. 5..
.15.... t
. Hwy-VJ.

n.’ .

.1.»

‘c~

1...... r .u.‘ 2..

E B. u “is p.
, v.3...

. ¢ n; [a
1| 1- Wkuaflxh. a:
I: V «
‘1. A. it: It
a»

Z .1 :- 1!!-
:.§
1.3.0....
. 1.1.1.14!
fill - ._.\~ vlv, I

3,: .1

1 ‘
.r1 1...:-

1i 5
3.2:..va :.
fisri)...
15.13.35
I )A 1. . .r
P bu. 3...! ...\

t.
brtuflbs. .g,
t 2. 9:9.5‘:

‘ 9. 5......
[2.3:
. avaifil.
u}... 1..

a- ‘

.
.19. In.
:I‘flh‘.

1L};
nip-1:2.
.o. E t

.125?

h“ 1 1 ‘
. . 13st

\ 55133.2};

.1..:[d.tu.):lu?

. A k \..
.. d: v .3.
"a. 2 f.)

 

 

.. ill: . b.

r .. ‘v‘ .5. I . , .

N:.D.:it;!1. at a 1.:
.l I... g It... .. 2 6 :n. p.

r a 5-2:... ..

.1.

:Azr:\:‘ ‘. ... .. 31.1 ...

 

 

 

 

(Mists:

QOOI

 

LIBRARY
Michigan State
University

W

 

 

This is to certify that the

dissertation entitled

ECONOMICS AND ELECTIONS

IN SOUTH KOREA AND TAIWAN
presented by

J UNHAN LEE

has been accepted towards fulfillment
of the requirements for

PILD- degree in WIEN CE

 

Wily/C

Major professor

DateQQJ’k’l/l 3/0?!

MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

‘ APR ~2 3709-5;

 

'JAN 1 6 2009

 

 

 

 

 

 

 

 

 

 

6’01 c:/ClFiC/DaI90u9.p65—p.15

 

 

ECONOMICS AND ELECTIONS IN SOUTH KOREA AND TAIWAN
By

Junhan Lee

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Political Science

2001

ABSTRACT
ECONOMICS AND ELECTIONS IN SOUTH KOREA AND TAIWAN
By

Junhan Lee

This study undertakes a two-fold task. First, I probe the relationship between
economic conditions and voter turnout at the national level elections in South Korea and
Taiwan after the democratic transition in the mid-19805. Second, I investigate the
association between economic conditions and electoral support in the two countries’
national elections in the 19905. This study is based on the assumption that voters in these
countries are not likely to be mobilized any longer as they have been under authoritarian
regimes.

Nonetheless, long lasting regional conflicts between Cholla province and
Kyungsang province still have a significant impact on the decisions of who to vote for in
South Korea. In Taiwan, traditional ethnic conflicts between the Taiwanese and the
Mainlanders have significantly affected voters’ choices. Regional and ethnic conflicts
have been frequently utilized and manipulated by political leaders in maximizing their
vote returns in the two nations. In this research project, I compare the effects of
economic conditions on voting behavior with those of old regional (or ethnic) cleavages
in each country. Also, I investigate the impacts of the socioeconomic factors, the social
psychological factors, the political mobilization factors, and the party identification

factors on voting behavior.

I utilize the survey data sets from the Institute for Korean Election Studies (IKES)
and the National Chengchi University (NCU). The IKES provides survey data for the
1992 and 1996 National Assembly elections and the 1992 and 1997 Presidential
elections. The NCU conducted surveys for the 1992 and 1995 Legislative Yuan elections
and the 1996 Presidential election. South Korea’s National Assembly is equivalent to
Taiwan’s Legislative Yuan. In analyzing the survey data, I employ probit for the two-
way choices, for instance, voters’ choice between vote and abstention or between two
parties (or candidates). For the three-way choices, I employ multinomial logit.

The main findings of the study are as follows: First, economic conditions appear
to have had nothing to do with voter turnout in South Korea and Taiwan. Second,
regional and ethnic cleavages were not significant for voter participation in South Korea
and Taiwan, respectively. Third, genuine economic voting was evident in the most recent
elections (the 1997 South Korean race and the 1996 Taiwanese election). Fourth,
regionalism was consistently significant throughout the South Korean elections under

study, while ethnic cleavages did not have a consistent impact on vote returns in Taiwan.

ACKNOWLEDGEMENTS

I would like to dedicate this dissertation to my parents. I cannot thank them
enough for their love and support. My sister and brothers deserve my deepest thanks for
their love and support, too. They also took good care of my parents while I was studying
in the US I am also very grateful to my adviser, Professor Brian D. Silver, for his
guidance throughout my graduate years and his comments on the drafts of this
dissertation at various stages. His thoughtful advice always improved the quality of this
research project. Professors Dennis P. Patterson, Jeffrey M. Riedinger, and David W.
Rohde (in alphabetical order) also provided me with ideas and encouragement whenever I
came across problems in writing this dissertation.

My thanks also go to those who greatly encouraged and helped me to complete
my study in many ways: Sangwoo Han, Seungwon Kim, Nam-young Lee, Taesup Lee,
and Chan Wook Park (in alphabetical order). In addition, I should list all the names who
nurtured me during my life at Michigan State University. However, the list may exceed
the length of the “Literature Cited” section at the end of this dissertation. So, I just note
the four very special names, Beobmo Koo, Hoa Sun Lee, Yonghee Lee, and Chansu Park
(in alphabetical order). I know for certain that they will always be in my wonderful
memories at Michigan State University for the rest of my life. Finally, I praise God for
guiding me all the way to this moment and making everything possible, as I believe he
will continue to do.

When I started to think about my dissertation many years ago (alas!), I decided to

write about electoral politics in Asia including my mother country, South Korea. The

question struck me, “Do Asian people also vote according to their perceptions about
economic conditions under democratic governments, as the voters do in other
democracies?” At the beginning, I wanted to cover all the democratic countries in Asia
but I realized that South Korea and Taiwan were the countries that had survey data
available and comparable to each other. When I conducted the statistical analyses, I was
frustrated to find the surveys unsystematic and inconsistent, unlike US. National Election
Study surveys. Sometimes, the surveys do not use exactly the same survey questions
with the same codes. Other times, Taiwan’s surveys lack some survey questions
compatible to those of South Korea’s surveys or vice versa. Nonetheless, without the
survey data sets, it would have been impossible for me to even start this study. Whereas I
am solely responsible for the analyses and interpretation of the survey data sets, I wish to

extend my thanks to the researchers at the IKES and the NCU.

TABLE OF CONTENTS

LIST OF TABLES ...................................................................... .

KEY TO ABBREVIATIONS ......................................................

CHAPTER I. INTRODUCTION .............................................. .
1. Hypotheses of the Dissertation .................................. .
2. Electoral Studies in South Korea and Taiwan .............
3. Scope of the Dissertation .......................................... .
4. Outline of the Dissertation ..........................................

CHAPTER II. LITERATURE REVIEW AND MODEL BUILDING

1. Models of Voter Turnout ............................................
A. Economic Condition Model ...........................
B. Socioeconomic Model ....................................
C. Social Psychological Model ............................
D. Political Mobilization Model ..........................
E. Close Election Model .....................................

2. Models of Economic Voting .......................................
A. Economic Voting Model .................................
B. Sociotropic Voting and Pocketbook Voting
C. Retrospective Voting and Prospective Voting

CHAPTER III. VOTER TURNOUT IN SOUTH KOREA ........

1. Model Specification and Methodology ........................
2. Statistical Models ....................................................... .

3. Voter Turnout in South Korea .....................................

A. The National Assembly Elections ....................

The March 1992 National Assembly Election
The April 1996 National Assembly Election

B. The Presidential Elections .................................

The December 1992 Presidential Election
The December 1997 Presidential Election
4. Conclusions .................................................................

CHAPTER IV. VOTER TURNOUT 1N TAIWAN ......................
1. Model Specification and Methodology ...........................
2. Statistical Models .......................................................... .
3. Voter Turnout in Taiwan ................................................
A. The Legislative Yuan Elections ..........................

The December 1992 Legislative Yuan Election
The December 1995 Legislative Yuan Election

vi

viii

B. The Presidential Election ................................... 77

The March 1996 Presidential Election ........ 77

4. Conclusions ................................................................... 80
CHAPTER V. ECONOMIC VOTING IN SOUTH KOREA ........ 82
1. Model Specification and Methodology ......................... 82

2. Statistical Models .......................................................... 87

3. Economic Voting in South Korea .................................. 88

A. The National Assembly Elections ....................... 88

The March 1992 National Assembly Election 88

The April 1996 National Assembly Election 93

B. The Presidential Elections .................................. 96

The December 1992 Presidential Election 96
The December 1997 Presidential Election 103

4. Economic Voting and Regionalism ................................ 108
5. Conclusions .................................................................. 114
CHAPTER VI. ECONOMIC VOTING IN TAIWAN ................... 117
1. Model Specification and Methodology ........................... 117

2. Statistical Models ........................................................... .

120

3. Economic Voting in Taiwan ........................................... 122
A. The Legislative Yuan Elections .......................... 122
The December 1992 Legislative Yuan Election 122
The December 1995 Legislative Yuan Election 125
B. The Presidential Election .................................... 131
The March 1996 Presidential Election ......... 131
4. Economic Voting and Ethnic Voting ............................... 135
5. Conclusions ................................................................... 138
CHAPTER VII. CONCLUSIONS ................................................ 141
APPENDIX A. SOUTH KOREAN SURVEY DATA .................... 152
APPENDIX B. TAIWANESE SURVEY DATA .......................... 154
APPENDIX C. SURVEY QUESTIONS ........................................ 156

APPENDIX D. MULTINOMIAL LOGIT ESTIMATES FOR THE THREE-
PARTY MODELS: THE 1995 LEGISLATIVE YUAN ELECTION
(COEFFICIENTS FOR THE KMT’S NORMALIZED TO ZERO),
TAIWAN ............................................................................ 160

LITERATURE CITED ................................................................... 161

vii

LIST OF TABLES

Table 1-1. Dates of National Elections in South Korea since Political Liberalization
................................................................................ 14
................................................................................ 15

Table 3-1. Probit Estimates for Voter Turnout: The National Assembly Elections, South
Korea ............................................................................... 54

Table 3-2. Joint Test Results of the Probit Estimates: The National Assembly Elections.
South Korea ................................................................................ 56

Table 3-3. Probit Estimates for Voter Turnout: The Presidential Elections, South Korea
................................................................................ 59

Table 3-4. Joint Test Results of the Probit Estimates: The Presidential Elections, South
Korea ................................................................................ 61

Table 4-1. Probit Estimates for Voter Turnout: The Legislative Yuan Elections, Taiwan
............................................................................... 74

Table 4-2. Joint Test Results of the Probit Estimates: The Legislative Yuan Elections,
- Taiwan ................................................................................. 76

Table 4-3. Probit Estimates for Voter Turnout: The Presidential Election, Taiwan
................................................................................ 78

Table 4-4. Joint Test Results of the Probit Estimates: The Presidential Election, Taiwan
............................................................................... 79
................................................................................ 82

Table 5-2. Multinomial Logit Estimates for the Three-Party Models: The 1992 National

Assembly Election (Coefficients for DLP’s Normalized to Zero), South Korea
................................................................................ 9O

............................................................................... 93

viii

Table 5-4. Multinomial Logit Estimates for the Three-Party Models: The 1996 National
Assembly Election (Coefficients for NKP’s Normalized to Zero), South Korea
............................................................................... 94

Table 5-5. Multinomial Logit Estimates for the Three-Candidate Models: The 1992
Presidential Election (Coefficients for KYS’s Normalized to Zero), South Korea
............................................................................... 97

Table 5-6. Multinomial Logit Estimates for the Three-Candidate Models: The 1997
Presidential Election (Coefficients for LHC’s Normalized to Zero), South Korea

................................................................................ 104
Table 5-7. First Differences of Economic Voting, Party Identification, and
Regionalism in the National Assembly Elections, South Korea
............................................................................... 110
Table 5-8. First Differences of Economic Voting, Party Identification, and
Regionalism in the Presidential Elections, South Korea
................................................................................ 112
Table 6-1. Vote Distribution in the Taiwanese Elections .
............................................................................... 118

Table 6-2. Probit Estimates for the Two-Party Models: The 1992 Legislative Yuan
Election, Taiwan .................................................................... 124

Table 6-3. Joint Test Results of the Probit Estimates: The 1992 Legislative Yuan
Election, Taiwan .................................................................... 125

Table 6-4. Multinomial Logit Estimates for the Three-Party Race: The 1995
Legislative Yuan Election (Coefficients for the KMT’s Normalized to Zero),
Taiwan ................................................................................. 127

Table 6-5. Joint Test Results of the Multinomial Logit Estimates: The 1995 Legislative
Yuan and the 1996 Presidential Elections, Taiwan ........................ 128

Table 6-6. Multinomial Logit Estimates for the Three-Candidate Race: The 1996
Presidential Election (Coefficients for LTH’s Normalized to Zero), Taiwan
................................................................................ 133

Table 6-7. First Differences of Economic Voting, Party Identification, and Ethnic
Background in the Legislative Yuan Elections, Taiwan .......... 136

Table 6-8. First Differences of Economic Voting, Party Identification, and Ethnic
Background in the 1996 Presidential Election, Taiwan ........... 138

Table 7-1. Summary of Voter Turnout in South Korea and Taiwan 143

Table 7-2. Summary of Voter Choice in South Korea and Taiwan 147

KEY TO ABBREVIATIONS

CJY: Chung Ju Young

CPS: the Center for Political Studies

CSDP: the Chinese Social Democratic Party
DJP: the Democratic Justice Party

DLP: the Democratic Liberal Party

DP: the Democratic Party

DPP: the Democratic Progressive Party

GNP: the Grand National Party

IKES: the Institute for Korean Election Studies
IND: Independents

KDJ: Kim Dae Jung

KMT: the Kuomintang

KYS: Kim Young Sam

LHC: Lee Hoe Chang

LIJ: Lee In Je

LTH: Lee Teng-hui

LYK: Lin Yang-kang

NCNP: the National Congress for New Politics
NCU: the National Chengchi University
NDRP: the New Democratic Republican Party

NBS: the National Election Studies

xi

NKP: the New Korea Party

NP: the Chinese New Party

NPRP: the New Political Reform Party
PCJ: Park Chan Jong

PDP: the Peace and Democratic Party
PMM: Peng Ming-min

RDP: the Reunification Democratic Party
SRC: the Survey Research Center

ULD: the United Liberal Democrats

UPP: the United People’s Party

xii

CHAPTER I

INTRODUCTION

1. Hypotheses of the Dissertation

1 launch an investigation into the relationship between economics and elections in
South Korea and Taiwan in the 19905. Since the mid-19805, South Korea and Taiwan
have undergone a genuine transition to democracy. By democracy, I mean a political
system where rulers are chosen in free and fair elections, which are regularly held and
participated in by people with considerable political rights and civil liberties.1
Democracy rests on the premise that no citizens compete for power in anything other than
elections or are excluded from participating in the selection of leaders. In these
democratic societies, individuals are more likely to vote for the party that may maximize
their interests.

Since the democratic transition in the mid-19803, elections in the two countries
are generally considered free, fair, and competitive. Compared to previous elections
under authoritarianism, voters also exhibited different behavior. In South Korea, for
instance, a peaceful power transfer from the ruling party to an opposition party took place

for the first time in its history in 1997. Considering a long electoral history, the power

 

' This study is based on Schumpeter’s electoral definition of democracy, instead of Diamond’s liberal
definition of democracy. The bottom line is that there is no democracy without elections. Schumpeter’s
notion of electoral democracy, however, has been criticized because elections can be held with the
systematic abuses of political rights and civil liberties (Diamond 1996; Karl 1995). In contrast, the liberal
definition of democracy does not allow us to find many democracies in the real world. Thus, I emphasize
the new practice of “free, fair, and competitive” elections in defining a democratic society.

alternation was a miraculous event.2 This power transfer was not possible without
political actors’ willingness to abide by the democratic rules of the game. Additionally, it
was as late as 1988 when the Taiwanese began to vote for the candidates who they liked
to support in the context of competing parties. Prior to the 1987 lifting of martial law,
there had been no legal opposition parties other than the satellite parties of the ruling
Kuomintang (KMT) in the political arena.3 Voter decisions became even more
complicated when a dissident group of the KMT came out to build the Chinese New
Party (NP) in 1993. Since then, there were three major parties (the KMT, the DPP, and
the NP) in Taiwan’s electoral politics. Furthermore, Taiwan also embraced a power
transfer in the 2000 Presidential election.

In this research project, I attempt to answer the following questions: What are the
features of the new voting behavior in South Korea and Taiwan after the democratic
transition? As in other democracies, do voters in these two countries cast their ballots for
the party that promotes their economic interests? More specifically, do economic factors
have an impact on voters’ choices? If so, to what extent are voters affected by economic
conditions? If not, why? The effects of economic conditions on elections can be two-
fold. In other words, economic circumstances have impacts on both the decisions of

whether people vote and how they vote. Consequently, I examine how the state of the

 

2 The first general election in South Korea was held on May 10, 1948.

3 It was in the 1977 local election that the Tangwai (meaning outside the [KMT] party) became the main
opposition by winning 35% of the seats in the Provincial Assembly and 20% of the county magistrates.
The Tangwai movement was transformed into a political party (the Democratic Progressive Party: DPP) in
September 1986. Although martial law was lifted in July 1987, it was not until April of 1988 that the KMT
regime officially legalized the DPP. The satellite parties were, for example, the China Youth Party and the
Chinese Democratic Socialist Party in the 19705 and 19805 (Tien 1989, 85).

economy affects voter turnout as well as the fate of the incumbent party in South Korea
and Taiwan in the 19905.

First, my hypothesis on voter turnout follows Rosenstone’s theory (1982) that if
the state of economic affairs is worse than before, voter turnout is likely to be lower.
Conversely, if the state of the economy is better than before, voter turnout is likely to be
higher. Because people with economic adversity cannot afford time and effort to manage
anything other than their daily lives, they are less likely to go to the polling places.
Similarly, when economic difficulties prevail in a society, voter turnout is likely to drop.

Second, my hypothesis about voter choice is that if economic conditions are
worse than before, individuals are less likely to vote for the incumbent party.
Conversely, if economic conditions are better than before, individuals are more likely to
vote for the incumbent party. According to Mueller (1970, 34), when the economy is in
trouble, people are likely to vote for an opposition party in order to punish the incumbent
party; which is responsible for the bad economy (the responsibility hypothesis, see also
Lewis-Beck and Paldam 2000). Thus, it is more often than not observed that the
incumbent party tries to prevent an economic downturn or improve the state of the
economy before the election campaign starts. This will not jeopardize electoral victory of

the incumbent party.

2. Electoral Studies in South Korea and Taiwan

In South Korea, some previous studies have suggested that mobilization factors

have a significant impact on voter participation. According to the mobilization theory,

voter mobilization was more likely to occur among rural residents, older people, and less
educated people than their urban, young, and educated counterparts. Mo, Brady, and R0
(1991, 21-2) attempt to solve the puzzle that although urban voters were likely to have
higher education and more income than rural voters, urban residents voted less than rural
residents.4 The 1988 National Assembly election survey by the Korea Gallup revealed
that rural voters were more likely to be mobilized to the polling booths than their urban
counterparts. Also, Park (1993b, 7) finds that younger people and more educated people
were unlikely to vote in the 1992 National Assembly election. On the other hand,
according to Lee (1993, 44-5), interest in election, age, political efficacy, and occupation
were the significant variables in explaining voter turnout in the 1992 National Assembly
election.

With survey data from the 1997 Presidential election, Kim (1998, 210) resolves
the discrepancies of the socioeconomic effects on voter participation. He emphasizes the
different impacts of the socioeconomic and mobilization factors on different voters.
Among voluntary voters, individuals with higher education and income were more likely
to go to the polling booths than individuals with lower education and income. In
contrast, voter mobilization occurred more frequently among people with lower education
and income than among people with higher education and income. Because the number
of people with lower education and income exceeded the number of people with higher

education and income, the former marked a higher turnout rate than the latter.’

 

‘ These findings contradict the socioeconomic model of voter turnout that will be presented in Chapter 11.

’ There is no English-written research done on voter turnout in Taiwan. The literature reviewed in this
section covers all that l have located in English-language academic journals thus far. 1 also cite a couple of
South Korean articles published in Korean.

Regarding the determinants of South Korean voters’ choices, regionalism has
been identified as the most significant factor (Lee 1993; Morriss 1996; Park 1993a,
1993b). By regionalism, I mean that voters are likely to cast their ballots for the party
that is led by the leaders from the same home province as the voters.“ Political leaders
have built their parties based heavily on their home province and have personalized their
parties. In addition, because political leaders merged and split their party so often and
changed their party’s name so frequently, there may have not been enough time for voters
to develop genuine party identification, as the voters have developed in the U.S.’ Instead,
voters have developed a strong attachment to their regional leader.

After the democratic transition in South Korea, the ruling party could no longer
mobilize voters, as it had under the military-backed authoritarian regimes, although
regionalism is still reported to have played a pivotal role in the recent elections. While
Park (1993a, 453) reaffirms the significance of regionalism, he also finds the impact of
issue voting with a statistical analysis of survey data from the 1992 Presidential election.
Voters took into account their retrospective evaluations and the campaign issues of the
candidates. Based on the statistical analyses of survey data, however, Kim (1993, 117-
18) uncovers an insignificant relationship between economic conditions and electoral

returns in the 1992 Presidential election.

 

° Since the 19605, the former presidents (Park Jung Hee, Chun Doo Hwan, Roh Tae Woo, and Kim Young
Sam) were originally from Kyungsang province (southeast 'area) and they overdeveloped the economy in
Kyungsang province compared to the economy in Cholla province (southwest area). The current president
(Kim Dae Jung) is from Cholla province and the other political leader (Kim Jong Pil) is based on
Choongchung province (central area). They were elected president by voters from their home province,
with the exception of Kim Jong Pil. He has not been elected president yet. The regional conflicts occur
most seriously between Cholla and Kyungsang provinces.

7 Perhaps no other countries have experienced party mergers and splits more frequently than South Korea.
Between January 1963 and January 2000, 82 parties have been formed. This means that about 2.2 parties

While Lee (1998) observes the significance of regionalism again in an analysis of
survey data from the 1997 Presidential election, he sheds new light on economic voting in
South Korea. Voters cast their ballots with sociotropic considerations and voted in
accordance with both retrospective evaluations and prospective promises of the
presidential candidates. In addition, Kim (1998) reports that economic voting was
significant in the regions where regionalism was weak in the 1997 Presidential election.

In Taiwan, ethnic divisions have played a critical role in electoral politics, as
regionalism has done in South Korea. The Chinese Mainlanders (the KMT) moved to
Taiwan in 1949 when they were defeated by the communist party. The KMT built a new
capital in Taipei in 1949 and ruled in Taiwan until 1999. The term “Mainlanders” refers
to those who came to the island after 1949, while the term “Taiwanese” means those who
came before the Japanese colonization of the island in 1895. While the KMT is based on
the Mainlanders, the DPP’s support mainly comes from the native Taiwanese islanders.
Such ethnic divisions have encouraged individuals to vote on the basis of their origin and
ethnic voting has characterized Taiwanese voting behavior (Cheng 1995, 1-2). The
ethnic divisions are also tightly connected to the issues of national identity (or
unification). The KMT is viewed as pro-unification, whereas the DPP is pro-
independence. The NP shares the Mainlanders’ votes with the KMT and is considered to
be ultra pro-unification.

Taiwan’s citizens recently decided for whom to vote often with respect to their

issue preferences. This is partly because of the fact that as the old generation from the

 

have been built every year (The Korea Central Daily January 21, 2000). Yet, the parties are not totally
new. They simply change their names while excluding or including some politicians.

mainland diminishes, so does ethnic voting. With a survey data analysis of the 1993 local
elections, Hsieh and Niou (N .d., 23) observe that issues (the economy, national identity,
the environment, social stability, local public works, and anti-corruption) affected voter
choice. In the 1994 local elections, Hsieh, Lacy, and Niou (1996, 69) also find that “the
economy can be every bit as important as Taiwan independence and national security
issues.” Voters were affected by both retrospective and prospective evaluations.

Hsieh, Lacy, and Niou (N .d., 20) find that the economy had a significant impact
on the 1996 Presidential election with a survey data analysis. Voters were prospective
rather than retrospective. Hsieh and Niou (1996, 546-47) also find that while national
identity was still the most significant variable, constitutional reform, the candidate’s
honesty and integrity, and religion were other significant issues in the 1996 Presidential
election. Hwang, however, finds an insignificant relationship between economic
conditions and voting decisions in a Legislative Yuan election (Hsieh, Lacy, and Niou
N.d., 7).

This review suggests that there is an increase in scholarly attention to economic
voting in South Korea and Taiwan. Yet, economic voting research has presented
contradictory findings, which I attempt to resolve in this research project. The
confirmation of economic voting in South Korea and Taiwan would enhance the
explanatory power of the economic voting model in the worldwide setting. Many
scholars of voting behavior have observed that the state of the economy affects voter

turnout as well as the electoral fortunes of the incumbent party, although there are still

disagreements over some specific issues of economic voting behavior. The literature of
economic voting has dealt mostly with old industrial democracies, especially the US.8
Outside the US, there are many studies done in old democracies such as
Australia (Carlsen 2000; Schneider and Pommerehne 1980), Britain (Carlsen 2000;
Clarke, Ho, and Stewart 2000; Clarke and Whitely 1990; Frey and Schneider 1978;
Goodhart and Bhansali 1970; Hibbs and Vasilatos 1981; Nadeau, Niemi, and Amato
1996; Pissarides 1980; Price and Sanders 1993; Sanders 1991, 1995, 1996, 1999, 2000;
Whitely 1980), Canada (Alvarez, Nagler, and Willette 2000; Black 1978; Carlsen 2000;
Clarke and Stewart 1994; Happy 1992; Monroe and Erickson 1986), Denmark
(Nannestad and Paldam 1996, 1997a, 1997b, 2000; Paldam and Nannestad 2000), France
(F leury and Lewis-Beck 1993; Hibbs with the assistance of Vasilatos 1981; Jerome and
Jerome-Speziari 2000; Jerome and Lewis-Beck 1999; Lewis-Beck 1980, 1983, 1984,
1993, 1996, 1997, 2000; Lewis-Beck and Bellucci 1982; Lewis-Beck and Nadeau 2000;
Rosa 1980; see also Frey and Schneider 1975; Miller and Mackie 1973; Pissarides 1980),
Germany (Alford and Legge 1984; Feld and Kirchgassner 2000; Frey and Schneider
1980; Goergen and Norpoth 1991; Kirchgassner 1985, 1986, 1991; Rattinger 1991),
Israel (Guttman and Shachmurove 1990); Italy (Bellucci 1984, 1991; Lewis-Beck and
Bellucci 1982), Japan (Inoguchi 1980; Reed and Brunk 1984), Norway (Miller and
Listhaug 1985), Spain (Lancaster and Lewis-Beck 1986), Sweden (Douglas 1978; Jonung

1981; Jonung and Laidler 1988; Jonung and Wadensjo 1979; Lybeck 1985; Madsen

 

8 See Chapter II for the detailed literature review on US. economic voting.

1980), and Switzerland (Schneider, Pommerehne, and Frey 1981).9 Anderson (2000),
Anderson and Ward (1996), Chappell and Veiga (2000), Kiewiet (2000), Lewis-Beck
(1986, 1988a), and Whitten and Palmer (1999) conduct cross-national analyses and find
significant economic voting in the advanced economies of European countries, which
include most of the above nations.

The economic voting model has recently been extended to developing countries or
new democracies. Remmer (1991) asserts that economic crisis introduced electoral
turnover in Latin America during the 19805, analyzing 21 Presidential elections of 12
Latin American countries. Paldam (1987) also reviews the relationship between inflation
and political instability in eight Latin American countries between 1946 and 1983.
Seligson and Gomez (1989) observe significant economic voting in Costa Rica. Panzer
and Paredes (1991) analyze the Chilean referendum focusing on the effect of economic
issues on elections. Rius (1993) also probes the political economy of elections in
Uruguay. Stokes (1996) investigates the relationship between economic reform and
public opinion in Peru between 1990 and 1995. Dominguez and McCann (1996)
examine the impact of the economic health on voter intent in the 1988 and 1991 Mexican

elections.

 

° See Paldam (1981) for a good review of the economic voting literature in industrial countries. His
conclusion (1981, 194; emphasis in original) is that “The very existence of the VP-function [vote-popularity
function] should no longer be doubted The main exceptions to the responsibility pattern occur for weak
and unstable governments in multi-party systems, and when external events clearly underlie economic
problems.” Paldam’s first review of the economic voting literature reflects the instability of economic
voting across the world. On the other hand, Lewis-Beck and Bellucci (1982, 102-3) assert that economic
voting is evident “in the multiparty systems of France and Italy.” However, Whiteley (1980) finds no
economic voting in Britain and Madsen (1980) observes negligible economic effects on voting in Denmark
and Norway. Peretz (1981) and Rattinger (1991) find no economic voting in Germany. In contrast,
Paldam’s second literature review (Nannestad and Paldam [1994]) suggests positive results of economic
voting studies. See also Lewis-Beck and Paldam (2000). '

In addition to the Latin American cases, Pacek and Radcliff (1995) note that
economic adversity impeded electoral success of the incumbent parties in the develOping
countries such as Botswana, Costa Rica, India, Jamaica, Sri Lanka, Trinidad and Tobago,
Uruguay, and Venezuela. Moreover, Pacek (1994) moves on to Bulgaria, the former
Czeck and Slovak Federated Republic, and Poland. Fidrmuc (2000) also analyzes the
impact of economic reforms on the electoral support in post-communist Czeck Republic,
Slovakia, Hungary, and Poland in the 19905. From the same perspective, Przeworski
(1996) focuses on the Polish public support for economic reforms. In addition, Bell
(1997) observes that unemployment had an effect on the Polish elections between 1990
and 1995. On the other hand, Silder (1994) and Colton (1996) find economic voting in
Russia’s 1993 and 1995 elections. White, Rose, and McAllister (1997) also confirm
economic voting behavior in Russia. Lafay (1981) even observes the interactions
between the economy and politics in Eastern European countries under the old system.‘0
Wilkin, Haller, and Norpoth’s (1997) ambitious piece does not make a strong case for
economic voting in the worldwide setting. Nonetheless, the economic voting model has

gained supportive evidence from both old and new democracies.

3. Scope of the Dissertation

The scope of this study is to examine the political effects of economic conditions

on the national level elections in South Korea and Taiwan after the democratic transition.

 

'0 On the other hand, Harper (2000, 1191) argues that “country-specific studies of electoral behavior in
Eastern Europe that observe that the return to parliamentary power of ex-communist parties in these
societies was not simply a function of economic voting.”

10

In other words, I investigate economic voting at the level of the Presidential elections and
the parliamentary (South Korea’s National Assembly and Taiwan’s Legislative Yuan)
elections under democratic governments. A new democratic era in the two countries was
marked by a “founding election.” Founding election is defined as the race that occurs
“for the first time after an authoritarian regime, elected positions of national significance
are disputed under reasonably competitive conditions” (O’Donnell and Schmitter 1986,
57).

There are five conditions for a founding election. First, a founding election
should be held “after an authoritarian regime.” To be more precise, the timing should be
after political liberalization. Since political liberalization eloquently suggests that there is
a general “relaxation of official controls on basic civil rights” (Bratton 1997, 74), the
election held after political liberalization is quite different from the elections under
authoritarianism. For instance, opposition parties are legalized or opposition leaders are
allowed to run for offices. In South Korea, political liberalization occurred with the
“June 29 Declaration” in 1987, which embraced most of the democratic demands from
the political opposition including direct Presidential elections. On July 15, 1987, the
lifting of martial law introduced political liberalization into Taiwan.

Second, a founding election should be contested for “nationally significant
positions.” People should form a new government by choosing a President or a Prime
Minister through a new parliament. A founding election is not necessarily a Presidential
election. Since a new Prime Minister is chosen or a new government is formed after a
parliamentary election, the parliamentary election can be a founding election as well. In

contrast, local level elections and elections for an electoral college are not races for

11

“nationally significant positions.” Additionally, a plebiscite for a new constitution is not
a founding election for the same reason.

Third, a founding election should be a free, fair, and competitive race. A
founding election should be held under fairly democratic conditions. Parties and
candidates are not to be prohibited from running for elections. If severe election-related
violence or coercion occurs, it is difficult to call the election democratic. Political parties
should have substantial access to the media and the weight of each vote should be the
same. There should be no electoral fraud. In reality, however, vote buying or other
forms of electoral fraud may have occurred in founding elections. Nonetheless, what is
more important is that the electoral fraud and violence were tolerable compared with the
historical standards built in the past in each country. It is an undeniable trend that
election fraud has decreased lately. A5 a result, a founding election should reflect the will
of its citizens as accurately as possible.

Fourth, there should be no election boycott and election results should be accepted
by candidates and their followers. Opposition parties may boycott an election, due to the
ruling party’s manipulation of the electoral laws or schedules. The manipulative actions
by the ruling party undermine the very “democraticness” of the election. Also, as Bratton
and van de Walle (1997, 194) adequately assert, losers’ willingness to accept the electoral
outcomes signifies a consensus on the democratic rules of the game. This is one of the
most important conditions for democracy. Whoever loses should accept the results of the
people’s choice. As mentioned at the onset of this study, democracy rests on the premise
that no citizens compete for power in anything other than elections or by vetoing

democratically elected leaders.

12

Finally, a founding election does not necessarily require a shift in power. In a
founding election, the old authoritarian leader can be reelected. This indeed casts
shadows over the new political system after the transition. The new government may be
a “hybrid” democracy still filtered by traces of authoritarianism (Karl 1995). Thus, it
may be claimed that the nature of the new political system is “soft authoritarianism.”
However, if the democratic rules of the game have been kept during the election, the
results should be respected by the citizens.

South Korea’s founding election was the Presidential race on December 16, 1987.
This election was first held after political liberalization in June 1987. The 1987 election
was monitored by international watchdog groups and South Korea’s civil society, who
assessed it as free and fair. The election was generally considered clean, fair, and
competitive. Some opposition groups claimed the election to be fraudulent but the
opposition parties accepted the results. Therefore, the new President was inaugurated as
scheduled.II The South Korean opposition did not win the direct Presidential election
because of the split bid of the opposition leaders. A former general, Roh Tae Woo (the
ruling Democratic Justice Party: DJP) won the first-past-the-post election by receiving
36.6% of the popular vote, while opposition leaders, Kim Young Sam (the Reunification
Democratic Party: RDP) and Kim Dae Jung (the Peace and Democracy Party: PDP)

earned 28.0% and 27.1% of the popular vote, respectively. Table l-1 displays all

 

" For example, a group of students and citizens who monitored the vote counting process at the Kuro Ward
Office, Seoul, believed some absentee ballot boxes to be suspicious. They demanded a serious
investigation by protesting in the office building. However, riot police stormed the office building two
days after the election.

