LOCAL POLICY CHOICES AND PARTY CONTROL: POLITICAL INSTITUTIONALIZATION AMONG FRENCH URBAN COMMUNES DISSERTATION FOR THE DEGREE OF DH. D MICHIGAN STATE UNIVERSITY PLATON N. RIGOS 1974 Date This is to certify that the thesis entitled LOCAL POLICY CHOICES AND PARTY CONTROL: POLITICAL INSTITUTIONALIZATION AMONG FRENCH COMMUNES presented by Platon N. Rigos has been accepted towards fulfillment of the requirements for Ph. D. Political Science degree in Major professor //- /f- 1974 0-7539 ABSTRACT LOCAL POLICY CHOICES AND PARTY CONTROL: POLITICAL INSTITUTIONALIZATION AMONG FRENCH URBAN COMMUNES by Platon N. Rigos Most hypotheses describing the relationships between political variables and policy have been formulated with little regard as to the historical characteristics of the political systems under scrutiny. This dissertation, in testing for the effects of party control on French local Social Welfare policy, attempts to explain such relation- ships in the context of the political institutionalization process which in the case of France, features a close correspondence between social and political cleavages and the local entrenchment of political parties. The testing uses two samples of French cities, one of which encompasses 36 cities of the Paris Basin where the Communist Party has developed such local power that the area has often been called the Paris Red Belt. The other sample includes nine cities from the Lyon urban area, as well as some outlying Parisian suburbs. The study compares the effects of political variables on redistri- butive policies such as Welfare with policies such as Education and Total Expenditures for the year 1963. The political variables are descriptive of the size of the Left parties (Socialist and Communist) representation, as well as of the size of the Communist Party alone. The analysis can be said to stand out from other previous efforts in that the spuriousness tests are structured so as to control for the Platon N. Rigos size of taxable resource levels on one hand and service need on the other. This was possible because one of the need indicators, personal income, was not highly correlated with taxable resources in the chosen samples (industrial cities with high tax base and poor populations). Suppressor effects were detected in that controlling for personal income highlighted the true impact of resource levels. The results confirmed income as a need indicator in that it was negatively related to all three policy areas. The basic hypothesis was confirmed in that party control had a very powerful effect on Social Welfare in the Paris Basin sample and a still significant effect in the provincial cities sample. In Paris, party control effects on Education and Total Spending were diminished considerably (often to the point of insignificance because of collinear interactions), but in the provinces no relationship existed to begin with. Educational Spending in Parisian cities was seemingly determined equally by the preference of people in rich communities for private education, as well as by the budgetary commitment of Communist municipalities. Among other findings was the apparent ability of Left administrations to qualify for more central grants-in-aid than their need indicated. The presence of Socialists on city councils was found to result in even greater increases in Social Welfare expenditures in Paris while in the provinces, their presence was a moderating influence. This was a by- product of divergent political strategies after the introduction in 1959 of the Two Ballot Electoral System, and the more consensual culture of the provinces. Vbter turnout was found to have no effect on policies and grants- in-aid were equally weak determinants in both samples. Finally, a Platon N. Rigos minor policy area, Business Taxation (symbolic of class conflict), was adequately explained by the same determinants used in explaining Social Welfare (also in both samples). In comparing these findings to similar efforts conducted in Europe, the author concluded that sharp discontinuities in resources and industrialization levels between the units of analysis must somehow be accounted for by the methodology used (i.e., sampling or dummy variables) if meaningful findings are to result. This conclusion emerged as the author observed that such discontinuities seemed to account for the divergent findings between the two samples. LOCAL POLICY CHOICES AND PARTY CONTROL: POLITICAL INSTITUTIONALIZATION AMONG FRENCH URBAN COMMUNES By Platon N. Rigos A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1974 To Jenny and Nick ii ACKNOWLEDGMENTS Every author of as monumental a task as a dissertation has often reflected on the number of people without whose help deadends would have remained unbreachable. I know that in my case, what was intuition and limited generalizations, took on the systematic nature of theoretical explanation after my friend and colleague, Rodney P. Stiefbold, acquainted me with the ramifications of cleavage theory. Yet there are debts that I owe to many others. Fbr testing ideas on a cool and precise intellect, I wish to thank my friend and companion in graduate study, James E. Piereson. Among the others, I want to express my sincere thanks to the members of my dissertation committee: To Professor Charles Press, who served as my chairman, for his invaluable help in solving the administrative and substantive problems; Professor Frank A. Pinner, who in his insistence on theoretical and methodological rigor insured that this dissertation was a product worthy of the Department of Political Science of Midhigan State University; and Professor Charles F. Cnudde, whose membership on the committee came only at the later stages of the dissertation, but nonetheless inspired me at first by his writings and later by his insights into the significance of my own findings. Finally, I wish to thank Monsieur Sauvet of the French Prefecture des Hauts-de-Seine in Nanterre, whose initial guidance through French municipal budgets insured a minimum of error. My special thanks go to iii Professor John F. Kramer who initiated me to quantitative political science and specifically, regression analysis, when I was negatively inclined towards such concerns. iv II. TABLE OF CONTENTS PagegNumber Chapter I: Introduction ..... . ....... . ...... . ........ ....1 A. B. C. D. Purpose Of the StudYO.OOOOOOOOOOOOOOOOOOO0.0.0.0000001 Concept FormationOOOOOOOO0.0.0.0...0.0.0.0000000000004 1. Some Theoretical Considerations,,,,,,,,,,,,,,,..,4 2. The Policy-Making Mbdels......... ......... .......8 3. me conceptSOOOOOOOOOOOOOIOOOO.00.0.00000000000011 The Setting and the Study Sample....................15 Implications of This Study..........................17 Chapter II: The Setting: Changes in the A. French Stalemate............... ..... .......20 Development and Institutionalization of CleavageSOOOOOOOOOOOOOOOQOOOOOCOOOOOOOOOOC.0.0.0.00023 Administrative Development and Local Autonomy.......32 1. Pre-Mbdern Period...............................33 2. Napoleonic System.................. ...... .......34 3. The Administrative Structure Today..............37 a. Tutelage and Other Controls.................42 b. SOCial welfareOOOOOOOOOOOOOOOCOOOOOOOOOO0.0.45 The Administrative Center in a Stalemated Society...49 1. Administration and Local Government in Paris....49 2. The Supply of National Services in Paris........54 Urbanization: Old and New..........................59 l. The Two Mbdels of Urbanization..... ..... . ..... ..59 The New Urban Political Patterns......... ..... ......62 serOOOOOOOOOOOOOOOO0.000000000000000I0.0.0.0....70 Page Number 111. Chapter III: Research Design... ........... .............72 A. Framework and Propositions.................. ....... .72 1. Models of Policy-Making Revisited.. ........... ..72 2. Ass‘mptionSOOOOOOOOO.COO......OOOOOOOOOOOOOOOOOO76 3. Hypotheses........ ........ . ................. ....80 B. MethOdOIOgy.OCIOOOOOOOOOOOOO ...... 0...... ........ .0081 1. Correlation and Regression......................82 2. Multicollinearity and Its Remedies... ....... ....87 3. Sampling Design.................................93 C. Sample Description.................................108 1. me CitieSOOOOOOOOOOOOOOOOOOOOOOOOOOOOO ....... .108 D. The Variables: Choice and Operationalization......113 . Data Sources: Budgets... ..... . ..... ...........113 . The Services.............. ........ .............114 . Revenue Sources................................ll7 . Taxation.......................................ll8 . Other Measures and Variables...................122 LnJ-‘wNI-l E. Operational Hypotheses.............................126 IV. Chapter IV: Findings (Part One).......................129 A. Expenditure Patterns........... ..... ...............132 B. Zero-Order Relationships...........................134 C. The Determinants of Social Welfare Spending........140 l. Spuriousness Tests.............................141 2. me Explanatory mde10000COO...0.00.00.00.00000146 D. The Determinants of Educational Spending...........153 l. Spuriousness Tests..... ....... .................153 2. T116 Explanatory mde1000000000000000.0.0.000000157 E. The Determinants of Total Operating Expenditures...l61 1. Spuriousness Tests.............................161 2. The Explanatory Model..........................163 F. Two Other Minor Policy Areas............ ....... ....168 1. Tt‘e BIISineSS TaXOOOOOOOO00.0.00...0.0.0.0000000168 2. Central Grants-in-Aid....... ........ . ....... ...l73 G. Conclusions................ ..... . ......... .........174 vi Page Number V. Chapter V: Findings (Part Two)...... ..... .............l77 A. The Validity of the Second Sample .............. ....178 1. The Sample and French Urbanization.............178 2. Methodological Limitations.....................l80 B. Expenditure Patterns.............. ............. ....l82 C. Zero-Order Relationships...........................184 D. The Determinants of Social Welfare Spending........190 E. The Determinants of Educational Spending...........l97 F. The Determinants of Total Operating Expenditures...l97 C. Other Policy Areas.................................l98 1. Grants for Operating Expenditures..............198 2. BUSineSS TaxationOOOOOOOOOOOOO0.00.00.00.000000198 H. Conclusions........................................201 VI 0 Chapter VI: conC1USionS O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O .202 1. Theoretical Considerations.....................208 2. SUbstantive Argumnts I I O O O O O O O O O O O O O O O O O O O O O O O .210 3 O ResearCh Imlications O O O O O O O O O I O O O O O O O O O O O O O O O O 212 4. Improvements and Future Research...............214 APPENDIXOOOOOOOOIO...OOOOOOOOOOOOOOO...OOOOOOOOOOOOO0000......217 BIBLIOGRAPHYOOOOOOCO...OOOOCOOOOOOO ...... 0.... ........ 0.0.0.00219 vii LIST OF TABLES Table Page II-l Communes by Population Size.............................36 II-2 Population of the Departements of the Paris kgion (1962)....OOOOOO00......0.000000000000000.0.00.0052 11—3 Population Growth by Regions and Type of City...........58 II-4 Densities and Other Characteristics of Selected Communes in 1962, in the Paris and Lyon Areas...........59 II-S Political Participation by City Size....................60 III-l Population Growth (1954-62) by Size of Commune's POPulation Among Suburban CommuneS......................94 III-2 Some Mean Per Capita Expenditures for the Satellite Communes of Selected Regions and Our Samples.,,,,,,,,,,,96 III-3 Means, Standard Deviations and Coefficients of variation for Selected Variables in Each of the Two sampleSOOOO...0.0.0.000...ICC.....OOOOOOOOOOOOOOOOOO97 III-4 Demographic Characteristics of Very Poor Communes (Poor in Resources and Poor in Personel Income) in1962.O...OOOOOOOOOOOOOOOOOOOO....0.0.0.0000000000000099 III-5 Demographic Characteristics of Poor Communes (Poor in Personal Income, Rich in Taxable ResourceS) in1962.00.00.0000OOOOOOOOOOOOOOOOOOO0.0.0.0100 III-6 Demographic Characteristics of Rich Communes (Rich in Resources and Personal Income) in 1962........101 III-7 Demographic Characteristics of Moderately Rich Communes (Poor in Resources, Rich in Personal Income) in1962.00.00.00000000OOOOOOOOOOOOOOOO.00.00...102 III-8 Social Structure and Political Variables of Very Poor Communes (Poor in Resources and Poor in Personal Incom) in19620....OOOOOOOOOOOOOOOOOOO.0.00.0103 viii Table Page III-9 Social Structure and Political Variables of Poor Communes (Poor in Income, Rich in Taxable ResourceS) in1962.00.00.00....OOOOOOCOIOOOOOOOOO0.000.104 III—10 Social Structure and Political Variables of Moderately Rich Communes (Poor in Resources, Rich in Personal Income) in 1962.......................105 III-ll Social Structure and Political Variables of Rich Communes (Rich in Resources and Rich in Personal Income) in1962.0.000000000000000OOOOOOOOOOOOOOOOOOOOO.106 III-12 LiSt Of variables and LabeISOOOI.0.000000000000000.124-125 IVel Selected Per Capita Spending Policy Areas of 54 French Cities, Grouped by Type of Partisan Control of Municipal Councils in 1963..........................l30 IV—2 Selected Spending Policy Areas of 36 Paris Core Cities, Grouped by Type of Partisan Control of the MuniCipal COWCilS in 1963.00.00.00000000000000.0.0131 IV-3 Intercorrelations Between Sixteen Independent variables (36 Paris Core Cities) in l963...............l35 IV44 Zero-Order Correlation Coefficients Between Inde- pendent Variables and Per Capita Expenditures Categories for 1963 (36 City—Sample)...................137 IV#5 Intercorrelations Between Per Capita Expenditure Categories for 1963 (36 Paris Core Cities)...... ...... .139 