Discreet   Critique   of   Discrete   Regime   Type   Data.”   Forthcoming  in  Comparative  Political  Studies.       Wimmer,   Andreas,   Brian   Min   and   Lars-­‐Eric   Cederman.   2009.   “Ethnic   Politics   and   Armed   Conflict:   A   Configurational   Analysis   of   a   New   Global   Data   Set.”   American   Sociological   Review  74:  316-­‐337.       Wintrobe,   Ronald.   1998.   The   Political   Economy   of   Dictatorship.   New   York:     Cambridge   University  Press.         296   Wood,   Elizabeth.   2000.   Forging   Democracy   from   Below:   Insurgent   Transitions   in   South   Africa  and  El  Salvador.  New  York:  Cambridge  University  Press.       Wooldridge,   Jeffrey.   2002.   Econometric  Analysis  of  Cross-­‐Section  and  Panel  Data.   The   MIT   Press.       Wright,  Joseph.  2008.  “Do  Authoritarian  Institutions  Constrain?  How  Legislatures  Affect   Growth  and  Investment.”  American  Journal  of  Political  Science  52-­‐2:  322-­‐343.       Wright,   Joseph.   2009.   “How   Foreign   Aid   Can   Foster   Democratization   in   Authoritarian   Regimes.”  American  Journal  of  Political  Science  53-­‐3:  552-­‐571.       Wright,  Joseph.  2011.  “Electoral  Spending  Cycles  in  Dictatorships.”  Working  Paper.       Wright,   Joseph,   Erica   Frantz   and   Barbara   Geddes.   2013.   “Oil   and   Autocratic   Regime   Survival.”  Working  Paper.       Zhovits,  Evgeny  A.  1999.  “Freedom  of  Association  and  the  Question  of  Its  Realization  in   Kazakhstan.”   In   Ruffin,   M.   Holt   and   Daniel   C.   Waugh,   Civil   Society   in   Central   Asia,   University  of  Washington  Press,  pp.57-­‐70.     Ziblatt,   Daniel.   2009.   “Shaping   Democratic   Practice   and   the   Causes   of   Electoral   Fraud:   The   Case   of   Nineteenth-­‐Century   Germany.”   American  Political  Science  Review   103-­‐1:   1-­‐ 21.           297