13

elections held in South Korea after political liberalization. There have been three direct

Presidential elections and four National Assembly elections.

Table 1-1. Dates of National Elections in South Korea since Political Liberalization

 

 

Presidential National Assembly
Elections Elections
December 16, 1987

 

April 26, 1988
March 24. 1992

 

 

December 18. 1992

 

April 11. 1996

 

GUI-tht-t

December 18. 1997
7 April 13, 2000
Note: Founding election is in bold. Underlined elections are under investigation.

 

 

 

 

 

 

In contrast, it is not easy to identify a founding election in Taiwan. The first
possibility for being a founding election is the 1989 Legislative Yuan election, the first
national election after political liberalization (see Table 1-2). Martial law was lifted two
years earlier (July 15, 1987) and opposition parties participated in the election. The
election was free and fair. The KMT won 70% of the contested seats and the DPP
became the major opposition party. However, the number of seats open for the election
was less than half of the total number of parliamentary seats. The Legislative Yuan
members elected in Taiwan was only 35.2% (101 out of 287) in 1989 (Domes 1999, 52).
The Mainlanders were over-represented in the Legislative Yuan.

The second possibility is the 1990 Presidential election. However, the election
was an indirect race by the KMT controlled-National Assembly (electoral college).

Consequently, the people’s choice could not be adequately reflected. The National

14

Assembly was formed in 1946 and only a small number of the National Assembly

members have been replaced several times afterwards.

Table 1-2. Dates of National Elections in Taiwan since Political Liberalization

 

Presidential National Assembly Legislative Yuan
Elections Elections Elections

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 December 2, 1989
2 March 21, 1990

3 December 21, 1991

4 December 19. 192;
5 December 2. 1995
6 _M_arch 23, 1996 March 23, 1996

7 December 5, 1998
8 March 18, 2000

 

Note: Founding election is in bold. Underlined elections are under investigation.

The third possibility is the 1991 National Assembly election. This was the first
full-scale election in Taiwan’s history. Many lifelong members were replaced by
Taiwanese residents. However, elections for the electoral college are not contests for
nationally significant positions. Additionally, the mission of the new National Assembly
was to reform the constitution.

The fourth possibility for a founding election is the 1992 Legislative Yuan
election. The election was not only a full-scale election, but also the first race for
nationally significant positions held under free and fair conditions. There was no dispute
over the election results. Thus, the 1992 Legislative Yuan election is defined as the
founding election in Taiwan (Chu 1996, 71; Tien and Chu 1996, 1141; Tien and Cheng
1997, 21). After the founding election, Taiwan has held two direct Presidential elections

and three Legislative Yuan elections.

15

 

Since a founding election may contain the direct impact of democratization, the
founding election itself cannot be used for the estimation of usual voting patterns (Pacek
and Radcliff 1995, 749). Also, a new democracy can revert to a non-democratic form of
government after the founding election. Yet, the two countries regularly held national
level elections after the founding election and the uninterrupted elections reflect
individuals’ ordinary voting behavior. This enables me to examine the normal voting
patterns in the newly democratized countries.

Unfortunately, the elections to be investigated in this research project are limited
by the data availability. Since aggregate data often commit the “ecological fallacy,”
survey data have been more commonly employed in electoral studies (Niemi and
Weisberg 1993a, 2). Individual voting behavior is best analyzed by individual level
survey data. The Institute for Korean Election Studies (IKES) has conducted nationwide
surveys since the 1992 National Assembly election. Thus, this study deals with the 1992
and 1996 National Assembly elections and the 1992 and 1997 Presidential elections. For
Taiwan, I use survey data collected by the National Chengchi University (NCU) in the
1992 and 1995 Legislative Yuan elections and the 1996 Presidential election (see
Appendices A and B for more about survey data).12 Because survey data are not available
at the time of this writing, the elections held in 1998 and 2000 are beyond the scope of

this research.

 

'2 The survey data sets to be used in this research do not include the survey for the founding election in
South Korea. In Taiwan, the democratic transition occurred in a very incremental way. So, this study is
not affected by the direct impact of democratization.

l6

4. Outline of the Dissertation

This dissertation consists of seven Chapters. Chapter I provides the introduction.
In Chapter II, I review the economic voting literature in the US. and other democratic
countries. The purpose of Chapter II is to examine the factors determining voter turnout
and to define the concepts of economic voting. Also, Chapter II leads to the development
of statistical models to be tested in the following Chapters, III through VI. In Chapter III,
I investigate the relationship between economic conditions and voter turnout in South
Korea. I also test the different models of voter turnout: a socioeconomic model, a social
psychological model, a political mobilization model, and a close election model.‘3 As a
consequence, I identify the factors that have had an effect on voter turnout. Chapter IV
examines the relationship between economic conditions and voter turnout in Taiwan. I
also make a comparison among the effects on voter turnout of an economic condition
model, a socioeconomic model, a social psychological model, and a political mobilization
model.

In Chapter V, I probe the effects of economic circumstances on voter choice in
South Korea in the 19905. Once economic voting is detected, 1 identify the nature of
economic voting. Are voters concerned about the national economic condition
(sociotropic voting) or their personal economic well-being (pocketbook voting)? Do

citizens vote according to retrospective evaluations (retrospective voting) or prospective

 

'3 In addition to these models, there would be an institutional model. The institutional model focuses on
electoral laws and institutional arrangements that determine voter turnout. However, there are not many
differences in the two countries’ institutional arrangements. Thus, 1 do not include the institutional model
in the list of the models under investigation. The only major difference is that the Single Nontransferable
Vote (SNTV) System is used for the Legislative Yuan elections.

17

expectations (prospective voting)? I also compare economic voting between the
Presidential races and the National Assembly elections. Additionally, I compare the
effects of economic voting on voter choice with those of regionalism. If there is no
economic voting. I explain why this is so. Chapter VI examines the nature of economic
voting in Taiwan in the 19905. Basically, the process of Chapter V1 is the same as in
Chapter V. Yet, Chapter VI makes a comparison between the effects of economic voting
and ethnic voting on voter choice. Chapter VII makes conclusions about the main
findings of the study and compares the nature and magnitude of economic voting in the

two countries.

18

CHAPTER II

LITERATURE REVIEW AND MODEL BUILDING

1. Models of Voter Turnout

A. Economic Condition Model

The economic condition model focuses on the economic situation that discourages
or encourages voter turnout. The first theory is that economic adversity discourages voter
turnout. According to Brody and Sniderrnan (1977, 344), people with economic
hardships are more likely to be preoccupied with their economic concerns than people
without economic grievances. The poor and the unemployed tend to spend most of their
time surviving. They are not expected to actively participate in political activities
including elections (Wolfinger and Rosenstone 1980, 25-9). With statistical tests at both
the individual and aggregate levels, Rosenstone (1982, 4]) concludes that “Turnout is
lower when short-term unemployment is high, prices are unstable, and a large proportion
of the population experience financial difficulties.”

In the comparative perspective, Radcliff (1992, 444; 1996, 718) contends that
turnout rates in industrial countries are affected by the state of the economy and that
economic adversity has different impacts on voter participation depending upon the
welfare level in a country. His multivariate regression reaffinns Rosenstone’s finding
that economic hardships in the US. discourage voter participation. In contrast, economic

adversity in developing countries encourages individuals to go to the polling booths.

l9

The second theory is that economic adversity encourages electoral participation.
This theory helps us to understand a positive relationship between economic hardship and
voter turnout in developing countries, which is suggested by Radcliff. As Lipset (1960,
192) has originally asserted, people under economic pressure are “expected to turn to
government action as a solution and to show a high voting average.” A surge of voter
turnout in the 1982 Congressional election in the US. amidst economic downturn may be
a case for this theory. Southwell (1988, 273) suggests that the increased voter
participation “among blacks, working class and the unemployed” attributed to the surge
of electoral participation in 1982.

This theory is partly supported by the “negative voting” theory. Campbell,
Converse, Miller, and Stokes (1960) contend that people are more sensitive to negative
information than positive information. Consequently, “As long as public affairs go well,
there is little to motivate the electorate to connect events of the wider environment with
the actors of politics [However] An economic or military or other forms of calamity
can force events across the threshold of political awareness” (Campbell et al. 1960, 556).
Therefore, individuals are more likely to go to the polling places in order to punish the
incumbent candidates for their mistakes than to reward the incumbent candidates for their
successes.

In explaining the decline in support for the incumbent party in midterm
Congressional elections, Kemell (1977, 44) also observes that voters’ motivation of
political punishment is greater than that of political reward. Moreover, Lau proposes
another negativity that affects voter decisions, utilizing data from the 1968, 1972, 1974,

1978, and 1980 Center for Political Science (CPS) National Election Studies (NES). The

20

motivational “cost orientation” negativity is that “people are more strongly motivated to
avoid costs than to approach gains” (Lau 1985, 132). Feldman (1982, 452) finds a
similar result with the 1972 CPS data. Furthermore, Fiorina and Shepsle (1989, 438)
assert that negative voting occurs among supporters with strong opinions about politics.
Incumbent candidates are more likely to have supporters with strong opinions about
politics than their challengers. Thus, negative voting occurs more frequently among in-
party incumbent seeking reelections.

The third theory is that economic adversity has nothing to do with voter
participation. Lane (1959, 330) proposes that voter turnout in Presidential elections is not
affected by economic downturns or uptums. Analyzing aggregate data from
Congressional elections between 1896 and 1970, Arcelus and Meltzer (1975, 1237-238)
also find an insignificant relationship between Congressional elections and the economic
variables of real compensation, unemployment, and inflation. In addition, using survey
data from the Survey Research Center (SRC) Election Studies between 1956 and 1974,
Fiorina (1978, 439) does not find “any systematic effect of economic conditions on

turnout.”

B. Socioeconomic Model”
The socioeconomic model emphasizes the role of social structures as determinants

of electoral participation. Verba and Nie (1972, 125) have suggested that those who have

 

" The socioeconomic (or sociological) model has mainly been developed by a Columbia University
research team (Lazarsfeld, Berelson, and Gaudet [1944]). They have provided an explanation for the 1940
election, “relating voters’ socioeconomic status (education, income, and class), religion, and place of
residence (urban or rural) to their vote” (N iemi and Weisberg 1993b, 8).

21

higher social and economic status are more likely to take part in politics than people with
lower social and economic status. In other words, people who have higher income,
education, and occupational status, living in urban areas tend to participate in elections
more than those with lower income, education, and occupational status, living in rural
areas. They have more resources to maintain their political activities than their
counterparts.

The educational factor, however, produces a puzzle. As Wolfinger and
Rosenstone (1980, 102) point out, “personal qualities that raise the probability of voting
are the skills that make learning about politics easier and more gratifying and reduce the
difficulties of voting.” Nonetheless, the increased level of educational attainment is not
accompanied by an increase in voter turnout (Brody 1978, 296). Since the educational
factor alone cannot explain electoral turnout, the “puzzle” may be solved by relating the
decline in electoral participation not to the educational factor but to other socioeconomic

'5 As far as the educational factor is concerned, higher educational attainment is

factors.
expected to produce a higher rate of voter turnout, other things being equal.

There is also a positive relationship between age and voter turnout (Wolfinger and
Rosenstone 1980, 102-3). Older people feel a sense of responsibility for their society
more than the younger generation. After a certain age, however, voter turnout is

depressed among older voters due to the cost and inaccessibility of the polling places.

Women are observed to participate in elections less than men (Fiorina 1978; Wolfinger

 

'5 For instance, Bennett (1990, 51) reports that “White grade schoolers outvoted those with some high
school elderly citizens, who vote at relatively high rates, are increasingly disproportionately represented
in the grade school category, and young Whites who are high school dropouts have very low turnout
records.”

22

and Rosenstone 1980). Since women may be less interested in elections than men,
female suffrage has had a significant negative impact on voter turnout rates in twenty

industrial democracies (Blais and Carty 1990, 176).

C. Social Psychological Model'6

The social psychological model is based on the premise that voters’ decisions of
who to vote for are largely affected by psychological factors. The social psychological
model emphasizes three attitudes: voters’ psychological orientation toward parties, issues,
and candidates. Among the three attitudes, the concept of party identification is
considered as the most important factor in accounting for the voting patterns in US.
elections, even though voting is affected by various factors through a “funnel of
causality” (Campbell et a1. 1960). Party identification refers to a psychological link,
“which can persist without legal recognition or evidence of formal membership and even
without a consistent record of party support” (Campbell et al. 1960, 121).

With respect to voter turnout, Campbell and his colleagues assert that a strong
psychological attachment to a party fosters electoral participation. Some of the important
conceptual devices are “political efficacy, sense of civic duty, strength of partisan
attachment, interest in politics, concern about the outcome of the election, and trust in
government” (Caldeira, Patterson, and Markko 1985, 497). From the political
psychological perspective, Abramson and Aldrich (1982, 502) suggest that the decline in

US. voter turnout is likely to result from the erosion of party loyalty and “external”

23

political efficacy.l7 In contrast, because “internal” political efficacy, sense of civic duty,
interest in politics, and concern about the outcome of elections did not dwindle during the
19605 and 19705, they cannot directly explain the turnout decline (Abramson 1983, 293-
94). Abramson, Aldrich, and Rohde (1995, 118) find that the combined attitudinal
variables of party loyalty and external political efficacy can explain 91% of the decline in
voter turnout between 1960 and 1992.

Furthermore, Aldrich (1993, 271-74) demonstrates the compatibility between the
two social psychological variables (party loyalty and political efficacy) and the rational

choice model.‘8 The basic function of voting in the rational choice model is

R=PB-C+D,

where R is net rewards to voting; P is the probability that citizens will affect the outcome;
B is the extent to which an individual feels that one candidate will benefit the voter more
than another; C is the cost of voting; and D is civic duty.

The B term is equivalent to political efficacy. The low efficacy refers to the
situation that the government in power is not willing or able to respond to voters’

demands and to solve various problems. The erosion of efficacy explicitly means that

 

'6 A group of the University of Michigan researchers (Campbell et a1. [1960]) was the pioneer of the social
psychological model. They find that “A person’s identification with a party became the core of the model.
It, in turn, affected the person’s attitude toward candidates and issues” (N iemi and Weisberg 1993b, 8).

'7 External efficacy refers to “views of system responsiveness” and internal efficacy means “an individual’s
belief about his or her own political abilities” (Timpone 1998, 148).

'8 During the 19705, a research group from Rochester University developed the rational choice model.
According to Niemi and Weisberg (1993b, 9), “voters decide whether or not to vote and which candidate
to vote for on some rational basis, usually on the basis of which action gives them greater expected
benefits.”

24

there is no big difference in the expected benefit from voting for a party versus other
parties. Pattie and Johnston (1998, 279) also provide a similar interpretation with the
British election data during the 19805 and 19905.

On the other hand, the D term has been extended to include the expression of
party affiliation by Riker and Ordeshook (1968) and Fiorina (1976) (Aldrich 1993).
Particularly, F iorina develops the concept of partisanship with the D term in the rational
choice model. The election specific variable employed in the voting formula that has
decreased with voter turnout is the question that asks voters how strongly they are
concerned about which party wins the election. This is claimed to be largely linked with

the decline in party attachment.

D. Political Mobilization Model

The political mobilization model mainly deals with the relationship between
campaign mobilization and voter participation. In research on gubernatorial elections,
Patterson and Caldeira find the influences of campaign activities and partisan
competitions to be good factors for electoral participation (Caldeira, Patterson, and
Markko 1985, 500). Analyzing the 1978 Congressional election, Caldeira and his
associates (1985, 507) also conclude that “accountings for participation in elections
which omit political mobilization are partial Active political mobilization can enlarge
and enrich participation in elections.”

In general, the total amount of campaign funds is a good indicator of the political
mobilization model. Campaign spending has a positive relationship with voter turnout.

Larger campaign funds enable candidates to buy advertisement spaces in newspapers and

25

time on TV and radio. Candidates can also employ more campaign staff with larger
contributions and campaign workers mobilize voters with repeated contacts.
Consequently, the candidates become more visible to voters during the campaign, which
may lead to higher rates of turnout.

In addition, Cohen, Krassa, and Hamman (1991, 176) assert that a president’s
campaign appearances helped a candidate to mobilize voters in Senate races between
1966 and 1986. Exploring various alternative explanations for the voter turnout increase
in the 1988 and 1992 Presidential elections with survey data and state level data, Knack
partly tests the political mobilization model. Among others, MTV’s “Rock the Vote”
registration campaign was anticipated to attract more electoral participation of the young
generation. Also, the introduction of a “motor voter” bill was expected to foster voter
participation. However, they collectively explain “only a fraction of 1 percentage point
of the turnout increase” (Knack 1996, 30). The abrupt turnout increase of six percentage
points in 1992 may be partially explained by the fact that “the low turnout of 1988 is at

least as great an anomaly as the high turnout of 1992” (Knack 1996, 17).

E. Close Election Model

The close election model theorizes the impacts of the closeness of races on voter
turnout. A close election enhances the political interest of voters and voters are more
likely to turn out at the polls than in a landslide campaign (Caldeira and Patterson 1982;
Patterson and Caldeira 1983). In a similar vein, Powell (1980, 19) asserts that when one
of the two US. parties dominates, political interests of citizens will decrease and voter

turnout is expected to decline. After reviewing the literature, Grofrnan (1993, 103)

26

asserts that the close election model does quite well in accounting for “change at the
margin.” Cox and Munger (1989, 217) also observe that a close election had an impact
on electoral participation at the levels of both the mass and elite in the 1982 House
election.‘9 Election closeness fosters campaign spending, which requires more campaign
contributions especially from the elite level and ultimately stimulates voter turnout (Cox
and Munger 1989, 226). Using data fiom the 1972 and 1976 NES Panel Study and Vote
Validation Survey in Presidential elections, Moon (1992, 136) also concludes that “Not
only can closeness of the election impact on turnout but the fact that the expected benefit
differential is important to some would-be voters suggests that the subjective probability
of affecting the election outcome is significantly greater than zero for at least some
voters.”

According to Shaffer (1981, 74), however, because voters’ perception of the
closeness is related to the individual election, the closeness of elections is not an
important determinant of voter turnout. Foster (1984) also finds the results of the close
election model limited, using a cross-sectional approach between 1968 and 1980. In the
comparative context, Powell (1993, 60) investigates the close election model in 12
industrial countries in the 19705 and finds no significant relationship between close
elections and voter turnout. Pattie and Johnston (1998, 278) draw a similar conclusion
from the British elections during the 19805 and 19905.

Thus, it is difficult to define the relationship between the closeness of elections

and voter turnout. The difficulty is partly embedded in the “paradox” of not voting as the

 

'9 See Cox and Munger (1989, 217-18) and Grofrnan (1993) for the literature review in the area of the
close election model.

27

rational choice model may suggest. The probability of turnout affecting the outcome (the
P term) lies at the center of the paradox. If a citizen feels an election is a landslide, the
citizen may not go to vote. If a citizen believes an election is close, the citizen is more
likely to go to the polling places. However, the problem arises when “very few contests
for public office are so close that they come down to a single vote. An individual citizen
i must usually expect that the probability of a preferred candidate winning is only
infinitesimally larger when he or she votes” (Uhlaner 1989, 394; emphasis in original).
According to Moon (1992, 125), one of the solutions to the paradox of not voting
is to show that the P term is neither normally zero nor relevant to the voting decision.
However, although there is no exact value of the P term, it appears to be close to zero
(Miller 1986, 7). Riker and Ordeshook (1968) attempt to reforrnulate the P term to take
into account the individual’s perceived closeness of elections. While Ferejohn and
Fiorina also admit that individuals have difficulties in evaluating the magnitude of the P
term, they suggest that voters are concerned about other situations: “My god, what if I
didn’t vote and my preferred candidate lost by one vote? I’d feel like killing myself”
(Ferejohn and Fiorina 1974, 535). The notion of minimax regret emphasizes the relative
magnitudes of the B term and the C term. Some voters perceive that the benefit of voting
is bigger than the cost of voting, thus voting is rational. Consequently, “minor changes in
the margin of either can affect turnout decisions, so get-out-the-vote drives can be

effective in increasing turnout” (N iemi and Weisberg 1993b, 17).

28

2. Models of Economic Voting

A. Economic Voting Model

At the heart of economic voting lies the premise that “as the economy worsens,
the government loses votes” (Lewis-Beck 1988a, 8). This is because “voters hold the
government responsible for economic events” (Lewis-Beck and Paldam 2000, 114).
Voters are responsive to the environments surrounding them and act upon their self-
interests. In the economic voting model, voters are assumed to be rational. A rational
actor is defined as an individual whose behavior is motivated by the actor’s self-interests
and conscious choices.20 Voters can compare one party with another and assess the
expected utility of a party’s winning elections.

According to Riker (1990), the rational actor assumption requires only that
individuals make a choice in order to maximize their expected utility with the available
information about circumstances and consequences. Although individuals are assumed to
be self-interested, “the content of the self-interest” is not specified in the rational choice
model.” Moreover, a rational actor sometimes makes mistakes in assessing and choosing
alternatives. It is also possible for a rational actor to take an action that makes it difficult
to accomplish the individual’s primary goals. In addition, a rational actor makes a choice

that would be different from what the actor would choose with more information.

 

2° Campbell and his colleagues (1960) have asserted that only 2-3% of the electorate think ideologically.
Individuals are also depicted to be unaware of government policies and the differences in party positions
(Campbell et al. 1960). This negative view of voters dominated in the 19505 and early 19605. Research
including The Changing American Voter (N ie, Verba, and Petrocik 1979) has documented that voters
became sophisticated during the 1970s. Having said that, V. 0. Key (1966) is the pioneer of the positive
view of voters. He assumes that individuals are not fools, so that they can vote with respect to their issue
preferences.

29

Therefore, a rational actor does not necessarily need an extensive set of
information, a solid system of preferences, or the sophisticated skills of calculation in
order to make the best choice among the available alternatives. Nor does the economic
voting model require voters to have the ability to conduct a sophisticated analysis of
economic situations or to be able to identify subtle differences in economic policies. In
an experiment, McKelvey and Ordeshook (1985, 55-8) demonstrate that although many
voters are not well informed about parties, candidates, or issues, they vote as if they had
full information. In addition, candidates do not fully know voters’ utility functions but
they act as if they knew everything.22

In fact, individuals are not willing to take pains to be completely informed about
anything. They acquire some information only if the benefits of the information are
greater than the costs of its acquisition (Ferejohn 1990, 12-4). Then, they will use the
information as a cue to make decisions in voting. From this perspective, food prices are
the best predictor of presidential approval ratings among other economic indicators (Beck
1991, 95). The ordinary citizens who go grocery shopping can easily get the information
about economic conditions. Virtually all individuals who have passed voting age have
access to such information. Even more, the costs of the informational acquisition and
process are quite low. Such information does not necessarily prevent individuals from
making reasoned decisions in voting (Aidt 2000; Lucas and Sargent 1981; Lupia 1994;

Lupia and McCubbins 1998).

 

2' A rational actor may take even altruistic or moral actions when the costs of such actions are small.

2’ Bartels (1996, 194; emphasis in original) does not agree that uninformed voters cast their ballots “as if
they were fully informed.” See also Lau and Redlawsk (1997) for their objection to the “fully informed”
voter assumption. ’

30

Sanders (2000) also argues that, in Britain, voters’ overall feelings of economic
uptums or downturns are considerably acute even with their partial knowledge about the
economy. Nonetheless, individuals’ general senses about economic health matter most in
making voter decisions. He documents that it is voters’ subjective perceptions about the
economy that matter at the polling booths, instead of their objective knowledge about the
indicators of inflation or unemployment. From a similar vein, Paldam and Nannestad
(2000, 388-89) report that Danish voters know more than expected and are more sensitive
to unemployment compared to any other economic indicators. In contrast, Danish voters
are ignorant of detailed information, for instance, on budget deficits but they make their
decisions in voting according to their economic evaluations.

It is Mueller who has set the direction for the study of economic voting in US.
Presidential elections. According to Mueller (1970, 34), “an economy in slump harms a
President’s popularity but an economy boom does not help his rating.” Mueller’s
dependent variable is the rates of presidential popularity by the Gallup Poll during the 24
years from President Truman to President Johnson. The economic slump variable is one
of his independent variables including “the length of time the incumbent has been in
office the influence on his rating of major international events and war” (Mueller
1970, 18). Mueller also views that presidential popularity generally erodes during the
term (“the coalition of minorities” [the length of time the incumbent has been in office]
variable).

Focusing on the general decline in presidential popularity, Stimson (1976)
develops the "expectation/disillusionment cycle” model. According to Stimson, new

Presidents take advantage of broad support from the citizens. The high approval rates

31

during the honeymoon period are based on expectations unaccompanied by the
Presidents’ performance and they gradually decrease as time progresses. Finally,
disillusioned people show some of their prior support at the end of the Presidents’ terms.
The parabolic pattern of presidential popularity during the term is also developed by
using the Gallup Poll data from President Truman to President Nixon (first term).

According to Stimson, however, other independent variables added by Mueller
(rally effects, economic slump, and war) into his expectation/disillusionment cycle model
do not significantly increase the multiple correlation. Thus, while Stimson puts much
significance on the general cycle of presidential popularity, he disproves the importance
of economic effects on presidential popularity. From the same vein, Campbell (1966, 61)
has asserted that the cycle of surge and decline accounts for the midterm loss of the
incumbent party’s seats in Congress. Erikson (1988, 1027) also views the midterm loss
as the cyclical punishment of the incumbent party (see also Alesina and Rosenthal 1989;
Born 1990, 642-43). In addition, Frey and Schneider (1978), Miller and Mackie (1973),
and Pissarides (1980) commonly support the electoral cycle model in Britain.

Norpoth takes issue with Stimson in explaining the erosion of presidential
popularity during the term. The inauguration effect boosts the presidential approval rate
by some 20 percentage points after the election, which is so high that the decline in the
approval rate is inevitable. Instead, Norpoth (1985, 184) observes that “Only if poor
economic performance, an unpopular war, or a scandal compound the erosion does
electoral defeat become likely.” By examining the effects of economic conditions on
presidential popularity between 1961 and 1980, Norpoth (1985, 182) finds significant

economic voting.

32

With survey data analyses, Fiorina (1981), Kiewiet (1983), and Lewis-Beck
(1988b) also observe significant economic voting in US. Presidential elections.
Especially, Lewis-Beck (1988a, 133-35) uncovers a clear association between voters’
choices and their perception of the economy in the 1984 Presidential election.
Respondents who believed the incumbent party was good at fighting unemployment and
inflation were 44% more likely to cast their ballots for President Reagan than those who
perceived the government was not good enough, regardless of their socioeconomic status
and party affiliation. In the 1984 Presidential election, Kiewiet and Rivers (1985) and
Lipset (1985) draw a similar conclusion with different data sets. They commonly detect a
positive relationship between an individual’s economic evaluation and voter decisions.

Lewis-Beck and Rice (1992, 47) develop a multivariate model including the
variables of percentage point change in real GNP over six months prior to Presidential
elections and approval rates for Presidents in July of election years. They find a
significant impact of the two variables on voters’ choices. Later, Lewis-Beck and Rice
incorporate the new variables of party strength and candidate appeal into the previous
model, so that they provide a more fully specified model for Presidential elections. With
the same set of data for 1948-88, they conclude again that all four variables have the
correct directions and have a statistically significant impact on voter decisions (Lewis-
Beck and Rice 1992, 52).

Concerning R2, Abramowitz’s (1988a) model stands out. Abramowitz utilizes the
three independent variables of the incumbent President’s popularity, economic condition,
and the length of time the incumbent has been in office. According to Abramowitz

(1988a, 844), “the longer a party has been in power, the more likely the public is to feel

33

that ‘it’s time for a change.”’ Abramowitz finds that all three variables are statistically
significant and display the expected directions, using data for the Presidential elections
from 1948 to 1984. His model has an adjusted R2 of .98. In addition, Fair (1978) and
Frey and Schneider (1980) have successfully demonstrated interaction between the
economy and the vote in US. Presidential elections.

At the level of US. Congressional elections, Kramer has inaugurated the main
line of research. He offers an explanation about the effect of short-term economic
fluctuations on Congressional elections between 1896 and 1964 with aggregate time
series analysis (excluding the controversial years of 1912, 1918, 1942, and 1944). His
independent variables include unemployment, income, inflation, incumbency, and
presidential coattails. Kramer (1971, 140-41) finds that “Economic fluctuations, in
particular, are important influences on congressional elections, with economic upturn
helping the congressional candidates of the incumbent party, and economic decline
benefiting the opposition.” Among others, real personal income appears to be the most
important economic variable. In contrast, there is no significant impact of the changes in
unemployment and inflation on Congressional elections (Kramer 1971, 141).

Kiewiet defines Kramer’s model as an incumbency-oriented hypothesis and
alternatively develops a policy-oriented hypothesis. In the policy-oriented hypothesis,
voter decisions are observed to be made with respect to economic problems, especially
the issues of unemployment and inflation (Kiewiet 1981, 448). He examines the
hypothesis with survey data from the CPS NBS between 1956 and 197 8. Kiewiet finds
important differences between unemployment and inflation. According to Kiewiet (1981,

458-59), “There were large differences between racial and occupational groups in

34

respondents’ propensity to name unemployment their most pressing personal problem.”
The same differences exist at the national level. Yet, these group differences do not exist
in the issue of inflation both at the personal level and the national level. In addition,
individuals who are concerned about unemployment tend to vote for the Democratic
Party, whereas voters who are concerned about inflation have the propensity to vote for
the Republican Party. However, the inflation factor attracts weaker support than the
unemployment factor does (Kiewiet 1981, 451-58).

On the other hand, Stigler (1973, 167) generally disproves the economic voting
theory with a statistical analysis of the Congressional elections between 1896 and 1970.
While Stigler reaffirms an insignificant relationship between the fluctuations in
unemployment and electoral support, he rebuts Kramer’s claim that real personal income
has a significant impact on electoral decisions. Arcelus and Meltzer (1975, 16) also
detect very little significance in the relationship between the short-term effects of
unemployment (and income) and Congressional elections. In response to Stigler’s
claims, Goodman and Kramer (1975, 1264) uncover that all three variables (real personal
income, inflation, and unemployment) are statistically significant in explaining
Congressional elections.

Tufte (1975) views that midterm Congressional elections are referenda on the
incumbent party’s performance. Voters are expected to respond to both economic and
political issues. The economic issue is operationalized by real disposable income per
capita. The impact of political issues is measured by the presidential approval rates
provided by the Gallup Poll just before elections. On the other hand, Marra, Ostrom, and

Simon (1990), Ostrom and Simon (1985, 1989), Ostrom and Smith (1992), Simon and

35

Ostrom (1988), and Nickelsburg and Norpoth (2000) investigate presidential popularity
with the variables of the economy and foreign policy. In addition, Atkeson and Partin
(1995, 104) confirm referendum voting but not economic voting in Congressional
elections.

Contrary to Tufte, however, Hibbs (1982, 410) reaffirms economic voting
behavior with a statistical test of midterm Congressional elections from 1946 to 1978.
Hibbing and Alford also observe that the fate of the incumbent party was most likely
affected by the fluctuations of the economy, utilizing the 1978 Congressional election
survey data. Furthermore, Hibbing and Alford (1981, 433-38) identify how voters link
economic fluctuations with candidates in different kinds of districts: “(1) an in-party
incumbent seeking reelection, (2) an out-party incumbent seeking reelection, or (3) no
incumbent seeking reelection.” They find that senior incumbents are more likely to be
related to the current economic conditions than junior incumbents in in-party incumbent
seeking reelections. In contrast, personal economic conditions have no significant impact
on electoral choices in the districts of out-party incumbent seeking reelections and no
incumbent seeking reelections.

Furthermore, Lewis-Beck and Rice revise Tufte’s referendum voting model.
They examine 18 Congressional elections from 1950 to 1984. According to Lewis-Beck
and Rice (1985), the net seat changes of the incumbent party in the House are affected by
economic grth rates, presidential approval rates, and the electoral calendar (midterm
election or not). Additionally, Campbell (1986), Jacobson and Kemell (1983), Lewis-
Beck and Rice (1992), and Oppenheimer, Stimson, and Waterman (1986) observe

significant economic voting in House elections.

36

Regarding the magnitudes of economic voting between Presidential elections and
Congressional elections in the US, Kramer (1971, 141) asserts Presidential elections to
be less responsive to economic conditions than Congressional elections. In contrast to
Kramer, many scholars of economic voting have reported that the electoral effects of
economic conditions are far greater in Presidential elections than Congressional elections
(see especially Feldman 1982, 454; Kiewiet 1981, 445; Lewis-Beck and Eulau 1985, 4).
Even more, Erikson (1990, 397-98) has found no significant impact of economic
conditions on midterm Congressional elections between 1946 and 1986, whereas he
uncovers a compelling relationship between economic conditions and Presidential
elections. The rationale behind this is that Presidents are believed to have more
responsibility for the national economy than the members of Congress. In addition,
Presidential races are more visible and competitive than Congressional elections.

A similar relationship is seen between Senate elections and House elections. In
other words, the electoral effects of economic conditions are stronger in Senate than
House elections (Abramowitz 1988b, 385; Kuklinski 1981, 445; Westlye 1984, 253-54).
From a slightly different angle, Atkeson and Partin (1995, 105) observe that Senate
elections were affected by the President’s popularity, whereas gubernatorial elections
were influenced by the status of the state’s economy, with data from the 1986 NES and
1990 Senate Election Study (see also Niemi, Stanley, and Vogel 1995).

In France between 1993 and 1995, voters punished or rewarded the Prime
Minister not the President, depending upon the state of the economy. It was the Prime
Minister who managed the national economy in France (Lewis-Beck 1997, 321). The

same was observed in France between 1960 and 1978 (Lewis-Beck 1980, 320).