IV—6 Partials, Betas, Regression Coefficients for the Effects of Resources and Left Representation on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris core Cities in 1963000000000.00000000000000000141 IV-7 Partials, Betas and Regression Coefficients for the Effects of Income and Left Representation on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in 1963...........................14l IV-8 Partials, Betas and Regression Coefficients for the Effects of Resources and Income on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in 1963....................................l42 IV-9 Partials, Betas and Regression Coefficients for the Effects of Resources and Workers on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in 1963.........................................l42 ix Table IVeIO IV-ll IV-12 IV¥13 IV-14 IV-15 IVel6 IV#17 IV418 IVel9 Page Partials, Betas and Regression Coefficients for the Effects of WOrkers and Left Representation on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in 1963........................143 Partials, Betas and Regression Coefficients for the Effects of Workers and Communist Representation, 1959, on Per Capita Social welfare Spending (S.W.Sp.) of 36 Paris Core Cities in 1963........................143 Partials, Betas, and Regression Coefficients for the Effects of Income and Communist Representation, 1959, on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in 1963........................147 Partials, Betas and Regression Coefficients for the Effects of Resources, Income and Left Representation on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in 1963...........................147 Partials, Betas and Regression Coefficients for the Effects of Resources, Income and Communist Representation, 1959, on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in 1963.....148 Partials, Betas and Regression Coefficients for the Effects of Resources, Income and Communist Representation, 1953, on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in l963,,,_.l48 Partials, Betas and Regression Coefficients for the Effects of Resources, Workers and Left Representation on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in l963..............149 Partials, Betas and Regression Coefficients for the Effects of Resources, workers and Communist Representation, 1959, on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in l963,,,,,149 Partials, Betas and Regression Coefficients for the Effects of Income and Resources and Left Representation, 1959, on the Logarithm of Per Capita Social Welfare Spending (LOG.S.W.Sp.) of 36 Paris Core Cities in 1963...........................152 Partials, Betas and Regression Coefficients for the Effects of Income, Resources and Communist Representation, 1953, Squared on Per Capita Social Welfare Spending (S.W.Sp.) of 36 Paris Core Cities in 19630....OOIOOOOOOOOOIOOOOOOOOIOOO00.00.00.0000.0... 152 Table IVFZO IV-21 IV-22 IV#23 IV-24 IV—25 IV-26 IV-27 IV-28 IV-29 Page Partials, Betas and Regression Coefficients for the Effects of Resources and Communist Representation, 1959, on Per Capita Educational Spending (Ed.Sp.) of 36 Paris Core Cities in 1963...........................154 Partials, Betas, Regression Coefficients for the Effects of Communist Representation, 1959, and Youth on Per Capita Educational Spending (Ed.Sp.) of 36 Paris Core Cities in l963........................154 Partials, Betas, Regression Coefficients for the Effects of Resources, Youth and Communist Representation, 1959, on Per Capita Educational Spending (Ed.Sp.) of 36 Paris Core Cities in 1963......155 Partials, Betas, Regression Coefficients for the Effects of Youth, Resources, Income and Communist Representation, 1959, on Per Capita Educational Spending (Ed.Sp.) of 36 Paris Core Cities in 1963......155 Partials, Betas, Regression Coefficients for the Effects of Resources and Youth on Per Capita Educational Spending (Ed.Sp.) of 36 Paris Core Cities in 1963.........................................156 Partials, Betas, Regression Coefficients for the Effects of Resources and Income on Per Capita Educational Spending (Ed.Sp.) of 36 Paris Core Cities in 1963.........................................156 Partials, Betas, Regression Coefficients for the Effects of Youth, Resources and Income on Per Capita Educational Spending (Ed.Sp.) of 36 Paris core Cities in1963.00.00.00...00.00.000.00...00.00.00.158 Partials, Betas, Regression Coefficients for the Effects of Income, Resources and Communist Representation, 1959, on Per Capita Educational Spending (Ed.Sp.) of 36 Paris Core Cities in 1963......158 Partials, Betas, and Regression Coefficients for the Effects of Income, Youth, and the Interaction of Resources and Communism (CISTRPSBxR) on Per Capita Educational Expenditures of 36 Paris Core Cities in 1963................................... ...... 160 Partials, Betas and Regression Coefficients for the Effects of Grants-Operating Expenditures (REPART), Resources, Income and Communist Representation, 1959, on Per Capita Educational Expenditures (Ed.Sp.) of 36 Paris Core Cities in 1963................................................16O xi Table IVF30 IV-31 IV-32 IV-33 IV—34 IV-35 IV-36 IV‘37 IV-38 Partials, Betas, Regression Coefficients for the Effects of Resources and Communist Representation, 1959, on Per Capita Total Operating Expenditures (Tot.Sp.) of 36 Paris Core Cities in 1963..............l62 Partials, Betas, Regression Coefficients for the Effects of Income and Communist Representation, 1959, on Per Capita Total Operating Expenditures (Tot.Sp.) of 36 Paris Core Cities in 1963....,.....,..,162 Partials, Betas, Regression Coefficients for the Effects of Resources and Income on Per Capita Total Operating Expenditures (Tot.Sp.) of 36 Paris Core Cities in 1963 164 Partials, Betas and Regression Coefficients for the Effects of Resources, Income and Communist Representation, 1959, on Per Capita Total Operating Expenditures (Tot.Sp.) of 36 Paris Core Cities in 1953................................................164 Partials, Betas and Regression Coefficients for the Effects of Resources, WOrkers, and Communist Representation, 1959, on Per Capita Total Operating Expenditures (Tot. Sp.) of 36 Paris Core Cities in 1963................................................165 Partials, Betas and Regression Coefficients for the Effects of Resources, Income and Workers on Per Capita Total Operating Expenditures (Tot.Sp.) of 36 Paris Core Cities in 1963........................l65 Partials, Betas and Regression Coefficients for the Effects of Resources, WOrkers, Population, Density, and Communist Representation, 1959, on Per Capita Total Operating Expenditures (Tot.Sp.) of 36 Paris Core Cities in 1963........................167 Partials, Betas and Regression Coefficients for the Effects of Resources, Income and the Interaction of Resources and Communism (CISTRPSBxR) on Per Capita Total Operating Expenditures (Tot.Sp.) of 36 Paris Core Cities in 1963...................... ...... ..167 Partials, Betas and Regression Coefficients for the Effects of Resources, WOrkers, Population and the Interaction of Resources and Communism (CISTRP53xR) on Per Capita Total Operating Expenditures (Tot.Sp.) of 36 Paris Core Cities in 1963........................169 xii Table IV-39 IV-40 IV-4l IV-42 Page Partials, Betas and Regression Coefficients for the Effects of Income, Resources, Communist Representation, 1959, on Per Capita Business Taxation Revenue (Bus.Tax.) of 36 Paris Core Cities in 1963.... ............... . ............................ 169 Partials, Betas and Regression Coefficients for the Effects of Income, Resources and the Interaction of Resources and Communism (CISTRP53xR) on Per Capita Business Taxation Revenue(Bus.Tax.) of 36 Paris Core Cities in 1963-00.000000000000000.0000000000171 Partials, Betas and Regression Coefficients for the Effects of Youth, Growth and Left Representation, 1959, on Per Capita Operating Expenditure Grants (REPART) of 36 Paris Core Cities in 1963...............l71 Partials, Betas and Regression Coefficients for the Effects of Youth and Growth on Per Capita Investment Expenditure Grants (SUBV) of 36 Paris Core Cities in 1963....................................l72 Selected Per Capita Policy Areas of 18 Provincial French Cities, Grouped by Type of Partisan Control of the Municipal Councils in 1963 ...................... 183 Intercorrelations Among Sixteen Independent Variables in 1963 (18 City Sample)......... ..... .......185 Simple Correlation Coefficients Between Independent Variables and Per Capita Expenditure Categories for 1963 (18 City Sample).......... ................. .......138 Intercorrelations Between Per Capita Expenditure Categories for 1963 (18 City Sample)...................189 Partials, Betas and Regression Coefficients for the Effects of Resources and Left Representation on Per Capita Social Welfare Spending (S.W.Sp.) of 18 Provincial French Cities in 1963.------..........191 Partials, Betas and Regression Coefficients for the Effects of Resources and Communist Representation, 1959, on Per Capita Social Welfare Spending (S.W.Sp.) of 18 Provincial French Cities in l963--~--- ----- ------------19l Partials, Betas and Regression Coefficients for the Effects of Resources and Communist Representation, 1953, on Per Capita Social Welfare Spending (S.W.Sp.) of 18 Provincial French Cities in 1963-----------------o-----192 xiii Table V-8 V-lO V411 V-12 V—13 V-14 V415 V¥16 Page Partials, Betas and Regression Coefficients for the Effects of Resources, Income and Communist Representation, 1959, on Per Capita Social Welfare Spending (S.W.Sp.) of 18 Provincial French Cities in 1963.......................................... ...... 192 Partials, Betas and Regression Coefficients for the Effects of Resources and WOrkers on Per Capita Social Welfare Spending (S.W.Sp.) of 18 Provincial French Cities in 1963..................................194 Partials, Betas and Regression Coefficients for the Effects of Resources, Workers and Communist Repre- sentation, 1959, on Per Capita Social Welfare Spending (S.W.Sp.) of 18 Provincial French Cities in 19630000000000.0000!a...00000000000000.0000... ...... 194 Partials, Betas and Regression Coefficients for the Effects of Youth, Resources, Income and Communist Representation, 1959, on Per Capita Educational Spending (Ed.Sp.) of 18 Provincial French Cities in 1963.00.00.00000000...COOIIIOOOOOO0.0.0.0000...0.00.195 Partials, Betas and Regression Coefficients for the Effects of Growth and Youth on Per Capita Educational Spending (Ed.Sp.) of 18 Provincial French Cities in 196300....OOOOOOOOOOIOOOOOOOOOOOOO0.000.000.0000.00.195 Partials, Betas and Regression Coefficients for the Effects of Growth and Resources on Per Capita Educational Spending (Ed.Sp.) of 18 Provincial French Cities in 1963.........................................196 Partials, Betas and Regression Coefficients for the Effects of Resources, Income and Communist Repre- sentation, 1959, on Per Capita Total Operating Expenditures (Tot.Sp.) of 18 Provincial Cities in 1963................................................l96 Partials, Betas and Regression Coefficients for the Effects of Youth, Resources and Communist Representation, 1959, on Per Capita Operating Expenditure Grants (REPART) of 18 Provincial French Cities in 1963--------199 Partials, Betas and Regression Coefficients for the Effects of Income, Resources, Communist Representation, 1959, on Per Capita Business Taxation Revenue (Bus.Tax.) of 18 Provincial French Cities in 1963-o..-------------199 xiv Table V-l7 Partials, Betas and Regression Coefficients for the Effects of Income, Resources and the Interaction of Resources and Communism (CISTRP53xR) on Per Capita Business Taxation Revenue (Bus.Tax.) of 18 Provincial French Cities in 1963............................. ............200 MAP I CHART I LIST OF FIGURES Page Number Reorganization of the Departments of the Region, 1964-6800000000000000 ooooooooooooooo 50 Territorial Jurisdictions, Paris Region, 1964......OOIOOOOOO OOOOOOOOOOOOOOO O ..... 51 CHAPTER ONE INTRODUCTION A. Purpose of the Study A study of policy choices implies that policies are dependent variables to be explained by a host of independent variables,1 one subset of which can be so-called political variables such as: inter- party competition, party control and voter turnout.2 Initial results from research on the expenditures of American states and localities 1Studies structured in this way include Richard E. Dawson and James A. Robinson, "Interparty Competition, Economic Variables and Welfare Policies in the American States," Journal of Politics, 25 (May, 1963), pp. 265-289; Thomas R. Dye, Politicsi Economics and the Public: Policy Outcomes in the American States (Rand McNally, 1966); Richard I. Hofferbert, "The Relation Between Public Policy and Some Structural and Environmental Variables in the American States," American Political Science Review, 60 (March, 1966), pp. 73-82. For similar research on city policy, see John H. Kessel, "Governmental Structure and Political Environment: A Statistical Note about American Cities," American Political Science Review, 56 (September, 1962), pp. 615-620; Lewis A. Froman, Jr., "An Analysis of Public Policies in Cities," Journal of Politics, 29 (1967), pp. 94-108. Similarly structured were studies of the economics of public policy; see John F. Due, Government Finance: Economics of the Public Sector, 4th Ed. (Irwin, 1968) and more specifically Julius Margolis, "The Demand for Urban Public Services," in Harvey S. Perloff and Lowdon Wingo, Jr. (eds.) Issues in Urban Economics (Johns Hopkins Press, 1968), pp. 548-552. 