37

According to Lewis-Beck and Nadeau (2000, 181), French economic voting “appears
strongest under unified government, in Presidential elections where an incumbent is
running for reelection.” In contrast, economic voting is weaker in legislative elections
and elections held under “cohabitation governments.” In Germany, Chancellors appear to
hold responsibility for economic conditions under the same logic (Alford and Legge
1984,1179)

Party systems also affect voter decisions. In the two party system, the
responsibility hypothesis nicely explains economic voting. In multiparty systems or
coalition governments, however, it is not always easy for voters to identify the sources of
responsibility for certain economic problems and punish them accordingly. Powell and
Whitten (1993), Whitten and Palmer (1999), and Kiewiet (2000) classify many coalition
governments depending upon the clarity of government responsibility and find a
connection between economic voting and the clarity of government. In addition,
Anderson (2000) observes that “when it is clear who the target is, when the target is
sizable, and when voters have only a limited number of viable alternatives to throw their
support to,” economic voting is enhanced in 13 European democracies. Carlsen (2000)
also finds a significant relationship between party ideology, economic policies, and
electoral support. In addition, rightist governments are more sensitive to changes in the
level of inflation, whereas leftist governments are more responsive to the issue of
unemployment. ’

On the other hand, following Mueller (1970), Bloom and Price (1975, 1244)
assert that “economic conditions have a strong asymmetric impact on the congressional

vote. Political parties are ‘punished’ by the voters for economic downturns but are not

38

‘rewarded’ accordingly for prosperity,” with an aggregate level data analysis between
1896 and 1970.23 In contrast, Kiewiet (1983, 49) contends that the asymmetry of
economic effects is not supported by the individual level CPS survey data. In the
comparative perspective, Lewis-Beck (1988a, 78) shows that the asymmetry hypothesis
does hold in Britain, France, Germany, Italy, and Spain. This is evident regardless of the
institutional and cultural differences in these countries. After reviewing the economic
voting literature in European countries, Paldam (1981, 194) reaches the same conclusion.
The responsibility hypothesis works asymmetrically for parties based on different
ideological backgrounds. For instance, there is a party that cares most about the inflation
policy. The “clientele hypothesis” suggests that voters support the party if the inflation
condition deteriorates. The “saliency hypothesis,” on the other hand, says that'voters
punish the incumbent party if the inflation condition gets worse. Carlsen (2000) observes
that the clientele hypothesis works better than the saliency hypothesis and rightist

governments are seen to be supportive of the clientele hypothesis.

B. Sociotropic Voting and Pocketbook Voting

One of the important issues of economic voting is whether voters are sociotropic
or pocketbook (or egocentric). Sociotropic voters are “influenced most of all by the
nation’s economic condition” (Kinder and Kiewiet 1981, 132; emphasis in original).
Sociotropic voters support the candidates who further national economic interest. In

contrast, pocketbook voters are “swayed most of all by the immediate and tangible

 

2’ The asymmetry of economic voting is closely related to negative voting introduced in the previous
Chapter.

39

circumstances of their private lives” (Kinder and Kiewiet 1981, 130). Thus, pocketbook
voters support the candidates who promote their personal economic well-being.
Nannestad and Paldam (1994) use the term of egotropic voters, which has the exact same
meaning as pocketbook voters. Both a sociotropic voter and a pocketbook voter are not
necessarily a sophisticated analyst of the economy.

According to Kinder and Kiewiet (1981, 132), the differences between sociotropic
and pocketbook voting are not the same as the differences between an altruistic and a
self-interested choice in voting. The motive behind both sociotropic and pocketbook
voting is self-interest. Sociotropic voters may emphasize concerns for national economic
conditions more than personal economic circumstances. Yet, they believe that the
national economic conditions are closely connected with their own economic interests.
The distinctions between sociotropic and pocketbook voting are not based on different
motivations but on different information. As Kinder and Kiewiet (1981, 132) note,
“Pocketbook voting reflects the circumstances and predicaments of personal economic
life; sociotropic voting reflects the circumstances and predicaments of national economic
life.”

With survey data from 1956 to 1976, Kinder and Kiewiet (1979, 504) observe that
the personal experiences of financial difficulties or unemployment did not affect
Congressional election outcomes. Instead, there was a statistically significant
relationship between sociotropic voting and Congressional elections (Kinder and Kiewiet
1981, 141). Their main point is “not that personal experiences are generally unimportant

but that they are, or seem to be, politically unimportant” (Kinder and Kiewiet 1979, 522).

40

According to them, such sociotropic voting was evident in the 1972 and 1976 Presidential
elections (Kinder and Kiewiet 1981, 148).

Alvarez and Nagler (1995, 728) also assert that sociotropic voting, instead of
pocketbook voting, had a statistically significant impact on voter decisions in the 1992
Presidential election. On the other hand, Kramer (1983, 106) asserts that there are no
“purely personal favors or benefits from office holders,” thus the measurements of and
the distinctions between pocketbook and sociotropic voting are artifactual. In sharp
contrast, Lewis-Beck (1985, 355) concludes that pocketbook voting was not artifactual
with the quasiexperimental research of the 1956-82 CPS SRC election surveys. In the
comparative context, Lewis-Beck (1988a; see also Lewis-Beck and Lockerbie 1989, 160)
finds no pocketbook voting in legislative elections in Britain, France, Germany, Italy, and
Spain.

F urtherrnore, Kinder, Adams, and Gronke (1989) enhance the sociotropic voting
model, with the CPS NES data in the 1984 Presidential election. In addition to
sociotropic and pocketbook voting, Kinder and his associates (1989) newly develop the
“group voting” hypothesis that individuals are influenced by the information on their
group’s economic conditions. Kinder and his colleagues (1989, 502) find that voters
make a good distinction among the three levels of economic concerns. Moreover, it is
their contention that sociotropic voting is the most important determinant (Kinder et a1.
1989, 512). On the other hand, with an analysis of pooled survey data from Presidential
elections, Markus (1988, 151-52) finds significant sociotropic and pocketbook voting
between 1956 and 1984. In addition, Abramowitz, Lanoue, and Ramesh find pocketbook

voting as a significant factor in the 1984 Presidential election. Abramowitz and his oo-

41

authors (1988, 860-61) claim that pocketbook voting prevails when the economy is bad,
especially when the media blames the government for economic downturns. In contrast,
when the economy is good, voters tend to take credit from economic uptums.

In the comparative setting, Lewis-Beck (1986, 342) finds weak pocketbook voting
in Britain, France, Germany, and Italy. Lancaster and Lewis-Beck (1986, 665) also
uncover weak pocketbook voting in Spain. The same is observed in Germany (Alford
and Legge 1984, 1179) and Japan (Reed and Brunk 1984, 62). In contrast, Nannestad

and Paldam (2000, 134) find significant pocketbook voting in Denmark.

C. Retrospective Voting and Prospective Voting

The other issue of economic voting hinges upon voters’ time frame. Do they vote
with regard to retrospective evaluations or prospective promises? Most of the above-
mentioned studies are based on retrospective economic voting. The retrospective voting
model is traced back to V. 0. Key. Since voters are “a rational god of vengeance and
reward,” voters are viewed to be involved in a punishment and reward system depending
upon the incumbent party’s economic performances (Key 1964, 567). The traditional
retrospective voting model is based on the assumption that voters are more concerned
about actual policy outcomes than policies themselves (Fiorina 1981, 8).

While V. O. Key does not view the past as a gauge for the future, Downs
emphasizes the expected utility of future policies (Chappell and Keech 1985, 11). The
Downsian theory of retrospective voting relies on a motivational ground. Retrospective
voting is a cost-saving cue for their decisions in voting. It may cost a lot less to acquire

information about past performances than future policies. In addition, past records are

42

more reliable to be evaluated than future promises. To Downs, “the citizen simply uses
the past as a guide to the policies the parties would implement in the future” (Fiorina
1981, 12). Thus, the Downsian retrospective voting theory is connected with prospective
voting. Fiorina (1981, 13) notes, “The Downsian citizen compares the challenger’s and
the incumbent’s platforms, interpreting the latter in light of the incumbent’s past
performance.”

It is Fiorina who takes issue with simple retrospective voting. Utilizing survey
data from the SRC Election Studies between 1956 and 1974, Fiorina (1978, 434) finds
that voter decisions were influenced by the health of the economy in Presidential
elections. In Congressional elections, however, the relationship was positive until 1960
but became negative afterwards (Fiorina 1978, 435). Therefore, Fiorina (1981) deve10p5
a mixed model of retrospective and prospective voting with data from the 1976
Presidential election and the 1974-76 Congressional elections.

Miller and Wattenberg also test the voter’s time perspective by using survey data
from Presidential elections between 1952 and 1980 and draw a similar conclusion. They
make a distinction among the four dimensions of retrospective performance, prospective
performance, personal retrospective evaluations, and prospective policy considerations.
As a result, Miller and Wattenberg (1985, 370) find that voters are not necessarily
retrospective or prospective. Voters are able to assess the incumbent’s past performances
and are also concerned about the future of policy promises.

Furthermore, Miller and Wattenberg (1985, 370) assert that respective emphasis
that was given by V. 0. Key, Downs, and Fiorina can provide better explanations under

different circumstances. V. O. Key’s stress on “retrospective performance” may account

43

well for the elections where the incumbent is running for reelection. In contrast,
Downsian retrospective voting based on the policies (the platform of the party) may
explain well how challengers interact with voters. Fiorina’s emphasis on “prospective
performance” may best fit elections where candidates are running in open seat races. To
summarize, “the public appears to act in a highly responsible fashion, basing their
evaluations on the most reliable and concrete information available” (Miller and
Wattenberg 1985, 367). On the other hand, Conover, Feldman, and Knight (1986) assert
that voters are more sensitive to unemployment than inflation issues in evaluating past
economic performance.

Compared to research done on retrospective voting, prospective voting has not
been studied as much. According to Chappell and Keech (1985, 11), Rabinowitz and
Zechman test the prospective voting theory in Presidential elections by employing actual
changes in income “during the year following the elections as an unbiased proxy for
expectations of future performance.” Rabinowitz and Zechman also argue that
“individuals apparently do have reasonably accurate expectations about short-term
fluctuations in the performance of the economy and they act upon these expectations”
(Chappell and Keech 1985, 11). Chappell (1983) and Chappell and Keech (1995) also
argue that voters are not naive but sophisticated and prospective.

Monroe and Levi (1983) also examine the prospective voting model in light of
presidential popularity. Survey data of leading businessmen and economists are used in
their model for the measurements of future economic growth and inflation. In addition,
Kuklinski (1981) observes that economic voting implies a relationship between future

financial conditions and voter choice. However, they find only weak prospective voting.

44

At the level of Congressional elections, Abramowitz ( 1985) contends that
prospective voting affected the 1974, 1978, and 1982 Congressional elections more than
retrospective voting did. Lockerbie presents a similar result, utilizing the NES survey
data from the 1956-88 Congressional elections. According to Lockerbie (1991, 256),
“Expectations of economic prosperity have a consistently stronger influence on voter
choice than do evaluations of past performance.”

Some survey data analyses indicate the prospectiveness of voters’ evaluations. By
employing data from the 1984 Surveys of Consumer Attitude, Lewis-Beck (1988b, 19)
concludes that “the prospective economic evaluations weigh at least as heavily as the
retrospective.” With an analysis of consumer survey data in 1960-91, Haller and
Norpoth’s (1994, 646) conclusion is that voters are “remarkable economic forecasters.”
However, Clarke and Stewart (1994, 1118) claim that the dominance of prospective
voting is problematic (see also Conover, Feldman, and Knight 1986). From a slightly
different perspective, Jacobson finds a consistent connection among national economic
conditions, candidate quality, and electoral outcomes. He suggests that “A party
expected to have a good year because of the economy fulfills expectations partly because
it recruits better challengers, which in itself produces more victories” (Jacobson 1991,
46).

Additionally, from a broader angle, MacKuen, Erikson, and Stimson (1992, 597)
define a “peasant” as a retrospective pocketbook voter and a “banker” as a prospective
sociotropic voter. Between 1954 and 1988, presidential approval ratings appeared to be a
function of voters’ collective evaluations of the prospective national economy. Thus,

MacKuen and his associates (1992, 606) conclude that “The electorate responds with the

45

sophistication of the banker.” From the extended data between 1954 and 1996, Erikson,
MacKuen, and Stimson (2000) reaffirm their banker theory. In contrast, Nickelsburg and
Norpoth (2000, 320) rebut the banker theory.

Using the 1984 Euro-Barometer public opinion surveys, Lewis-Beck (1988a, 64)
uncovers that “none is statistically more secure than the prospective item.” The
prospective voting variable is statistically significant in Britain, France, Germany, Italy,
and Spain. Later, Lewis-Beck and Lockerbie (1989) maintain the same conclusion as the
Euro-Barometer public opinion surveys conducted in later years. Canadian retrospective
voting is also observed by Alvarez, Nagler, and Bowler (2000). In some post-communist
countries, prospective expectations about economic reforms have a significant impact on

voter decisions (Fidrmuc 2000, 215).

46

CHAPTER III

VOTER TURNOUT IN SOUTH KOREA

1. Model Specification and Methodology

In this Chapter, I examine the effects of economic conditions on voter turnout in
South Korea. I also compare the impacts of economic conditions on voter turnout with
those of many other factors (socioeconomic, social psychological, political mobilization,
and anticipated closeness). In addition, I probe the effect of the unique factor
(regionalism) on voter turnout in South Korea. The dependent variable of the statistical
models to be tested is voter turnout. Voter turnout is coded 1 if a self-reported voter;
otherwise 0. Since the dependent variable is binary, I utilize probit. The coefficient
estimates are the probabilities of voting. The independent variables of the basic (or
restricted) model include the measurements of economic conditions, socioeconomic
factors, social psychological factors, political mobilization factors, and the close election
factor. The combined (or unrestricted) model includes the basic model, home province,
and individual election specific events. The combined model tests the effect of
regionalism on voter turnout. In order to compare the power of the combined model with
the basic model, I utilize likelihood ratio tests.

To test the economic condition hypothesis, I employ the independent variables of
national economic conditions, inflation, personal economic conditions, and
unemployment. The translated or recoded survey questions are in Appendix C. If a

respondent perceived the national economy as being worse than before, the respondent

47

was less likely to vote. The 1992 National Assembly election survey does not have the
national economic condition survey item. Instead, the national economic conditions in
1992 are indirectly measured by the dummy variable of inflation. The survey asked,
“What is the most important problem that the country is facing?” If a respondent
considered inflation to be the nation’s most important problem, the respondent was less
likely to vote. The inflation item is also available in the two Presidential election surveys
but not in the 1996 National Assembly election survey.

In 1997, since South Korea was placed under the most serious economic crisis in
its history, the survey did not ask the usual question, “How would you evaluate national
economic conditions during the past couple of years?” Instead, the survey used the
question, “Which political party did you think was mainly responsible for the current
economic crisis [the International Monetary Fund (IMF) bailout]?”2" In this study, I
utilize this survey question for the national economic condition variable. If a respondent
answered that the ruling party was responsible for the financial crisis, the respondent
correctly understood the crisis was less likely to vote. In addition, I utilize the personal
economic condition and unemployment variables in all the elections. If a respondent
answered that personal economic condition was worse than before or that the individual
was unemployed, the respondent was less likely to vote.

For the socioeconomic hypothesis, I utilize the variables of residential area, age,

gender, and educational attainment. Individuals living in urban areas and with higher

 

2" Unfortunately, there is no sociotropic voting item that directly asks the national economic situations in
1997. Thus, a researcher (Lee 1998) also employs this survey question for his economic voting model in
South Korea. 4.3% of the respondents believed the incumbent party to be responsible for the financial
crisis, while 2.8% (1.2%) of the respondents thought the first (second) opposition party to be responsible
for the crisis. 41.8% of the respondents answered “don’t know.”

48

education are expected to vote more than individuals living in rural areas and with lower
education. In this research project. I recode the age variable as 1 = 20-29, 2 = 30-39, 3 =
40-49, and 4 = 50 and older.25 Older people are more likely to vote than the younger
generation. In addition, I use the dummy variable of gender (male) in order to capture a
positive relationship between men and voter turnout, with women being the excluded
category.

For the measurement of the social psychological factors, I employ only party
identifier and election interest. A party identifier is indicated by a self-reported supporter
for a party. South Korea’s surveys have a straightforward question asking voters’
election interest (How interested were you in the election?) The two social
psychological variables are expected to have a positive impact on voter turnout. South
Korea’s surveys often lack survey questions for political efficacy or political trust.

For political mobilization, I use the frequencies of participating in campaign
rallies (or watching TV advertisements) and campaign contacts. TV campaigns were
legalized in the Presidential elections since 1992, thus the TV advertisement variable is
used for the Presidential elections, instead of the frequencies of participating in campaign
rallies. The more frequently the political mobilization occurred, the higher the expected

voter turnout.

 

2’ As I noted in Chapter II, voter turnout may be depressed afier a certain age due to the cost and
inaccessibility of the polling places. Nonetheless, the category of 4 includes the wide age range of voters
in their 505, 605, 705, 805, and over. Thus, there is no dramatic decrease in voter turnout across the
categories. In addition, this kind of code is more appropriate for the voter choice hypothesis in the
following Chapters V and VI (Alvarez and Nagler 1995).

49

In addition, voters’ anticipated election closeness was asked only in South
Korea’s Presidential election surveys. As an individual perceives the results of the
election as being close, the voter is more likely to vote.

In order to evaluate the impact of South Korea’s regionalism on voter turnout, I
first employ respondents’ home province as dummy variables (Cholla province and
Kyungsang province), leaving the other provinces as the omitted category. Regionalism
is the most serious between these two provinces. In the 1997 Presidential election, for
instance, Kim Dae Jung from South Cholla province received 94.6% of the votes from
South Cholla province and 92.3% from the North Cholla province, while he garnered
11.0% of votes from South Kyungsang province and 13.7% from North Kyungsang
province. His rival Lee Hoe Chang based on Kyungsang province picked up 55.1% of
votes from South Kyungsang province and 61.9% from North Kyungsang province,
whereas he collected 3.2% of votes from South Cholla province and 4.5% from North
Kyungsang province. Regionalism is expected to have a positive impact on voter
participation. If regionalism is strong, individuals are more likely to go to the polling
places to cast their votes for their regional party.

South Korea’s regionalism is also measured by the individual election specific
event of a party merger (or coalition).26 The mergers were made by political leaders’
calculations mainly based on regional differences and drew support from their followers,
by stimulating regional sentiments. Thus, a merger was preferred by certain people and
they were more likely to vote. So, dummy variables (coalition supporters and coalition

non-supporters) are utilized in the statistical models, leaving “don’t know” as the base

50

category. The 1996 National Assembly election survey did not ask about the preference
over a party merger.

It is immediately noticed that the above mentioned hypotheses are involved with
more than one parameter, with the exceptions of the close election hypothesis. Thus, I
need to conduct joint hypothesis tests to see whether the group of variables jointly has a
significant effect on voter turnout. If the group of variables turns out to be jointly
significant, I conclude that the hypothesis is significant at the 95% (or 90%) level. Even
though a single variable of a hypothesis appears to be significant, the joint hypothesis test
may disprove the hypothesis because of the other insignificant variables. On the other
hand, even if joint test confirms the significance of the hypothesis, it is still possible that
every single variable of the group is not significant (Wooldridge 2000, 144-45).
Therefore, joint tests are required to confirm the significance of the hypothesis under

investigation.
2. Statistical Models

A basic voter turnout function and a combined voter turnout function are,

respectively, as follows:

P (VT0=1\x) =<r> (E, 5013, PSY, and M [and C]) and (1)

P (VT,=l\x) =<r> (E, SOE, PSY, M [and C], HP, and IE), (2)

 

2" The dynamics and backgrounds of party merger will be explained in detail in Chapter V.

51

where VT0 and VTl are voter turnout; E is economic conditions; SOE is socioeconomic
factors; PSY is social psychological factors; M is political mobilization factors; C is
anticipated election closeness in the Presidential elections; HP is home province; and IE
is the individual election specific event.

The basic model is transformed as follows:

P (VT0=1\X) =(D (30+ Brxr + pzxz + I33X3 + B4X4 + psxs + fisxo + B7X7 + Bsxs

+ p9x9+BlOX10+ I311)(II+DIZ)(12+ B|3Xl3+ Ui)’ (3)

where VT0 = voter turnout; [30 = intercept; XI = national economic conditions (N/A in the
1992 National Assembly election); X2 = inflation (N/A in the 1996 National Assembly
election); X3 = personal economic conditions; X4 = unemployment; X5 = residential area;
X6 = age; X7 = gender (male); X8 = educational attainment; X9 = party identifier; Xlo =
electidn interest; Xll = participating in campaign rallies (or watching TV advertisements
in the Presidential elections); Xl2 = campaign contact; X” = anticipated closeness (in the
Presidential elections); and Ui = error term.

The combined model estimates the impact of regionalism on South Korea’s voter

turnout.

P (VT1=1\X) =<D(BIOO+ [5101x101 + p102x102+ 5103X103+ [3104x104+ BIOSXIOS
+[3106)(1064’.[3107)(107-II-BlO8XIO8+I3109X109+[3110)(110fli-I3lll)(lll+pllZ)(112

+ Bll3xll3+pll4xll4+ BIISX115+I3116X116+ B117X117+ Ur), (4)

52

where VT, = voter turnout; [3,00 = intercept; X,,,, = national economic conditions (N/A in

the 1992 National Assembly election); Xm = inflation (N/A in the 1996 National

Assembly election); X,03 = personal economic conditions; X,“ = unemployment; X,,,_S
residential area; X,06 = age; X,,,, = gender (male); X“,8 = educational attainment; X,09 =
party identifier; X,,0 = election interest; X,,, = participating in campaign rallies (or
watching TV advertisements in the Presidential elections); X, ,2 = campaign contact; X, ,3 =
anticipated closeness (in the Presidential elections); X, ,4 = home province (Cholla); X, ,5 =
home province (Kyungsang); X,,6 = coalition (Support); X,,7 = coalition (Not Support);

and U, = error term.

3. Voter Turnout in South Korea

A. The National Assembly Elections

The estimates of probit models for the National Assembly elections (the
Presidential races) are presented in Table 3-1 (Table 3-3). The left two columns next to
the independent variable column are for the 1992 National Assembly election (the 1992
Presidential race) and the right two columns are for the 1996 National Assembly election
(the 1997 Presidential race). The left column of each election displays the parameter
estimates of the basic model. The right column of each election has the coefficient

estimates of the combined model. Standard errors are given in parentheses.

53

Table 3-1. Probit Estimates for Voter Turnout: The National Assembly Elections, South

 

 

Korea
Independent 1992 1 992 l 996 1 996
Variables Basic Combined Basic Combined
Economics
National N/A N/A -.02 -.02
economy (.08) (.08)
Inflation -.16 -.17 N/A N/A
(. l 1) (.1 1)
Personal .12 .10 -.03 -.02
economy (.08) (.08) (.08) (.08)
Unemployment N/Aa N/A"l . 16 .1 5
(.22) ( .22)
Socioeconomic
Residential -.19* -.18* -.16* -.l7*
area (.08) (.08) (.08) (.08)
Age .28* .28* .28* .28*
(.06) (.06) (.05) (.05)
Gender -.09 -.07 .06 .05
(Male) (.11) (.12) (.10) (.10)
Educational .03 .05 -.09* * -.09* *
attainment (.07) (.07) (.05) (.05)
Psychological
Party .02 .04 -.04 -.04
identifier (.11) (.11) (.10) (.10)
Election .62* .63 * .64* .64*
interest (.07) (.07) (.06) (.06)
Mobilization
Campaign .45 * .46* .26* * .24* *
rallies (.16) (.16) (.14) (.14)
Campaign .07 .09 .07 .03
contact (.12) (.12) (.10) (.10)
Regionalism
Home province .17 .04
(Cholla) (.15) (.13)
Home province . 12 .21 * *
(Kyungsang) (.13) (.11)
Coalrtron .00 N/A N/A
(Support) (.19)
Coalition -. 1 7 N/A N/A
(Not Support) (.14)
Intercept -1.68* -1.81* -l.10* -1.09*
(.46) (.48) (.40) (.40)
Log Likelihood -348.19 -343 .59 -463 .06 -461.23
Prob > chi2 0.00 0.00 0.00 0.00
Pseudo R2 0.17 0.18 0.20 0.20
Number of obs. 991 977 1 197 1 197

 

Note: * (**) indicates an estimate significant at the p < .05 level (p < .10 level).
a unemployment is dropped out. It predicts self-reported voter perfectly.

54

The March 1992 National Assembly Election. As the estimates of the basic
model suggest, economic conditions did not affect voter turnout (see Table 3-1). Of the
four socioeconomic factors, only residential area and age affected the decisions of
whether an individual voted or not. Contrary to the hypothesis, rural voters were more
likely to vote, ceteris paribus. Yet, age had the anticipated impact on voter turnout in that
older people appear to have been more likely to go to the polling booths. Also, election
interest, a social psychological factor, affected voter turnout. The more interested in the
election an individual was, the more likely it was that the person went to the polling
places. In addition, voters’ mobilization into campaign rallies fostered voter turnout, as
expected.

The coefficient estimates of the combined model illustrate almost the same
estimations as in the basic model. Rural residents and older people were more likely to
vote, controlling all other factors. In addition, election interest and campaign rallies were
positively related with voter turnout. However, the combined model indicates that the
province and party merger variables did not have significant influences on voter turnout,
ceteris paribus.

Likelihood ratio (Pr [R _>_ 9.2, df = 4]) tests reject, at a .1 level of significance, the
null hypothesis that there is no difference between the basic model and the combined
model. Instead, the tests approve the alternative rival hypothesis that the combined
model is significantly different from the basic model. In other words, the addition of the
regionalism variables into the basic model generates more power in explaining voter

turnout in 1992. Yet, the better model (the combined model) suggests that the economic

55

condition hypothesis and the regionalism hypotheses, respectively, are not jointly
significant (see Table 3-2). On the other hand, the results of joint tests confirm that all

the hypotheses in the basic model are jointly significant.27

Table 3-2. Joint Test Results of the Probit Estimates: The National Assembly Elections,
South Korea

 

 

Hypothesis 1992 1 992 1 996 1996
Basic Combined Basic Combined
Economic .098“ * .1 1 .83 .87
conditions
Socioeconomic .00* .00* .00* .00*
Social .00* .00* .00* .00*
psychological
Political .02* .01 * .13 .21
mobilization
Regionalism .45 .16

 

Note: Prob > chi2
* (**) indicates an estimate significant at the p < .05 level (p < .10 level).

. The April 1996 National Assembly Election. The estimates of the basic model
for the 1996 election reveal that none of the economic condition factors reached statistical
significance (see Table 3-1). Among the socioeconomic factors, residential area, age, and
educational attainment appear to be significant but only the sign of age is in the
hypothesized direction. Older people were more likely to vote. In contrast, urban
residents and the educated were less likely to vote. Nonetheless, the signs of the

residential area and educational coefficients are consistent with the conclusions of prior

research (Kim and Park 1991; Mo, Brady, and R0 1991; Park 1993b). A social

 

27 Mostly, the economic condition hypothesis and the regionalism hypothesis (or the ethnic voting

hypothesis) are not jointly significant, with respect to voter turnout. Moreover, the variables of these

56

psychological factor, election interest appears to have the anticipated effect on voter
turnout. Also, the participation in campaign rallies was positively correlated with voter
turnout, which suggests voter mobilization occurred in 1996.

The coefficients of the combined model are very similar to those of the basic
model. The residential area, age, educational attainment, election interest, and campaign
rally variables had a significant impact on voter participation. They are either in the
expected directions (the age, election interest, and campaign rally variables) or correctly
reflect the characteristics of South Korea’s electoral politics (the residential area and
educational attainment variables). In addition, home province played an important role in
the decisions of whether to vote or not. Kyungsang people were more likely to go to the
polling booths.

Likelihood ratio (Pr [R _>. 3.66, df = 2]) tests approve the null hypothesis that there
is no difference between the basic model and the combined model. On the other hand,
joint tests confirm the joint significance of the socioeconomic hypothesis and the social
psychological hypothesis, respectively (see Table 3-2). Joint tests disprove the joint
significance of the economic condition hypothesis and the political mobilization
hypothesis, respectively. These are true both in the basic and combined models. Also,

the regionalism hypothesis in the combined model appears not to be jointly significant.

 

hypotheses are not, frequently, significant. Thus, it is not meaningful to calculate “first differences,” as in
Chapters V and VI.

57

B. Presidential Elections

The December 1992 Presidential Election. The estimates of the basic model for
the 1992 Presidential election suggest that economic conditions did not have significant
influences on voter turnout (see Table 3-3). Among the four socioeconomic factors, only
residential area and age made a significant difference in deciding whether to vote or not.
Rural residents and older voters tended to vote, all other things being equal. Election
interest was the only social psychological factor that had a significant impact on voter
turnout. Election interest was positively correlated with electoral participation. In
contrast, the political mobilization factors did not reach statistical significance. The
anticipated closeness of the election also affected voter turnout. However, the election
was far from a close race. The election ended up with a comfort margin of approximately

eight percentage points between the first two contenders.

58

Table 3-3. Probit Estimates for Voter Turnout: Presidential Elections, South Korea

 

 

Independent 1992 1992 1997 1997
Variables Basic Combined Basic Combined
Economics
National .13 .13 -.03 -.02
economy (.11) (.11) (.14) (.15)
Inflation N/A3 N/A21 .01 .02
(.14) (.15)
Personal -.03 -.04 .08 .13
economy (.10) (.10) (.13) (.13)
Unemployment .23 .26 -.32 -.32
(.22) (.22) (.33) (.35)
Socioeconomic
Residential -. 14’” -.1 1 -.02 -.06
area (.08) (.08) (.10) (.10)
Age .18* .18* .25* .25*
(.07) (.07) (.07) (.08)
Gender .04 .01 .09 .12
(Male) (.13) (.13) (.14) (.14)
Educational .03 .02 -.04 .01
attainment (.08) (.09) (.10) (.10)
Psychological
Party -.08 -.15 .24M .30*
identifier (.14) (.15) (.14) (.15)
Election .60* .59* .64* .70*
interest (.08) (.08) (.08) (.09)
Mobilization
TV .10 .08 .37 .35
advertisements (.14) (.14) (.33) (.34)
Campaign .17 .15 -.43* -.40*
contact (.18) (.18) (.16) (.17)
Closeness
Anticipated .15** .16** .12 .13
closeness (.09) (.09) (.1 1) (. 12)
Regionalism
Home province .46* .25
(Cholla) (.21) ( .22)
Home province -.03 .33"
(Kyungsang) (.14) (.17)
Coalition .31 * * -.28
(Support) (.17) (.21)
Coalition .20 -.26
(Not Support) (.16) (.17)
Intercept -1.23** -1.29** -1.95* -2.35*
(.65) (.67) (.86) (.88)
Log Likelihood -243.42 -238.89 -209.95 -201.73
Prob > chi2 0.00 0.00 0.00 0.00
Pseudo R2 0.14 0.15 0.21 0.23
Number of obs. 1177 1175 1051 1044

 

Note: * (**) indicates an estimate significant at the p < .05 level (p < .10 level).
a inflation is automatically dropped out. It predicts self-reported voter perfectly.

59

The story does not differ much in the combined model. No economic condition
factors played a significant role in the citizens’ decisions of whether to vote or not. Age
was the only socioeconomic factor that reached statistical significance for voter turnout.
As expected, older people were more likely to vote. Election interest, among the social
psychological factors, did matter in deciding whether to go to polling booths or not.
Election interest was positively connected with voter turnout, controlling all other things.
Also, the close election variable had a positive impact on voter participation.

Moreover, a province factor was significant for voter turnout. Cholla people
tended to go to the voting places. The individual election specific event of the three party
merger also affected the decisions of whether to vote or not. Supporters for the three
party merger were more likely to vote. It is also worth noting that preferences over the
party merger may have been affected by respondents’ home provinces, particularly
Cholla or Kyungsang province. However, the preference was examined among voters
from other regions of the country, thus there are no serious multicollinearity problems
between the two variables in the model.

Likelihood ratio (Pr [R 2 9.06, df = 4]) tests approve, at a .1 level of significance,
the alternative rival hypothesis that there is a significant difference between the basic
model and the combined model. In addition, the more powerful combined model
suggests that the economic condition hypothesis and the political mobilization
hypothesis, respectively, are not jointly significant. In contrast, each of the
socioeconomic hypothesis, the social psychological hypothesis, the close election

hypothesis, and the regionalism hypothesis appears to be jointly significant (see Table 3-

60

4). The socioeconomic hypothesis, the social psychological hypothesis, and the close

election hypothesis, respectively, are also jointly significant in the basic model.

Table 3-4. Joint Test Results of the Probit Estimates: Presidential Elections, South Korea

 

 

Hypothesis 1992 1992 1997 1 997
Basic Combined Basic Combined
Economic .47 .44 .84 .76
conditions
Socioeconomic .02* .04* .00* .01*
Social .00* .00* .00* .00*
psychological
Political .50 .62 .02* .04*
mobilization
Close Election .09** .09** .29 .25
Regionalism .08" .11

 

Note: Prob > chi2
* (**) indicates an estimate significant at the p < .05 level (p < .10 level).

The December 1997 Presidential Election. On the decisions of whether to vote
in the 1997 Presidential election, economic conditions in the basic model did not generate
any significant impacts (see Table 3-3). Also, age was the only socioeconomic factor that
reached statistical significance for voter participation. Older people were more likely to
vote, as hypothesized. Party identifier and election interest, the two social psychological
variables, had significant influences on voter turnout. As expected, the two variables
were positively connected to voter turnout. Contrary to the hypothesis and the reality, the
anticipated closeness of the election did not produce any difference'in voter participation.
This is interesting because the election was a neck-and-neck race. The first two

candidates ended up with a 1.6 percentage point difference.

61

Among the mobilization factors, the campaign contact variable reached statistical
significance. However, the sign of the coefficient displays the direction opposite to the
hypothesis, which suggests the negative relationship between campaign contact and voter
turnout. This is partly because of the widespread disrespect and disbelief towards politics
and politicians in South Korea. Presidents were omnipotent but had spent more time in
political maneuvering than political reform and economic development even after the
democratic transition. Their families and senior staff members were frequently linked to
various corruption cases and scandals. Party leaders fought each other over personal
causes and this sometimes paralyzed the democratic process of political bargaining or
negotiation. The dominance of the presidency in South Korean politics often jeopardized
the autonomy and stability of the National Assembly. The members of the National
Assembly also used up their effort for political maneuver in order to back up their
regional leader, instead of legislative bills and policy matters (Park 1996). Election
campaigns were often marred by name calling and finger pointing.

The 1992 National Assembly election survey indicates to what extent the disbelief
toward politicians was spread. The vast majority of the respondents (77.4%) had
negative evaluations of politicians. Also, about the same portion of the respondents
(74.2%) had negative evaluations of the members of the National Assembly. In addition,
the 1992 Presidential election survey illustrates that the majority of the respondents
(68.5%) disapproved of the way the government had been handling its job. Moreover,
75.5% of the respondents had negative evaluations of politicians. In the 1996 National
Assembly election survey, 32.8% of the respondents believed that their vote had no

impact or a negligible impact on the National Assembly elections. Nearly a quarter

62

(25.7%) believed that their vote had no impact or a negligible impact on Presidential
elections. In the 1997 Presidential election survey, 86.7% of the respondents believed
that politicians were only interested in how many votes they received.28

The combined model reveals almost the same results as the basic model. Older
people were more likely to participate in the election. The party identifier and election
interest variables appear to have been positively connected to voter turnout, ceteris
paribus. Also, the campaign contact variable has a negative impact on voter turnout.
Voters’ home provinces were significant for voter participation. Kyungsang people
tended to go to the polling booths. Yet, the strategic coalition between Kim Dae Jung
and Kim Jong Pil did not attract voter participation.

Likelihood ratio (Pr [R 2 16.44, df = 4]) tests approve, at a .005 level of
significance, the alternative rival hypothesis that there is a significant difference between
the basic model and the combined model. As the results of joint tests for the combined
model display, however, neither the economic condition hypothesis, nor the regionalism
hypothesis and the close election hypothesis, respectively, are jointly significant (see
Table 3-4). In contrast, the socioeconomic hypothesis, the social psychological
hypothesis, and the political mobilization hypothesis, respectively, appear to have joint

significance both in the basic and combined models.

 

2’ Because the surveys did not always ask the same question, 1 cannot analyze the changes in a particular
voter perception throughout the elections. But voters usually held a negative image about politics and
politicians.

63

4. Conclusions

From the above statistical analyses, I draw the following conclusions. First, the
most striking new pattern of voter turnout is that the mobilization effect appears to have
dwindled. Although the political mobilization effect did not decrease overnight after the
democratic transition, the political mobilization hypothesis lost joint significance after the
1992 National Assembly election. The negative relationship between campaign contact
and voter turnout in 1997 suggests that the joint significance of the political mobilization
hypothesis cannot verify the traditional voter mobilization.

As a result, official voter turnout rates in the elections under study have gradually
decreased, compared to those in past elections. For instance, the first direct Presidential
election in 1987 after about two-decade-long military controlled authoritarian elections
marked record high 89.2%. Since then, voter turnout rates in the Presidential elections
dropped to 81.9% (1992) and again to 80.6% (1997). Decrease in voter turnout is more
dramatic in the National Assembly elections than in the Presidential elections. In the
National Assembly elections, voter turnout rates were 78.4% (1981) and 84.6% (1985)
but they dropped to 75.7% (1988), 71.9% (1992), and again to 63.9% (1996).

During the South Korean authoritarian era, mobilized voters were more likely to
be older people, women, less educated people, and rural residents (Kim 1980). Rapid
economic development in the 19605 and 19705 had many young and educated male
workforces move from their home to urban industrial areas. As a result, the ruling party
won among rural areas where older people, females, and less educated people were more

likely to live, while opposition parties won among urban areas where younger people,

64

males, and more educated people were more likely to live. The term of “yea chon ya do”
(the ruling party based on rural areas, opposition parties based on urban areas) succinctly
captured the traditional relationship between parties and supporters between the 19605
and the early 19805.

In the 19905, the older generation, rural residents, and less educated people still
tended to vote more than their young, urban, and educated counterparts. Nevertheless,
this must have had little to do with the traditional voter mobilization. Previously, voter
mobilization occurred with wide spread vote buying, implicit or explicit coercion, and the
like. After the democratic transition, elections became remarkably free and fair and
election laws became tighter against various electoral fraud than ever. Although this does
not imply that the electoral fraud did not exist in the 1990, elections in the 19905 have
certainly been cleaner than past elections. Thus, traditional voter mobilization was
challenged by the people and the new democratic governments as well. Moreover, the .
power transition between the ruling party to an opposition party orderly took place for the
first time in its history as the result of the 1997 Presidential election. Therefore, the
history-long “yea chon ya do” division could no longer be applied to electoral politics.
Then. why did the older generation, rural residents, and less educated people still tend to
vote more than their counterparts?

The younger generation’s lower turnout is associated with generational
differences in South Korea. As hypothesized at the beginning of this Chapter, the older
generation felt responsibility for the society more than the younger generation did, thus
the older generation was more likely to vote. According to Inglehart and Abramson

(1994; 349), the difference between the values of young and old in South Korea with

65

differing rates of economic growth (1950-90) is the strongest among their twenty cases.
Consequently, the younger generation was more concerned about postrnaterialist values,
such as “freedom, self-expression, and the quality of life” than traditional materialistic
politics (Inglehart and Abramson 1994; 336).

On the other hand, the younger generation’s lower turnout seems not to be
correlated to political distrust or disbelief in South Korea. It was said that political
cynicism induced the decrease in the younger generation’s voting participation.
However, there is no political cynicism survey item. Thus, political cynicism is
indirectly indicated by the political distrust and disbelief items. Gamma test value (-.260)
between age and disbelief toward politicians suggests that there was a weak correlation
between them in the 1992 National Assembly election. Gamma test values between age
and the similar perceptions in different elections indicate a weak or no correlation
between them. The examples of Gamma test values are the following: the disbelief about
the members of the National Assembly (-.092) in the 1992 National Assembly election
survey; the belief that their vote had no impact or a negligible impact on the National
Assembly elections (.073), the belief that their vote had no impact or a negligible impact
on Presidential elections (.002) in the 1996 National Assembly election survey; the
evaluations of politicians (-.247) in the 1992 Presidential election survey; and the belief
that politicians were only interested in how many votes they received (-.106).

In addition, the traditional voter mobilization, if any, occurred less frequently
among rural residents and less educated people in the 19905’ elections than in previous
elections. The residential area variable was significant only in the two National

Assembly elections and the educational attainment variable was significant only in the

66

1996 National Assembly election. Also, political distrust or disbelief (political cynicism)
did not affect voting behavior with regard to residential area and educational attainment.
Gamma tests suggest a weak or no correlation between residential area (and educational
attainment) and the above mentioned survey items. Voters generally had negative
feelings about politicians and politics regardless of their residential area and educational
attainment. Nonetheless, the negative evaluation did not create a pattern about the vote
participation.

The second conclusion is that voters’ economic evaluation has not affected their
decisions of whether to vote or not in South Korea. Surprisingly enough, even the most
significant economic crisis in South Korea (the IMF bailout in November 1997) did not
significantly change voter turnout in the December 1997 Presidential election. The
official turnout rate of the 1992 Presidential election was 81.9%, while the actual voter
turnout of the 1997 Presidential election was 80.6%. This conclusion a bit contradicts
Radcliffs findings (1992; 1996). According to him, turnout rates in industrial countries
are affected by the state of the economy and that economic adversity has different
impacts on voter participation depending upon the welfare level in a country. His
multivariate regression reaffirms Rosenstone’s finding that economic hardships in the
US. discourage voter participation. In contrast, economic adversity in developing
countries encourages individuals to go to the polling booths.

Third, regionalism had no significant impact on voter turnout in South Korea.
Joint tests suggest that the regionalism hypothesis is not jointly significant, with the sole
exception of the 1992 Presidential election. There is no systematic pattern between

voter’s home province and election participation.

67

Fourth, the socioeconomic hypothesis is jointly significant across the elections.
Particularly, the age variable reached statistical significance throughout the elections.
The age variable was also in the anticipated direction. On the other hand, urban residents
and the educated were less likely to vote, which disproves the hypothesis.

Fifth, the social psychological hypothesis is also jointly significant for voter
participation throughout the elections. Particularly, the election interest variable reached
statistical significance across the elections. Election interest was positively correlated
with voter participation. This is consistent with Lee’s (1993) analysis on a National
Assembly election. In addition, party supporters were more likely to vote than

independent voters.

68

CHAPTER IV

VOTER TURNOUT IN TAIWAN

1. Model Specification and Methodology

In Chapter IV, I investigate the impacts of economic conditions on voter turnout
in Taiwan. I also compare the effects of economic conditions on voter turnout with those
of the socioeconomic, social psychological, and political mobilization factors. The
comparisons will, in part, be made with joint significant tests. There is no survey item in
Taiwan’s surveys that asks voters’ anticipated election closeness unlike in South Korea’s
surveys. Moreover, I examine the effect of ethnic voting inherent in Taiwan on voter
turnout. As in Chapter III, the dependent variable of the statistical models to be tested is
the binary variable of voter turnout. Thus, I employ probit. The independent variables of
the basic model include the measurements of economic conditions, socioeconomic
factors, social psychological factors, and a political mobilization factor. The combined
model includes the basic model, ethnic background, and the individual election specific
event (tension between Taiwan and China). The combined model is designed to explain
the relationship between ethnic background and voter turnout.

In general, the hypotheses and the coding systems are the same as the South
Korean counterparts in Chapter III. Otherwise, I specify the differences. The translated
or recoded Taiwanese survey questions are also in Appendix C. The economic condition
hypothesis is tested with the variables of national economic conditions, personal

economic conditions, and unemployment. The state of the national and personal

69

economies are positively correlated with voter turnout, while unemployment is negatively
related to voter turnout. Taiwanese data lack the survey item for inflation.

The socioeconomic factors are age, gender, and educational attainment. These
three variables have exactly the same coding system as in Chapter III and they are
hypothesized to have a positive relationship with voter turnout. Taiwanese surveys do
not have the residential area item that is compatible with South Korean surveys.

Also, the measurements of the social psychological factors include party identifier
and election interest. A party identifier is indicated by a self-reported supporter for a
party. Because Taiwan’s surveys do not have a straightforward question about election
interest, I employ the survey question, “How interested were you in the election-related
articles in newspapers?” The interest in the election-related articles is assumed to be
related to election interest. These two social psychological variables are expected to have
a positive impact on voter turnout. Taiwan’s surveys lack questions for political efficacy
or political trust.

For the variables of political mobilization, I use the frequencies of contact with
party campaign staffers. The frequencies are also positively related to voter turnout. On
the other hand, although TV commercials were also legalized for the first time in 1992,
Taiwan surveys do not have survey questions for TV advertisements.

Ethnic voting is specified by voters’ ethnic background. Ethnic background is
measured by the dummy variables of the Taiwanese and the Mainlanders, with the other
choices being the excluded category. If ethnic voting is strong, then voter turnout is more
likely to increase. Citizens are willing to go to the polling booths to show their support

for their ethnic party. In addition, the tension variable is included in the 1995 and 1996

70

elections. Encountering the military threat from outside (the mainland), citizens were
more likely to vote in order to demonstrate their support for the ruling party. The 1995
election survey asked, “How do you think current cross-strait relations are going?” If a
respondent perceived the relations as improved (this is interpreted as support for the
ruling party), the respondent was more likely to vote. If a respondent disagreed with the
1996 survey question (this is also interpreted as support for the ruling party), “Did Lee
Teng-hui provoke the current cross-strait tension?” the ' ‘spondent was more likely to go

to the polling booths.

2. Statistical Models

A basic voter turnout function and a combined voter turnout function are,

respectively, as follows:

P (VT2=l\x) =<r>(1~:, SOE, PSY, and M) and (5)

P (VT_,=1\x) =<r> (E, s05, PSY, M, ETH, and IE), (6)

where VT2 and VT3 are voter turnout; E is economic conditions; SOE is socioeconomic
factors; PSY is social psychological factors; M is a political mobilization factor; ETH is
ethnic background; and IE is the individual election specific event.

The basic probit model is transformed into as follows:

71

P (VT2=1\X) = (D (6200+ [3201x201 + p202x202 + B203x203 + p204X204 + 6205x205

+ p206x206 + p207x207 + B208X208 + 5209x209 + U1), (7)

where VT2 = voter turnout; [3200 = intercept; X20, = national economic conditions; X202 =
personal economic conditions; X203 = unemployment; X20, = age; X205 = gender (male);
X206 = educational attainment; X207 = party identifier; X208 = election interest; X209 =
campaign contact; and U, = error term.

The combined model estimates the impact of ethnic cleavages on Taiwan’s voter

turnout.

P (VT3=1\X) = (D “5300+ p301x301 + fisozxmz + 5303x303 + 5304x304

+ B305X305 + B306X306+ fi307X307 + B308X308 + B309X309

+ p310X310+p311X311+ p3l2x312+ Ui)’ (8)

where VT3 == voter turnout; [5300 = intercept; X30, = national economic conditions; X302 =
personal economic conditions; X303 = unemployment; X3,), = age; X3,), = gender (male);
X306 = educational attainment; X,07 = party identifier; X308 = election interest; X309 =
campaign contact; X3,O = ethnic background (Taiwanese); X3” = ethnic background
(Mainlander); X3,2 = tension (N/A in the 1992 Legislative Yuan election); and U, = error

term.

72

3. Voter Turnout in Taiwan

A. The Legislative Yuan Elections

The estimates of the probit models for the Legislative Yuan elections are
presented in Table 4-1. The left two columns next to the independent variable column are
for the 1992 election and the right two columns are for the 1995 election. The left
column of each election displays the parameter estimates of the basic model. The right
column of each election has the coefficient estimates of the combined model. Standard

errors are presented in parentheses.

The December 1992 Legislative Yuan Election. The estimates of the basic
model for the 1992 election suggest that economic conditions reached statistical
significance for voter turnout (see Table 4-1). However, the effects of the national
economic conditions and personal economic conditions on voter participation
contradicted each other. The state of the national economy was positively related to voter
turnout, while the state of the personal economy was negatively associated with electoral
participation. Age is the only socioeconomic factor that had a significant impact on voter
turnout. Older people were more likely to vote. Among the social psychological
variables, election interest played a significant role in electoral participation. The more
interested an individual was, the more likely it was that the person went to the polling
places. Since the campaign contact variable did not reach statistical significance, voter

mobilization was not likely to have occurred in 1992.

73

Table 4-1. Probit Estimates for Voter Turnout: The Legislative Yuan Elections, Taiwan

 

 

Independent 1 992 1992 1 995 l 995
Variables Basic Combined Basic Combined
Economics
National .18* .20* .08 .06
economy (.08) (.08) (.09) (. 10)
Personal -.18** - .19** .20* .19**
economy (.10) (.10) (.09) (.10)
Unemployment -.3 1-.33 .40
(.46) (45) (.33) (. 37)
Socioeconomic
Age .39* .41* .31 * 3.0“
(. 06) (. 07) (.06) (. 07)
Gender .01.03 - .02 -.30
(Male) (.101) (.11) (.1013) (.1022)
Educational -.00
attainment (. 06) (.06) (. 063) (. O72)
Psychological
Party .10 .10 .19 .27*
identifier (.11) (.12) (. 12) (.13)
Election .11* .1 1* .07
interest (.05) (.05) (. 05) (. 05)
Mobilization
Campaign .18 .16 .33 .33
contact (.13) (.13) (.21) (.21)
Ethnic
Taiwanese .27 .00
(.17) (.13)
Mainlander -.10 - .02
(.13) (.15)
Tension N/A N/A -.05
' (07)
Intercept -.19 -.33 -l.58* -l .37
(.37) (.39) (.77) (.88)
Log Likelihood -321.32 -312.37 -346.10 -307.44
Prob > chi2 0.00 0.00 0.00 0.00
Pseudo R2 0.11 0.12 0.08 0.08

Number of obs.

Note: * (**) indicates an estimate significant at the p <

1104

1075

74

1129

.05 level (p < .10 level).

944

The combined model presents more or less the same estimations. The national
economic situation positively affected voter turnout, while the personal economic
situation negatively affected voter participation. Older people tended to vote but gender
and education were not the determining factors. Election interest was the only social
psychological factor that reached statistical significance. As expected, election interest
was positively associated with electoral participation, all other things being equal.
Contact with party campaign workers did not generate differences in voter participation.
In addition, the ethnic cleavage variables did not reach statistical significance for voter
turnout.

Likelihood ratio (Pr [R _>. 17.9, df = 2]) tests approve the alternative rival
hypothesis that the combined model is significantly different from the basic model at a
.005 level of significance. Additionally, the results of joint tests for the statistical models
confirm that the economic condition hypothesis, the socioeconomic hypothesis, and the
social psychological hypothesis. respectively, are jointly significant (see Table 4-2). Yet.
the political mobilization hypothesis and the ethnic cleavage hypothesis, respectively, are
not jointly significant. However, since the signs of the national economy and the
personal economy are opposite to each other, the joint significance of the economic
hypothesis does not confirm any meaningful relationship between the economic

conditions and voter turnout.

75

Table 4-2. Joint Test Results of the Probit Estimates: The Legislative Yuan Elections,
Taiwan

 

 

Hypothesis 1992 1992 1995 1995
Basic Combined Basic Combined
Economic .07" .04* .03 * .12
conditions
Socioeconomic .00* .00* .00* .00*
Social .06** .07” .08" .03*
psychological
Political .16 .21 .1 l .12
mobilization
Ethnic .15 .90

 

Note: Prob > chi2
* (**) indicates an estimate significant at the p < .05 level (p < .10 level).

The December 1995 Legislative Yuan Election. The estimates of the basic
model for the 1995 election indicate that the coefficient for the personal economic
condition variable reached statistical significance for electoral participation (see Table 4-
1). Personal economic condition appears to have had a positive impact on voter turnout.
Among the socioeconomic variables, only age had a significant influence on voter
participation. Consistent with expectations, older people were more likely to participate
in the election, ceteris paribus. In 1995, neither the social psychological factors, nor the
political mobilization factors worked as significant variables.

The combined model exhibits almost the same pattern as the basic model.
Personal economic conditions and age were positively associated with voter participation.
In addition, party identifiers were more likely to go to the polling places. On the other
hand, the ethnic background and tension variables did not induce any differences in voter

participation.

76

Likelihood ratio (Pr [R 2 77.32, df = 3]) tests approve the alternative rival
hypothesis that there is a significant difference between the basic model and the
combined model at a .005 level of significance. Joint tests for the combined model
disprove the joint significance of the economic condition hypothesis and the ethnic voting
hypothesis, respectively (see Table 4-2). In the basic model, the results of joint tests
confirm the joint significance of the economic condition hypothesis. Both in the basic
and combined models, the socioeconomic hypothesis and the social psychological
hypothesis, respectively, are jointly significant, while the political mobilization

hypothesis is not jointly significant.

B. Presidential Election

The March 1996 Presidential Election. Table 4-3 illustrates the probit
estimations for the 1996 Presidential election. The middle column displays the parameter
estimates for the basic model. The right column illustrates the coefficients estimates for
the combined model. The national economy was significant among the economic
condition variables (see Table 4-3). Of the socioeconomic factors, age and gender had
significant impacts on voter turnout. Older people were more likely to vote. Yet,
contrary to the hypothesis, females were also more likely to vote in the 1996 election. In
addition, the party identifier and election interest variables had positive effects on voter
participation. In contrast, voter mobilization appears not to have occurred, which is

suggested by the insignificant campaign contact variable.

77

Table 4-3. Probit Estimates for Voter Turnout: Presidential Election, Taiwan

 

 

Independent 1 996 l 996
Variables Basic Combined
Economics
National .19*"‘ .16
economy (.10) (.1 1)
Personal .02 .08
economy (.10) (.1 1)
Unemployment -.35 -.46
(.28) (.31)
Socioeconomic
Age .25* .21 *
(.06) (.07)
Gender -.31* -.30*
(Male) (.12) (.14)
Educational .09 . l 0
attainment (.07) (.07)
Psychological
Party .34* .39*
identifier (.13) (.15)
Election .15 * . 13 "‘
interest (.05) (.06)
Mobilization
Campaign .12 .12
contact (.26) (.27)
Ethnic
Taiwanese .30"‘*
(.16)
Mainlander .06
(.18)
Tension .00
(.08)
Intercept -.75 -.86
(.48) (.58)
Log Likelihood -264.35 -213.61
Prob > chi2 0.00 0.00
Pseudo R2 0.08 0.08
Number of obs. 1100 927

 

Note: *(**) indicates an estimate significant at the p < .05 level (p < .10 level).

78

The combined model reveals that the economic condition variables are not
significant, however. The age and gender variables reached statistical significance for
voter turnout. As in the basic model, the sign of the age variable is in the expected
direction, whereas the sign of the gender variable is not in the expected direction. The
social psychological variables, party identifier and election interest, had significant
impacts on voter turnout. Moreover, an ethnic variable reached statistical significance for
the first time in the Taiwan’s survey data. Native Taiwanese were more likely to vote in

1996.

Table 4-4. Joint Test Results of the Probit Estimates: Presidential Election, Taiwan

 

 

Hypothesis 1996 1996
Basic Combined
Economic .1 8 . 1 8
conditions
Socioeconomic .00* .01 *
Social .00* .00*
psychological
Political .66 .66
mobilization
Ethnic .3 1

 

Note: Prob > chi2
* indicates an estimate significant at the p < .05 level (p < .10 level).

Likelihood ratio (Pr [R 2 101.48, df = 3]) tests approve the alternative rival
hypothesis that the combined model is significantly different from the basic model at a
.005 level of significance. Joint tests for the two models confirm that the socioeconomic

hypothesis and the social psychological hypothesis, respectively, are jointly significant

79

(see Table 4-4). In contrast, the economic condition hypothesis, the political mobilization

hypothesis, and the ethnic voting hypothesis, respectively, are not jointly significant.

4. Conclusions

The statistical analyses of the probit models for Taiwan’s voter turnout lead to the
following conclusions. First, the economic condition hypothesis does not have a
consistent impact on voter turnout. Joint tests confirm that the economic condition
hypothesis is jointly significant in 1992 and 1995. However, the economic condition
hypothesis appears to have had an contradictory impact on voter turnout in 1992. The
national economic condition variable and the personal economic condition variable have
the signs opposite to each other. In 1995, only the personal economic condition variable
in the basic model had the expected impact on voter turnout. In 1996, the national
economy variable had a statistical impact on voter participation but did not produce the
joint significance of the economic condition hypothesis.

Second, likelihood ratio tests suggest that the combined model is better than the
basic model across the elections. However, joint tests suggest that the ethnic voting
hypothesis appears not to be jointly significant throughout the elections under
investigation. Only the dummy variable of the Taiwanese ethnicity in 1996 reached
statistical significance for electoral turnout.

Third, the socioeconomic hypothesis is the one that had a consistent impact on
voter turnout throughout the elections. Also, age is the only variable that has had a

consistent effect on voter participation in Taiwan. The sign of the age variable is also in

80

the expected direction. In addition, females were more likely to vote than males in
Taiwan in 1996, which is opposite to the hypothesis. This may be one of the
characteristics of Taiwan’s electoral politics.

Fourth, the social psychological hypothesis also appears to be jointly significant
for voter turnout. Election interest was significant in 1992 and 1996. In addition, party
attachment was significant in 1995 (the combined model) and 1996. The signs of these
variables are also in the expected directions. In contrast, the political mobilization

hypothesis never reached statistical significance.

81

CHAPTER V

ECONOMIC VOTING IN SOUTH KOREA

1. Model Specification and Methodology

In this Chapter, I investigate the relationship between economic conditions and
voter choice in South Korea. I also make a comparison among the impacts of economic
conditions, socioeconomic factors, party identification, and regionalism on voter choice.
Thus, the dependent variable is self-reported voter choice. As in Table 5-1, there have

always been three main parties (or candidates) in South Korea’s elections under study.

Table 5-1. Vote Distribution in the South Korean Elections

 

 

Vote Share
National Assembly Elections Presidential Elections
1992 1996 1992 1997

lst Party DLP-38.5% GNP-34.5% KYS-42.0% LHC-38.7%
2nd Party DP-29.2% NCNP-25.3% KDJ-33.8% KDJ-40.3%
3rd Party UPP-17.4% ULD-16.2% CJY-16.3% LlJ-19.2%
4th Party NPRP-1.8% DP-11.2% PCJ-6.4%
5th Party IND-11.5% IND-12.9%

Total“ 85.1% 76.0% 92.1% 98.2%

 

Note: * The percentages in the bottom cells are the total vote shares of the first three
parties.
Sources: Asian Survey (various issues) and Journal of Democracy (various issues).

The dependent variable has three unordered possibilities. Such three-way races
prevent me from using an ordered probit (or logit) technique. Consequently, I use a

multinomial logit method for the three-way races, normalizing coefficients for the ruling

party (or candidate) to zero. The coefficients for the two opposition parties in the

82

estimates are the impact of each variable on voter choice, relative to a vote for the ruling
party. When the sign of a coefficient is negative, respondents are interpreted to vote for
the ruling party. Otherwise, respondents are interpreted to opt for one of the opposition
parties.

However, Alvarez and Nagler take issue with multinomial logit for the three-way
races. First, they point out that multinomial logit “is computing estimates of precisely the
same parameters as is binomial logit” (Alvarez and Nagler 1998, 85; emphasis in
original). Second, they assert that “if one is interested in more strategic questions about
politics then multinomial logit is the wrong model to use” (Alvarez and Nagler 1998,
85). Third, “multinomial logit [is] all quite limited in that they impose the IIA
restriction upon voters” (Alvarez and Nagler 1998, 85). The IIA condition refers to the
property of Independence of Irrelevant Alternatives on individual voters, which implies
that “the ratio of the probability of choosing one party to the probability of choosing a
second party is unchanged for individual voters if a third party enters the race” (Alvarez
and Nagler 1998, 57). Therefore, they suggest that multinomial probit works better with
more than three choices than multinomial logit.

However, the multinomial logit methodology provides almost the same parameter
estimates as Alvarez and Nagler’s alternative and thus, is utilized for analyzing three-way
elections (Dominguez and McCann 1996; Hsieh, Lacy, and Niou N. D.; King, Tomz, and
Wittenberg 2000; Lee 1998). This research project is also focused on self-reported voter
choices in post-election surveys rather than the issue movements by parties. In other
words, this study does not probe issue voting in the spatial context. Instead, I investigate

the relationship mainly between economic conditions (or regionalism) and voter choice.

83

Therefore, I utilize multinomial logit that provides greater efficiency than multinomial
probit.

The independent variables of the basic model include the measurements of
sociotropic voting, pocketbook voting, retrospective voting, prospective voting,
socioeconomic factors, and party identification. The combined model consists of the
basic model, home province, and the individual election specific event of a party merger
(or coalition). This combined model measures the impact of regionalism on voter choice.

To measure sociotropic voting, I employ the national economic condition and
inflation variables, depending upon the data availability. If a respondent perceived the
national economy as being worse than a couple of years earlier, the respondent was more
likely to support an opposition party. The 1992 National Assembly election survey does
not have the national economic condition item. To indirectly measure the national
economic condition for the election, I employ the dummy variable of inflation. The
survey asked, “What is the most important problem that the country is facing?” If a
respondent considered inflation to be the nation’s most important problem, the
respondent was more likely to opt for an opposition party. The inflation item is also
available in the two Presidential election surveys but not in the 1996 National Assembly
election survey.

Also, South Korea experienced its most serious economic crisis during the 1997
election campaign, so the survey did not ask the usual question, “How would you
evaluate national economic conditions during the past couple of years?” For the national
economic condition in 1997, I use the question, “Which political party did you think was

mainly responsible for the current economic crisis [the International Monetary Fund

84

(IMF) bailout]?” If a respondent answered that the ruling party was responsible for the
IMF bailout, the respondent was expected to support an opposition party.

For pocketbook voting, I use the personal economic condition and unemployment
variables in all the elections. If a respondent answered that the personal economic
condition was worse than earlier or that the individual was unemployed, the respondent
was expected to prefer an opposition party. These sociotropic and pocketbook voting
variables are, in nature, the retrospective economic voting survey items. Additionally,
there is a retrospective question, “How well do you think the government has been
running things?” which does not exactly capture retrospective economic voting. Rather,
the question deals with the evaluations of Presidents’ overall performances. However,
“Perceptions of presidential performance in economic policy had strong relationships
with more general evaluations of the president” (Edwards 1992, 140). Thus, it is
reasonable to employ the question as retrospective economic voting. If a respondent
answered negatively to the question, the person was more likely to vote for an opposition
party.

The IKES had not included prospective survey items until the 1997 Presidential
election. The 1997 survey asked, “Who do you think will best be able to resolve the
current national problems?” There would be very few respondents able to understand the
current national problems as anything other than the IMF bailout. Since the candidates
(Lee Hoe Chang, Kim Dae Jung, and Lee In Je) are nominal, I treat them as dummy
variables, leaving other minor candidates as the excluded category. Respondents were

more likely to opt for the candidate who would best be able to handle the economic crisis.

85

In order to measure the impact of the socioeconomic factors on voter choice, I use
the variables of residential area, age, gender, and educational attainment. Because
individuals living in urban areas and more educated people are more likely to be liberal
compared to their rural and less educated counterparts, they are expected to choose an
opposition party over the ruling party. Because the younger generations are more
receptive to change and the new democratic values than older people, they are also more
likely to support an opposition party than the ruling party. I recode the age variable as l
= 20-29, 2 = 30-39, 3 = 40-49, and 4 = 50 and older (Alvarez and Nagler 1995). Since
men are likely to be more liberal than women, men are expected to vote for an opposition
party. I use gender (male) as a dummy variable.

I also include party identification as a factor of voter choice. Party identification
in South Korea is equivalent to a categorical attachment to the ruling party versus
opposition parties. The IKES has the survey item asking if the respondents were close to
the ruling party or opposition parties. I treat party identification as dummy variables (the
ruling party and opposition parties), with independents being the omitted category.

To evaluate the impact of regionalism on voter choice, I first employ respondents’
home provinces as dummy variables (Cholla province and Kyungsang province), leaving
other provinces as the omitted category. Regionalism is expected to have a positive
impact on voter choice. If a respondent was born in the same province as the party’s
leader, the respondent is more likely to vote for this leader’s party. Also, regionalism
affects the individual election specific event of a party merger. Party leaders orchestrated
party mergers in order to win the election and a certain merger was supported by certain

regional voters. So, dummy variables (coalition supporters and coalition non-supporters)

86

are used in the statistical tests, with “don’t know” being the excluded category. The 1996
election survey did not ask about the preference over a party merger. There had been no

party merger during the 1996 election campaign.

2. Statistical Models

A basic voter choice function and a combined voter choice function are,

respectively, as follows:

VC0 = f (S, P, RETRO, PRO, SOE, and PID) and (9)

VC, = f (S, P, RETRO, PRO, SOE, PID, HP, and IE), (10)

where VC0 and VC, are voter choicef’ S is sociotropic voting; P is pocketbook voting;
RETRO is retrospective voting; PRO is prospective voting; SOE is socioeconomic
factors; PID is party identification; HP is home province; and IE is the individual election
specific event. Function f is assumed to be polychotomous.

The combined model, for example, is transformed into the following equation:

VCI = f(BO+ plxl + BZX2+ p3x3 + p4x4 + BSXS+ B6X6+ B7X7 + [38x8
+ p9)(9+ BIOXIO+ flllxll + B12xl2+ Bl3xl3+Bl4xl4+ filSXIS

+ BI6XI6+ B17Xl7+618xl8+Ui)9 (11)

 

2" VC0 is the brief form of P(Y, = j)/P(Y, = k), where Y, is voter i’s vote, k is the base party (or candidate),
and j is each of other two parties (or candidates). '

87

where VC, = voter choice; I30 = intercept; X, = national economic condition (N/A in the
1992 National Assembly election); X2 = inflation (N/A in the 1996 National Assembly
election); X3 = personal economic condition; X,, = unemployment; X5 = retrospective
evaluation; [X,, = prospective evaluation (Lee Hoe Chang); X7 = prospective evaluation
(Kim Dae Jung); X,, = prospective evaluation (Lee In Je) in the 1997 Presidential
election]; X9 = residential area; X,,, = age; X,, = gender (male); X,2 = educational
attainment; X, 3 = party identification (the ruling party); X,4 = party identification
(opposition parties); X,5 = home province (Cholla); X,,, = home province (Kyungsang);
[X,7 = merger (support); X,8 = merger (not support) (N/A in the 1996 National Assembly

election)]; and Ui = error term. The basic model has only X, through X, 4.

3. Economic Voting in South Korea

A. The National Assembly Elections

The March 1992 National Assembly Election. In the 1992 National Assembly
election, there were three leading parties competing for seats, Kim Young Sam’s ruling
Democratic Liberal Party (DLP), Kim Dae Jung’s Democratic Party (DP), and Chung Ju
Young’s United People’s Party (UPP). The estimates for the multinomial logit models
are in Table 5-2. The left two columns next to the independent variable column are the
basic model and the right two columns are the combined model. The left column of each

model displays the coefficient estimates for the first opposition party relative to the ruling

88

party. The right column of each model illustrates the parameter estimates for the second
opposition party relative to the ruling party. Standard errors are in parentheses. The
following Tables 5-4, 5-5, and 5-6 have the same format as Table 5-2.

The basic model suggests that voters were sociotropic. Contrary to expectations,
those who perceived inflation as being the most important national problem were more
likely to vote for the ruling DLP than the opposition DP. This contradicts the hypothesis
and keeps appearing in the estimates for the 1996 National Assembly and the 1992
Presidential races. The analyses of this discrepancy will be presented in the 1992
Presidential election section. Also, voters appear to have voted retrospectively. The
retrospective evaluation had a significant impact on voter choice between the DP and the
DLP. As expected, the citizens with the favorable evaluations of the ruling party were
more likely to opt for the ruling DLP. Between the UPP and the DLP, only the
retrospective evaluation of the government, among the economic condition variables, had
a significant effect on voter choice. The favorable evaluations of the incumbent

government led to support for the incumbent DLP.

89

Table 5-2. Multinomial Logit Estimates for the Three-Party Race: The 1992 National
Assembly Election (Coefficients for DLP’s Normalized to Zero), South Korea

 

 

Independent Basic Basic Combined Combined
Variables DP UPP DP UPP
Economics
Inflation -.56* -.14 -.42** -.14
(.21) (.24) (.23) (.25)
Personal .08 .04 . l 7 .02
economy (.15) (.18) (.17) (.19)
Unemployment -.59 -.51 -.26 -.55
(.57) (.71) (.60) (.70)
Retrospective -.65* -1.07* -.36 -.93*
evaluation (.25) (.32) (.29) (.35)
Socioeconomic
Residential .19 -.29** .59* -.27
area (.14) (.16) (.17) (.18)
Age -.26* .10 -.50* .14
(.12) (.13) (.13) (.14)
Gender -.17 -.20 -.24 -.34
(Male) (.22) (.25) (.24) (.26)
Educational -.07 .63* -.22 .54*
attainment (.13) (.16) (.15) (.17)
PID
Ruling Party -1.53* -l.16* -l.23* -1.23*
(.28) (.32) (.31) (.33)
Opposition 2.83* 1.91* 2.40* 1.76*
(.29) (.32) (.31) (.34)
Regionalism
Home province 1.81 * -1.38*
(Cholla) (.31) (.46)
Home province -.83* -1.11*
(Kyungsang) (.28) (.29)
Merger .19 .41
(Support) (.37) (.40)
Merger 1 .09* .78*
(Not Support) (.30) ( 34)
Intercept 1.01 -.85 -.42 -.85
(.74) (.86) (.86) .93)
Log Likelihood -601.73
Prob > chi2 0.00
Pseudo R2 0.26

Number of obs.

788

Note: * (**) indicates an estimate significant at the p < .05 level (p < .10 level).