2Dawson and Robinson, _p, cit.; Dye, gp._gi£. See also Otto A. Davis and George H. Haines, "A Political Approach to a Theory of Public Expenditure: The Case of Municipalities," National Tax Journal, Vol. 19 (September, 1966), pp. 259-275; and William C. Birdsall, "A Study of the Demand for Public Goods," in Richard A. Musgrave (ed.) Essays in Fiscal Federalism (Brookings, 1965), pp. 235-294, for a different conceptualization of "political" variables. 2 have shown such variables to have little, if any, independent effect when that of socioeconomic variables has been controlled.3 Interpretations of such findings4 have often revolved around 5 the "End of Ideology" debate which has had its counterpart in Europe 3Dawson and Robinson, 92, gig}; Dye, 22,.git.; Hofferbert, _p,.gi£.; Ira Sharkansky, Spending in the American States (Rand MCNally, 1966). Exceptions include Charles F. Cnude and Donald J. MCCrone, "Party Competition and welfare Policies in the American States," American Political Science Review, Vol. 63 (September, 1969), pp. 858-866; and Ira Sharkansky and Richard Hofferbert, "Dimensions of State Politics, Economics and Public Policy," American Political Science Review, Vol. 63 (September, 1969), pp. 867-879, where some effect of political "dimensions" is detected among so-called redistributive policies,most notably welfare. 4The results of Dye and others have also been used by Dye himself in his common effort with Harmon Zeigler, The Irogy of Democragy: An Uncommon Introduction to American Politics (Wadsworth Publishing Company, 1970) as evidence of policy control by a non-political (the military- industrial complex) alliance of elites. Similarly, T. J. Lowi points to the irrelevance of the Conservative-Liberal debate in the major policy decisions of the recent years in his End of Liberalism, (Norton, 1969). Policy is viewed as the result of an unstructured bargaining between interest groups and governmental bureaucracies, and therefore devoid of any political ideological pattern. 5The debate as to the end of ideology is an old one and is best known from the work of Daniel Bell, The End of Ideology: On the Exhaustion of Political Ideas of the Fifties, new ed. rev. (Collier Books, 1962). An updating of the literature, arguments and research findings can be found in Harvey Waterman, Political Change in Contemporary France (Charles E. Merrill Publishing Company, 1969). As for a rebuttal to the hypothesis, see Chaim I. Waxman (ed.), The End of Ideology Debate (Funk and Wagnalls, 1968) especially the papers by C. Wright Mills, Henry D. Aiken, William Delaney, Robert A. Haber and Irving Louis Horwitz. 3 H ’ - "6 in the phenomenon described by some French scholars as depolitisation. It is therefore pertinent to ask whether the same process can be said to have affected European (French) budgetary patterns, and whether these also reflect mainly socioeconomic forces. This study aims to do just that. We hypothesize that in France, regardless of any so-called signs of depoliticization7 among the electorate, the budgets of French municipalities for any one year, still reflect strong partisan and political choices, more specifically, we expect that the variable "party control" has an important (statistically significant) effect on the spending policies, (i.e., welfare) of 6See Jean-Yves Colvaz, "L'idée de dépolitisation: Jugement de valeur," in Georges Vedel (ed.) La Dépolitisation (Caliers de la Fondation nationale des sciences politiques, N:120: Paris, Armand Colin, 1962), pp. 181-182; and other articles in the same source by Georges Lavau "Les aspects socioculturels de la depolitisation," and by Henri Mendras "Politisation, dépolitisation, repolithation du milieu rural." For formulations of the same problem in England, see Norman Birnbaum, "Great Britain: The Reactive Revolt," in Morton A. Kaplan (ed.) Revolution in World Politics (J. E. and Sons, 1962), pp. 46-47. Part of this debate focuses on the class basis of politics and the contention that sections of the working class are voting conservative in Europe; see Seymour M. Lipset, "The Changing Class Structure and Contemporary European Politics," in Mattei Dogan and Richard Rose (eds.) European Politics (Little and Brown, 1971), pp. 146-158; and in the same volume, articles by Otto Kirchheimer, "The Waving of Opposition in Parliamentary Regimes," pp. 280-295, and Richard Rose, "Class and Party Divisions: Britain as a Test Case," pp. 159-182. Also see Ralf Dahrendorf, "Recent Changes in the Class Structure of European Societies," Daedalus (93, L, 1964), pp. 232-244. 7vedel in his "Rapport Introductif," 22, gi£., p. 20, denies that electoral participation and organizational membership as indicators of depoliticization have declined in France or Britain. Moreover, there is a question as to how much politicization has already existed in the French system. Philip E. Converse and Georges Dupeux, "Politicization of the Electorate in France and the United States," Public Opinion Quarterl , 26 (Spring, 1962), pp. 1-23, show that party identification is weakly transmitted among French families. Similarly, Georges Lavau has claimed that France has fewer associations than other Western societies in his essay, Partis politigues et realites sociales (Paris, Colin, 1953). This has more recently been contradicted by Duncan MacKae in Parliaments, Parties and Society in France, 1946-58 (St. Martin's, 1967). 4 French Communes even after "socioeconomic" variables have been controlled. Through regression analysis, we expect to find that the effect of party control is much more pronounced on Social Welfare policies than on either Education Spending or Total Spending. This also means that Communist and Socialist administrations tend to spend more in general, than equally wealthy non-Left administrations. B. Concept Formation 1. Some Theoretical Considerations Imbedded within the "End of Ideology" debate are numerous problems of conceptualization, the most acute of which is the meaning of the concept of "ideology" itself.8 All too often it has been used inter- changeably to describe the political belief systems of mass publics as well as those of certain elites and organizations. We concur here with the previous writers on the subject9 that if ideology is to mean anything it must exhibit constraint; that is, stand for a "belief system that is 8We must differentiate between ideology as a problem in cognition and social context derived from Karl Manheim's Ideology and Utopia, translated by Louis Wirth and Edward Shils (Harvest, 1936); see Willard A. Mullins, "On the Concept of Ideology in Political Science," American Political Science Review, 66 (June, 1973), pp. 498-510, for the most recent treatment of an age-old problem -- and ideology in politics. While the two are related as Giovanni Sartori shows in "Politics, Ideology and Belief Systems," American Political Science Review, 63 (June, 1969), pp. 398-411, it is the latter which is our concern. 9Sartori, 923‘gi£., p. 401; see also, David W. Minar,"Ideology and Political Behavior," Midwest Journal of Political Science, 5 (1961), pp. 371-373, and Philip E. Converse, "The Nature of Belief Systems in Mass Publics," in David Apter (ed.) Ideology and Discontent (The Free Press of Glencoe, 1964). 5 internally consistent and consciously held."lO Moreover, it is reasonable to assume that mass publics seldom exhibit constraint in their belief systems and that it is through organizations (i.e., political parties) that the ideologies of elites affect the belief systems of mass publics on one hand and the outputs of governmental structures on the other.11 In fact, the latter (outputs) can in turn become a key tool in the perpetuation of the organization and its ideology.12 The degree to which organizations such as parties will continue translating their belief systems into bUdgetary patterns congruent with 10Samuel H. Barnes, "Ideology and the Organization of Conflict: On the Relationship Between Political Thought and Behavior," Journal of Politics, 28 (1966), pp. 513-530:514. See also Robert D. Putnam, "Studying Elite Political Culture: The Case of Ideology," American Political Science Review, 65 (September, 1971), pp. 651-681. 11Barnes, 92, gi£,, p. 515. For empirical validation we have of course the by now classic Herbert MCCloskey, et. a1. "Issues and Followers," American Political Science Review, 54 (June, 1969), pp. 406-427. For France, see Duncan MacRae, Jr., Parliament, Parties and Sociegy in France 1956-58 (St. Martin's Press, 1967). MacRae has been criticized by Howard Rosenthal for overemphasizing the role of ideology in the motivations of party militants, see "The Electoral Politics of Gaullists in the Fourth French Republic: Ideology or Constituency Interests?" American Political Science Review, 63 (June, 1969), pp. 476-487, and his Review Article, "Commentary on Duncan MacRae, Jr.'s Parliament, Parties and Society in France 1946-58," American Political Science Review, 63 (September, 1969), pp. 907-914. 12This feedback cycle is what Mark Kesselman characterizes as Over-Institutionalization in "Over-Institutionalization and Political Constraint: The Case of France," Comparative Politics, 3 (October, 1971), pp. 21-44. 6 . 13 . such beliefs, will vary between organizations but it Will depend also on the setting that saw their emergence -- the political institu- tionalization14 of past conflicts and cleavages and the nature and rate of social change -- political institutionalization is conceptualized here as a feedback cycle which includes as elements: the initial degree of hostility or consenses generated by the configuration of cleavage 15 patterns, at the early stages of modernization; the opportunity 13This will depend on the organization's size, membership, function and initial structure. See Anthony Downs Inside Bureaucragy (Little and Brown, 1966) for a good theorization of such factors. For applications to political parties, see Philip Selznick The Organizational Weapon (MoGraw-Hill, 1952). Finally, Richard M. Cyert and James G. March A Behavioral Theggy of the Firm (Prentice-Hall, 1964) have shown how the internal coalition formation within an organization will determine its decision-making. It must be noted that many explanations of organizational structure use exogenous variables. 4For definitions of institutionalization, see Talcott Parsons Esssys in Sociological Theory (rev. ed., Free Press of Glencoe, 1954), pp. 143-239. For a definition of institutionalization which takes the role of bureaucracy in a historical perspective, see S. N. Eisenstadt, "Initial Institutional Patterns of Political Modernization," Civilizations, 12 (1962), pp. 461-472, and "Institutionalization and Change," American Sociological Review, 24 (April, 1964), pp. 235-247. Finally, the most recent theorist on institutionalization is Samuel P. Huntington in Political Order in Changing Societies (Yale university Press, 1968), pp. 12-27. Huntington, however, does not dwell enough on the importance of the cleavage patterns in a society undergoing modernization. 15The literature here is quite voluminous although only recently has the linkage between cleavage patterns and policy outcomes been made. See Bingham Powell Social Fragmentation and Political Hostiligy (Stan- ford University Press, 1971) and B. Guy Peters, "The Development of Social Policy in France, Sweden and the United Kingdom: 1950-1965," in Martin O. Heisler (ed.) Politics in Eurgpe (MoKay, 1974), pp. 257-292. The earliest treatment can be traced to Karl Deutsch Nationalism and Social Communication (M.I.T. Press, 1953). See also, Seymour M. Lipset and Stein Rokkan (eds.) Party Systems and VOter Alignments (The Free Press, 1967); Stein Rokkan, Citizens, Elections and Parties (David McKay Company, 1970); and E. Allardt and Stein Rokkan (eds.) Mass Politics (The Free Press, 1969). Finally, Robert Dahl uses similar explanations of American historical development in a comparative perspective in three articles, "The American Oppositions: Affirmation and Denial," "Patterns of Opposition," and "Some Explanation," in Political Qppgsitions in Western Democracies (Yale University Press, 1966). Finally, in the same tradition is Robert E. Alford Party and Society (Rand-McNally, 1965). 7 . l6 . structures created by policy outputs and the degree to which social change is exogenous to this process.17 From various theoretical efforts we derive two models of political institutionalization,l8 one that is characterized as Fluid and the other as Structured.19 They in turn lead us to formulate two corresponding 30 models of policy-making which can guide our hypothesizing. 16 See Joseph Schlesinger Ambition and Politics (Rand-McNally, 1966) and Lewis Edinger (ed.) Political Leadership_in Industrialized Societies (Wiley and Sons, 1967). 7This implies that not only the rate of change, but iussources and channels are of importance. Moreover, the nature of previous institutionalization (i.e., monarchies, nobilities and merchant classes) will determine the patterns by which change itself may be institutionalized. Huntington, og._gi£. On a smaller scale, Anthony Downs in Inside Bureau- cracy,‘_p, 315,, pp. 5-10, exemplifies this in his concept of age lump phenomenon. 