90

Among the four socioeconomic variables in the basic model, only age had
significant influences on voter choice between the DP and the DLP. Older people were
more likely to support the ruling DLP than the DP. For the choice between the UPP and
the DLP, residential area and educational attainment reached statistical significance.
Rural residents and the educated were more likely to support the opposition UPP than the
DLP. Additionally, party identification turns out to have played a significant role in voter
decisions. For the DP versus DLP comparison, the ruling party supporters tended to vote
for the DLP and the opposition party sympathizers were more likely to support the DP.
On the other hand, between the UPP and the DLP, the ruling party identifiers tended to
opt for the DLP and the opposition party identifiers were more likely to prefer the UPP.
These are all consistent with the hypothesis.

The combined model indicates that sociotropic voting occurred in the DP versus
DLP comparison. Those who perceived inflation as being the nation’s most serious
problem were more likely to opt for the ruling DLP than the DP. Between the UPP and
the DLP, only retrospective voting was significant. The individuals with the good
evaluations of the Kim Young Sam government were more likely to support the
incumbent DLP than the opposition UPP.

Additionally, residential area and age of the four socioeconomic factors had
significant effects on voter choice between the DP and DLP. The DP appealed to urban
residents and younger voters more than the DLP, ceteris paribus. Also, the UPP did
better among educated people than the DLP. Party identification appears to have had the
anticipated effects on voter choice. Between the DP and the DLP, the ruling party

identifiers were more likely to vote for the DLP and the opposition party supporters were

91

more likely to opt for the DP. Also, between the UPP and the DLP, the ruling party
sympathizers tended to support the DLP and the opposition party identifiers were more
likely to opt for the UPP.

Regionalism also had the expected influences on voter choice. Cholla people
were more likely to vote for Cholla province-based DP than Kyungsang province-based
DLP, whereas Kyungsang people were more likely to vote for their DLP than the DP.
Moreover, the merger variables also turn out to be significant for voter choice. The
voters who disliked the three party merger supported Kim Dae Jung’s DP. Between the
UPP and the DLP, Kyungsang people were more likely to cast their votes for their
regional DLP than the UPP. Yet, Cholla people tended to support the rival DLP rather
than the UPP. The coefficients are the relative probability only between the DLP and the
UPP. Chung Ju Young is from a North Korean area so that he could attract neither
Kyungsang nor Cholla voters.

Likelihood ratio (Pr [R 2 180.88, df = 4]) tests reject the null hypothesis that there
is no difference between the basic model and the combined model. In other words, the
tests approve the alternative rival hypothesis that the combined model is significantly
different from the basic model at a .005 level of significance. Joint tests for the combined
model suggest that the regionalism hypothesis turns out to be jointly significant for voter
choice in the combined model (see Table 5-3). In the combined model, however, the
economic voting hypothesis is not jointly significant for the DP versus DLP comparison,
while the hypothesis is jointly significant for the UPP versus DLP competition. As the

results of joint tests suggest, the economic voting hypothesis is also jointly significant in

92

making voter decisions in the basic model. In addition, the party identification

hypothesis is jointly significant throughout the models.

Table 5-3. Joint Test Results of the Multinomial Logit Estimates, South Korea

 

 

Election Hypothesis Basic Basic Combined Combined
Model Model Model Model
1 992 DP/DLP UPP/DLP DP/DLP UPP/DLP
National Economics 0.01 * 0.02* 0.23 0.08* *
Assembly Regionalism 0.00* 000*
Party ID 000* 0.00* 0.00* 0.00*
1996 NCNP/N KP ULD/N KP NCNP/N KP ULD/N KP
National Economics 000* 0097* * 0.00* 0.07* *
Assembly Regionalism 0.00* 0.47
Party ID 000* 0.00”“ 0.00* 0.00*
1 992 KDJ/KYS CJY/KYS KDJ/KYS CJY/KYS
Presidential Economics 000* 0.57 0.00* 0.68
Regionalism 0.00* 000*
Party ID 000* 0.00* 0.00* 0.00““
1997 KDJ/LHC LIJ/LHC KDJ/LHC LIJ/LHC
Presidential Economics 000* 0.00* 0.00* 0.00*
Regionalism 0.00* 001*
Party ID 0.00”“ 0.00* 0.00* 0.00*

 

Note: Prob > chi2
* (**) indicates an estimate significant at the p < .05 level (p <.10 level).

The April 1996 National Assembly Election. The election was held under
President Kim Young Sam. He led Kyungsang province’s New Korea Party (NKP). Kim
Dae Jung’s Cholla province-based National Congress for New Politics (NCNP) and Kim
Jong Pil’s Choongchung province-based United Liberal Democrats (ULD) were the next
leading parties in the race. The estimates of the multinomial logit models for the 1996

election are presented in Table 5-4.

93

Table 5-4. Multinomial Logit Estimates for the Three-Party Race: The 1996 National
Assembly Election (Coefficients for NKP’s Normalized to Zero), South Korea

 

 

Independent Basic Basic Combined Combined
Variables NCNP ULD NCNP ULD
Economics
National .83* .60* .73* .67*
economy (.21) (.25) (.23) (.26)
Personal .22 .08 .08 .03
economy (.22) (.27) (.25) (.27)
Unemployment . 1 7 .06 .45 .19
(.55) (.84) (.66) (.86)
Retrospective -.59* -.33 -.29 -.27
evaluation (.27) (.35) (.30) (.36)
Socioeconomic
Residential .68* 1.01* 1 . 12* 1.12*
area (.19) (.27) (.22) (.28)
Age -.27* -.49* -.29* -.48*
(.12) (.15) (.13) (.16)
Gender -.22 -.27 .02 -.26
(Male) (.24) (.31) (.27) (.31)
Educational -.23 . l 6 -.09 . 14
attainment (.14) (.16) (.16) (.16)
PID
Ruling Party -1.42* -.84* -1.36* -.92*
(.34) (.42) (.37) (.42)
Opposition 2.93* 2.38* 2.44* 2.28*
(.29) (.38) (.33) (.38)
Regionalism
Home province 2.06* .56
(Cholla) (.33) (.46)
Home province -1 .74* .04
(Kyungsang) (.42) (.34)
Intercept -2.58* -4.26* -4.21* -4.71*
(.87) (1.17) (1.01) (1.21)
Log Likelihood -407.08 -354.20
Prob > chi2 0.00 0.00
Pseudo R2 0.30 0.39
Number of obs. 622 622

 

Note: * indicates an estimate significant at the p < .05 level.

94

The basic model reveals that voters were more likely to be concerned about the
sociotropic economy than the pocketbook economy. Contrary to expectations, the signs
of the national economic condition are positive, which indicates that the good evaluations
of the national economic condition led to support for the opposition parties (the NCNP
and the ULD). In addition, retrospective voting occurred in the NCNP versus NKP
competition. As expected, the good evaluations of the govemment’s past performance
were related to support for the ruling NKP rather than the NCNP, controlling all other
factors. The other economic voting variables appear to have had no effects on the choice.

The socioeconomic coefficients indicate that the NCNP and the ULD did better
among urban residents and the younger generation than the ruling NKP, ceteris paribus.
The other socioeconomic variables did not have any influences on voter choice. On the
other hand, party identification had the expected impacts on voter choice. The ruling
party identifiers were more likely to vote for the ruling NKP than the two opposition
parties, whereas the opposition party sympathizers were more likely to vote for one of the
two opposition parties than the NKP.

In the combined model, the economic voting, socioeconomic, and party
identification factors had similar effects on voter choice as their counterparts in the basic
model had, with the exception of the retrospective evaluation coefficient. The
retrospective evaluation factor did not reach statistical significance for voter choice
between the NCNP and ULD competition. I

With regard to regionalism, Cholla people were more likely to vote for Cholla
province-based NCNP than the NKP, whereas Kyungsang people were more likely to

vote for Kyungsang province-based NKP than the NCNP. In contrast, the regionalism

95

coefficients indicate no significant influences on the choice between the ULD and the
NKP. The ULD was based on the third region of Choongchung province. These
coefficient estimates may confirm that regional conflicts were most serious between
Cholla and Kyungsang provinces.

Likelihood ratio (Pr [R 2 105.76, df = 2]) tests reject, at a .005 level of
significance, the null hypothesis that there is no difference between the basic model and
the combined model. Joint tests for the combined model suggest that the economic
voting hypothesis and the party identification hypothesis, respectively, are jointly
significant for voter choice (see Table 5-3). On the other hand, the regionalism
hypothesis is jointly significant for the NCNP versus NKP comparison, whereas this is

not the case for the ULD and NKP comparison.

B. Presidential Elections

The December 1992 Presidential Election. In the 1992 Presidential race, Kim
Young Sam, Kim Dae Jung, and Chung Ju Young competed for office. The two Kims
were prominent opposition leaders but Kim Young Sam became the ruling party
candidate after the three party merger in 1990.30 Chung Ju Young was an entrepreneur

turned-politician, who owned Hyundai.

 

3° Kim is the most common last name in Korea.

96

 

 

Table 5-5. Multinomial Logit Estimates for the Three-Candidate Race: The 1992
Presidential Election (Coefficients for KYS’s Normalized to Zero), South Korea
Independent Basic Basic Combined Combined
Variables KDJ CJY KDJ CJY
Economics .
National .51 * .21 63 * .15
economy (.16) (.18) (.20) (.19)
Inflation -.80* - .14 -.80* -.15
(.21) (. 23) (.26) (.24)
Personal -.42* .0 -.43* .07
economy (.16) (.18) (.20) (.19)
Unemployment - .50 - .19 -.76** -.27
(35) (41) (.43) (.42)
Retrospective -.38 -40 -. l 6 -.3 8
evaluation (. 24) (. 27) (.31) (.28)
Socioeconomic
Residential - .22** .05 -.20 .02
area (.1135) (.15) (.16) ( 15)
Age -.19 - .39*
(.115) (.12) (.13) (139)
Gender -.41* - .08 -.13 - .10
(Male) (20) (. 23) (25) (. 24)
Educational -.02 .2 .31 .21
attainment (.12) (.14) (.15) (. 15)
PID
Ruling Party -1.70* -.77* -l .36* -.63*
(.27) (.27) (.33) (.28)
Opposition 2.77* 1.63 * 2.07* 1.44*
(.25) (.30) (.30) (.31)
Regionalism
Home province 3.20* -1.80**
(Cholla) (.37) (1.05)
Home province -1 .03* - .73*
(Kyungsang) (~29) ( 25)
Merger -1.05* -.39
(Support) (.35) (. 30)
Merger .89* .83
(Not Support) (31) (31)
Intercept 1.09 -1.57** .9
(. 73) (.86) (.96) (.91)
Log Likelihood -645.24 —497.30
Prob > chi2 0.00 0.00
Pseudo R2 0.30 0.45
Number of obs. 966 962

 

Note: * (**) indicates an estimate significant at the p < .05 level (p < .10 level).

97

The basic multinomial logit estimates indicate that the 1992 Presidential election
was affected by sociotropic voting as well as pocketbook voting (see Table 5-5). The
sociotropic variables (national economic condition and inflation) had significant impacts
on voter choice between Kim Dae Jung and Kim Young Sam. The signs of the two
sociotropic voting coefficients, however, are opposite to the hypothesis. The favorable
evaluations of the national economy were associated with support for the opposition
candidate, Kim Dae Jung. If individuals perceived inflation as the nation’s most
important problem, they tended to vote for the incumbent party’s Kim Young Sam,
instead of Kim Dae Jung. In contrast, voters correctly cast their ballots with respect to
their personal economic considerations. This also indirectly suggests that individuals cast
their votes retrospectively. On the other hand, the economic voting factors appear to have
had no effects on voter choice between Chung Ju Young and Kim Young Sam.

Among the four socioeconomic variables, residential area and gender reached
statistical significance for voter choice between Kim Dae Jung and Kim Young Sam.
Urban residents and men tended to support the ruling party’s candidate, Kim Young Sam.
Kim Dae Jung had been accused of no military records during the Korean War and
thereafter. Because the military service is mandatory for men in South Korea, males
tended to dislike Kim Dae Jung. In contrast, the party identification coefficients had the
expected impacts on voter choice. Opposition party identifiers were more likely to opt
for Kim Dae Jung, while the ruling party sympathizers were more likely to vote for Kim
Young Sam. For the choice between Chung Ju Young and Kim Young Sam, party

identification was the only coefficient that reached statistical significance. Kim Young

98

Sam appealed to the ruling party supporters, whereas Chung Ju Young appealed to the
opposition party sympathizers.

In the combined model of the 1992 Presidential race, sociotropic and pocketbook
voting appear to have had significant influences on voter choice between Kim Dae Jung
and Kim Young Sam. Again, the signs of the two sociotropic voting variables are in the
unexpected direction, while the signs of the two pocketbook voting variables are in the
conflicting directions. The unemployed appeared to have voted for the ruling party’s
Kim Young Sam.

Let’s move on to the origins of the odd signs of the national economic condition,
inflation, and unemployment coefficients that persist in the elections held between 1992
and 1996. To understand such statistical results, I suggest to take into consideration the
roles of the ruling party that played in South Korea’s economic development. Because
the miraculous economic development had been orchestrated by the ruling party since the
19605, voters may have considered the ruling party as the manager of the economy,
despite their authoritarian nature. In sharp contrast, opposition parties had never had a
chance to show their abilities to manage the economy. They also lacked human resources
with economic specialties and bureaucratic experience. Thus, voters may have thought
that the ruling party would be better than opposition parties in handling the inflation and
unemployment problems.

Moreover, the South Korean elections had been a place where the politicians
struggled against each other centering on the issues of democratic reform, without the
realistic chance of a power alternation until 1997. Thus, the opposition leaders had felt

relatively low demand for identifying themselves as an attractive alternative from an

99

economic perspective. Instead, they had conducted campaigns focused on political
issues. Although South Korea embraced the democratic transition in 1987, such decades-
long perceptions needed more time and momentum in order to be changed. The
momentum was finally provided in the 1997 Presidential election.

Of the four socioeconomic variables, only age had a significant impact on voter
choice. As expected, Kim Young Sam did better than Kim Dae Jung among older people.
The party identification factors appear to have had the expected influences on voter
choice between Kim Dae Jung and Kim Young Sam. The party identification factor was
the only coefficient that reached statistical significance for the Chung Ju Young versus
Kim Young Sam comparison. The signs of the party identification variables are also in
the expected direction.

Clearly, regionalism significantly affected voter choice in the 1992 Presidential
election. Between Kim Dae Jung and Kim Young Sam, Cholla people were more likely
to opt- for their regional leader Kim Dae Jung, while Kyungsang people were more likely
to support their regional leader Kim Young Sam. Additionally, those who preferred the
three party merger supported Kim Young Sam, while Kim Dae Jung did better than Kim
Young Sam among those who did not like the three party merger. This is consistent with
the expectations. In the Chung Ju Young versus Kim Young Sam comparison, Cholla
people turn out to have supported Kim Young Sam more than Chung Ju Young. The
exact same thing was observed in the 1992 National Assembly election. As expected,

Kim Young Sam appealed to his Kyungsang people more than Chung Ju Young.

100

Likelihood ratio (Pr [R 2 295.88, df = 4]) tests approve the alternative rival
hypothesis that the combined model is significantly different from the basic model at a
.005 level of significance. Joint tests for the combined model suggest that economic
voting was significant for the voter choice between Kim Dae Jung and Kim Young Sam
(see Table 5-3). In contrast, the regionalism and party identification factors played a
distinct role in voting decisions across the three leading contenders. This is the same in
the basic model.

The estimates of the multinomial logit models for the 1992 Presidential race
suggest that the party merger worked as a significant factor determining voter choice
particularly between Kim Young Sam (DLP) and Kim Dae Jung (DP). The merger was
created by political actors’ calculations to win the elections. In order to ensure the, future
security of their military group, President Roh Tae Woo and his Democratic Justice Party
(DJP, 125 seats) attempted to change the political system from presidentialism to
parliamentarism in 1989. In addition, they hoped to let old politicians (including Kim
Young Sam, Kim Dae Jung, and Kim Jong Pil: the three Kims) share the political power
through a parliamentary system and then, have them retire from politics. They wanted to
make a generational change in the political arena. However, the DJP fell far short of the
two-thirds of the 299 National Assembly seats, which is required to change the
constitution.

The goal of the opposition leaders, Kim Dae Jung (the Peace and Democratic
Party: PDP, 71 seats), Kim Young Sam (the Reunification Democratic Party: RDP, 59

seats), and Kim Jong Pil (the New Democratic Republican Party: NDRP, 35 seats), was

101

to win the presidency. However, the three Kims were well aware of the fact that there
was no realistic chance of winning alone. If a parliamentary system would increase their
chances of sharing (or winning) power, they were willing to adopt a parliamentary
system. Since Kim Dae Jung’s image was the most progressive and the least
compromising among the three Kims, Roh Tae Woo formed the DLP, merging his DJP
with Kim Young Sam’s RDP and Kim Jong Pil’s NDRP in 1990. The merger was based
on the pact of parliamentarism. As a result, Roh Tae Woo was able to hold more than
two-thirds of the National Assembly. However, the DLP failed to manage the “over two-
thirds” status in the 1992 National Assembly election. The DLP secured only 49.8% of
the seats, while the DP and the UPP held 32.4% and 10.4% of the seats, respectively.

After winning a series of inner party struggles, however, Kim Young Sam
nullified the pact of adopting a parliamentary system and became the candidate of the
ruling DLP for the 1992 Presidential race. In fact, his ultimate interest was to not share
power with others. He knew that there was no figure with President caliber in the DJP
and that the military legacy would hurt its chances of winning another election.
Therefore, he took the chance of becoming the presidential candidate of the ruling party
and was elected president with 42.0% of the vote. Kim Dae Jung and Chung Ju Young
garnered 33.8% and 16.3% of the vote, respectively."

It was the party merger that seemed Kim Young Sam’s 8.2 percentage point
margin of victory. Previously, Kim Young Sam'(28.0%) and Kim Dae Jung (27.1%)

received almost the same electoral returns in the 1987 Presidential election. In addition,

 

3' The actual processes and detailed story are documented in Oh (1995). In addition, Kim (1997) gives an
interesting analysis about the party merger process based on the rational choice model.

102

Kim Young Sam’s RDP (23.8%) received slightly more votes than Kim Dae Jung’s PDP

(19.3%) in the 1988 National Assembly election.

The December 1997 Presidential Election. The 1997 Presidential election was
also a three-way race among Lee Hoe Chang (the Grand Nation Party, GNP), Kim Dae
Jung (the National Congress for New Politics, NCNP), and Lee In Je (the New Party for
the People). The election was quite different from previous elections in many ways.
Above all, the election was held during a severe financial crisis, which constituted the
worst economic situation in South Korea’s history. The ruling party’s Lee Hoe Chang
experienced a very difficult campaign due mainly to President Kim Young Sam’s
mismanagement of the economy. Economic issues became one of the most salient issues.
Citizens were seriously concerned about the future of the economy admist the IMF
bailout (November 21, 1997). Under these circumstances, the election survey included
prospective voting items for the first time. The 1997 survey provides a new opportunity
to assess more comprehensive economic voting behavior than ever.

The 1997 election was characterized by sociotropic voting (see Table 5-6). As the
coefficients of the national economic condition indicate, those who perceived the ruling
party as being responsible for the IMF bailout were more likely to vote for Kim Dae Jung
than Lee Hoe Chang, which also indirectly confirms retrospective voting. Yet, people
appear to have voted prospectively as well. Those who believed Lee Hoe Chang would
best be able to resolve the current national problems were more likely to vote for him
than Kim Dae Jung. Additionally, individuals with good expectations for Kim Dae Jung

tended to opt for him.

103

 

 

Table 5-6. Multinomial Logit Estimates for the Three-Candidate Race: The 1997
Presidential Election (Coefficients for LHC’s Normalized to Zero), South Korea
Independent Basic Basic Combined Combined
Variables KDJ LIJ KDJ LIJ
Economics
National 1.11* .96* .78* .79*
economy (.28) (. 32) (.31) (.33)
Inflation -.31-.33 -.20 -.41
(.27) (. 30) (.30) (.32)
Personal . 12 2.7.33 -.1 1
economy (. 24) (. 27) (. 27) (.29)
Unemployment -.89 -1.06 -.72 -l.08
(. 70) (. 83) (.78) (.86)
Retrospective .33 .94 .83 .50
evaluation (28) (. 31) (. 32) (.32)
Prospective -3. 14* -2. 60* -3. 10* -2.67*
(LHC) (.79) (. 73) (.84) (.74)
Prospective 1.84* .52 l. 54* .19
(KDJ) (.53) (. 55) (. 57) (.57)
Prospective .57 3.17* .88 3.20*
(LID (.68) (.62) (. 72) (.63)
Socioeconomic
Residential - .09 - .27 - .09 - .30
area (. 20) (22) (. 21) (22)
Age .08 -.31 - .14
(.14) (15) (15) (.16)
Gender .08 - .160 -.10
(Male) (.27) (. 30) (. 30) (.31)
Educational .20 .40 .1 5
attainment (. 17) (. I9) (.19) (.20)
PID
Ruling Party -1.88* -1.54* -1.91* -1.47*
(.35) (.39) (.39) (.39)
Opposition 1.85* 1.44* 1.34* 1.30*
(.33) (.37) (.36) (.39)
Regionalism
Home province 2.05* 1.53*
(Cholla) (.63) (.67)
Home province -1.62* - .33
(Kyungsang) (.34) ( 33)
Merger .77“ .31
(Support) ( 39) ( 42)
Merger -.63 - .69**
(Not Support) (. 36) (.37)
Intercept -1.83** - .86 -l .60 -.44
(1.07) (1.16) (1.18) (1.21)
Log Likelihood . -4l4.74
Prob > chi2 0.00 0.00
Pseudo R2 0.52 0.56
Number of obs. 918 912

 

Note: * (**) indicates an estimate significant at the p < .05 level (p < .10 level).

104

Kim Dae Jung’s economic policies had been considered too radical against the
chaebol (conglomerates), which may have worked against him in previous elections. In
1997, however, his reforrn-minded economic policies became very attractive to voters.
People believed that the chaebol-oriented economic system of South Korea caused the
financial crisis. Kim Dae Jung took advantage of his image as the pioneer of democracy
and economic reform as well. His publications on the economy also appealed to voters
but neither Lee Hoe Chang, nor Lee In Je could show their economic specialties. In brief,
the economic crisis caused people to draw a different evaluation for Kim Dae Jung.

The same pattern seen in voter choice between Kim Dae Jung and Lee Hoe Chang
is observed in the Lee In Je versus Lee Hoe Chang competition. The national economic
condition and prospective voting variables reached statistical significance for voter choice
between them. The directions of the coefficients also turn out to be correct.

In contrast, no socioeconomic coefficients appear to have made differences in
voter decisions across the three leading candidates. On the other hand, party
identification played a significant role in voter choice. The ruling party sympathizers
were more likely to vote for the incumbent party’s Lee Hoe Chang, while the opposition
party identifiers were more likely to support the opposition contenders.

The combined model exhibits almost the same effects of economic voting,
socioeconomic factors, and party identification, as in the basic model. In addition, the
combined model confirms that regionalism had a statistically significant influence on
voter choice. Kim Dae Jung did better than Lee Hoe Chang among Cholla people, while
Lee Hoe Chang did better than Kim Dae Jung among Kyungsang people. Lee Hoe Chang

was the presidential candidate of Kyungsang province-based ruling party. Between Lee

105

Hoe Chang and Lee In Je, Cholla people were more likely to opt for Lee In Je than Lee
Hoe Chang but Kyungsang people’s support for Lee Hoe Chang was not significant. In
other words, while Cholla people’s regionalism-induced voting behavior persisted,
Kyungsang people’s support for the ruling party became insignificant. This is one of the
main causes of Lee Hoe Chang’s defeat. Also, those who supported the coalition
between Kim Dae Jung and Kim Jong Pil tended to vote for Kim Dae Jung, instead of
Lee Hoe Chang. Those who disliked the strategic alliance cast their votes for Lee Hoe
Chang rather than Lee In Je.

Likelihood ratio (Pr [R 2 90.5, df = 4]) tests reject the null hypothesis that there is
no difference between the basic model and the combined model at a .005 level of
significance. The results of joint tests indicate a substantial effect of economic voting
throughout the models (see Table 5-3). Also, the regionalism hypothesis and the party
identification hypothesis, respectively, were jointly significant in making voter decisions.

In understanding the dynamics of the 1997 election, it is necessary to consider the
impact of the strategic coalition between Kim Dae Jung and Kim Jong Pil. The alliance
between the two candidates during the campaign was also motivated by their desire to
win. Since the constitution prohibits Presidents from running for a second term, the
situations became very favorable to Kim Dae Jung. His life long rival Kim Young Sam
was not allowed to run for office again and the ruling party chose Lee Hoe Chang as its
presidential candidate. Although he was just a beginner in politics after serving as a chief
Supreme Court justice, Lee Hoe Chang had enormous advantages as the ruling party’s

(GNP) presidential candidate. The financial crisis, however, backfired on him. To make

106

matters worse, Lee In Je defected from the GNP after he had lost the primary election and
he formed his own party. The election campaign became a neck-and-neck race between
Lee Hoe Chang and Kim Dae Jung.

Under these circumstances, a party merger or coalition with other candidates was
most wanted by Lee Hoe Chang and Kim Dae Jung. In contrast, Kim Jong Pil had no
realistic chance of winning but could manage to be in the position of tipping the balance
in favor of one of the two leading contenders. Kim Jong Pil garnered 8.1% of the popular
vote in the 1987 Presidential election and his parties earned 15.6-16.2% of the vote in
previous National Assembly elections. Kim Jong Pil left Kim Young Sam’s DLP in
1995, thus it was not easy for him to join the ruling party again. Also, Lee Hoe Chang
was not interested in a coalition with an old generation politician like Kim Jong Pil.
Instead, he made a coalition with Cho Soon, a Seoul National University Economics
professor who later became a politician. He had a good reputation as a professor and was
considered adept at handling the economic crisis.

On the other hand, Kim Dae Jung was more than willing to do anything to win the
election. Since he was over seventy, the 1997 election was likely to be his last chance to
run for the presidency. He chose to form a coalition with Kim Jong Pil. This was ironic
because Kim Jong Pil, a former director of the Korean Central Intelligence Agency, even
attempted to kill Kim Dae Jung decades earlier. Moreover, Kim Dae Jung (an advocate
of presidentialism) agreed with Kim J ong Pil (a proponent of parliamentarism) to adopt a
parliamentary system during his tenure in office, if elected. The constitutional change
would give Kim Jong Pil the Prime Minister position in the coming years. The strategic

alliance with Kim Jong Pil was viewed favorably by Kim Dae Jung because this would

107

also help him to dampen his progressive image. These were the best payoffs the two
politicians could offer one another.

Kim Dae Jung finally won the presidency with 40.3% of the vote after failing
three campaigns. Lee Hoe Chang and Lee In Je received 38.7%, and 19.2% of the vote,
respectively. The razor-thin difference between the two front-runners’ electoral returns
suggests that if there had been no coalition between Kim Dae Jung and Kim Jong Pil,
Kim Dae Jung would have lost again. Kim Jong Pil also became the Prime Minister
under Kim Dae Jung. Nevertheless, the promise of parliamentarism has not been kept

thus far.

4. Economic Voting and Regionalism

I make a comparison between the magnitudes of economic voting and regionalism
in South Korea’s elections. Since the multinomial logit model is curvilinear, the effect of
a coefficient depends upon other coefficients. The magnitude of the marginal impact of
an independent variable is indicated by “first differences.” With all other variables being
held constant, first differences are calculated by the value changes of the interested
coefficient from one end of the code to the other. I use the program, Clarify (Tomz,
Wittenberg, and King 1999) in order to efficiently calculate first differences. I set this
program to calculate the change in the simulated probabilities, given a change in an
interested variable from its lowest code to its highest code (see Appendix C). Tables 5-7
and 5-8 display first differences of electoral returns for each party (or candidate). The far

left column of Tables 5-7 and 5-8 illustrates the independent variables of interest. The

108

next three columns illustrate first differences of support for each of the three parties in the
1992 National Assembly election (Table 5-7) and the 1992 Presidential election (Table 5-
8). The last three columns contain first differences of support for each of the three parties
in the 1996 National Assembly election (Table 5-7) and the 1997 Presidential election
(Table 5-8).

First differences enable us to understand the probability that a hypothetical
individual would vote for each of the three parties with regard to a certain variable. The
hypothetical individual, here, is the average respondent in the South Korean surveys. The
person is a male in his thirties who has graduated from high school and lives in an urban
area. In the 1992 National Assembly election, for example, the hypothetical individual
had a .42 probability of voting for the ruling DLP, a .40 probability of voting for the DP,
and a .18 probability of voting for the UPP (see Table 5-7). If the hypothetical voter
moved from the perception that the personal economy worsened to the perception that it
improved over the years, the voter’s probability of supporting the ruling DLP and the
opposition UPP would have increased by 6% and 2%, respectively. His probability of

voting for the genuine opposition DP would have decreased by 8%.

109

Table 5-7. First Differences of Economic Voting, Party Identification, and Regionalism
in the National Assembly Elections, South Korea

 

 

1992 1996
DLP DP UPP NKP NCNP ULD
Baseline .42 .40 .18 .67 .21 .12
Prob.
Economics
National N/A N/A N/A .30 -.20 -.10
economy
Inflation .08 -.09 .01 N/A N/A N/A
Personal .06 -.08 .02 .03 -.02 .00
economy
Unemploy- .07 -.01 -.06 -.10 .10 .00
ment
Retro- -.13 .03 .10 -.06 .04 .02
spective
PID
Ruling .30 -.20 -.09 .23 -.17 -.06
Opposition -.46 .41 .06 -.52 .35 .17
Regionalism
Cholla -.24 .48 -.24 -.37 .38 -.01
Kyungsang .22 -.13 -.10 .18 -.22 .04
Merger -.O7 .02 .05 N/A N/A N/A
(Support)
Merger -.24 .20 .04 N/A N/A N/A
(N 0 Support)

 

Note: I use Clarify (Tomz, Wittenberg, and King 1999) in calculating first differences.

The magnitudes of economic voting’s marginal effects are meager and the signs
of first differences of inflation, unemployment, and retrospective evaluation are not in the
expected directions. In sharp contrast, first differences of regionalism stand out. To be
more specific, Cholla people were more likely to be driven by regionalism. If the
hypothetical voter was from Cholla province, he was 24% less likely to vote for
Kyungsang province-based DLP and the UPP, respectively. Yet, he was 48% more likely

to vote for his DP. If the hypothetical voter was from Kyungsang province, he was 22%

110

more likely to support his DLP. Yet, he was 13% and 10% less likely to support Cholla
province-based DP and the UPP, respectively. Also, the changes in his perceptions
regarding the three party merger generated a decline in support for the ruling DLP and an
increase in support for the opposition parties, which contradicts the hypotheses.
Nonetheless, this may not be problematic because the three party merger was not
unanimously favored by their regional voters. Some may have disliked the merger
because of their unique preferences about other regional parties. Also, the survey was
conducted from all around the country except the Island of Jeju.

Additionally, party identification had a very large marginal effect. If the
hypothetical voter became a ruling party sympathizer, his probability of supporting the
ruling DLP would have increased by 30%. His probability of supporting the opposition
DP and UPP would have dropped by 20% and 9%, respectively. If the voter became an
opposition party sympathizer, he was 41% and 6% more likely to vote for the DP and the
UPP, respectively, whereas he was 46% less likely to vote for the ruling DLP.

This pattern remains similar in the 1996 National Assembly election (see Table 5-
7). First differences of the economic voting variables are smaller than those of the
regionalism and party identification variables. Also, the signs of the economic voting
variables are not always in the expected directions. The sign of the retrospective voting
variable is in the wrong direction but the regionalism and party identification variables
have the expected signs. Moreover, Cholla people were more likely to be affected by

regionalism than Kyungsang people.

111

Table 5-8. First Difference of Economic Voting, Party Identification, and Regionalism in
the Presidential Elections, South Korea

 

 

1992 1997
KYS KDJ CJ Y LHC KDJ LIJ
Baseline .59 .28 .13 .33 .40 .27
Prob.
Economics
National .24 -.26 .02 -.17 .10 .07
economy
Inflation .14 -.15 .01 .06 .00 -.06
Personal -.12 .17 -.05 .06 -.17 .11
economy
Unemploy- .12 -.1 l -.01 .19 -.06 -.13
ment
Retrospective -.05 .02 .04 .l 5 -.05 -.1 1
Prospective N/A N/A N/A .60 -.38 -.22
(LHC)
Prospective N/A N/A N/A -.20 .32 -. 12
(KDJ)
Prospective N/A N/A N/A -.32 -.25 .57
(LIJ)
PID
Ruling .25 -.23 -.02 .39 -.27 -.11
Opposition -.43 .36 .07 -.28 .17 .1 l
Regionalism
Cholla -.48 .69 -.21 -.32 .28 .03
Kyungsang .21 -.17 -.05 .23 -.31 .08
Merger .19 -.18 -.01 -.11 .17 -.06
(SUPport)
Merger -.18 .17 .01 .ll -.01 -.10
(No Support)

 

Note: I use Clarify (Tomz, Wittenberg, and King 1999) in calculating first differences.

As in the National Assembly elections, first differences of the economic voting
variables in the 1992 Presidential election are smaller than those of the regionalism and
party identification variables (see Table 5-8). The signs of some economic voting
variables (inflation, the personal economy, unemployment, and retrospective voting) are

not in the expected directions. On the other hand, if the hypothetical voter came from

112

Cholla province, he was 48% less likely to vote for Kim Young Sam and 69% more
likely to vote for Kim Dae Jung. These high percentage points suggest not only that
regionalism was one of the strongest factors but also that Cholla people were more likely
to be regionalism-driven voters. If he came from Kyungsang province, he was 21% more
likely to vote for Kim Young Sam and 17% less likely to vote for Kim Dae Jung. Also,
the three party merger factors generate the anticipated effects. If the hypothetical voter
supported the three party merger, he was 19% more likely to vote for the merged party’s
Kim Young Sam and 18% less likely to vote for Kim Dae Jung. If he did not support the
merger, he was 18% less likely to vote for Kim Young Sam and 17% more likely to vote
for Kim Dae Jung. Additionally, the party identification variables have very large first
differences, which are also in the expected directions.

The same pattern appears in the 1997 Presidential race (see Table 5-8). First
differences of the economic voting variables are smaller than those of the regionalism and
party identification variables. The exception to this is first differences of prospective
economic voting. If the hypothetical voter changed his perception from Lee Hoe Chang
as being the worst candidate for resolving the economic crisis to the opposite, he was
60% more likely to vote for Lee Hoe Chang. He was also 38% and 22% less likely to
vote for Kim Dae Jung and Lee In Je, respectively. In addition, the same change in his
perceptions toward Kim Dae Jung would have generated a 32% increase in Kim Dae
Jung’s support. In this case, there would be a 20% and a 12% decrease in Lee Hoe
Chang’s and Lee In Je’s support, respectively. On the other hand, first differences of
other economic voting are smaller and sometimes have the wrong signs (the national

economic condition, inflation, and unemployment variables). In contrast, as expected, if

113

the hypothetical voter had the perception that the ruling party was responsible for the
economic crisis, he was 17% less likely to vote for Lee Hoe Chang. He also 10% more
likely to vote for Kim Dae Jung and 7% more likely to vote for Lee In Je.

As in other elections, Cholla people were slightly more regionalism-induced
voters than their Kyungsang counterparts in 1997. If the hypothetical voter came from
Cholla province, he was 28% more likely to vote for Kim Dae Jung, while he was 32%
less likely to vote for Lee Hoe Chang. In contrast, if he was from Kyungsang province,
he was 23% more likely to vote for Lee Hoe Chang and 31% less likely to vote for Kim
Dae Jung. The merger variables had weaker impacts than the home providence variables.
The signs of merger variables’ first differences are also in the hypothesized directions. If
the hypothetical voter supported the alliance between Kim Dae Jung and Kim Jong Pil, he
was 17% more likely to vote for Kim Dae Jung and 11% less likely to vote for Lee Hoe
Chang. Otherwise, he was 1% less likely to vote for Kim Dae Jung and 11% more likely
to vote for Lee Hoe Chang. In addition, party identification had a strong and consistent

impact.

5. Conclusions

The statistical analyses lead me to the following conclusions. First, economic
voting seems to have occurred throughout the elections under study. In the elections held
between 1992 and 1996, however, the signs of the economic voting coefficients
sometimes have the directions opposite to the hypotheses. This is because of voters’

perceptions regarding the ruling party’s roles that were played in South Korea’s economic

ll4

development. They perceived that the ruling party was more capable of managing the
economy and resolving economic problems. In other words, economic voting occurred
with voters’ limited rationality between 1992 and 1996. Also, joint tests does not
approve of the economic voting hypothesis in some elections, for instance, the 1992
National Assembly election (the DP versus DLP comparison in the combined model) and
the 1992 Presidential election (the Chung Ju Young versus Kim Young Sam competition
in the basic and combined models). However, genuine economic voting is observed in
1997. This confirms Lee’s (1998) conclusions about economic voting in 1997. All the
coefficients are in the expected directions and the economic condition hypothesis is
jointly significant on voter choice among the candidates. In brief, voters held the ruling
party responsible for the financial crisis in 1997.

Second, sociotropic voting was more likely to occur than pocketbook voting
throughout the elections under investigation. Pocketbook voting is observed only in the
1992 Presidential election. On the other hand, retrospective voting is directly and
indirectly found in the elections under investigation. Moreover, prospective voting is
significant in 1997. These findings are also consistent with Lee’s (1998) conclusions
about the nature of economic voting in 1997.

Third, it is not safe to say that economic voting was stronger in Presidential race
than National Assembly elections. There are no significant differences in the joint test
chi square values. In contrast, economic voting may have been stronger between the
ruling party versus the first opposition party competition than otherwise. Two out of the
three insignificant joint test results regarding the economic voting hypothesis are

observed in the competition between the ruling party and the second opposition party.

115

 

Fourth, regionalism was stronger and more consistent than economic voting. This
conclusion is more or less the same as in previous research (Lee 1993; Morriss 1996;
Park 1993a, 1993b). First differences of the regionalism variables are bigger than those
of economic voting across the elections. The signs of the home province and party
merger variables are not in unexpected directions. Kim Dae Jung or his parties collected
Cholla voters’ support more than the candidates from Kyungsang province collected
Kyungsang voters’ support throughout the elections. In addition, joint tests for the
regionalism hypothesis are not significant only between the NKP (Kyungsang province)
and the ULD (Choongchung province) in the 1996 National Assembly election.
Moreover, likelihood ratio tests suggest that the combined model is significantly better
than the basic model across the elections.

Finally, party identification also induced a consistent impact on voter choice,
while the socioeconomic factors did not generate such effects. Although South Korea’s
party identification is not similar to genuine party identification as in other countries, the
identification between the ruling party versus opposition parties appears to have always
worked as a significant cue for voter decisions. In contrast, different socioeconomic

factors were significant in different elections.

116

_ _m... ml

CHAPTER VI

ECONOMIC VOTING IN TAIWAN

1. Model Specification and Methodology

Chapter VI investigates economic voting in Taiwan. In addition, this Chapter
compares the impact of economic voting on voter choice with that of ethnic voting. The
dependent variable of the statistical models is voter choice. As can be seen in Table 6-1,
there were only two major parties (the KMT and the DPP) in the 1992 Legislative Yuan
election. I use probit for this election. The choice for the ruling KMT is coded as 1 and
the opposition DPP is coded as 0. The coefficient estimates are viewed to be the
probabilities of voting for the KMT. When the sign of a coefficient is negative,
respondents are interpreted to vote for the DPP. When the sign of a coefficient is
positive, respondents are interpreted to vote for the KMT. On the other hand, I use
multinomial logit for the three-way races in 1995 and 1996, normalizing coefficients for
the ruling party (the KMT) to zero. The coefficients for the two opposition parties in the
estimates are the impact of each variable on voter choice, relative to a vote for the ruling
party. When the sign of a coefficient is negative, respondents are interpreted to vote for
the ruling party. Otherwise, respondents are interpreted to support one of the opposition
parties.

The independent variables of the basic model include the measurements of
sociotropic voting, pocketbook voting, retrospective voting, prospective voting,

socioeconomic factors, and party identification. The combined model consists of the

117

basic model, ethnic background, and the individual election specific event (tension
between Taiwan and China). This combined model probes the impact of ethnic voting on

voter choice.

Table 6-1. Vote Distribution in the Taiwanese Elections

 

 

Vote Share
Legislative Yuan Elections Presidential Election

1992 1995 1996
lst Party KMT-53.0% KMT-46. 1% LTH-54.0%
2nd Party DPP-31.0% DPP-33.2% PMM-2l.l%
3rd Party CSDP-1.3% NP-13.0% LYK-14.9%
4th Party IND-5.6% IND-7.8% CLA-10.0%

Total* 84.0% 92.3% 90.0%

 

Note: * The first percentages in the bottom cells are the total vote shares of the first two
(three) parties (or candidates) in 1992 (1995 and 1996).
Sources: Asian Survey (various issues) and Journal of Democracy (various issues).

To measure sociotropic voting, I employ the national economic condition
variable. If a respondent perceived the national economy as being worse than before, the
respondent was more likely to support an opposition party. For pocketbook voting, I use
the personal economic condition and unemployment variables. If a respondent answered
that the personal economic condition was worse than before or that the individual was
unemployed, the respondent was expected to prefer an opposition party. These
sociotropic and pocketbook voting variables are, in nature, the retrospective economic
voting survey items.

The NCU surveys first included the prospective economic voting question in the
1995 Legislative Yuan election survey. So, for the 1992 election, I use the question, “Is

the democratic development in Taiwanese politics promising?” to indirectly measure

118

.11" ”W.

prospective economic voting by asking the prospective expectation of the government’s
general performance. On the other hand, the prospective voting questions for the 1995
Legislative Yuan election aim at the two levels of sociotropic and pocketbook voting,
“How do you think national (or household) economic conditions are going to be in one
year?” For the 1996 Presidential race, I utilize the question, “Who do you think will best
be able to generate economic development?” I treat the three candidates (Lee Teng-hui,
Peng Ming-min, and Lin Yang-kang) as dummy variables, leaving the other choices as
the omitted category.

In order to measure the impact of the socioeconomic factors on voter choice, I use
the variables of age, gender, and educational attainment. Since more educated people are
more likely to be liberal than their less educated counterparts, they are expected to choose
an opposition party over the ruling party. Since the younger generations are more
receptive to change and the new democratic values than older people, they are more likely
to support an opposition party than the ruling party. I recode the age variable as l = 20-
29, 2 = 30-39, 3 = 40-49, and 4 = 50 and older. Since men are more likely to be liberal
than women, men are expected to vote for an opposition party. I use gender (male) as a
dummy variable. I also include party identification, treating party identification (the
KMT, the DPP, and the NP) as dummy variables, leaving the other choices as the base
category.

Ethnic voting is measured by the dummy variables of the Taiwanese and the
Mainlanders, with the other choices being the omitted category. If a respondent shares
the same ethnic background as the party (or the leader), the respondent is expected to vote

for this party. The individual election specific event is the cross-strait tension, which is

119

included in the statistical model for the 1996 election. The military threat from the
Mainland would generate citizens’ support for the ruling party. This is the “rally behind
the flag effect,” which was originally named by Mueller (1970). The opponents of a
government turn to the government if there is an imminent crisis involving foreign
affairs. If a respondent answered “disagree” to the 1996 survey question, “Did
[incumbent] Lee Teng-hui provoke the current cross-strait tension?” the respondent was
more likely to opt for him than his opponents. The answer was regarded as support for
the ruling party. If a respondent answered “agree” to the question, the respondent was

less likely to vote for him than his opponents.

2. Statistical Models

A basic probit function and a combined probit function for the two-way 1992

election are, respectively, as follows:

P (VC2=l\x) =<I> (S, P, PRO, SOE, and PID) and (12)

P (VC3=l\x) =<D (S, P, PRO, SOE, PID, ETH, and IE), (13)

where VC, and VC2 are voter choice; S is sociotropic voting; P is pocketbook voting;
PRO is prospective voting; SOE is socioeconomic factors; PID is party identification;
ETH is ethnic background; and IE is the individual election specific event.

The combined probit model, for example, is written as the following equation:

120

P (VC3=1\X) = (D “3300+ B301X30| + B302x302 + p303x303 + B304X304 + B305X305

+ p306x306 + p307X30-7 + B308x308+ B309X309+ fi310x310+ p3llx3ll + U1), (14)

where VC3 = voter choice; [3300 = intercept; X3,,, = national economic condition; X302 =
personal economic condition; X303 = unemployment; X304 = prospective evaluation; X305 =

age; X306 = gender (male); X307 = educational attainment; X3,08 = party identification (the

 

KMT); X309 = party identification (the DPP); X3,0 = ethnic background (Taiwanese); X,,,
= ethnic background (Mainlander); and U, = error term. The basic model includes only
X30, through X309.

A basic multinomial logit function and a combined multinomial logit function for

the three-way race in 1995 and 1996 are, respectively,

vc, = f(S, P, PRO, 5015, and PID) and (15)

vc, = f(S, P, PRO, 5013, PID, ETH, and IE), (16)

where VC4 and VCs are voter choice; S is sociotropic voting; P is pocketbook voting;
PRO is prospective voting; SOE is socioeconomic factors; PID is party identification;
ETH is ethnic background; and IE is the individual election specific event. Function f is
assumed to be polychotomous.

The combined model, for example, is transformed into the following equation:

121

VCS = f(fi500+ flSOIXSOI + [5502x502 + p503X503 + B504x504 + 0505x505
+ [3506x506 + [3507x507 + [3508x508 + B509X509 + BSIOXSIO + B51 IXSII

+ BSI2XSIZ+ BSI3XSI3+ 5514X514+ B515X515+ Ui)’ (17)

where VC5 = voter choice; [3500 = intercept; X,,,, = national economic condition; X502 =
personal economic condition; X503 = unemployment; [X504 = the prospective evaluation of
national economic condition; X5,), = the prospective evaluation of personal economic
condition in the 1995 election]; [X,,, = prospective evaluation (Lee Teng-hui); X5,), =
prospective evaluation (Peng Ming-min); X506 = prospective evaluation (Lin Yang-kang)
in the 1996 election]; X507 = age; X5,), = gender (male); X,,, = educational attainment; X,,0
= party identification (the KMT); X,,, = party identification (the DPP); X,,,2 = party
identification (the NP); X5,3 = ethnic (Taiwanese); X5,4 = ethnic (Mainlander); [X,,5 =
tension in the 1996 election]; and U, = error term. The basic model includes only X50,

through X,,)

3. Economic Voting in Taiwan

A. The Legislative Yuan Elections

The December 1992 Legislative Yuan Election. The probit estimates for the
1992 Legislative Yuan election are presented in Table 6-2. The center column displays
the coefficient estimates for the basic model and the right column displays the parameter

estimates for the combined model. Standard errors are in parentheses.

122

The basic model indicates that no economic voting coefficients have made
significant differences in choosing the KMT. Of the three socioeconomic factors, only
age had a significant influence on voter choice for the KMT. As expected, older people
were more likely to vote for the ruling KMT than the DPP, ceteris paribus. Also, party
identification generated the expected impact on voter choice. KMT supporters were more
likely to opt for the KMT than the DPP. DPP sympathizers were less likely to vote for
the KMT than the DPP.

The estimates for the combined model suggest a similar pattern as seen in the
basic model. The economic voting variables do not have a significant impact on the
choice between the KMT and the DPP. Older people were more likely to vote for the
ruling KMT than the opposition DPP. In addition, party identification had the anticipated
influence on voter choice, ceteris paribus. KMT supporters were more likely to vote for
the KMT than the DPP. DPP identifiers were less likely to support the KMT than the
DPP. On the other hand, ethnic cleavages did not play a significant role in 1992 in

deciding which party to vote for, which does not confirm the hypothesis.

123

Table 6-2. Probit Estimates for the Two-Party Race: The 1992 Legislative Yuan

Election, Taiwan

 

 

Independent 1 992 l 992
Variables Basic Combined
Economics
National .15 .15
economy (.09) (.09)
Personal -.1 1 -.09
economy (.12) (.12)
Unemployment .26 .23
(.45) (.45)
Prospective -.07 -.07
evaluation (. l 0) (. l 0)
Socioeconomic
Age .18* .19*
(.07) (.07)
Gender .01 .01
(Male) (.14) (.14)
Educational . 1 0 .09
attainment (.07) (.07)
PID
KMT 1.60* 1.57*
(.16) (.16)
DPP -l .70* -1.71*
(.19) (.20)
Ethnic
Taiwanese -.08
(.19)
Mainlander -.05
(.16)
Intercept -.23 -.22
(.42) (.44)
Log Likelihood -220.77 -2l6.23
Prob > chi2 0.00 0.00
Pseudo R2 0.49 0.49
Number of obs. 717 707

 

Note: * indicates an estimate significant at the p < .05 level.

124

Likelihood ratio (Pr [R 2 9.08, df = 2]) tests approve, at a .05 level of
significance, the alternative rival hypothesis that the combined model is significantly
different from the basic model (see Table 6-3). Joint test results for the combined model
display that the economic voting hypothesis and the ethnic voting hypothesis,
respectively, are not jointly significant. Also, the joint effect of economic voting was not
significant in the basic model. In contrast, the party identification hypothesis appears to

be jointly significant throughout the models.

Table 6-3. Joint Test Results of the Probit Estimates: The 1992 Legislative Yuan
Election, Taiwan

 

 

Hypothesis Basic Combined
KMT DPP
Economics 0.50 0.57
Ethnic 0.90
Party ID 000* 0.00*

 

Note: Prob > chi2
* indicates an estimate significant at the p < .05 level.

The December 1995 Legislative Yuan Election. The multinomial logit
estimates for the 1995 Legislative Yuan election are in Table 6-4. It should be noted,
however, that the 1995 survey data may be problematic. Although the total number of
the respondents is 1485, 964 of the interviewees did not answer their voter choice.
Moreover, the valid responses were only 89 and 87 in the basic model and the combined
model, respectively. The format of Table 6-4 (and 6-5) are the same as the multinomial

logit estimates for the three-way races in South Korea.

125

The estimates of the basic multinomial logit model indicate that voters were not
affected by the economic conditions (see Table 6-4). No economic voting factors reached
statistical significance for voter choice. In addition, no socioeconomic factors appear to
have had a significant influence on voter choice, with the exception of educational
attainment. The opposition NP did better than the ruling KMT among educated people.
In fact, the NP came out of the KMT in 1993 and was based on voters with higher
education.32

The party identification variables were significant for voter choice. Between the
DPP and the KMT, KMT sympathizers were expected to vote for the KMT and DPP
identifiers were more likely to support the DPP. For the NP versus KMT comparison, the
NP received support from their NP identifiers, as expected. Nonetheless, the results of
joint tests suggest that the economic voting hypothesis in the basic model is jointly
significant for the NP versus KMT competition, while this is not the case for the DPP

versus KMT comparison (see Table 6-5).

 

32 The NP “has drawn more support from the metropolitan areas college-educated people the younger
generation female the middle class [than the KMT]” (Tien and Chu 1996, 1159).

126

Table 6-4. Multinomial Logit Estimates for the Three-Party Race: The 1995 Legislative

Yuan Election (Coefficients for the KMT’s Normalized to Zero), Taiwan

 

 

Independent Basic Basic Combined Combined
Variables DPP NP DPP NP
Economics
National .23 22.80 .07 34.25
economy (1.21) (7456.66) (1.57) (7432.78)
Personal -. -33.44 - 92 -44.35
economy (.94) (.) (1.15) (.)
Unemployment -2.67 -54.76 -3.60 -74.81
(2.16) (3728.36) (2.44) (3716.42)
Prospective -.66 -4.47 - 96 -4.39
National (.80) (4.55) (1.01) (4.69)
Prospective -.17 6.16 - 40 6.32
Personal (.86) (7.54) (1.04) (7.66)
Socioeconomic
Age -.66 .41 -.52 .59
(.68) (1.53) (.82) (1.56)
Gender .39 3.96 .06 3.77
(Male) (.99) (4.47) (1 .12) (4.56)
Educational .19 36.31* .84 47.56*
attainment (.71) (3.40) (.84) (3.43)
PID
KMT -3.75* -5.79 -3.65* -5.87
(1.36) (5.89) (1.41) (5.99)
DPP 3.23* -71 .79 4.13* -63.51
(1.35) (2.29e+09) (1.71) (l .95e+08)
NP .59 39.08* -40.42 49.62*
(1.60) (5.58) (9.57e+08) (5.63)
Ethnic
Taiwanese 1.57 -l 1.84
(1.19) (7433.01)
Mainlander -.97 .01
(2.82) (2.50)
Intercept 8.40 .19 8.84 5.10
(5.94) (6.83) (.)
Log Likelihood -25.57 -22.02
Prob > chi2 0.00 0.00
Pseudo R2 0.63 0.67
Number of obs. 89 87

 

Note: * indicates an estimate significant at the p < .05 level.

127

The estimates of the combined model are almost the same as those of the basic
model. The combined model suggests that no economic condition factors reached
statistical significance for voter choice. Yet, joint tests of the economic condition
hypothesis confirm its joint significance for the NP versus KMT comparison (see Table
6-5). Educational attainment was the only socioeconomic variable that had a statistically
significant impact on voter choice. For the choice between the NP and the KMT, the NP
did better among educated people. In addition, the party identification variables were
significant for voter choice. As expected, KMT identifiers were more likely to vote for
the KMT than the DPP. DPP supporters were more likely to opt for the DPP than the
KMT. Between the NP and the KMT, NP sympathizers were more likely to support the

NP than the KMT.

Table 6-5. Joint Test Results of the Multinomial Logit Estimates: The 1995 Legislative
Yuan and the 1996 Presidential Elections, Taiwan

 

 

Election Hypothesis Basic Basic Combined Combined
Model Model Model Model

1995 DPP/KMT NP/KMT DPP/KMT NP/KMT
Legislative Economics 0.58 0.00* 0.51 0.00*

Yuan Ethnic 0.38 1 .00

Party ID 000* 0.00* 0.01* 0.00*

1 996 PMM/LTH LYK/LTH PMM/LTH LYK/LTH
Presidential Economics 000* 0.00* 0.01 * 0.00*
Ethnic 0.08** 000*
Party ID 000* 0.00* 0.00* 0.00*

 

Note: Prob > chi2
* (**) indicates an estimate significant at the p < .05 level (p < .10 level).

Likelihood ratio (Pr [R 2 7.1, df = 2]) tests approve the alternative rival

hypothesis that the combined model is significantly different from the basic model at a

.05 level of significance. Contrary to the expectations, the estimates of the combined
model indicate the insignificant impact of the ethnic cleavage hypothesis on voter choice
(see Table 6-5). In contrast, the party identification hypothesis reached joint significance.

In accounting for the 1995 election, however, the individual specific event of the

’3 The cross-strait tension between Taiwan

cross-strait tension deserves more emphasis.
and China has resulted from a rather long historical relationship between the two
governments. Ever since Chang Kai-shek moved to Taiwan from the Mainland in 1949,
Taiwan has proclaimed to recover the Mainland and restore the previous ruling system
over the nation’s old territory. On the other hand, China continues to claim to be the only
legitimate government of the Mainland. China has constantly pursued the diplomatic
isolation of Taiwan and has repeatedly threatened to use military force, if Taiwan seeks
formal independence. In contrast, Taiwan did everything to maintain international ties
with other countries in the world.

The irreconcilable conflicts between the two governments on both sides of the
Taiwan Strait began to abate after 1987. Starting in October 1987, the Mainlanders who
moved to Taiwan in the 19405 were allowed to meet their families on the Mainland.
There had been an enormous increase in trade, tourism, family visits, and official contacts
as well as sports and cultural exchanges between the two governments. Moreover,
China’s President Jang Zemin called for a meeting to negotiate an end to the conflicts in

January 1995. Nevertheless, the peaceful atmosphere broke down when Lee Teng-hui

went to the US. in June of 1995.

 

3’ The inclusion of the tension variable generates strange estimates (see Appendix D). Thus, I take out the
variable from the combined model.

129

lt“ “

Although Lee Teng-hui, born in Taiwan, appeared to be sympathetic of the notion
of Taiwan’s independence, he could not outspokenly pursue independence as the leader
of the Mainlander-based KMT. Yet, he felt the need to ease pressure from the DPP’s
independence policy by taking a more aggressive stance toward independence than the
traditional position taken by the KMT. In June 1995, he attempted to enhance Taiwan’s
international visibility by visiting the US. As a result, China feared that he would
declare independence based on the increasing U.S. sympathy toward Taiwan after his
visit. Moreover, China suspected that “Washington had shifted to a policy of
‘containing’ China and playing the Taiwan card” (Tien 1996, 36). Thus, China tried to
disrupt Taiwanese ethnic Lee Teng-hui’s reelection by a military threat.

The People’s Liberation Army of China began to conduct a series of military
exercises in the Taiwan Strait between July and November of 1995. The cross-strait
relationship seriously worsened after the visit. Nevertheless, those who supported Lee
Teng-hui’s KMT did not complain about the increased tension. Instead, his supporters
may have shown their backing of the leader and supported his inter-China policies.
Contrary to China’s expectations, Lee Teng-hui’s popularity went up and the KMT
mustered the majority of seats (85 seats out of 164) in the 1995 December Legislative
Yuan election (see Table 6-1). In contrast, the DPP won 54 seats and the NP earned 21
seats.

Table 6-1 indicates that there was a decline in support of the KMT between 1992
and 1995. Yet, the decline was affected by the formation of the NP in 1993, not by the
decrease in support for Lee Teng-hui. The base of the NP used to be that of the KMT.

The NP garnered its support from the old Mainlanders who believed that Lee Teng-hui

130

was leading the country on a path toward war and away from unification. Considering
that the vote and seat shares of the DPP remained almost the same between 1992 and
1995, the NP’s vote share of 13 percentage points may have come from the KMT’s
supporters (see Table 6-1). Thus, the seven percentage point decline in the KMT support

may not be serious.

B. Presidential Election

The March 1996 Presidential Election. As a result of the 1992 constitutional
reform, a direct Presidential election was held for the first time in Taiwan’s history in
1996. There were three leading contenders in the race. Lee Teng-hui was the successor
of the late President Chang Ching-kuo in 1988 and took over the presidency through an
indirect election in the National Assembly in 1993. Peng Ming-min was known for his
life long pursuit of Taiwan’s independence and returned home in 1992 after a more than
two decade exile. He became the presidential candidate of the DPP. Lin Yang-kang had
been the vice chairperson of the KMT but left the party to run against the incumbent Lee
Teng-hui in 1995. Chen Li-an also came out of the KMT and attacked Lee Teng-hui’s
cross-strait policies. However, the NP endorsed Lin Yang-kang as the party’s
presidential candidate, instead of Chen Li-an.

The multinomial logit estimates of the basic model demonstrate that voters cast
their votes according to their sociotropic concerns (see Table 6-6). Between Lin Yang-
kang and Lee Teng-hui, the national economic condition worked as a positive factor for
Lee Teng-hui. Those who perceived that the national economy was better than before

were more likely to vote for the incumbent Lee Teng-hui than Lin Yang-kang. In

131

if?" . I

addition to such retrospective evaluations, individuals were also concerned about the
prospective promises of the candidates. Those who believed that Lee Teng-hui would
best be able to generate economic development tended to choose him over his opponents.
On the other hand, the other economic voting coefficients did not reach statistical
significance for voter choice.

Age and educational attainment were the socioeconomic factors that had
significant effects on voter choice. Lee Teng-hui appealed to older people more than
Peng Ming-min, as expected. Also, the two opposition candidates did better than the
incumbent candidate among educated voters, which is consistent with the expectations.
Party identification also played a significant role in choosing candidates. KMT
identifiers were more likely to vote for Lee Teng-hui and DPP supporters were more
likely to choose Peng Ming-min, between the two candidates. Also, NP supporters were
more likely to vote for Lin Yang-kang than Lee Teng-hui, with all other things being

equaL

132

 

 

Table 6-6. Multinomial Logit Estimates for the Three-Candidate Race: The 1996
Presidential Election (Coefficients for LTH’s Normalized to Zero), Taiwan
Independent Basic Basic Combined Combined
Variables PMM LYK PMM LYK
Economics
National - .25 -1. 13* -.08 -1. 03*
economy (. 24) (. 37) (.27) (40)
Personal 36 -.49** 35
economy (. 24) (. 32) (.26) (. 33)
Unemployment -1 .20 - .94 -.51 -1.05
(.90) (1.27) (1.06) (1.45)
Prospective -l. 16* -1.78* -1.38* -l .64*
(LTH) (. 36) (.45) (.40) (. 514)
Prospective .05 .12 - 4.2
(PMM) (. 58) (.92) (. 60) (.97)
Prospective 71 1 .43 .58 .76
(LYK) (1 .23) (.90) (1.22) (.94)
Socioeconomic
Age -.31* -.12 -.37* -.18
(.16) (.20) (.17) (.22)
Gender .57 .49
(Male) (. 31) (. 39) (. 33) (.42)
Educational .43* .41* .31** .14
attainment (. 15) (.19) (.16) (.22)
PID
KMT -3.l4* -.69 -3.20* -.84
(.76) (.47) (.78) (.52)
DPP 2.52* -.21 2.22* -.20
(.34) (.65) (.39) (.68)
NP -.04 4.45* -.04 4.00*
- (1.11) (.61) (1.14) (.66)
Ethnic
Taiwanese .75* -.35
(.36) (.50)
Mainlander - .37 - .60
(63) ( 56)
Tension .17 .97*
(.18) (.23)
Intercept - .79 -1.50 - .46 -2.75*
(. 87) (1.07) (1.03) (1.29)
Log Likelihood -271.18 -229.20
Prob > chi 0.00 0.00
Pseudo R2 0.56 0.58
Number of obs. 852 712

 

Note: * (**) indicates an estimate significant at the p < .05 level (p < .10 level).

133

The economic voting factors in the combined model have almost the same effects
on voter choice as observed in the basic model. In addition, the personal economic
condition variable reached statistical significance. The favorable evaluations of the
personal economic condition led to support for Lee Teng-hui more than Peng Ming-min.
This is consistent with the expectations. Also, sociotropic voting is observed between the
Lin Yang-kang and Lee Teng-hui competition. The sign of sociotropic voting is. in the
anticipated direction. In contrast, educational attainment loses its statistical significance
for the Lin Yang-kang versus Lee Teng-hui comparison, as opposed to the basic model.
Otherwise, the socioeconomic factors have the same effects on voter choice in the
combined model as their counterparts in the basic model. Party identification factors
induced almost the same effects on voter choice as in the basic model.

Clearly, ethnic cleavages had a significant impact on voter choice. The
Taiwanese were more likely to prefer Peng Ming-min (the DPP) than Lee Teng-hui (the
KMT). Moreover, the cross-strait tension induced a significant impact on voter choice.
The tension between Taiwan and China became more serious when China even
conducted missile tests near Taiwan, as the Presidential election day was getting closer.
As expected, those who blamed Lee Teng-hui for the 1996 cross-strait crisis would vote
for Lin Yang-kang more likely than him. Lin Yang-kang’s supporters perceived that Lee
Teng-hui provoked the crisis and as a result, there would be a war between the two
governments.

Likelihood ratio (Pr [R 2 83.96, df = 3]) tests reject the null hypothesis that there

is no difference between the basic model and the combined model at a .005 level of

134

significance. Joint tests for the combined model confirm the joint significance of the
economic voting hypothesis, the ethnic voting hypothesis, and the party identification
hypothesis, respectively (see Table 6-5). In addition, this is the same in the basic model.
Despite the military intimidation, Lee Teng-hui won a landslide victory by
receiving 54.0% of the popular vote in the 1996 Presidential election (see Table 6-1).
Peng Ming-min and Lin Yang-kang earned 21.1% and 14.9% of the vote, respectively.
As Table 6-1 displays, Lee Teng-hui’s vote share of 54.0% was the highest. China’s
missile tests in the Taiwan Strait generated a rally behind the flag effect. Also, Lee Teng-
hui’s independence approach may have attracted some of the DPP’s traditional
supporters. It was the DPP (Peng Ming-min) that experienced a decline in support in

1996. On the other hand, the NP’s vote share (Lin Yang-kang) did not decrease.
4. Economic Voting and Ethnic Voting

First differences of the 1992 Legislative Yuan election indicate that the marginal
effects of economic voting and ethnic voting were meager, while those of party
identification were substantial (see Table 6-7). Also, the signs of party identification’s
first differences are in the correct direction but some signs of economic and ethnic voting
are in the wrong directions. In other words, first differences of the personal economy,
unemployment, prospective evaluation, and Mainlander variables display in the directions

opposite to the hypotheses.

135

Table 6-7. First Differences of Economic Voting, Party Identification, and Ethnic
Background in the Legislative Yuan Elections, Taiwan

 

 

1992 1995
KMT DPP KMT DPP NP
Baseline 1.00 .00 .00
Prob.
Economics
National .08 -.08 .00 .00 .00
economy
Personal -.04 .04 .00 .00 .00
economy
Unemploy- .03 -.03 .00 .00 .00
ment
Prospective* -.04 .04 .00 .00 .00
National
Prospective N/A N/A .00 .00 .00
Personal
PID
KMT .37 -.37 .00 .00 .00
DPP -.55 .55 -.05 .00 .05
NP N/A N/A -.15 .15 .00
Ethnic '
Taiwanese -.02 .02 .00 .00 .00
Mainlander -.01 .01 .00 .00 .00

 

Note: I use Clarijy (Tomz, Wittenberg, and King 1999) in calculating first differences.
* prospective evaluation for the 1992 Legislative Yuan election.

First differences of the 1995 Legislative Yuan election suggest that economic
voting and ethnic voting did not generate marginal effects on voter decisions at all (see
Table 6-7). In contrast, party identification had some marginal impacts on voter choice.
If the hypothetical voter changed his party loyalty to the DPP, he was 5% less likely to
vote for the KMT and 5% more likely to vote for the NP. The hypothetical individual is
also a male in his thirties who has graduated from high school and lives in an urban area,
as in the South Korean case. Additionally, if he changed his loyalty to the NP, he was

15% less likely to vote for the KMT and 15% more likely to opt for the DPP with no

136

change in the NP’s support. So, these marginal effects may not well reflect the party
identification hypothesis. Overall, first differences of the 1995 Legislative Yuan election
exhibit odd results. The hypothetical individual has a 1.00 probability of voting for the
ruling KMT, while he had a .00 probability of voting for the two opposition parties.
Again, this may come from the low number of .valid responses.

First differences of the 1996 Presidential election may also be problematic. The
signs of many first differences are not in the expected directions. Only first differences of
unemployment and prospective voting (Peng Ming-min) display the anticipated sign (see
Table 6-8). If the respondent became unemployed, he was 4% less likely to vote for Lee
Teng-hui and 4% more likely to support Lin Yang-kang. If he perceived that Peng Ming-
min would best be able to manage the economy, he was 1% more likely to vote for him
and 4% less likely to opt for Lee Teng-hui.

Also, if he believed that Lee Teng-hui provoked the tension between Taiwan and
China, he was 27% more likely to opt for Lee Teng-hui and 1% (26%) less likely to vote
for Peng Ming-min (Lin Yang-kang). However, these first differences do not have the
anticipated signs. Party identification had much stronger marginal effects on voter choice
than economic voting and ethnic voting did. Nonetheless, the signs of party

identification’s first differences are not in the expected directions.

137

Table 6-8. First Difference of Economic Voting, Party Identification, and Ethnic
Background in the 1996 Presidential Election, Taiwan

 

 

LTH PMM LYK
Baseline .89 .05 .05
Prob.
Economics
National -.07 .00 .07
economy
Personal -.01 .05 -.04
economy
Unemployment -.04 .00 .04
Prospective -.20 .09 .1 1
(LTH)
Prospective -.04 .01 .03
(PMM)
Prospective .03 .02 -.05
(LYK)
PID
KMT -.24 .2 1 .03
DPP . l 9 -.2 1 .02
NP .6] .03 -.64
Ethnic
Taiwanese .02 -.04 .02
Mainlander -.04 .01 .03
Tension .27 -.01 -.26

 

Note: I use Clarify (Tomz, Wittenberg, and King 1999) in calculating first differences.

5. Conclusions

The above statistical analyses enable me to draw the following conclusions. First,
economic voting is observed in the 1995 Legislative Yuan and 1996 Presidential
elections. In 1995, the economic voting hypothesis is jointly significant between the NP
and KMT competition in the basic and combined models. Yet, no single economic

voting variable is confirmed to be individually significant. Thus, it is impossible to tell

138

iff‘wfimk In

whether the responsibility hypothesis is confirmed or not. The statistical analyses of the
1995 election are also limited by the number of valid respondents. In contrast, genuine
economic voting occurred in the 1996 election. Joint tests confirm that the economic
voting hypothesis is jointly significant across the models in 1996. This is consistent with
the findings about the 1996 election by Hsieh, Lacy, and Niou (N .d.) and Hsieh and Niou
(1996). Nonetheless, it is not safe to conclude that economic voting was more likely to
occur in the Presidential elections than the Legislative Yuan elections. The number of
elections under study is too small to draw a meaningful generalization.

Second, voters have been affected by their concerns about the sociotropic
economy and the pocketbook economy in 1996. The national economic conditions
affected voter choice between Lin Yang-kang and Lee Teng-hui in the basic model as
well as the combined model. In addition, the personal economic conditions had a
significant impact on voter choice between Peng Ming-min and Lee Teng-hui in the
combined model.