18 The inspirations for these come from Robert Dahl's "Patterns of Oppositions," in Political Oppositions in Western Democracies (Yale University Press, 1966), pp. 332-347. See Val Lorwin, "Segmented Pluralism: Ideological-Cleavages and Political Cohesion in the Small European Democracies," Comparative Politics (January, 1971), pp. 141-175; Bingham Powell, _p,‘oi£.; and particularly Arendt Lijphart, The Politics of Accommodation: Pluralism and Democracy in the Netherlands (university of California Press, 1968). For a recent elaboration of a model that summarizes much of the findings of the above cited literature and which bears striking resemblance to our Structured Policy-Making Model, see "Patterns of European Politics: The European Policy Model," by Martin O. Heisler in collaboration with Robert B. Kvavik in Politics in Europe, _p, gig}, pp. 27-90. 19The main variable that can be used to differentiate between the two models is the so-called class vote as operationalized by Robert R. Alford, 22:.EEE3: Ch. 5. But they can also be distinguished by the degree of party voting in the legislative roll-calls of each political system. The Structured Model includes political systems with or without ethnic cleavages, i.e., Austria, France, the United Kingdom, as well as Belgium. When ethnic cleavages exist, they are cumulative to class cleavages. (See Note 18, 15 and 12.) 0These models are not elements of competing and alternative theories of policy-making, they are analytical constructs with empirical referents (political systems) which may vary in their actual functioning from their respective model. Elements of both models may be present in specific subsystems of each society. 8 2. The Policy-Making Models The Fluid Policy-Making Model is characterized by constantly 2 fluctuating structures, 1 where bargaining occurs even over decision rules. Political leaders are often left to their own devices, subject to competing influences among which we find that of their constituents, the major interest groups (industries, unions) in their constituency, their party's ideology and their own perception of the national interest.22 As described in the American State Politics literature, community resources appear to exert such a pervasive influence that the range and number of policy options available to the community's political leadership are severely circumscribed. At the local level, those limitations are mani- fested partly through fears about the mobility of the community's tax 2 base, in the event of high tax burdens. 3 More importantly, however, 21By fluctuating structure we mean the changing coalitions with- in each of the major parties on one hand the constant formation of new regulatory bureaucracies. Both phenomena change the acors involved in the bargaining. See T. J. Lowi, op, gig}, pp. 29-97. 22See Warren Miller and Donald Stokes, "Constituency Influence in Congress," American Political Science Review, 57 (1963), pp. 45-56. For a testing of such formulations, see Raymond A. Bauer, Ithiel de Sola Pool and Lewis Dexter American Business and Public Polioy (Atherton Press, 1963). 3 Studies of the impact of taxation on the locational considerations of businesses are numerous and even though most locational decisions can be more readily explained by other factors (i.e., transportation, raw materials, labor force, etc.) and only marginally by taxes; the political use of the argument goes on. See Glenn W. Fisher Taxes and Politics: Study of Illinois Finance (University of Illinois Press, 1969), Ch. 6. 9 are the model's assumptions of a media (T.V.) informed, cross- pressured24 electorate, which gives preference to pragmatic vote maximizing political leaders. The latter are organized into part- time political parties with little discipline. Bureaucracies, private and public have much more permanence. The Structured Policy-Making Model assumes that political leaders are well shepherded by well organized political parties with tight control over distinct constituences25 seeking to maximize policy benefits for that constituency.26 The level of resources is limiting to some extent but the political party in power (party control) may chose to reorder priorities so as to maximize a policy of greater importance to its constituency. In this model, slow social change has strengthened the parties and the bureaucracies and made these all the more able to stem or alterate it. In fact, these institutions 4See Dahl, op, oi£,; also.David Truman The Governmental Process (Knopf, 1951) was among the first scholars to make the point that multiple memberships in groups or categories with opposite orientations, contributes to a less intense form of participation. 25See Anthony Downs, An Economic Thegry of Democracy, _p, gig. Downs makes the point that in multi-party systems there is an advantage for parties in keeping their ideologies distinct and representing only a small but well defined constituency. French constituencies have been characterized as "political chapels" and the working class as a "political ghetto." See Stanley Hoffman, "Paradoxes of the French Political Community" in Stanley Hoffman, et. al.,In Search of France (Harvard university Press, 1963), pp. l-ll7; and Mattei Dogan "Political Cleavage and Social Stratification in France and Italy," in Lipset and Rokkan Party Systems...op, gi£., pp. 129-195. See also Mattei Dogan, "Le VOte Ouvrier en France, Analyse ecologique des elections de 1962," Revue Francaise de Sociologie, 6 (December, 1965), pp. 435-471. 2 6It must be pointed out that hostility between classes limits the ability of parties to appeal to larger sections of the electorate, thus a stalemate ensues. 10 . . 27 are able to control the flow of most extra-systemic inputs better than is done in the Fluid Model. It is this ability of political parties to alter (even if it is to slow down) the impact of some basic socioeconomic forces that leads us to expect that party control in France will explain large parts of the variance in local Social Welfare spending. The French political behavior deviates somewhat from the Structured Model in that organizational membership is not as closely coincident with political party lines as in Austria.28 On the other hand, the existence of a Communist Party and of a particularly Stalinist Party at that29 increases the degree of rigidity in the attitudes of a large 30 part of the electorate and in the policy-making of certain cities. 27 See Kesselman, op, cit.;extra—systemic inputs for French local government would include policies of the central government (i.e., grants-in-aid) or of large corporations wishing to locate in the community. 28See Val Lorwin, op, oio, '29The existence of a Communist Party means the importation of extra-systemic tensions and the isolation of a large political party from national policy-making. Variations in the circumstances surrounding the growth of such a party will produce a more or less Kremlin relying organization. See Thomas H. Greene, "The Communist Parties of Italy and France: A Study in Comparative Communism," World Politics, 21 (October, 1968), pp. 1-38. On the pragmatism of Italian Communism, see Sidney G. Tarrow, "Political Dualism and Italian Communism," American Political Science Review, 61 (March, 1967), pp. 39-53. For the meaning of this pragmatism in the fiscal policy of Italian Communist municipalities, see Robert C. Fried, "Communism, Urban Budgets and the Two Italics: A Case Study in Comparative Urban Government," Journal of Politics (July, 1971). 30See Richard Hamilton Affluence and the French Worker in the Fourth Republic (Princeton University Press, 1967) for the persistent radicalism of the French working class; and Henry W. Ehrman, Politics in France (Little, Brown, and Company, 1968), p. 215. 11 The French Communist rigidity in policy-making is best exempli- fied by the comparison between the Italian Communist Party's effort at making Bologna a model of fiscal responsibility,31 and the fears of a French Communist mayor that his efforts at making Aubervilliers (a low income suburb in our sample) a model of slum clearance and low income housing, may attract a middle-class migration which may put him out of office. Thus, the Structured Model may still be useful in developing hypotheses about French policy-making, at the local level. 3. The Concepts 33 Policies vary in salience to the public. Budgetary policies such as taxation and expenditures are not too meaningful to the electorate. But it has been found that the organizations (i.e., interest groups) who compete for the benefits of such policies usually 31See Fried, op. cit. 2For comments on Communist policy in the municipalities, see Paul Thibaud, "Le Communisme municipal," Esprit, 34 (October, 1966), pp. 413-422. For a more partisan viewpoint, see Jacques Duclos, "Les Municipalités Communistes au Service des Populations," Nouvelle Revue Internationale, 9 (May, 1966), pp. 89-99. 33The concept of salience is mainly to be found in voting and public opinion studies, but the first research results are to be found in Angus Campbell, Philip E. Converse, Warren E. Miller and Donald E. Stokes, The American Voter (John Wiley and Sons, l960)and by the same authors in Elections and the Political Order (John Wiley and Sons, 1966). For research relating salience to party identification, see V. 0. Key, Jr. Public Opinion and American Democracy (Alfred A. Knopf, 1961), p. 445; and John G. Pierce, "Party Identification and the Changing Role of Ideology in American Politics," Midwest Journal of Political Science, 14 (February, 1970), pp. 25-42. 12 behave differently according to how redistributive a policy is. Since V. O. Key's formulations, it is those so-called have-not policies which are expected to be more determined by political variables. Resources and pesos are two concepts which help us understand how the so-called socioeconomic variables or the level of economic development impinge on budgetary policies.36 They are also derived from the Eastonian systems framework37 (demands and supports), which guides most studies of policy choice. While resource indicators (total amount of goods available for taxation) are usually highly correlated with oooo_indicators (level of income), a distinction is always worthwhile given the differing patterns of income distribution, but it is all the more useful when the 34For a classification of policies, see Robert Salisbury and John Heinz, "A Theory of Policy Analysis and Some Preliminary Applica- tions" in Ira Sharkansky (ed.) Polioy Analysis in Political Science (Markham, Chicago, 1970), pp. 39-59. See also the earlier typology by T. J. Lowi in his "American Business, Public Policy, Case Studies and Political Theory," world Politics, 14, #4 (July, 1964), pp. 677-715. 35V. 0. Key, Jr., Southern Politics (Vintage Books, 1949), p. 307. 36Dye, 22:.EiE- More recently Ira Sharkansky has attempted to conceptualize what he calls resource-polioy as opposed to need:polioyo linkages in "Economic Theories of Public Policy: Resource-Policy and Need-Policy Linkages Between Income and Welfare Benefits," Midwest Journal of Political Science (June, 1972), pp. 722-739. We do not, however, agree with his approach. 37David Easton, The Political System (Alfred A. Knopf, 1953) is the main work cited by policy analysts, but Karl Deutsch's systems model should also be considered. See his "Social Mobilization and Political Development," American Political Science Review, 54 (September, 1961), pp. 493-514. 13 testing universe is made up of urban units where migration from place of residence to place of work results in sharp disparities between resource levels and personal income for the same city. If we are to measure the true impact of party control on spending for any one policy, we must first know how much a city spends in relation to an existing need and given a specific level of resources. Personal income is a valid indicator of need in Social Welfare as well as education since poverty should always result in a greater demand for public services. In education, private schools are available to the rich so that cities with poor populations will need more and larger public schools. Similarly, the size of the school-age population will also reflect the need for that service. The concept of social structure is also imbedded within that of economic development, through the usual linkage between social class and personal income. A community's social structure is usually measured by the number of people belonging to specific socioeconomic categories. Among industrial societies the proportion of the community's population belonging to the so-called working class is one commonly used measure of social structure. Distinguishing between social class and personal income may be difficult.38 It is nevertheless useful to include it in policy analysis for if its impact on policy can be distinguished from that of personal 38Difficult because of problems in multicollinearity that arise in regression analysis when two or more independent variables are highly correlated. See Hugh Donald Forbes and Edward R. Tufte, "A Note of Caution in Causal Modeling," American Political Science Review, 62 (December, 1968), pp. 1258-1264; and Hubert M. Blalock, Jr., "Correlated Independent Variables: The Problem of Multicollinearity" in Edward R. TUfte Theoguantitative Analysis of Social Problems (Addison-Wesley Company, 1970). 14 income, we would be able to know if policy makers respond to objective needs or to socially defined needs (since social classes differ in their definition of the same objective need). Social structure is also usually correlated with party control in that a high proportion of the working class vote for a Left party. Again, distinguishing their effect on policy may be difficult, but nonetheless useful to the interpretation of our results. Three possibilities can be anticipated. If the effects of the variables cannot be differentiated,39 then we must accept a null hypothesis that party structure is essentially a reflection of the social structure. If, however, social structure is a more powerful explainer, then we would have to discount the importance of parties as organizations shaping policy alternatives. If party control seems to have a noticeably stronger effect, then we can conclude the political organizations amplify the demands of their constituents, as described in the Structured Policy-Making Model. Party control is the concept which will reflect the strength of parties on the city council. Since time reinforces the effects of party control on budgetary patterns, we will use more than one election to derive indicators of party strength. Ideally, we would have liked to have had the size of party membership or the size of politically meaningful organizations (i.e., unions and church-going populations) in order to 9 If through substitution of one variable for the other, there is no noticeable (significant) increase in explained variance. 15 test out other elements of the Structured Model but unfortunately 40 such data is hard to come by. Party control is also operationalized through a measure of the size of the Left representation (Socialist and Communist councilors are added up) on the city council because we assume a certain continuity in the intensity and consistency of policy choices (more spending) between the two parties. We are, thus, still measuring the degree to which the ideology has been institutionalized. C. The Setting and the Study Sample The study sample which will be used for the testing of hypotheses includes 54 urban communes of over 10,000 population. Thirty-six of these are situated in the immediate Paris urban area (excluding Paris proper which is centrally administered), nine are in the outlying suburban area, eight of them represent the Lyon urban area in the Southern part of France (including the city of Lyon proper). Except for the cities of Lyon and Clermont-Ferrand, which can be called agglomeration centers, all the other communes are essentially satellites of such centers. The distribution by population size for these satellite communities is: 10,000 - 20,000 9 communes ( 3 in Paris) 20,000 - 50,000 29 communes (21 in Paris) 50,000 - 99,000 12 communes (11 in Paris) 100,000 - 120,000 2 communes ( l in Paris) Lyon and Clermont-Ferrand have populations respectively of 524,000 and 134,000. 40Communist unions such as C.G.T. are very secretive about membership and even when such data is given it is not broken down by communes (cities). Moreover, the reliability is questionable. 16 The budgetary year chosen is 1963, midway between the municipal elections of 1959 and 1965. The data were collected by me from French archives in the summer of 1970. Population figures used are those of the 1962 Census. The commune is the only unit of French local government whose executive officer, the mayor, is elected (indirectly by the municipal councilmen whose own election comes every six years). In 1953 the elections took place under the Proportional Representation System, which results in municipal councils reflecting the approximate strength of each major political party's vote. In 1959, the Two-Ballot Majority System, which encourages coalitions to form.before elections, and even more urgently before the second ballot, was used. While absolute majority is necessary on the first ballot, a simple plurality is all that is needed on the second. While the French central government has important statutory and supervisory powers, the right of a commune to establish new services as new needs arise has always been accepted.41 Central influence is most important in the fields of education and security. Grants-in-aid are for the most part designed to finance new investments. Loans are also available but regulated by a central credit organization. Despite the fact that French municipalities cannot float bonds and are not For works on French local government, see F. Ridley and J. Blondel, Public Administration in France (Routledge and Kegan, Paul, London, 1964); Brian Chapman, Introduction to French Local Government (George Allen and Unwin, London, 1953); and Mark Kesselman, The Ambiguous Consensus: A Study of Local Government in France (A. Knopf, 1967). l7 supposed to go beyond certain limits of indebtedness, in practice the variation in contracted debts is considerable.42 Similarly, the form and incidence of local taxation varies substantially, even though here again formal guidelines do exist. Thus, we can see that French communes are quite suited for the type of analysis we wish to undertake. D. Implications of this Study This study straddles many fields, sub-fields and frameworks. As a policy analysis study it of course has relevance to this mushrooming field, which itself has links to the comparative study of administration and practical policy-making (i.e., Planning, Programming, Budgeting).43 42 Joseph Duplouy, Le Credit aux Collectivities Locales (Berger- Levrault, Paris, 1967). 43See Notes 1 and 2, and of course Solomon Fabricant, The Trend in Government Activity in the United States Since 1900 (National Bureau of Economic Research, 1952); and Glenn W. Fisher, "Determinants of State and Local Government Expenditures: A Preliminary Analysis," National Tax Journal (December, 1961), pp. 349-355. Also by the same author, "Interstate Variations in State and Local Government Expenditures," National Tax Journal (March, 1964), pp. 57-74. Finally, see Seymour Sachs, Robert Harris and John T. Carroll, The State and Local Government...The Role of State Aid, Comptroller's Studies in Local Finance, #3 (State Department of Audit and Control, 1963). For more pragmatically oriented, see Austin Ranney (ed.) Political Science and Public Policy (Markham, 1968). This literature was first spurred by Charles J. Hitch and Roland N. McKean, The Economics of Defense in the Nuclear Age (Harvard university Press, 1960) and E. E. Quade (ed.), Analysis for Militapy Decisions (Rand McNally, 1964) and furthered by Fremont J. Lyden and Ernest G. Miller (eds.), Planning,_Programming, Budgeting_(Harvard university Press, 1965). For a more recent and more critical view of this approach, see Robert H. Haveman and Julius Margolis (eds.), Public Expenditures and Policy Analysis (Markham, 1970). 18 As a study of local government, its findings can be compared to those in the field of American state and local government, but even more to those in the new cross-national study of local governments.44 As a case study of French local government, it is likely to be of interest to those scholars specializing in the study of whole systems (Western European democracies, and more specifically, France). Since quantitative studies of French institutions are few,45 an important contribution can be made by demonstrating their feasibility. For the specific field of French local government, we hope that our study demonstrates that despite the high degree of centralization presumed to prevail in France, local governments still exercise their policy options and differ widely in their policy choices. 44 See James Alt, "Some Social and Political Correlates of County Borough Expenditures," British Journal of Political Science, 1 (January, 1971), pp. 49-62, and NOel T. Boaden and Robert R. Alford, "Sources of Diversity in English Local Government Decisions," Public Administration, 47 (Summer, 1969), pp. 203-23. Also, F. R. Oliver and J. Stanyer, "Some Social Aspects of the Financial Behavior of County Boroughs," Public Administration, 47 (Spring, 1969), pp. 169-184. Robert C. Fried, "Politics, Economics and Federalism: Some Aspects of Urban Government in Mittel-Europa," unpublished paper delivered at the conference on "Problems of EurOpean Bureaucracy" of the Europe Committee of the Comparative Administration Group, American Society of Public Administration, university of Indiana, October, 1970. For a summation of such findings and an exhortation to the creation of a field of comparative local politics, see Mark Kesselman and Donald B. Rosenthal, "Local Power and Comparative Politics: Notes Toward the Study of Comparative Local Politics," in Rodney P. Stiefbold (ed.) Frontiers in Urban Research (university of Miami Press, forthcoming). 45See Jean Meynaud and Alain Lancelot, Les Attitudes Politiques (Paris: Presses Universitaires, 1964) and finally, Howard Rosenthal "The Electoral Politics of Gaullists in the Fourth French Republic: Ideology or Constituency Interest?", American Political Science Review, 63, #2 (June, 1969), pp. 476-87; "voting and Coalition Models in Election Simulation" in William Coplin (ed.) Simulation and the Study of Politics (Markham, 1968), and "Size and Winning Coalitions in the Fourth French Republic" in Sven Groennings et. al. (eds.), _p, oi£,; and more recently, Howard Rosenthal and Subrata Sen, "Electoral Participation in the French Fifth Republic," American Political Science Review, 67 (March, 1973), pp. 11-54. 19 In this chapter we have delineated the main problem, explained the conceptual tools to be used, and described the sample and the setting. In Chapter 2, we focus on a thorough description of the French setting; the political institutionalization patterns and their mani- festations at the local level at the time when the French system is undergoing unprecedented change. A description of the sampling universe and the sample itself through some basic characteristics will hopefully situate the reader. Chapter 3 will describe the research design, and as such link theory to hypothesis formulation through the construction of the variables as required by the methodology to be used. Coverage of specific literatures and previous research efforts will be undertaken as it becomes necessary to the buttressing of descriptions, characterizations or explanations of specific choices (i.e., choices in research design). Needless to say, this author (as others before him) will not shy away from comments, on substance or methodology, which although seemingly peripheral to the actual experi- ment, may in his opinion be useful in fleshing out the implications of his more central arguments. Chapter 4 will report on the testing of the hypotheses in the immediate Paris area. Chapter 5 will examine how the same hypotheses fare in the remainder of the sample and draw from these findings the qualifications that can legitimately be made. Finally, Chapter 6 will summarize our findings and suggest possible improvements for any future replications. CHAPTER TWO THE SETTING: CHANGES IN THE FRENCH STALEMATE In this chapter we will describe French economic and social development in the context of the Structured Model of Policy-Making and the particular institutionalization process which it entails. In the first part of this description, we will focus on the institutionalization of cleavages, then turn to the growth of the administrative system and explain its general principles of operation today. The special status of the Paris area will be of central concern here, since it is to be the site of the testing of our major hypotheses. This will raise the question of the area's representa- tiveness of French urban life and the justification for our choices in research design. The implication is that Parisian urban settlement and politics, particularly as embodied in the suburban inner ring of communities, display the closest approximation of the Structured Model of Policy-Making to be found in France. It must be specified, however, 1 that even in that area major departures from the ideal model exist. lwe have throughout the course of this description of the French political system, as well as in the conceptualization outlined in the previous chapter, relied extensively on the model of institutionalization as described by Rodney P. Stiefbold in "Segmented Pluralism and Consociational Democracy in Austria, Problems of Stability and Change," in Martin O. Heisler, et. al. (eds.), Politics in Europe, Structure and Processes (David McKay, 1973), pp. 117-177. 20 21 The existence of a multitude of parties, the relative ambiguity of ideological boundaries between parties of the Right, and the relatively weak organizational encapsulation of an individual's life are some of the more easily observable deviations. Yet the Structured Policy- Making Model still applies, because political parties find it in their mutual interest to perpetuate its features in the patterns of policy outputs. If some variations from the Model have always existed in Paris, they are all the more numerous in the new areas of urban settlement. This is what we will describe in the last part of this chapter, in an attempt to show how the cities of our sample fit within the context of these changes. This in turn will set the stage for some of the choices in sampling design to be more thoroughly described and explained in Chapter Three. There is no doubt that the French nation has undergone some radical changes since the end of world War II. Works on France have had to be revised as if they dealt with a "third world" nation. Various authors citing the periodicity of past French crises clearly predicted one for 2 Apres-Gaullisme or Post-Gaullism. A major crisis did occur in 1968, 3 but De Gaulle was still in power. Since his retirement and death, The substance of these predictions varied from expectations of a return to Fourth Republic ministerial crises to major confrontations. See Ch. 12, "Aprés De Gaulle," in Lowell G. Noonan France: The Politics of Continuity in Change (Holt, Rinehart and Winston, 1970). 