Third, individuals cast their votes retrospectively as well as prospectively in 1996.
On the contrary to Hsieh, Lacy, and Niou (N .d.), retrospective voting is also observed in
1996. The national economic condition and the personal economic condition variables
are, by definition, the retrospective economic voting survey items and the variables
appear to have a significant impact on voter choice. Also, voters were consistent in
connecting the good prospective expectation of Lee Teng-hui with electoral support. Lee
Teng-hui had continuously managed economic prosperity and stability until 1996.

Fourth, ethnic cleavages played at least as significant role as economic voting did

in 1996. Joint test results suggest that the joint test chi square values of ethnic voting are

139

about the same as those of economic voting. Likelihood ratio tests suggest that the
addition of the ethnic cleavage variables into the basic model appear to be more powerful
than the basic model across the elections.

Fifth, party identification consistently played a significant role in voter decisions,
whereas the socioeconomic factors did not have a consistent effect on voter choice. In
fact, party identification is the only consistent factor determining voter choice across the
elections under investigation. In contrast, different socioeconomic factors had different

impacts on voter choice.

140

‘1

5 fl
1‘.

CHAPTER VII

CONCLUSIONS

This study has examined, on the one hand, the association between economic
conditions and voter turnout, and on the other hand, the relationship between economic
conditions and electoral support in South Korea and Taiwan in the 19905. In other words,
I have investigated the new voting behavior at the level of the national elections held after
the democratic transition in these Asian nations. from a political economic perspective.
Under democratic governments, what would determine voter participation and electoral
returns in South Korea and Taiwan? Are there differences in South Korea’s and
Taiwan’s voting behavior between when they were under authoritarianism and after the
democratic transition? History-long regional conflicts between Cholla and Kyungsang
provinces had been a die-hard determinant for voters’ decisions in South Korea. Does
this still hold for the recent democratic elections? Also, how did the Taiwanese begin to
vote under the new circumstances of the two (or three) party system? In fact, there had
been only one party (the KMT) until the mid-19805. Did ethnic conflicts between the
Taiwanese and the Mainlanders play a significant role in electoral politics? Or did people
in South Korea and Taiwan begin to be more concerned about economic conditions when
they were at the polling booths, as voters were in many other democracies?

My hypothesis on voter turnout is that if voters perceive the state of the economy
as being worse than before, voter turnout is likely to decrease (Rosenstone 1982).
Conversely, if voters perceive the state of the economy as being better than before, voter

turnout is likely to increase. I have also tested the hypOthesis about voter choice that if

141

I.”
-
'I

people perceive economic conditions as being worse than before, they are less likely to
vote for the incumbent party. Voters hold the incumbent party responsible for the
economic downturns (Lewis-Beck and Paldam 2000). Conversely, if people perceive
economic conditions as being better than before, they are more likely to vote for the
incumbent party. In addition to the economic condition variables, I build the basic (or
restricted) model including socioeconomic factors, social psychological factors, political
mobilization factors, the close election factor, and party identification into the statistical
models, if available. The combined (or unrestricted) model consists of the basic model
and regionalism (or ethnic) factors. The combined model tests the impact of regional or
ethnic conflicts on voter decisions.

I have used the survey data sets collected by the IKES (the Institute for Korean
Election Studies) and the NCU (the National Chengchi University). Since the democratic
transitions in the mid-19805, South Korea has held three (1987, 1992, and 1997) direct
Presidential elections and four (1988, 1992, 1996, and 2000) National Assembly elections
and Taiwan has held two (1996 and 2000) direct Presidential elections and three (1992,
1995, and 1998) Legislative Yuan elections. However, the IKES has conducted the
nationwide surveys since the 1992 National Assembly election and the 2000 National
Assembly election survey data are not available at the time of this writing. Thus, this
study inevitably deals with the 1992 and 1996 National Assembly elections and the 1992
and 1997 Presidential elections. The NCU has conducted the election surveys in the
19905 but the NCU did not make the survey data available to the public since the 1998
Legislative Yuan election. Thus, only the 1992 and 1995 Legislative Yuan elections and

the 1996 Presidential elections are under investigation.

142

“'1

ill",
I I
l ..

For the methodology of this research project, first, I have employed probit. The
probit model is used for the two-way choices as in the decisions between vote and not-
vote or the decisions between two candidates. Second, I have also employed multinomial
logit. The multinorrrial logit model is utilized for the three-way choices as in the
decisions among three candidates. 1 have conducted likelihood ratio tests to see which
model is better between the basic model and the combined model. Additionally, joint
hypothesis tests are used to see if a group of variables for a hypothesis is jointly

significant.

Table 7-1. Summary of Voter Turnout in South Korea and Taiwan

 

 

South Taiwan
Korea .
Hypothesis 1992 N 1996 N 1992 P 1997 P 1992 L 1995 L 1996 P
Economic S
conditions
Socioeconomic S S S S S S S
Social S S S S S S S
psychological ’
Political S S
mobilization
Close N/A N/A S N/A N/A
election
Regionalism S
or Ethnic

 

Note: N - National Assembly election, P - Presidential election, L - Legislative Yuan
election, and S - jointly significant in the combined model

As a result, I have found some similarities and differences of voter turnout in
South Korea and Taiwan in the 19905. Table 7-1 summarizes the results of statistical

tests, with respect to voter turnout. Above all, economic conditions are likely to have

143

 

nothing to do with voter turnout in South Korea and Taiwan. This is consistent with
Fiorina’s (1978) findings, rather than Lipset’s (1960) and Rosenstone’s (1982)
conclusions in the US. context. Rosenstone suggests that there is a positive association
between economic conditions and voter turnout. Economically deprived people are
viewed to be preoccupied with their daily lives. They are unlikely to be involved in
political actions, for instance, voting (Rosenstone 1982). In contrast, Lipset has
speculated that a high voting average is more likely to be associated with the bad
economy. Economic depression drives people to turn to the government through various
political activities (Lipset 1960). Furthermore, although Radcliff (1992; 1996) concludes
that economic adversity encourages electoral participation in developing countries, which
appears not be the case in South Korea and Taiwan.

While joint tests disapprove of the joint significance of economic condition
hypothesis in the combined model in South Korea, joint tests approve the joint
significance of the economic condition hypothesis on voter turnout only in the 1992
National Assembly election in the basic model. Nonetheless, no single economic
condition variable reached statistical significance. Thus, I cannot confirm whether good
economic conditions or bad economic conditions generated the significant impact on
voter turnout. In Taiwan, joint tests suggest that the economic condition hypothesis on
voter turnout is jointly significant in 1992 (the basic and combined models) and 1995 (the
basic model). However, the signs of the national and personal economic condition
variables are in contradiction in 1992. Only personal economic conditions had the

hypothesized impact on voter turnout in 1995.

I44

Second, traditional regionalism and ethnic conflicts are not significant for voter
participation in South Korea and Taiwan, respectively. Likelihood ratio tests suggest that
the combined model is better than the basic model in explaining voter turnout in South
Korea, with the exception of the 1996 election. The more powerful combined model
approves the joint significance of the regionalism hypothesis only in the 1992
Presidential election. There was no systematic pattern in the relationship between home
provinces and voter turnout in South Korea. Voters’ home provinces were not significant
in the 1992 National Assembly election, whereas Kyungsang people tended to participate
more in the 1996 National Assembly election. In the 1992 Presidential election, Cholla
people were more likely to vote, whereas Kyungsang people were more likely to vote in
the 1997 Presidential election. On the other hand, the election specific party merger
variable was significant for electoral participation only in the 1992 Presidential election.

In Taiwan, likelihood ratio tests suggest that the combined model is better than
the basic model in explaining voter turnout. The more powerful combined model
disproves the joint significance of the ethnic voting hypothesis across the elections. Also,
there was no pattern between voters’ ethnic backgrounds and voter turnout. Only those
Taiwanese origin were more likely to vote in 1996.

Third, as can be seen in Table 7-1, the socioeconomic hypothesis appears to be
jointly significant across the elections in South Korea and Taiwan. Surprisingly enough,
it is age that commonly reached statistical significance throughout the elections in the two
countries. Older voters were more likely to show up at the polling places.

Fourth, the social psychological hypothesis is also jointly significant for voter

participation throughout the elections in South Korea and Taiwan (see Table 7-1).

145

—'.m

ire-m.

Specifically, it is election interest that had a statistically significant impact on voter
turnout commonly in these two countries. The only exception for this is Taiwan’s 1995
election. Election interest was positively correlated with voter participation.

Fifth, the effects of the political mobilization factors were not the same as in
previous elections in the 19705 and 19805 where authoritarian leaders mobilized voters to
the polling booths in South Korea. As a result, official voter turnout rates have
consistently decreased in the 19905. Similarly, the mobilization hypothesis never reached
statistical significance in Taiwan. Previously, the KMT regime did not have to mobilize
voters to win elections unlike the old regimes in South Korea because there used to be
just one party until the mid-19805. Also, there is no clear tendency in official voter
turnout rates. The voter turnout rates were 72% (1992), 69.8% (1995), and 76.0% (1996).
After the democratic transition, elections became remarkably free and fair in Taiwan,
compared to the standards of past elections. Also, the new TV commercials did not have
a significant impact on voter participation in the South Korean Presidential elections,
which disproves the hypothesis about the positive relation between the TV
advertisements and electoral participation.

There are also some differences in voter participation between South Korea and
Taiwan in the 19905. First, it is unique in South Korea that urban residents and the
educated are less likely to vote. In Taiwan, females were more likely to vote than males
in 1996. These are all opposite to the hypothesis.

Second, the close election hypothesis in South Korea exhibits odd results. The
close campaign in the 1997 Presidential election, which ended up with a razor-thin 1.6

percentage point difference, did not attract voter participation but the decent margin

146

1"“--

campaign in the 1992 Presidential race, which ended up with a 8.2 percentage point
difference, appears to have fostered voter turnout. In contrast, there is no compatible

survey question to test the close election hypothesis in Taiwan.

Table 7-2. Summary of Voter Choice in South Korea and Taiwan

 

 

South Taiwan
Korea
Hypothesis 1992 N 1996 N 1992 P 1997 P 1992 L 1995 L 1996 P
Economic /S S/S S/ S/S IS S/S
conditions
PID S/S S/S S/S S/S S S/S S/S
Regionalism S/S S/ S/S S/S S/S
or Ethnic

 

Note: N - National Assembly election, P - Presidential election, L - Legislative Yuan
election, (the incumbent party v. the first opposition party)/(the incumbent party v. the
second opposition party), and S - jointly significant in the combined model

Next, South Korea and Taiwan share some interesting features regarding
economic voting in the 19905. Table 7-2 summarizes the statistical test results of
economic voting behavior. First, genuine economic voting was evident in the 1997 South
Korean and the 1996 Taiwanese elections. These two elections were the last Presidential
races held in the 19905. Particularly, in South Korea, the worst economic crisis occurred
in 1997 and voters held the incumbent party accountable for the crisis in the presidential
election. As a result, South Korea embraced a power transfer for the first time in its
history. However, similar economic voting is not observed in other elections. As Lewis-

Beck and Paldam (2000, 119) adequately note, the instability of the economic voting

147

model has been a discouraging factor for researchers. Results differ from nation to nation
and from time to time.

In South Korea’s earlier elections held between 1992 and 1996, economic voting
is observed to have occurred with voters’ “limited rationality.” In other words, the signs
of the economic voting coefficients sometimes display in the directions opposite to the
hypotheses. Voters believed that the ruling party played an enormous role in South
Korea’s rapid economic development. Because the miraculous economic development
had been orchestrated by the ruling party since the early 19605, voters may have
considered the ruling party as the only manager of the economy, despite their
authoritarian nature. In sharp contrast, opposition parties had never had a chance to show
their abilities to manage the national economy. Opposition parties also lacked human
resources with economic specialties and bureaucratic experiences. Voters may have
thought that the ruling party would be better than opposition parties in handling the
national economy and resolving the unemployment or inflation problems. Therefore, if
people perceived inflation and unemployment as being worse, they were more likely to
vote for the ruling party.

In Taiwan, economic voting did not occur in 1992 and the statistical analyses of
the 1995 election are limited by the small number of valid respondents. In contrast, joint
tests approve the joint significance of the economic voting hypothesis in the 1996
election.

Second, sociotropic voting was more likely to occur than pocketbook voting in
South Korea’s 1997 race and Taiwan’s 1996 election. This conclusion is the same as that

of Kinder and Kiewiet (1981) and many others in the US. context. At the same time,

148

individuals cast their vote retrospectively as well as prospectively in these two elections.
Sociotropic and pocketbook voting are also retrospective voting survey items in this
study. Prospective survey questionnaires were employed for the first time in these
elections and prospective voting was loud and clear."

Third, as shown in Table 7-2, joint tests suggest that the party identification
hypothesis is one of the strongest hypotheses throughout the elections in South Korea and
Taiwan. The party identification variables are always in the hypothesized directions.
Because 82 parties have been formed between January 1963 and January 2000, it was not
easy for South Korea’s voters to form a genuine party identification, as voters do in other
countries. Nonetheless, there had been no power alternation until the 1997 election, thus
the identification between the ruling party versus opposition parties appears to have
always worked as a significant cue for decisions in voting. In contrast, Taiwan’s stable
party system may have established party identification stronger than that of South Korea.
The KMT had ruled in Taiwan until 2000. The DPP emerged as a genuine opposition
party in the mid-19805 and the NP was formed by those who left the KMT in 1993.

Fourth, the socioeconomic factors did not have a consistent effect on voter choice
in South Korea and Taiwan. Different socioeconomic factors were significant in the
different elections.

There are also notable differences in economic voting between South Korea and
Taiwan in the 19905. First, it is not safe to say that economic voting was stronger in the

Presidential races than the parliamentary elections. In South Korea, there are no

 

3‘ It is not safe to conclude that the two countries’ voters were “bankers” who were prospective and
sociotropic at the same time (MacKuen, Erikson, and Stimson 1992).

149

 

significant differences in the joint test chi square values of economic voting between the
Presidential races and the National Assembly elections. Nonetheless, economic voting
may have been stronger between the ruling party and the first opposition party than
between the ruling party and the second opposition party. The competition between the
ruling party and the first opposition party may be more visible and significant than the
competition between the ruling party and the second opposition party. Such a pattern was
not observed in Taiwan’s elections.

Second, regionalism was stronger and more consistent than economic voting in
South Korea but ethnic conflicts were as significant as economic voting in Taiwan in
1996. Likelihood ratio tests approve that the combined model in each country is better
than the basic model in accounting for economic voting. In South Korea, joint tests
approve that the regionalism hypothesis is jointly significant throughout the elections,
while the economic voting hypothesis is not jointly significant in the 1996 National
Assembly election (the combined model). Also, first differences of the regionalism
variables are bigger than those of economic voting across the elections. The home
province and party merger variables are not in the unexpected directions. Furthermore,
Cholla origin Kim Dae Jung or his parties collected Cholla voters’ support more than the
candidates from Kyungsang province collected Kyungsang voters’ support throughout the
elections.

In Taiwan, joint test results suggest that the joint significance of ethnic voting is
about the same as those of economic voting in the 1996 election. First differences are not

significantly varied between economic voting and ethnic voting.

150

To summarize, new voting behavior certainly emerged in the two countries in the
19905. With regard to voter turnout, the traditional mobilization factors were not as
significant as in previous authoritarian elections. Age and election interest were the
factors that had significant impacts on voter turnout throughout the elections in the two
countries. Economic conditions and regionalism (or ethnic cleavages) were not
significant for voter participation. Economic voting became significant in the decision
processes of who to vote for in the last presidential elections in each country in the 19905.
Voters were sociotropic as well as retrospective and prospective. At the same time,
regional or ethnic cleavages remained as a significant factor for voter decisions in South
Korea and Taiwan. In the near future, as survey data sets from more recent elections in
the two countries become available, I will be able to extend the number of cases. This
will improve the external validity of the conclusions of the research project. Also, as
economic issues become more important than ever in the two countries, I will be able to
see if there is a tendency of economic voting behavior with the increased number of

cases.

151

APPENDIX A

SOUTH KOREAN SURVEY DATA

The South Korean survey data were collected by the Institute for Korean Election
Studies (IKES). The IKES conducted the nationwide election surveys (except the Jeju
Island) under the auspices of the Central Election Management Commission and other
funding organizations in South Korea. The IKES employed the face-to-face interview
method by visiting every household drawn.

Survey data for the two National Assembly elections were collected by the
multistage probability sampling method. The 1992 election survey (N=1206) was
conducted between March 26 and 30 after the election on March 24, 1992. The 1992
National Assembly election survey is the first survey project done by the IKES. The
1996 election survey (N=1200) took place between April 11 and 18 after the election on
April 11, 1996. At the first stage, the total sample size of 1200 was drawn with
probability proportionate to the electorate of the cities and communities throughout the
nation. At the second stage, electoral constituencies within which thirty eligible voters
would be drawn were selected.

The sample for the 1992 Presidential election survey (N=1200) was also drawn on
the basis of the multistage probability sampling method. The interviews took place
between December 19 and 27 after the election on December 18, 1992. The sampling
method for the 1997 Presidential election survey (N=1179) was the quota sampling. At

the first stage, the total sample size of 1200 was drawn with probability proportionate to

152

the electorate of the cities and communities. At the second stage, samples were drawn
according to the ratio of sex and age, with regional characteristics taken into
consideration. At the third stage, fifteen eligible voters per two polling areas were
randomly selected to represent a constituency. The IKES interviewed respondents
between December 19 and 22 after the election on December 18, 1997.

In the 1992 National Assembly election survey, 44.6%, 38.4%, and 16.9% of the
respondents answered to have voted for the DLP, the DP, and the UPP, respectively, and
86.0% of the respondents reported to have participated in the election. The actual vote
returns are in Table 5-1 and the official turnout was 71.9%. The NKP, the NCNP, and
the ULD mustered 56.3%, 32.8%, and 10.9% of the respondents, respectively, in the 1996
National Assembly election survey. Also, 81.2% of the respondents reported to have
voted. The official turnout rate was 63.9%.

The 1992 Presidential election survey indicates that Kim Young Sam, Kim Dae
Jung, and Chung Ju Young received support from 45.3%, 27.3%, and 9.6 % of the
respondents, respectively, among the 93.4% self-reported voters. The official turnout rate
was 81.9%. _In addition, the 1997 election survey displays that Kim Dae Jung, Lee Hoe
Chang, and Lee In Je received support from 37.0%, 33.5%, and 15.7% of the
respondents, respectively, among the 92.5% self-reported voters. The actual voter

turnout was 80.6%.

153

APPENDIX B

TAIWANESE SURVEY DATA

The Election Study Center of the National Chengchi University (NCU) has
conducted nationwide surveys since the 1986 Legislative Yuan elections. The surveys
employed the face-to-face interview method like South Korea’s surveys. The NCU used
the multistage stratified sampling method. The population was divided into four strata
(Taipei City, Kaohsiung City, Provincial Cities, and Counties) and quotas for each
stratum were drawn with probability proportionate to the electorate of the cities and
districts. The NCU conducted pre-election surveys and post-election surveys throughout
the elections under investigation. I use only the post-election surveys, which are
equivalent to the South Korean surveys. The sizes of the samples are 1523 (the 1992
Legislative Yuan election), 1485 (the 1995 Legislative Yuan election), and 1396 (the
1996 Presidential election).

In the 1992 Legislative Yuan election survey, the KMT and the DPP earned
71.0% and 29.0% of the respondents, respectively, and 89.3% of the respondents reported
to have voted in the election. The actual vote returns are in Table 6-1 and voter turnout
was 72.0%. In the 1995 Legislative Yuan election survey, the KMT, the DPP, and the
NP collected 69.9%, 24.1%, and 6.0% of the respondents, respectively. 87.3% of the
respondents answered to have participated in the election but the official voter turnout

was 69.8%.

154

In the 1996 Presidential election, Lee Teng-hui, Peng Ming-min, and Lin Yang-
kang received 74.9%, 14.5%, and 10.5% of respondents, respectively, and 91.8% of the

respondents reported to have voted in the election. The official voter turnout was 76.0%

155

APPENDIX C

SURVEY QUESTIONS”

<Vote> Did you vote in the election?

(0) n0 (1) yes [A] [B] [C] [D] [E] [F] [G]

<Whom to Vote> Whom (which party) did you vote for?

( l) the Democratic Justice Party (2) the Democratic Party (3) the United People’s Party
[A]

(l) the New Korea Party (2) the National Congress for New Politics (3) the United
Liberal Democrats [B]

(1) Kim Young Sam (2) Kim Dae Jung (3) Chung Ju Young [C]

(1) Lee Hoe Chang (2) Kim Dae Jung (3) Lee In Je [D]

(0) the Democratic Progressive Party (1) the Kuomintang [E]

( 1) the Kuomintang (2) the Democratic Progressive Party (3) the New Party [F]

(1) Lee Teng-hui (2) Peng Ming-min (3) Lin Yang-kang [G]

<National Economy> How would you evaluate national economic conditions during the
past few years?

( 1) worsened (2) about the same (3) improved [B] [C] [E] [F] [G]

 

3" The following questionnaires are translated from Korean and Chinese. They are sometimes reworded
and recoded. [A], the 1992 South Korean National Assembly election; [B], the 1996 South Korean
National Assembly election; [C], the 1992 South Korean Presidential election; [D], the 1997 South Korean
Presidential election; [E], the 1992 Taiwanese Legislative Yuan election; [F], the 1995 Taiwanese
Legislative Yuan election; and [G], the 1996 Taiwanese Presidential election.

156

<National Economy in 1997> Which political party do you think was mainly
responsible for the current economic crisis (the IMF bailout)? [D]

(0) other parties (1) the ruling party (the Grand National Party)

<Personal Economy> How do you feel about household economic conditions during the
past few years?

(1) worsened (2) about the same (3) improved [A] [B] [C] [D] [E] [F][G]

<Inflation> What is the most important problem that the country is facing?36

(0) others (1) consumer price [A] [C] [D]

<Unemployment> Are you employed?

(0) yes (1) no [A] [B] [C] [D] [E] [F] [G]

<Retrospective Evaluation> How well do you think the government has been running
things?

(1) poorly (2) well [A] [C]

(1) disapprove (2) approve [B]

(1) disapprove (2) neither approve nor disapprove (3) approve [D] ’7

<Prospective Evaluation> Who do you think will best be able to resolve the current
national problems?

Lee Hoe Chang, Kim Dae Jung, and Lee In Je are dummies [D]

(1) not promising at all (2) a bit promising

 

3" The original question in the 1992 National Assembly election survey is “What do you think are the most
important problems that this country is facing? Please choose two in the order of importance.” For the
1992 and 1997 presidential election surveys, “Which issue do you have in mind as the most important
thing in selecting the candidate? Please choose two.” 1 consider the first choice as the most serious
national problems.

’7 The original question for [B] and [D] is “In general, do you approve or disapprove of the way the Kim
Young Sam government has been running things?”

157

(3) somewhat promising (4) absolutely promising [E]38

(1) will be worsened (2) will be about the same (3) will be improved [F]39

Lee Teng-hui, Peng Ming-min, and Lin Yang-kang are dummies [G]"0
<Residential Area> Where do you live?

(1) rural area (2) suburban (3) urban area [A] [B] [C] [D]

<Age> How old are you?

(1) 20-29 (2) 30-39 (3) 40-49 (4) 50 and older [A] [B] [C] [D] [E] [F] [G]
<Educational Attainment> What is your education level ?

(1) elementary school graduation and below (2) middle school graduation and below
(3) high school graduation and below (4) college education and more

[A] [B] [C] [D] [E] [F] [G]

<Party Identifier> Do you think of yourself as a supporter of any political party?
( 1) n0 (2) yes [A] [B] [C] [D] [E] [F] [G]

<Election Interest> How interested were you in the election?

(1) not interested at all (2) not so much interested

(3) somewhat interested (4) very much interested [A] [B] [C] [D]

(1) not interested at all (2) not so much interested

(3) it depends (4) somewhat interested (5) very much interested [E] [F] [G]"1

<Campaign Rallies> Have you ever been to a campaign rally?

(1) n0 (2) yes [A] [BIIE]

 

3” The original question is “Is the democratic development in Taiwanese politics promising?”

’9 The question is “How do you think national (household) economic conditions are going to be in one
year?”

‘0 The original question is “Who do you think will best be able to generate economic development?”

4' The original question is “How interested were you in election-related articles in newspapers?”

158

 

<TV Advertisement> Have you seen candidates’ TV advertisements during the
campaign?

(1)110 (2)yes [C] [D]

<Campaign Contact> Did anyone talk to you about supporting specific candidates in
this last election?

(1) no (2) yes [A] [B] [C] [D]

(1)110 (2) yes [E] [F] IGI’2

<Anticipated Election Closeness> Before the election, how did you forecast the
election outcome?

(1) one candidate winning by a big margin

(2) a fairly close race

(3) too close to forecast [C] [D]

<Coalition> Do you like the recent party merger? [A] [C] [D]

<Tension: Cross-Strait Relations> How do you think the current cross-strait relations
are?

(1) worsened (2) about the same (3) improved [E]

(1) disagree (2) agree [F]43

 

‘2 The original question is “Have any party campaign staffers talked to you about supporting specific
candidates in this last election? ”

‘3 The question is “Do you agree with the opinion that Lee Teng-hui provoked the current cross-strait
tension?”

159

APPENDIX D

MULTINOMIAL LOGIT ESTIMATES FOR THE THREE-PARTY RACE:

THE 1995 LEGISLATIVE YUAN ELECTION, TAIWAN

(COEFFICIENTS FOR THE KMT’S NORMALIZED TO ZERO)

 

 

Independent Basic Basic Combined Combined
Variables DPP NP DPP NP
Economics
National .2288744 22.79843 -3239.172 2.44e+16
economy (1.211) (7456.66) (.) (.)
Personal -.9841409 -33.43616 2260.409 -1.54e+18
economy (.935) (.) (.) (.)
Unemployment -2.672969 -54.76324 10341.58 -2204.579
(2.157) (3728.364) (.) (.)
Prospective -.6609297 -4.473 525 1.28e+13 1.37e+18
National (.795) (4.548) (.) (.)
Prospective -.1747721 6.164358 -5214.458 -9.65e+17
Personal (.856) (7.540) (.) (.)
Socioeconomic
Age -.6617601 .4143291 -8.94e+12 -l .35e+18
(.681) (1.530) (.) (.)
Gender .390294 3.958573 -1.28e+13 -679.9748
(Male) (.989) (4.466) (.) (.)
Educational . 1904859 36.31153* 1.28e+12 2.25e+17
attainment (.709) (3.397) (.) (.)
PID
KMT -3.747806* -5.785387 2762.068 1.87e+18
(1.358) (5.888) (.) (.)
DPP 3.231433* -71.79447 -1939.051 -71 .93991
(1.348) (2.29e+09) (.) (.)
NP .5862797 39.07793 * -46.68033 7.52e+18
(1.602) (5.557) (.) (.)
Ethnic
Taiwanese ~1497.538 4.64e+18
() (-)
Mainlander -4.86e+1 1 -2.46e+18
(-) (-)
Tension 55.95962 2.20e+18
(-) (-)
Intercept 8.399248 .1920403 -32.2036 408.0936
(5.935) (.) (.)
Prob > chi2 0.0000 .
Pseudo R2 0.6337 i.nf0e-01
Number of obs. 89 74

 

160

LITERATURE CITED

Abramowitz, Alan I. 1985. “Economic Conditions, Presidential Popularity, and Voting
Behavior in Midterm Congressional Elections.” Journal of Politics 47:31-43.

Abramowitz, Alan I. 1988a. “An Improved Model for Predicting Presidential Election
Outcomes.” Political Science and Politics 21:843-47.

Abramowitz, Alan I. 1988b. “Explaining Senate Election Outcomes.” American
Political Science Review 82:385-403.

33‘!

Abramowitz, Alan 1., David J. Lanoue, and Subha Ramesh. 1988. “Economic
Conditions, Causal Attribution, and Political Evaluations in the 1984 Presidential
Election.” Journal of Politics 50:848-63.

1:" 1

Abramson, Paul. 1983. Political Attitudes in America: Formation and Change. CA:
Freeman.

Abramson, Paul, and John Aldrich. 1982. “The Decline of Electoral Participation in
America.” American Political Science Review 76:502-21.

Abramson, Paul, John Aldrich, and David Rohde. 1995. Change and Continuity in the
1992 Elections. Washington, DC: Congressional Quarterly, Inc.

Aidt, Toke S. 2000. “Economic Voting and Information.” Electoral Studies 19:349-62.

Aldrich, John. 1993. “Rational Choice and Turnout.” American Journal of Political
Science 372246-78.

Alesina, Alberto, and Howard Rosenthal. 1989. “Partisan Cycles in Congressional

Elections and the Macroeconomy.” American Political Science Review 83:
373-98.

Alford, John R., and Jerome S. Legge, Jr. 1984. “Economic Conditions and Individual
Vote in the Federal Republic of Germany.” Journal of Politics 46:1168-181.

Alvarez, R. Michael, and Jonathan Nagler. 1995. “Economics, Issues, and the Perot
Candidacy: Voter Choice in the 1992 Presidential Election.” American Journal of
Political Science 39:714-44.

Alvarez, R. Michael, and Jonathan Nagler. 1998. “When Politics and Models Collide:

Estimating Models of Multiparty Elections.” American Journal of Political
Science 42:55-96.

161

Alvarez, R. Michael, Jonathan Nagler, and Shaun Bowler. 2000. “Issues, Economics,
and the Dynamics of Multiparty Elections: The British 1987 General Election.”
American Political Science Review 94: 13 1-49.

Alvarez, R. Michael, Jonathan Nagler, and Jennifer R. Willette. 2000. “Measuring the

Relative Impact of Issues and the Economy in Democratic Elections.” Electoral
Studies 19:237-53.

Anderson, C. J. 2000. “Economic Voting and Political Context: A Comparative
Perspective.” Electoral Studies 192151-70.

Anderson, C. J., and D. S. Ward. 1996. “Barometer Elections and Comparative
Perspective.” Electoral Studies 15:447-60.

Arcelus, Francisco, and Allen M. Meltzer. 1975. “The effect of Aggregate Economic
Variables on Congressional Elections.” American Political Science Review 69:
1232-239.

Asian Survey Various Issues.

Atkeson, Lonna Rae, Randall W. Partin. 1995. “Economic and Referendum Voting: A
Comparison of Gubernatorial and Senatorial Elections.” American Political
Science Review 89299-107.

Bartels, Larry M. 1996. “Uninforrned Votes: Information Effects in Presidential
Elections.” American Journal of Political Science 40:194-230.

Beck, Nathaniel. 1991. “The Economy and Presidential Approval: An Information
Theoretic Perspective.” In Economics and Politics: The Calculus of Support, eds.
Helmut Norpoth, Michael S. Lewis-Beck, and Jean-Dominique Lafay. Ann
Arbor: The University of Michigan Press.

Bell, J. 1997. “Unemployment Matters: Voting Patterns During the Economic
Transition in Poland, 1990-1995.” Europe-Asia Studies 49:1263-291.

Bellucci, Paolo. 1984. “The Effect of Aggregate Economic Conditions on the Political
Preferences of the Italian Electorate, 1953-1979.” European Journal of Political
Research 12:387-401.

Belluci, Paulo. 1991. “Italian Economic Voting: A Deviant Case or Making a Case for a
Better Theory?” In Economics and Politics: The Calculus of Support, eds.
Helmut Norpoth, Michael Lewis-Beck, and Jean-Dominique Lafay. Ann Arbor:
University of Michigan Press.

162

Bennett, Stephen Earl. 1990. “The Education-Tumout “Puzzle” in Recent US National
Elections.” Electoral Studies 9251-8.

Black, Jerome H. 1978. “The Multicandidate Calculus of Voting: Application to
Canadian Federal Elections.” American Journal of Political Science 22:609—38.

Blais, Andre, and R. K. Carty. 1990. “Does Proportional Representation Foster Voter
Turnout?” European Journal of Political Research 18:167-81.

Bloom, Howard S., and H. Douglas Price. 1975. “Voter Response to Short-Run

Economic Conditions: The Asymmetric Effect of Prosperity and Recession.”
American Political Science Review 69: 1240-254.

Born, Richard. 1990. “Surge and Decline, Negative Voting, and the Midterm Loss

Phenomenon: A Simultaneous Choice Analysis.” American Journal of Political
Science 34:615-45.

Bratton, Michael. 1997. “Deciphering Afiica’s Divergent Transitions.” Political
Scientific Quarterly 112:67-93.

Bratton. Michael, and Nicolas Van de Walle. 1997. Democratic Experiments in Afiica.
MA: Cambridge University Press.

Brody, Richard A. 1978. “The Puzzle of Political Participation in America.” In
The New American Political System, ed. Anthony King. DC:
American Enterprise Institute for Public Policy Research.

Brody, Richard A., and Paul M. Snidennan. 1977. “From Life Space to Polling Place:
The Relevance of Personal Concerns for Voting Behavior.” British Journal of
Political Science 7:337-60.

Caldeira, Gregory A., and Samuel C. Patterson. 1982. “Contextual Influences on
Participation in US. State Legislative Elections.” Legislative Studies
Quarterly 7:3 59-81.

Caldeira, Gregory A., Samuel C. Patterson, and Gregory A. Markko. 1985. “The
Mobilization of Voters in Congressional Elections.” Journal of Politics 47:490-
509.

Campbell, Angus. 1966. “Surge and Decline: A Study of Electoral Change.” In
Elections and the Political Order, eds. Angus Campbell, Philip E. Converse,
Warren E. Miller, and Donald E. Stoke. NY: John Wiley and Sons, Inc.

Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960.
The American Voter. IL: The University of Chicago Press.

163

‘7'

ll'

Campbell, James E. 1986. “Forecasting the 1986 Midterm Elections to the House of
Representatives.” Political Science 16283-7.

Carlsen, Fredrik. 2000. “Unemployment, Inflation and Government Popularity - Are
There Partisan Effects?” Electoral Studies 191141-50.

Chappell, Henry W. Jr. 1983. “Economic Performance and Presidential Popularity: Are
Voters Really So Naive?” Review of Economics and Statistics 65:385-92.

Chappell, Henry W. Jr., and William R. Keech. 1985. “A New View of Political
Accountability for Economic Performance.” American Political Science Review
79:10-27.

Chappell, Henry W. Jr., and Linda Goncalves Veiga. 2000. “Economics and Elections in
Western Europe.” Electoral Studies 19:183-97.

Cheng, You-ping. 1995. “The Interaction of Ethnicity and Party Politics in Taiwan.”
Issues and Studies 31 :1-15.

Chu, Yin-wah. 1996. “Democracy and Organized Labor in Taiwan: The 1986
Transition.” Asian Survey 36:495-510.