3Various commentators in L'Express have made the point that Apres- Gaullisme has been so successful because in reality it began while De Gaulle was still in power and thus a transition period was provided. Some date the start of Aprés-Gaullisme to the Presidential Elections of 1965 when De Gaulle's showing in the first round was somewhat disappointing. See Philip Williams, "The French Presidential Elections of 1965," Parliamentary Affairs, XIX:1 (1965-66), pp. 14-30 and Francois Goguel, "L'Election Presidentielle Francaise de Decembre 1965," Revue Francaise de Science Politigue, XVI:2 (1966), pp. 242-44. 22 the French system has gone through two presidential, one municipal and one legislative election and despite some problems, neither the Fifth Republic nor the General's Party seem about to be repudiated by the citizens of this old nation.4 French politics today are becoming more and more like those of other industrial nations concerned with pollution, growth, the quality of life, political apathy and corruption in high places.5 The focus is also on new emerging groups like the young and women.6 The issues of monarchy as opposed to Republicanism, the separation of religion from the State7 have almost disappeared and even the old fear of Communist participation in govern- ment seems to have receeded extensively.8 The Communist Party itself, which as we described in this study has been a paragon of rigidity and 4The most recent test for the Gaullist "system" was the legislative election of 1973 where the UDR (ex-UNR) was challenged by the largest Left Wing alliance since the Popular Front. 5 The career of Jacques Chaban Delmas, the first Prime Minister to serve President Pompidou was seriously damaged by disclosures that he had not paid taxes for years. 6 More recently an important issue of French public life has been the right to abortion for the women of Catholic France. 7See Rene Remond, et. a1., Forces Religieuses et attitudes politiques daus la France contemporaine (Paris, Colin, 1965) for a series of surveys on French religious attitudes and the problems of Left Wing Catholics. The last major clash on religious matters was over governmental aid to Catholic schools. There have been skirmishes on matters of censorship and the recent abortion issue, but not much else. 8The literature and the number of inquires made through surveys on this subject are numerous. See Institut Francais d'Opinion Publique, Les Francais devant 1e Communisme (mimeographed, Paris, February, 1966) "Questions au Parti Communists." Most findings show that some 60% of Frenchmen find the Party more conciliatory and only 30% would oppose its participation in government. See Sondages, N-l (1966), pp. 65-67. 23 atavism, seems to be undergoing revitalization and change under its young new leader, Georges Marchais.9 A. Development and the Institutionalization of Cleavagos If we accept that Stanley Hoffman's depiction of French society as "Stalemated,"10 was relevant until recently, now it can be said that the Stalemate is in process of dissolution. The shock of World War II; the collapse of 1940; and the occupation and soul searching11 that followed it made its continuation impossible even though at first glance the Fourth Republic seemed like a repeat performance of the Third. Nevertheless, many of the institutions of that Stalemate remain and in 1963, the year for which we have data reflective of policy-making, few major administrative changes had yet gone into effect.12 Most observers of French history will agree that the Stalemate lasted so long, partly because it could be afforded by a nation that had essentially solved some basic problems of national-building that seem unsolvable for most political communities.13 First, an abundant agricultural production from some of the richest soil in Europe, minimized the need to trade and industrialize. Second, The recent Popular Front alliance is credited to have been partly the result of his efforts and diplomacy. 10In Search of France (Harvard University Press, 1963), pp. 3-20. 1Most indicative of that soul searching is the recent excellent motion picture documentary "Le Chagrin et la Pitié" (The Sorrow and the Pity). Its scholarly quality and fairness to both sides has surprised many. 12The choice of 1963 was imposed on this researcher by the lack of uniformity in the adoption of certain budgetary innovations which would have created problems in comparability. 13Ehrman, _p, cit., p. 1. 24 the national identity has not been a question of concern to Frenchmen since Protestant King Henry IV converted to the Catholic faith in order to capture Paris and said that the city was well worth a mass.1 Even though some anti-Semitism still lingers on from the late 19th centuryls when the word MetEQue was used to describe a small wave of migrants from Eastern Europe,16 such manifestations of hostility to other Frenchmen for ethnic reasons are relatively mild and scarce. The Stalemate also lasted because it supported a blend of traditional and modern values and goals -- a "Civilisation" that has captivated even the imagination of non-Frenchmen.17 French modernity was and still consists of a series of no-trespassing signs to basic principles of industrial organization such as the division of labor and competition.18 These step signs took the form of: 1. The inheritance Systeml9 and the perpetuation of a multitude of small family farms in agriculture. 4 See Ernst Robert Curtius, The Civilization of France (Vintage Books, 1962), p. 85. 15 C There was a recent occurrence in the City of Orleans. 16See Charles Micaud, The French Right and Nazi Germany (Durban, 1943). The word Metegue is part of French slang today but in the 1930's it was used by many of the Right Wing groups such as the Maurassiens and Action Francaise. 17A most recent manifestation of such a phenomenon was the initiative of President Senghor of Senegal to foster a French-speaking community. 18It is interesting to note that other countries are today questioning the universality of these principles. 19The French system of inheritance forces the distribution of land to all the heirs of the deceased. 25 2. The small size of the non—agricultural labor force resulting from the greater attractiveness of farm life.20 3. The low rates of geographical mobility and urban migration.21 4. The slower urbanization and the few truly metropolitan centers. 5. The low rates of population and economic growth.23 6. The family firm in industry with antiquated business methods and a resulting weak investment market. 0Here again it is worth noting that the system of family farms is part of the American ideal of yeomanry that motivated the Homestead Acts of the 1860's. In America, however, Richard Hofstadfer tells us that this ideal was only partly fulfilled and not for long. See The Age of Reform (Vintage Books, 1955), pp. 1-23. 1 See Niles M. Hansen French Regional Planning (Indiana University Press, 1968), pp. 26-31. 22This feature has recently been sloganized to "Paris and the French Desert," the title of a book by J. F. Gravier, Paris et la désert francais (Paris, Le Portulan, 1947). 23The low rate of population growth (often negative) has been cited as a troublesome morale problem for the whole nation, provoking moreover a sense of insecurity in the face of the rise of a strong and populous Germany from 1870 on. Conversely the upward trend in population growth after WOrld war II has been cited also as an indication of a healthy development and dynamism. Strangely again, the limiting of population growth is a major concern in industrial nations today. In the light of other characteristics, one must ask if it is French society that has become modern, or the industrial nations that have been converted to its philosophical premises. 24Cited by various economists, this feature has come to be questioned lately as causing slow economic development. See Charles P. Kindleberger Economic Growth in France and Britain 1851-1950 (Har- vard University Press, 1965). There is no doubt, however, that French workers have particularly been given additional cause for alienation, when their bourgeois bosses (patrons) displayed a narrow conservatism that often resulted in bankrupcies and inefficiences. 26 7. The Protectionism which perpetuated these features both in agriculture and industry.25 8. The late and slow unionization of the labor force.2 The above mentioned features explain why it is difficult to specify any one date as the starting point for the Stalemate. Most observers have always focused on the unresolved conflicts brought about by the Revolution of 1789, and the inconclusive end of the other outbursts (i.e., the failure of 1848, the temporary character of the Third Republic of 1871). If we, however, observe the pattern of cabinet stability in the Third Republic, we cannot help but notice the contrast between the period preceding the Dreyfus Affair, and what was to follow. From 1896 to 1940 the Stalemate becomes less of a compromise and more of an overriding fear of change. The social and political meaning of the Stalemate was the pre- dominance of a bourgeoisie and of a consensus of values (stability, harmony, moderation and equilibrium, craftsmanship and esthetic consi- derations) derived partly from the defeated aristocracy and the independent peasantry.27 The urban proletariat was kept in a social and political ghetto from which, as Hoffman tells us, periodic thrusts 28 would be attempted to overturn the status-qua. But that, to some 25Protectionism commands much more respect in a country where mercantilism flourished under the "ancien regime." 26This is a feature of French society that is not easily deducible from the present strong position of some unions. But even today more than 75% of French workers do not belong to any trade union organization at all. Stanley Rothman European Society and Politics (Bobbs-Merill, 1970), pp. 171-174. . 27Hoffman, op, cit., pp. 4-5. 28Ibid; p. 7. 27 extent has been the experience of the urban proletariat in a large number of other nations in the process of industrialization.29 What gave special meaning to the French social and political ghetto was a series of factors that in combination increased the level of hostility and the feeling of entrapment of the classes. Patterns of social mobility were among the first. Even when it occurred, French social mobility has been described as one "where class barriers could be crossed but not destroyed. When one jumped over such a barrier, one had to leave one's previous way of thinking and living...."30 That such patterns can produce distinct subcultures among classes, even in the face of relatively fast rates of economic growth, can be easily observed among England's working class. As if this low rate of social mobility was not enough to increase hostility among groups, the religious conflict in France was to feature 9It is only in some of the underdeveloped nations today that industrialization is taking place in a climate seemingly emphasizing proletarian or peasant values. But most cases of Western industriali- zation display a dominance of bourgeois or aristocratic values, even where as in the United States, egalitarianism and a belief in the common man also prevailed. And, there is little doubt as to the class origins of major decision-makers in such systems. 30Edmond Goblot, La Barriere et le Niveau (Paris, 1925), p. 6. 28 a clerical/anti-clerical cleavage which is considerably more dis- ruptive in that it divides nations into communities of believers and unbelievers.31 Ironically it seems that countries where only one organized religion predominated, the Catholic bastions of the Counter Reformation,32 are those that were more affected by this conflict. In France, the Church was the subject of violent attacks from the time of the Revolution when it sided with the Monarchy.33 It suffered its worst defeats after the 34 Dreyfus Affair, when a complete separation of Church and State was 31Studies that find this type of cleavage as giving a special imprint to French political beliefs and conflicts include Philip Converse, "The Nature of Belief Systems in Mass Publics," in Ideology and Discontent, ed. David Apter (Free Press, 1964), pp. 206-261; Mattei Dogan, "Le VOte Ouvrier en France," Revue Francaise de Sociologie, 6 (October, 1966), pp. 435-471; and Vincent McHale, "Religion and Electoral Politics in France: Some Recent Observations," Canadian Journal of Political Science, 2 (September, 1969), pp. 292-311. Val Lorwin made the same distinction we make here in "Segmented Pluralism: Ideological Cleavages and Political Cohesion in the Small European Democracies," Comparative Politics (January, 1971), pp. 141-175; but it is in Derek Urwin and Richard Rose, "Social Cohesion and Political Strains in Regimes," Comparative Political Studies, 2, #1 (April, 1969), pp. 7-67, that this cleavage is found to account for most variations in political stability. 2France, Austria, Spain, we well as some of the American colonies of the latter (Mexico notably) have displayed a virulence in their anti-clericalism which culminated (as in the Spanish Civil War and the Mexican Revolution) with the execution of priests. 33The Church resisted the takeover of Church property and of Papal prerogatives (i.e., the nomination of bishops) by the State. The Church was also involved in Vendée in an insurrection against the Revolution. 34The Dreyfus Affair exemplified the cumulation of cleavages at its most intense level, in that anti-semitism was found to correspond with devout Catholicism and pro-militarism, while the forces Opposed to Dreyfus's incarceration had exactly opposite beliefs on all three levels. Thus, a miscarriage of justice came to symbolize the political struggle of two distinct forces. See D. W. Brogan, The Development of Modern France (revised edition, Harper and Row, 1966). 29 35 legislated and previous Concordats revoked. This was a time when even the groups on the Right were progressively parting company with the Church. The correspondence between the religious and class cleavages has been recently explained by two outstanding works which depict the roots and sources of radical discontent among workers in France and Italy.36 Both show how the anti-clericalism of the rural areas (agricultural workers angry at the Church's landholdings) has in Italy, as well as France, helped the recruitment efforts of unions and Left parties when new migrants arrive in the medium-sized, unemployment ridden cities of the French and Italian provinces. The religious conflict, moreover, creates on the part of those rebelling against Church authority a need for an associational life as rich and as all encompassing as that provided by the Church for its faithful.37 It is this need that the Communist Party has been able 35Adrien Dansette, Religious History of Modern France, VOl. 11: Under the Third Republic (Herder and Herder, Inc., 1961). See also Aline Controt et Francois Dreyfus, Les Forces religieuses dans la société francaise (Colin, 1965). See Sidney G. Tarrow, Peasant Communism in Southern Italy (Yale University Press, 1967), and Richard Hamilton, Affluence and the French Worker in the Fourth Republic (Princeton University Press, 1967). Also showing the same is the article by Mattei Dogan, "Political Cleavage and Social Stratification in France and Italy," in Seymour M. Lipset and Stein Rokkan (eds.) Party Systems and Voter Allignments (The Free Press, 1967). 37The similarities between the value structures of Catholic and Communist voters has been noticed since Gabriel Almong, The Appeals of Communism (Princeton University Press, 1954) propounded the idea that Communist strength resulted from the alienation of the working class. 30 to meet, and its eventual triumph over the Socialist Party in France must be seen in this perspective.38 In Austria, the Socialist Party quickly developed the needed parapolitical structures and this is why the Communist Party never grew even in circumstances of deeper hostilities than in France.39 Among the other political implications of the religious conflict was its linkage with the monarchy-republic debate which in my opinion is one of the most futile conflicts to have preoccupied Europe in the 19th Century."0 The institutionalized Stalemate could only be broken by exogenous factors. The Depression and the discontents it evoked broke the solidarity of the "status quo groups," the peasantry and the intellectuals. The various humiliations that led to Munich and Danzig dissolved the one 8 3 The French Communist Party underwent at first so many purges and splits that it was on the verge of extinction. It is understood, however, that this has also resulted in a more disciplined party, which was from the start an oddity in France. 39Even though the so-called Austrian Lager was a civil war fought by the paramilitary organizations belonging to each of the parties, the Communists were singularly unimportant throughout this period. Thus, it cannot be argued that their insignificance is due to the nearness of the Russian threat--there was no such threat in the 1930's. See Frederick Engelman, "Austria: The Pooling of Opposition," in Robert Dahl (ed) Political Qppositions in Western Democracies (Yale University Press, 1966), pp. 260-283. 40The futility of the conflict over support or opposition for a monarchy can be best appreciated by a view of the various divisions of the Right, making it impossible for responsible rule to take place. We discuss this particular problem later in this chapter. 41Hoffman, op, cit., p. 22. 41 31 truly nation-wide consensus on the country's international posture -- the so-called "grandeur." Others saw in the Communist Party an outside threat as unpalatable as the legions of Hitler. The collapse of 1940 made it plain that the Stalemate and its institutions could no more be afforded in a world where change exogenous or endogenous cannot be controlled or predicted. Laurence wylie's villagers will be buying items on credit the second time he visits them, not because they are confident in the economy, but because the alternative is just as risky. If the Stalemate was a series of compromises with the principles of modern organization, it did not apply to the French concept of the State, its system of representation of interests, and even less to its system of administration. Rationalized unitary control with all sovereignty vested in a parliament and exercised by a well trained pyramidal civil service, was put into application with so few departures from the theory43 that distortions appeared more because of this excess of rigorousness. The bureaucracy came to be a most important policy- making structure, in contradiction to Weberian principles, because of a 4 strict adherence to a Rousseau-inspired theory of democracy.4 The System of 42 Hoffman, op, cit., pp. 159-235. 3We refer to the seemingly strong French insistence that all shades of popular opinion be represented regardless of any costs in governmental stability, through the use of electoral systems that accentuate the number of parties and factions. 44The implications of Rousseau's idea of the "general will," namely that it somehow arises out of full discussion of all facets of a proposal, have influenced France's constitution makers, and particularly the Revolutionaries of 1789, 1848 and 1870. 32 Proportional Representation not only accurately reflected the community's "general will" (or lack of it), but institutionalized the socioeconomic Stalemate into a parliamentary immobilism which made policy-making by administrators inevitable.45 It must be observed that it is the adequate functioning of such a unitary national administration that made the Stalemated conflict, a tolerable reality in the National Assembly and in the municipalities where we will study its budgetary expression. It is because the national bureaucracy functioned with little ambiguity (expressing the values of the dominant middle-class), that local administrations could express their ideologies with lesser restraint. But unitapy control raises certain questions as to the degree of local autonomy and the uniqueness of areas where the administrative center is located. Such questions have a certain relevance to the appropriateness of the units of analysis chosen for this study. Let us, therefore, consider these two questions in order to rise above the superficial generalizations that have been made about French local autonomy and the uniqueness of Paris. B. Administrative Development and Local Autonomy It has been said that in France, political orders may come and go but administratively little has changed since Napoleon instituted his strongly centralized system. But in reality there have been constant changes though not easily detected because they rarely took 5It must be specified that France did not stay with a Proportional Representation system for long periods. Mbst of the Third Republic operated under the system of Two Ballot Majority, which today seems to lead to a bipolarity in the party system. 33 the form of new formal structures, but that of changes in administrative practice. Mere recently, however, there have been actual changes in structures and procedures and even one might say in the philosophy of French administration.46 1. Pre-Modern Period The pluralism of medieval France has been revived and is quickly gaining in legitimacy in a debate with the centralist and unitary conception of the State.47 Medieval France featured a pluralism with genuine self-government at the local level. Provincial "Parliaments" could impose taxes and exact ordinances while also acting as law courts. "Chartered towns had considerable powers of administering their own affairs through their own offices." Alongside charter towns were feudal fiefs and ecclesiastical domains and they each had distinct rights and duties. In fact, some provinces charged duties on goods and persons coming into their territory. The absolute monarchy of the Sun King and his successors gradually whittled these rights away and by the 18th Century representatives of the King (the Intendents) exercised fiscal, police and political powers in the provinces. "Appointed officials replaced elected officers in the towns and these were controlled by subordinates of the Iopendants, known as sous-délégués."48 46 This point has often been made. De Gaulle's Cabinets have included a large number of technocrats, and much of French moderni- zation after the War took place under the watchful eye of planners and Common Market bureaucrats. 7See Jean de Savigny, L'Etat Contre les Communes? (Seuil, 1970). 48F. Ridley and J. Blondel, Public Administration in France (Barnes and Noble Company, 1969), p. 85n. 34 The Revolution did not show any zeal for the idea of local autonomy. The latter would have meant accepting the legitimacy of pockets of royalist and ani-revolutionary sentiment. In fact, revolutionaries in France could rarely count on the support of the provinces."9 Ideologically the nation was viewed as an indivisible community with a Rousseau-type social contract. The State was its embodiment and there were to be no intermediaries between the citizen and his government. All interest groups of factions were viewed as dangerous in that they needlessly divide citizens and impede the process by which a "general will" appears. Thus, autonomous local governments were seen as just another divisive faction. This fear of intermediary mechanisms culminated in the "Loi Le Chapelier"of 1791 which prohibited associations.50 2. The Napoleonic System Napoleon made sure there were no such intermediaries between him and the citizen, at least none that he could not control! Napoleon was often out of France and needed a more predictable and of course controllable administrative system. He divided the country into 89 geographical areas which had no relationship to any previous political units and called these de'partements.51 9Time and again it will be Conservative rural France, the provinces that will send troops to recapture their capital city. In 1940 Marshall Petain will think it his duty to quickly conclude the War in order to come back with troops to squash another potential insurrection of the Capital, another Commune, as in 1870. 50See Ridley and Blondel, op, cit., "Introduction: The State and Society in France," by Peter Campbell, p. xiii. 51Ridley and Blondel, op, cit., p. 88. 35 He appointed an executive to administer them, borrowed the Roman title of prefect, made each of these officials capable of appointing every mayor in the area, and he had a system.which he could directly control through 89 men.52 It has been observed that this system resembled that of the Intendants, but with republican trappings.53 One of the important features of the French system of local government that must be grasped is that "unincorporated" areas do not exist. And conversely, incorporations cannot occur. This has not impeded the fractionalization of local government, which has occurred more because the multitude of units already existed. The territory was subdivided (under Napoleon) into 38,000 communes which can vary from a few hundred population to half a million (See Table l). The commune is the only unit that has survived from the "ancien regime" and while at first even their councils were appointed by the prefect in an advisory capacity, by the 1830's councils became elected and later received more powers (i.e., the law of 1871). By 1882, mayors ceased to be appointed and were elected by the councils of the commune as they still are today. Here we see evidence of an evolution from the Napoleonic system and become aware of the fact that true local autonomy dates from the Third Republic. 52Ibid, p. 92. 53Idem. 36 TABLE II-l COMMUNES BY POPULATION SIZE Population Size Number of 1962 Communes Population from O - 399 21,012 4,265,181 400 - 699 7,120 3,759,340 700 - 999 3,079 2,549,523 1,000 - 1,999 3,670 4,999,534 2,000 - 4,999 1,904 5,738,620 5,000 - 9,999 590 3,993,280 10,000 - 19,999 305 4,143,210 20,000 - 49,999 199 5,948,547 50,000 - 99,999 51 3,409,365 100,000 - 299,999 28 4,289,957 300,000 and higher 3 1,649,492 TOTALS 37,961 44,746,625 Figures compiled from a report called "Les Finances des Communes de plus de 5,000 habitants" (Exercise, 1965); Les Communes a l'Heure de la Region; Ministers de l'Interrieur; Direction Générale des Collectivites Locales Service de Statistiques et d'Analyses Financieres. From here on we will abreviate to "Les Finances des.... "54 Brian Chapman, Introduction to French Local Government (George Allen and Unwin, 1953), pp. 19-22. 37 3. The Administrative Structure Today The Units. Much of the administrative structure has, however, I 55 - survived. Ninety departements are governed by 90 centrally appointed prefects, presiding over locally elected councils (conseils generaux) located in the capital of the department.