Clarke, Harold D., Karl Ho, and Marianne C. Stewart. 2000. “Major’s Lesser (Not
Minor) Effects: Prime Ministerial Approval and Governing Party Support
in Britain Since 1979.” Electoral Studies 19:255-73.

Clarke, Harold D., and Marianne C. Stewart. 1994. “Prospections, Retrospections, and
Rationality: The ‘Bankers’ Model of Presidential Approval Reconsidered.”
American Journal of Political Science 38:1 104-123.

Clarke, Harold D., and Paul Whitely. 1990. “Perceptions of Macroeconomic
Performance, Government Support and Conservative Party Strategy in Britain
1983-87.” European Journal of Political Research 18:97-120.

Cohen, Jeffrey E., Michael A. Krassa, and John A. Hamman. 1991. “The Impact of
Presidential Campaigning on Midterm US. Senate Elections.” American
Political Science Review 852165-78.

Colton, Timothy J. 1996. “Economics and Voting in Russia.” Post-Soviet Affairs 12:
289-317.

Conover, Pamela Johnston, Stanley F eldman, and Kathleen Knight. 1986. “Judging

Inflation and Unemployment: The Origins of Retrospective Evaluations.”
Journal of Politics 482555-88.

164

Cox, Gary W., and Michael C. Munger. 1989. “Closeness, Expenditures, and Turnout in
the 1982 US. House Elections.” American Political Science Review 83:217-31.

Diamond, Larry. 1996. “Is the Third Wave Over?” Journal of Democracy 7:20-37.
Domes, Jurgen. 1999. “Electoral and Party Politics in Democratization.” In
Democratization in Taiwan: Implications for China, eds. Steve Tsang and Hung-

mao Tien. London: MacMillan Press.

Dominguez, Jorge I., and James A. McCann. 1996. Democratizing Mexico: Public
Opinion and Electoral Choices. Baltimore: The Johns Hopkins University Press.

Douglas, Roger. 1978. “Occupation, Economic Fluctuations and Support for the
Government.” Politics 13:31 1-19.

rmm “'1..- ~

Edwards, George C. 1992. Presidential Approval: A Source Book. Baltimore: Johns
Hopkins University Press.

Erikson, Robert S. 1988. “The Puzzle of Midterm Loss.” Journal of Politics 50:1011-
29.

Erikson, Robert S. 1990. “Economic Conditions and the Congressional Vote: A Review
of the Macrolevel Evidence.” American Journal of Political Science 342373-99.

Erikson, R. S., M. B. MacKuen, and J. A. Stimson. 2000. “Bankers or Peasants
Revisited: Economic Expectations and Presidential Approval.” Electoral Studies
19:295-312.

Fair, Ray C. 1978. “The Effect of Economic Events on Votes for President.” The
Review of Economics and Statistics 60: 1 59-73.

Feld, Lars P., and Gebhard Kirchgassner. 2000. “Official and Hidden Unemployment
and the Popularity of the Government: An Econometric Analysis for the Kohl
Government.” Electoral Studies 192333-47.

Feldman, Stanley. 1982. “Economic Self-Interest and Political Participation.” American
Journal of Political Science 26:446-66.

F ererjohn, John. 1990. “Information and the Electoral Process.” In Information and
Democratic Processes, eds. John Ferejohn, and James Kuklinski. IL: University
of Illinois Press.

Ferejohn, John A., and Morris P. F iorina. 1974. “The Paradox of Not Voting: A
Decision Theoretic Analysis.” American Political Science Review 68:525-3 6.

165

Fidrmuc, Jan. 2000. “Economics of Voting in Post-Communist Countries.” Electoral
Studies 19:199-217.

Fiorina, Morris P. 1976. “The Voting Decision: Institutional and Expressive Aspects.”
Journal of Politics 38:390-415.

Fiorina, Morris P. 1978. “Economic Retrospective Voting in American National
Elections: A Micro-Analysis.” American Journal of Political Science 222426-43.

Fiorina, Morris P. 1981. Retrospective Voting in American National Elections. New
Haven: Yale University Press.

Fiorina, Morris P., and Kenneth A. Shepsle. 1989. “15 Negative Voting an Artifact?”
American Journal of Political Science 33 2423-39.

Fleury, C. J., and Michael S. Lewis-Beck. 1993. “Deja Vu All Over Again: A Comment
on the Comment of Converse and Pierce.” Journal of Politics 55:1118-126.

Foster, Carroll B. 1984. “The Performance of Rational Voter Models in Recent
Presidential Elections.” American Political Science Review 78:678-90.

Frey, Bruno S., and Friedrich Schneider. 1975. “On the Modeling of Politico-Economic
Interdependence.” European Journal of Political Research 32339-60.

Frey, Bruno S., and Friedrich Schneider. 1978. “A Politico-Economic Model of the
United Kingdom.” The Economic Journal 88:243-53.

Frey, Bruno S., and Friedrich Schneider. 1980. “Popularity Functions: The Case of the
US and West Germany.” In Models of Political Economy, ed. Paul Whiteley.
London: Sage Publications.

Goergen, Christian, and Helmut Norpoth. 1991. “Government Turnover and Economic
Accountability.” Electoral Studies 102191-207.

Goodhart, C. A. E., and R. J. Bhansali. 1970. “Political Economy.” Political Studies
18243-106.

Goodman, Saul., and Gerald H. Kramer. 1975. “Comment on Arcelus and Meltzer:
The Effect of Aggregate Economic Conditions on Congressional Elections.”

American Political Science Review 6921255-265.

Guttman, J ., and Y. Shachmurove. 1990. “Dynamic Optimization of Political Popularity
in an Empirical Macroeconomic Model.” Public Choice 66:51-66.

166

 

Haller, H. Brandon, and Helmut Norpoth. 1994. “Let the Good Times Roll: The
Economic Expectations of US. Voters.” American Journal of Political Science
382625-50.

Happy, J. R. 1992. “The Effect of Economic and Fiscal Performance on Incumbency
Voting: The Canadian Case.” British Journal of Political Science 222119-30.

Harper, Marcus A. G. 2000. “Economic Voting in Postcommunist Eastern Europe.”
Comparative Political Studies 33:1191-227.

Hibbing, John R.. and John R. Alford. 1981. “The Electoral Impact of Economic
Conditions: Who Is Held Responsible?” American Journal of Political Science
252423-39.

Hibbs, Douglas A., Jr. 1982. “President Reagan’s Mandate from the 1980 Elections: A
Shift to the Right?” American Politics Quarterly 10:387-420.

Hibbs. Douglas A. Jr., and Nicholas Vasilatos. 1981. “Macroeconomic Performance and
Mass Political Support in the United States and Great Britain.” In Contemporary
Political Economy: Studies on the Interdependence of Politics and Economics,
eds. Douglas A. Hibbs, Jr., and Heino Fassbender, with the assistance of R.
Douglas Rivers. NY: North-Holland Publications Co.

Hibbs, Douglas A., Jr. with the assistance of Nicholas Vasilatos. 1981. “Economics and
Politics in France: Economic Performance and Mass Political Support for

Presidents Pompidou and Giscard d’Estaing.” European Journal of Political
Research 92133-45.

Hsieh, John Fuh-sheng, and Emerson M.S. Niou. 1996. “Issue Voting in the Republic of
China on Taiwan’s 1992 Legislative Yuan Election.” International Political
Science Review 17:13-27.

Hsieh, John Fuh-sheng, and Emerson M. S. Niou. N.d. “Salient Issues in Taiwan’s
Electoral Politics.” Working Papers in Taiwan Studies #3 American Political
Science Association Conference Group on Taiwan Studies.

Hsieh, John Fuh-sheng, Dean P. Lacy, and Emerson M.S. Niou. 1996. “Economic
Voting in the 1994 Taiwan Elections.” American Asian Review 14:51-70.

Hsieh, John Fuh-sheng, Dean Lacy, and Emerson M. S. Niou. N.d. “Retrospective and
Prospective Voting in a One-Party-Dominant Democracy: Taiwan’s 1996
Presidential Election.” Working Papers in Taiwan Studies #21 American
Political Science Association Conference Group on Taiwan Studies.

167

 

Inoguchi, Takashi. 1980. “Economic Conditions and Mass Support in Japan, 1960-
1976.” In Models of Political Economy, ed. Paul Whiteley. London: Sage
Publications.

Jacobson, Gary C. 1991. “The Economy in US. House Elections.” In Economics and
Politics: The Calculus of Support, eds. Helmut Norpoth, Michael S. Lewis-Beck,
and Jean-Dominique Lafay. Ann Arbor: The University of Michigan Press.

Jacobson, Gary C., and Samuel Kemell. 1983. Strategy and Choice in Congressional
Elections. New Heaven: Yale University Press.

Jerome, Bruno, and Veronique Jerome-Speziari. 2000. “The 1998 French Regional
Elections: Why So Much Political Instability.” Electoral Studies 192219-36.

Jerome, Bruno, and Michael S. Lewis-Beck. 1999. “15 Local Politics Local? French
Evidence.” European Journal of Political Research 352181-97.

Jonung, L. 1981. “Perceived and Expected Rates of Inflation in Sweden.” American
Economic Review 712964-68.

J onung, L., and D. Laidler. 1988. “Are Perceptions of Inflation Rational? Some
Evidence From Sweden.” American Economic Review 7821080-87.

Jonung, Lars, and Eskil Wadensjo. 1979. “The Effect of Unemployment, Inflation and
Real Income Growth on Government Popularity in Sweden.” The Scandinavian
Journal of Economics 81 2343-53.

Journal of Democracy Various Issues.

Karl, Terry Lynn. 1995. “The Hybrid Regimes of Central America.” Journal of
Democracy 6:72-86.

Kemell, Samuel. 1977. “Presidential Popularity and Negative Voting: An Alternative
Explanation of the Mid-Term Congressional Decline of the President’s Party.”

American Political Science Review 71:44-66.

Key, V. O. Jr. 1964. Politics, Parties, and Pressure Groups. New York: Thomas Y.
Crowell.

Key, V. 0. Jr. 1966. The Responsible Electorate: Rationality in Presidential Vote 1936-
1960. Cambridge: The Belknap Press of Harvard University.

Kiewiet, D. Roderick. 1981. “Policy-Oriented Voting in Response to Economic Issues.”
American Political Science Review 752448-59.

168

[ET—"mil

Kiewiet, D. Roderick. 1983. Macroeconomics and Micropolitics: The Electoral Effects
of Economic Issues. Chicago: The University of Chicago Press.

Kiewiet, D. Roderick. 2000. “Economic Retrospective Voting and Incentives for
Policymaking.” Electoral Studies 192427-44.

Kiewiet, D. Roderick, and Douglas Rivers. 1985. “A Retrospective on Retrospective
Voting.” In Economic Conditions and Electoral Outcomes: The United States
and Western Europe, eds. Heinz Eulau, and Michael S. Lewis-Beck. New York:
Agathon.

Kim, HeeMin. 1997. “Rational Choice Theory and Third World Politics: The 1990 Party
Merger in Korea.” Comparative Politics 30283-100.

IL“ ‘Wfi’fi

Kim, Jae Han. 1993. “The Economy and the 14th Presidential Election in South
Korea.” Korean Political Science Review 27299-120 (in Korean).

Kim, Kwang-Woong, and Chan Wook Park. 1991. “Public Attitudes toward the
National Legislature and Its Members in the 6th Republic of Korea.”
Korea Journal 31278-92.

Kim, Wook. 1998. “Voting and Nonvoting: Who and Why Vote?” In Elections in South
Korea 11: An Analysis of the 15th Presidential Election, ed. Nam-young Lee.
Seoul: Blue Road (in Korean).

Kinder, Donald R., and D. Roderick Kiewiet. 1979. “Economic Discontent and Political
Behavior: The Role of Personal Grievances and Collective Economic Judgments
in Congressional Voting.” American Journal of Political Science 23:495-527.

Kinder, Donald R., and D. Roderick Kiewiet. 1981. “Sociotropic Politics: The American
Case.” British Journal of Political Science 22129-61.

Kinder, Donald R., and Gordon S. Adams, and Paul W. Gronke. 1989. “Economics and
Politics in the 1984 American Presidential Election.” American Journal of
Political Science 33 2491 -5 l 5.

King, Gary, Michael Tomz, and Jason Wittenberg. 2000. “Making the Most of
Statistical Analyses: Improving Interpretation and Presentation.”
American Journal of Political Science 442341-55.

Kirchgassner, Gebhard. 1985. “Rationality, Causality, and the Relation Between
Economic Conditions and the Popularity of Parties.” European Economic Review

28:243-68.

Kirchgassner, Gebhard. 1986. “Economic Conditions and the Popularity of West

169

German Parties: A Survey.” European Journal of Political Research 142421-39.

Kirchgassner, Gebhard. 1991. “Economic Conditions and the Popularity of West
German Parties: Before and After the 1982 Government Change.” In
Economics and Politics: The Calculus of Support, eds. Helmut
Norpoth, Michael S. Lewis-Beck, and Jean-Dominique Lafay. Ann
Arbor: The University of Michigan Press.

Knack, Stephen. 1996. “The Reappearing American Voter: Why Did Turnout Rise in
’92?” Electoral Studies 16:17-32.

Korea Central Daily.

Kramer, Gerald H. 1971. “Short-Term Fluctuations in US. Voting Behavior, 1896-
1964.” American Political Science Review 652131-43.

Kramer, Gerald H. 1983. “The Ecological Fallacy Revisited: Aggregate- versus
Individual-level Findings on Economics and Elections, and Sociotropic Voting.”
American Political Science Review 77292-111.

Kuklinski, James H. 1981. “Economic Expectations and Voting Behavior in United
States House and Senate Elections.” American Political Science Review 752436-
47.

Lafay. J. D. 1981. “Empirical Analysis of Politico-Economic Interaction in East
European Countries.” Soviet Studies 33:1173-204.

Lancaster, Thomas D., and Michael S. Lewis-Beck. 1986. “The Spanish Voter:
Tradition Economics, and Ideology.” Journal of Politics 482648-74.

Lane, Robert E. 1959. Political Life. New York: Free Press.

Lau, Richard R. 1985. “Two Explanations for Negative Effects in Political Behavior.”
American Journal of Political Science 2921 19-38.

Lau, Richard R., and David P. Redlawsk. 1997. “Voting Correctly.” American Political
Science Review 912585-98.

Lazarsfeld, Paul, Bernard Berelson, and Helen Gaudet. 1944. The People ’5 Choice.
New York: Columbia University Press.

Lee, Hyun Woo. 1998. “Economic Voting in South Korea” In Elections in South

Korea 11: An Analysis of the 15th Presidential Election, ed. Nam-young Lee.
Seoul: Blue Road (in Korean).

170

Lee, Nam-young. 1993. “Voting and Nonvoting: An Analysis of the 14th National
Assembly Election.” In Elections in South Korea, ed. Nam-young Lee. Seoul:
Nanam (in Korean).

Lewis-Beck, Michael S. 1980. “Economic Conditions and Executive Popularity: The
French Experience.” American Journal of Political Science 242306-23.

Lewis-Beck, Michael S. 1983. “Economics and the French Voter: A Microanalysis.”
Public Opinion Quarterly 472347-60.

Lewis-Beck, Michael S. 1984. “France: The Stalled Electorate.” In Electoral Change in
Advanced Industrial Societies: Realignment or Dealignment? eds. Russell
Dalton, Scott Flanagan, and Paul Allen Beck. Princeton: Princeton University
Press.

Lewis-Beck, Michael S. 1985. “Pocketbook Voting in US. National Election Studies:
Fact or Artifact?” American Journal of Political Science 292348-56.

Lewis-Beck, Michael S. 1986. “Comparative Economic Voting: Britain, France,
Germany, Italy.” American Journal of Political Science 302315-46.

Lewis-Beck, Michael S. 1988a. Economics and Elections: The Major Western
Democracies. Ann Arbor: The University of Michigan Press.

Lewis-Beck, Michael S. 1988b. “Economics and the American Voter: Past, Present,
Future.” Political Behavior 1025-21.

Lewis-Beck, Michael S. 1993. “The French Voter: Steadfast or Changing?” In The
French Voter Decides, eds. D. Boy and N. Mayer. University of Michigan Press.
Ann Arbor.

Lewis-Beck, Michael S. 1996. “Cross-Nation Election Surveys: A French Pre-Test.”
Electoral Studies 1 525 13-28.

Lewis-Beck, Michael S. 1997. “Who’s the Chef?2 Economic Voting Under a Dual
Executive.” European Journal of Political Research 31 2315-25.

Lewis-Beck, Michael S. ed. 2000. How France Votes. NJ: Chatham House.

Lewis-Beck, Michael S., and Paolo Bellucci. 1982. “Economic Influences on
Legislative Elections in Multiparty Systems: France and Italy.” Political
Behavior 4293-107.

Lewis-Beck, Michael S., and Heinz Eulau. 1985. “Economic Conditions and Electoral
Behavior in Transnational Perspective.” In Economic Conditions and Electoral

171

Outcomes: The United States and Western Europe, eds. Heinz Eulau, and
Michael S. Lewis-Beck. New York: Agathon.

Lewis-Beck, Michael S., and Brad Lockerbie. 1989. “Economics, Votes, Protests:
Western European Cases.” Comparative Political Studies 222155-77.

Lewis-Beck, Michael S., and Richard Nadeau. 2000. “French Electoral Institutions and
the Economic Vote.” Electoral Studies 192171-82.

Lewis-Beck, Michael S., and Martin Paldam. 2000. “Economic Voting: An
Introduction.” Electoral Studies 1921 13-21 .

Lewis-Beck, Michael S., and Tom Rice. 1985. “Are Senate Election Outcomes
Predictable?” Political Science 182745-54.

 

Lewis-Beck, Michael S., and Tom W. Rice. 1992. Forecasting Elections. Washington
DC: Congressional Quarterly, Inc.

Lipset, Seymour Martin. 1960. Political Man. New York: Doubleday.

Lipset, Seymour Martin. 1985. “The Elections, The Economy, and Public Opinion:
1984.” Political Science 18228-3 8.

Lockerbie, Brad. 1991. “Prospective Economic Voting in US. House Elections, 1956-
88.” Legislative Studies Quarterly 16:239-61.

Lucas, R. E. Jr., and T. J. Sargent. 1981. Rational Expectations and Econometric
Practice. Minneapolis: The University of Minnesota Press.

Lupia, Arthur. 1994. “Shortcuts Versus Encyclopedias: Information and Voting
Behavior in California Insurance Reform Elections.” American Political
Science Review 88:63-76.

Lupia, A., and M. D. McCubbins. 1998. The Democratic Dilemma: Can Citizens Learn
What They Need to Know? NY: Cambridge University Press.

Lybeck, J. 1985. “A Simultaneous Model of Politico-Economic Interaction in Sweden.”
European Journal of Political Research 13,: 135-51 .

MacKuen, Michael 8., Robert S. Erikson, and James A. Stimson. 1992. “Peasants or

Bankers? The American Electorate and the US. Economy.” American Political
Science Review 86:597-611.

172

Madsen, Henrik J. 1980. “Electoral Outcomes and Macro-Economic Policies: The
Scandinavian Cases.” In Models of Political Economy, ed. Paul Whiteley.
London: Sage Publications.

Markus, Gregory B. 1988. “The Impact of Personal and National Economic Conditions
on the Presidential Vote: A Pooled Cross-Sectional Analysis.” American Journal
of Political Science 322137-54.

Marra, Robin F ., Charles W. Ostrom, Jr., and Dennis M. Simon. 1990. “Foreign Policy
and Presidential Popularity: Creating Windows of Opportunity in the Perpetual
Election.” Journal of Conflict Resolution 34:588-623.

McKelvey, Richard, and Peter Ordeshook. 1985. “Elections with Limited Information:
A Fulfilled Expectations Model Using Contemporaneous Poll and

Endorsement Data as Information Sources.” Journal of Economic Theory
36255-85.

Miller, A. H., and Ola Listhaug. 1985. “Economic Effects and the Voting in Norway.”
In Economic Conditions and Electoral Outcomes, eds. Michael S. Lewis-Beck
and Heinz Eulau. NY: Agathon.

Miller, Arthur H., and Martin P. Wattenberg. 1985. “Throwing the Rascals Out: Policy
and Performance Evaluations of Presidential Candidates, 1952-1980.” American
Political Science Review 792359-72.

Miller, Nicholas B. 1986. “Public Choice and the Theory of Voting: A Survey.” Annual
Review of Political Science 121-36.

Miller, W. L., and M. Mackie. 1973. “The Electoral Cycle and the Asymmetry of
Government and Opposition Popularity: An Alternative Model of the

Relationship between Economic Conditions and Political Popularity.” Political
Studies 212263-79.

Mo, Jongryn, David Brady, and Jaehun R0. 1991. “Urbanization and Voter Turnout in
Korea: An Update.” Political Behavior 13:21-31.

Monroe, Kristen, and Lynda Erickson. 1986. “The Economy and Political Support: The
Canadian Case.” Journal of Politics 482616-47.

Monroe, Kristen R., and M. D. Levi. 1983. “Economic Expectations, Economic
Uncertainty, and Presidential Popularity.” In The Political Process and Economic
Change, ed. Kristen R. Monroe. NY: Agathon.

Moon, David. 1992. “The Determinants of Turnout in Presidential Elections: An
Integrative Model Accounting for Information.” Political Behavior 142123-40.

173

 

Morriss, Peter. 1996. “Electoral Politics in South Korea.” Electoral Studies 152550-62.

Mueller, John E. 1970. “Presidential Popularity from Truman to Johnson.” American
Political Science Review 64218-34.

Nadeau, R., Richard G. Niemi, and T. Amato. 1996. “Prospective and Comparative or
Retrospective and Individual? Party Leaders and Party Support in Great Britain.”
British Journal of Political Science 262245-58.

Nannestad, Peter, and Martin Paldam. 1994. “The VP-function: A Survey of the
Literature on Vote and Popularity Functions After 25 Years.” Public Choice
792213-45.

Nannestad, Peter, and Martin Paldam. 1996. “It’s the Govemment’s Fault! A Cross-
Section Study of Economic Voting in Denmark, 1990-93.” European Journal of
Political Research 28:33-65.

Nannestad, Peter, and Martin Paldam. 1997a. “From the Pocketbook of the Welfare
Man: A Pooled cross-section Study of Economic Voting in Denmark, 1986-92.”
British Journal of Political Science 2721 19-36.

Nannestad, Peter, and Martin Paldam. 1997b. “The Grievance Asymmetry Revisited: A
Micro Study of Economic Voting in Denmark, 1986-92.” European Journal of
Political Economy 1 3 :8 1-99.

Nannestad, Peter, and Martin Paldam. 2000. “Into Pandora’s Box of Economic
Evaluations: A Study of the Danish Macro VP-Function.” Electoral Studies
192123-40.

Nickelsburg, Michael, and Helmut Norpoth. 2000. “Commander-in-Chief or Chief
Economist? The President in the Eye of the Public.” Electoral Studies 192
313-32.

Nie, Norman, Sidney Verba, and John Petrocik. 1979. The Changing American Voter.
MA: Harvard University Press.

Niemi, Richard G., and Herbert F. Weisberg. eds. 1993a. Controversies in Voting
Behavior. Washington, D. C.: Congressional Quarterly, Inc.

Niemi, Richard G., and Herbert F. Weisberg. eds. 1993b. Classics in Voting
Behavior. Washington, D. C.: Congressional Quarterly, Inc.

174

Niemi, Richard G., Harold W. Stanley, and Ronald J. Vogel. 1995. “State Economies
and State Taxes: Do Voters Hold Governors Accountable?” American Journal of
Political Science 392936-57.

Norpoth, Helmut. 1985. “Economics, Politics, and the Cycle of Presidential Popularity.”
In Economic Conditions and Electoral Outcomes: The United States and
Western Europe, eds. Heinz Eulau, and Michael S. Lewis-Beck. New York:
Agathon.

O’Donnell, Guillermo, and Philippe C. Schmitter. 1986. Tentative Conclusions about
Uncertain Democracies. Baltimore: Johns Hopkins University Press.

Oh, Byungsang. 1995. A Blue House Story (Chongwadae Biseosil). The Central News
Daily (in Korean).

3"-

Oppenheimer, Bruce 1., James A. Stimson, and Richard W. Waterman. 1986.
“Interpreting US. Congressional Elections: The Exposure Thesis.” Legislative
Studies Quarterly 112227-47.

Ostrom, Charles W. Jr., and Dennis M. Simon. 1985. “Promise and Performance: A
Dynamic Model of Presidential Popularity.” American Political Science Review
792334-58.

Ostrom, Charles W. Jr., and Dennis M. Simon. 1989. “The Man in the Teflon Suit: The
Environmental Connection, Political Drama, and Popular Support in the Reagan
Presidency.” Public Opinion Quarterly 532353-87.

Ostrom, Charles W. Jr., and Renee M. Smith. 1992. “Error Correction, Attitude
Persistence, and Executive Rewards and Punishments: A Behavioral Theory of
Presidential Approval.” Political Analysis 42127-82.

Pacek, Alexander C. 1994. “Macroeconomic Conditions and Electoral Politics in East
Central Europe.” American Journal of Political Science 382723-44.

Pacek, Alexander, and Benjamin Radcliff. 1995. “The Political Economy of
Competitive Elections in the Developing World.” American Journal of
Political Science 392745-59.

Paldam, Martin. 1981. “A Preliminary Survey of the Theories and Findings on Vote and
Popularity Functions.” European Journal of Political Research 92181-99.

Paldam, Martin. 1987. “Inflation and Political Instability in Eight Latin American
Countries, 1946-83.” Public Choice 522143-68.

175

Paldam, Martin, and Peter Nannestad. 2000. “What Do Voters Know About the
Economy? A Study of Danish Data, 1990-1993.” Electoral Studies 192363-91.

Panzer, J ., and R. D. Paredes. 1991. “The Role of Economic Issues in Elections: The
Case of the Chilean Referendum.” Public Choice 71251-9.

Park, Chan Wook. 1993a. “Korean Voters’ Candidate Choice in the 1992 Presidential
Election: A Survey Data Analysis.” Korea and World Affairs 172432-58.

Park, Chan Wook. 1993b. “The Fourteenth National Assembly Election in Korea: A
Test for the Ruling Democratic Liberal Party.” Korea Journal 3325-16.

Park, Chan Wook. 1996. “Characteristics of Korean Parliamentary Politics.” Korea
Focus: On Current Topics 4224-38.

Patterson, Samuel C ., and Gregory A. Caldeira. 1983. “Getting Out the Vote:
Participation in Gubernatorial Elections.” American Political Science Review 77:
675-89.

Pattie, Charles, and Ron Johnston. 1998. “Voter Turnout at the British General Election
of 19922 Rational Choice, Social Standing or Political Efficacy?” European
Journal of Political Research 332263-83.

Peretz, Paul. 1981. “The Effect of Economic Change on Political Parties in West
Germany.” In Contemporary Political Economy: Studies on the Interdependence
of Politics and Economics, eds. Douglas A. Hibbs, Jr., and Heino Fassbender,
with the assistance of R. Douglas Rivers. NY: North-Holland Publications Co.

Pissarides, Christopher A. 1980. “British Government Popularity and Economic
Performance.” The Economic Journal 902569-81.

Powell, G. Bingham, Jr. 1980. “Voting Turnout in Thirty Democracies: Partisan, Legal,
and Socio-Economic Influences.” In Electoral Participation: A Comparative
Analysis, ed. Richard Rose. London: Sage Publications.

Powell, G. Bingham, Jr. 1993. “American Voter Turnout in Comparative Perspective.”
In Controversies in Voting Behavior, eds. Richard G. Niemi and Herbert F.
Weisberg. Washington, D. C.: Congressional Quarterly, Inc.

Powell, G. Bingham, Jr., and Guy D. Whitten. 1993. “A Cross-National Analysis of

Economic Voting: Taking Account of the Political Context.” American Journal
of Political Science 372391-414.

176

..~*— ."';A 1. Ln

Price, Simon, and David Sanders. 1993. “Modeling Government Popularity in Postwar
Britain-A Methodological Example.” American Journal of Political Science
3 723 17-34.

Przeworski, Adam. 1996. “Public Support for Economic Reforms in Poland.”
Comparative Political Studies 292520-43.

Radcliff, Benjamin. 1992. “The Welfare State, Turnout, and the Economy: A
Comparative Analysis.” American Political Science Review 862444-54.

Radcliff, Benjamin. 1996. “Turnout and the Economy: A Comment on Jackman and
Miller’s ‘Voter Turnout in the Industrial Democracies in the 19805.’”
Comparative Political Studies 292718-23.

Rattinger, Hans. 1991. “Unemployment and Elections in West Germany.” In Economics
and Politics: The Calculus of Support, eds. Helmut Norpoth, Michael
Lewis- Beck, and Jean-Dominique Lafay. Ann Arbor: University of Michigan
Press.

 

Reed, Steven, and Gregory G. Brunk. 1984. “A Test of Two Theories of Economically
Motivated Voting: The Case of Japan.” Comparative Politics 17:55-66.

Remmer, Karen L. 1991. “The Political Impact of Economic Crisis in Latin America in
the 19805.” American Political Science Review 85:777-800.

Riker, William. 1990. “Political Science and Rational Choice.” In Perspectives on
- Positive Political Economy, eds. James Alt and Kenneth Shepsle. Cambridge:
Cambridge University Press.

Riker, William H., and Peter C. Ordeshook. 1968. “A Theory of the Calculus of
Voting.” American Political Science Review 62:25-42.

Rius, A. 1993. “El Gobiemo, la Economia y el Hombre de la Calle.” SUM 7-132
7-35.

Rosa, Jean Jacques. 1980. “Economic Conditions and Elections in France.” In Models
of Political Economy, ed. Paul Whiteley. London: Sage Publications.

Rosenstone, Steven J. 1982. “Economic Adversity and Voter Turnout.” American
Journal of Political Science 26225-46.

Sanders, David. 1991. “Government Popularity and the Next General Election.”
Political Quarterly 622235-61.

177

Sanders, David. 1995. “Forecasting Political Preferences and Election Outcomes in the
UK: Experiences, Problems and Prospects for the Next General Election.”
Electoral Studies 142251 -72.

Sanders, David. 1996. “Economic Performance, Management Competence and the
Outcome of the Next General Election.” Political Studies 442203-31.

Sanders, David. 1999. “Conservative Incompetence, Labour Responsibility and the
Feelgood Factor: Why the Economy Failed to Save the Conservative in 1997.”
Electoral Studies 182251-70.

Sanders, David. 2000. “The Real Economy and the Perceived Economy in Popularity
Functions: How Much do Voters Need to Know? A Study of British Data 1974-
97.” Electoral Studies 192275-94.

Schneider, Friedrich, and Werner W. Pommerehne. 1980. “Politico-Economic
Interactions in Australia: Some Empirical Evidence.” The Economic Record
562113-31.

Schneider, Friedrich, Werner W. Pommerehne, and Bruno S. Frey. 1981. “Politico-
economic Interdependence in a Direct Democracy: The Case of Switzerland,” In
Contemporary Political Economy: Studies on the Interdependence of Politics and

Economics, eds. Douglas A. Hibbs, Jr., and Heino Fassbender, with the assistance
of R. Douglas Rivers. NY: North-Holland Publications Co.

Seligson, Mitchell A., and Miguel Gomez B. 1989. “Ordinary Elections in
Extraordinary Times: The Political Economy of Voting in Costa Rica.”
In Elections and Democracy in Central America, eds. John H. Booth and
Mitchell A. Seligson. Chappel Hill: The University of North Carolina Press.

Shaffer, Stephen D. 1981. “A Multivariate Explanation of Decreasing Turnout in
Presidential Elections, 1960-1976.” American Journal of Political Science 25:
68-95.

Silder, Darrell. 1994. “Political Tendencies in Russia’s Regions: Evidence from the
1993 Parliamentary Elections.” Slavic Review 532711-32.

Simon, Dennis M., and Charles W. Ostrom, Jr. 1988. “The Politics of Prestige: Popular
Support and the Modern Presidency.” Presidential Studies Quarterly 182741-59.

Southwell, Priscilla L. 1988. “The Mobilization Hypothesis and Voter Turnout in
Congressional Elections, 1974-1982.” Western Political Quarterly 412273-87.

Stigler, George J. 1973. “Micropolitics and Macroeconomics: General Economic
Conditions and National Elections.” American Economic Review 632160-67.

178

Stimson, James A. 1976. “Public Support for American Presidents: A Cyclical Model.”
Public Opinion Quarterly 4021-21.

Stokes, S. C. 1996. “Economic Reform and Public Opinion in Peru, 1990-95.”
Comparative Political Studies 292544-65.

Tien, Hung-mao. 1989. Political and Social Change in the Republic of China. CA2
Hoover Institution Press.

Tien, Hung-mao. 1996. “Taiwan in 1995: Electoral Politics and Cross-Strait Relations.”
Asian Survey 36:33-40.

Tien, Hung-mao, and Tun-jen Cheng. 1997. “Crafting Democratic Institutions in
Taiwan.” The China Journal 3721-27.

Tien, Hung-mao, and Yun-han Chu. 1996. “Building Democracy in Taiwan.” The
China Quarterly 14821141-170.

Timpone, Richard J. 1998. “Structure, Behavior, and Voter Turnout in the United
States.” American Political Science Review 922145-58.

Tomz, Michael, Jason Wittenberg, and Gary King. 1999. Clarify: Software for
Interpreting and Presenting Statistical Results Version 1.2.1. MA: Harvard
University.

Tufte, Edward R. 1975. “Determinants of the Outcomes of Midterm Congressional
Elections.” American Political Science Review 692812-26.

Uhlaner, Carole Jean. 1989. “Turnout in Recent American Presidential Elections.”
Political Behavior 11:57-79.

Verba, Sidney, and Norman H. Nie. 1972. Participation in America: Political
Democracy and Social Inequality. New York: Harper and Row.

Westlye, Mark C. 1984. “Competitiveness of Senate Seats and Voting Behavior in
Senate Elections.” American Journal of Political Science 272253-83.

White, Stephen, Richard Rose, and Ian McAllister. .1997. How Russia Votes. Chatham,
NJ 2 Chatham House.

Whiteley, Paul. 1980. “Politico-Econometric Estimation in Britain: An Alternative

Interpretation.” In Models of Political Economy, ed. Paul Whiteley.
London: Sage Publications.

179

Whitten, Guy D., and Harvey D. Palmer. 1999. “Cross-national Analyses of Economic
Voting.” Electoral Studies 18:49-67.

Wilkin, Sam, Brandon Haller, and Helmut Norpoth. 1997. “From Argentina to Zambia:
a World-Wide Test of Economic Voting.” Electoral Studies 162301 -1 6.

Wolfinger, Raymond E., and Steven J. Rosenstone. 1980. Who Votes? New Haven:
Yale University Press.

Wooldridge, Jeffrey M. 2000. Introductory Econometrics: A Modern Approach. South-
Westem College Publishing.

180