* The offices of the prefect are called the prefecture. Each département is subdivided into arrondissements (three or four in each, 450 in all), and each arrondissement is governed by a sub-prefect who has his offices in a city that is designated as the site of the sous-prefecture. It is a purely administrative unit and is used partly to supervise and advise the communes (100 to 150 in each) of the area. They also bring govern- ment closer to the people; they are the units of national legislative elections. They usually have about 100,000 citizens. Communes must, / for instance, send their budgets to the sous-prefecture for audits, and to insure that compulsory functions are adequately funded. These remain there for a few years until they are finally deposited with I the Archives Departementales, 5 5 The departements are classified into three classes. This is a device which helps give incentives to civil servants by making status differentiations and thus stimulating greater discipline and compliance lest a departementis rating be revised downward. The same grading system.exists for arrondissements. Finally, when we speak of 90 departements, it must be remembered that this was under the old system (changed in 1964), and that even then there were Overseas départements in addition to the metropolitan units (France Metropolitaine & France d'Outre Mer). 56That is where this researcher was able to secure the budgetary data for the analysis. ifWhen not underlined, "department" will be the anglicized version of departement. 38 Finally, each arrondissement is divided into cantons (35 in each departement and 3,000 in all). They are used for administrative purposes by the army and the judiciary and serve as units of election for the department's conseil general. The Officials. The two most important local officials are the prefect and the mayor. Brian Chapman says of the prefect: "At the head of the departmental administration and the hierarchic superior of all State officials in the departement, no matter what their 57 seniority, is the prefect." Ridley and Blondel classify his powers into four categories: (1) He is like the Lord-Lieutenant in an English country, the representative of the State: he wears a uniform on ceremonial occasions (recently drastically simplified by General De Gaulle), and takes precedence over all other persons in the department.... (2) He is the representative of the whole government (i.e., not merely of his own ministry, the Ministry of the Interior). He is officially the head of all govern- ment services in the department and is responsible for their coordination. His functions are political as well as administrative. He is the 'eyes and ears' of the government, and even now, periodically writes confi- dential reports describing the morale of the population.--- (3) He is the representative of the Ministry of the Interior, which is the ministry directly responsible for the supervision of local authorities. He is the middle link of the chain which goes from the central administration to the commune; helped by the sub-prefects, he supervises the activities of mayors and municipal councils. 57 Ridley and Blondel, op, cit., pp. 93-94. 39 (4) He is the chief executive of local government at the level of the department. Since he is appointed by and responsible to the central government, his position is independent of the departmental council of which he is an agent: it resembles that of the executive in a system of separation of powers. The fact that the same man can be an official of the central government and the chief executive of a local authority is of major importance in the working of French local government and in particular, for the nature of central-local relationships. They point out that: ....Governments have had and, theoretically at least, still have complete discretion in appointing prefects; they may even choose men from outside the civil service. They can dismiss them at discretion. In practice, however, the prefectoral career is now a regular civil service career. It is open to graduates of the Ecole Nationale d' Administration and, as one of the grands corps, it attracts some of the very ablest of them. A member of the corps starts as a sub-prefect in a minor sub-prefecture and gradually moves up to more important sub-prefectures before becoming a prefect; the same process is then repeated unless he moves to a post in the Ministry of the Interior. Prefects are frequently transferred, since governments have always felt it best to prevent them from forming too close ties with a particular area.59 Chapman describes how the prefect is assisted by other officials: The prefect is assisted in his duties by certain officers who are members of the copps ppefectoral. Firstly, he has a personal assistant, who is the most junior member of the copps, called the chef de cabinet. His right-hand man in the prefecture is the secretary-general, who is the head of all the administrative organizations of the department, and who also has certain responsibilities for the arrondissement in which the department capital is situated.... An important point about the career of the copps prefectoral is that there is no automatic direct line of promotion from the ranks of the sub-prefects to those of the prefects.... 58Idem. 59Idem. 40 The career of sub-prefect is today comparatively clear and straightforward, with entry as chef de cabinet and promotion by steps to the grade of hors classe sub- prefect. There it may stop. An hors classe,%ub-prefect may or may not achieve promotion to prefect.O Of particular importance for the smooth functioning of French administration is the posting of officials in charge of external services of particular ministries under the prefect's authority so as to facilitate coordination at the local level, while those, who recruited locally, remain there. Both types of personnel, however, belong to the national civil service. The external services which the French describe as "technical services" are called directions departementales,but some have retained their older names, inspections de l'Académie (education), service des Ponts et Chaussée (roads), trésoreries-paieries generales (finance). Ridley and Blondel point out that their officials in practice enjoy a considerable independence, and are generally "members of specialized civil service corps. They move from one department to another and also spend part of their career in Paris."61 The mayor is like the prefect, a representative of the State as well as head of his community and can often simultaneously be a deputy to the National Assembly. Elected by the municipal council every six years, mayors often remain in office for many decades. They enjoy a series of independent powers. Able to appoint officials and make by-laws, the mayor dominates above all because of a certain charisma. Modern urban problems have actually increased the glamour of the office. 0 Chapman, op, cit., pp. 23-24. 619p. cit., p. 96. 41 The set of interactions between the prefect and the mayor will determine to a large extent the content of policy at the local level. In this relationship, the mayor clearly enjoys most of the advantages if he cares to use them. The mayor has greater permanence, local support and popularity and multiple contacts with the central administration, through the local elected leaders and members of various ministries.6 If he happens to be a deputy, he can appeal to his fellow parliamentarians who may solicit his vote in various logrolling situations. The prefect, on the other hand, is the top law enforce- ment official of the area (he is in charge of the National Police). In a divided France, at least half of the population will see in him the repressive arm of the bourgeoisie. All the same, he is a bureaucrat and if he hopes to move up in the hierarchy, he had better have his There are numerous works on the subject of French mayors and their roles (official and unofficial at the local level). In the English language the most pertinent is Mark Kesselman's The Ambiguous Consensus: A Study of Local Government in France (Alfred A. Knopf, 1967). Also useful are some of the ethnographic studies in English of French village life, the most well known of which is Lawrence Nylie's Village in the Vaucluso_(Harper and Row, 1964); and also by Wylie, Chanzeaux: A Village in Anjou (Harvard University Press, 1966). See also Robert T. Anderson and BarbanaG. Anderson, Bus Spop for Paris: The Transformation of a French Village (Doubleday, 1965)- For a French view of the office of the mayor in French cities and towns, there cannot be more indicative a title than Robert Buron's Le Plus Beau des Métiers (Plon, 1963). See also Charles Schmidt, Le Maire de la Commune Rurale (Berger-Levrault, 1959); Jean Singer, Le Maire et ses Pouvoirs de Police (Berger-Levrault, 1959); Jean- Louis Quereillac, Un tel....maire (France-Empire, 1962); and Pierre \ wantzenriether, "Le Maire rural et son sous-prefet," Etude des problemes municipaux, #6 (June, 1965), p. 27. 42 63 . . . house in order. To achieve this goal, he W111 have to apply just the right kind of diplomacy mixing the carrots and sticks which his office allows him to use. a. Tutelage and Other Controls under the category of tutelage powers, Ridley and Blondel distinguish two types -- those exercised by virtue of the adminis- trative subordination of the mayor, whereby the latter can be suspended for not carrying out his duties or for acting improperly; and those powers of tutelage in which the prefect supervises in the very exercise of the function. Chapman specifies that while a prefect can have a mayor sus- pended for a month, only the Minister of the Interior has power to suspend a mayor for up to three months or dismiss him from office. This would, of course, happen upon advice to that effect by the local prefect. Chapman goes on to explain that: ....Normally, consistent refusal to perform their duties properly comes from mayors of communes which are firmly in the hands of an extremist party, and suspensions or dis- missal involve no change of political control, since the senior assistant mayor is generally designated to take the mayor's place. A strong majority of one party in a commune 3For works on the roles of the prefect, see Brian Chapman, The Prefects and Provincial France (George Allen and Unwin, 1955), and Jean PierreWorms, "Le Prefet et ses notables," Sociologie du Travail, #8 (July, 1966), pp. 249-275. 669p,.oi£., p. 104. See also Roland Maspetiol and Pierre Laroque, La tutelle administrative (Sirey, 1930); Jean Cathlineau, Le Controls des Finances Communales en France (Libraries Generales du Droit et de Jurisprudence, 1963); and on the subject of local autonomy, see Paul Lambin, "Projet de Defense de 1'autonomie des communes rurales," in Departements et Communes (July-August, 1960). 43 can make this form of tutelage extremely difficult to exercise; for example, when the Communist mayors of seven communes in the Department of the Seine (Paris) were suspended for three months, all the assistant mayors (who were also Communists) refused to take office, and eventually the Minister of the Interior had to nominate a councilor of the opposition to look after current business.... Chapman gives us some excellent illustrations of other sorts of clashes between prefect, mayor and municipal councilors. The prefect can also require the mayor to take steps which the prefect considers to be necessary in the interests of public order, and if the mayor refuses to obey the prefect can take the necessary measures in his own name, even though the law reserves the police municipale to the mayor. Several laws explicitly grant the prefect power to inter- vene in communal affairs if the communal authorities fail to perform their duties. For example, conseils municipaux who fail to provide the necessary facilities for public assistance institutions or who refuse to provide accommodation for State primary schools, may find their powers of acquisition used by the prefect. The latter provision has been a cause of frequent disturbance in the west, where there is a predominantly Catholic population which resents having to spend communal funds on State schools when it is forbidden to make grants towards the maintenance of Church schools. Otherwise, well- administered communes with sober conseils have come into conflict with prefectoral authority on such an issue, and a sub-prefect, supported by a squad of gendarmerie, has had to brave local wrath in order to extract the keys from the mayor and let the instituteur into the school. He was immediately faced with mass-resignations by the local conseils munioipaux, who used the new elections to mark popular disapproval.b6 These examples are of particular relevance given our choice of policy areas. Tutelage can be exercised in any of three ways: 1. The simple power of supervision allows the prefect to use delays and litigation in administrative courts to make any local decision more costly. 5 Chapman, _p, cit., p. 126. 66Ibid., p. 128. 44 2. Financial supervision means also that loans and grants that the community needs can be hopelessly delayed. If opposition by the commune creates certain administrative costs, they can be written into the latter's budget. 3. Certain large transactions require previous approval by the prefect. we will review the powers of tutelage more closely when we examine the financing of services among communes of our sample. Nevertheless, we can conclude that both the importance of the two local executives and the provisions for tutelage provide ample illustration of the unitary concept of the State, where local authorities exist only at the sufferance of the State. / The organization of services. Communes and departements are general-purpose units of government, but even though some administra- tive duplication occurs, departements have a greater role in some areas (highways) and communes in other areas (education). The commune, being nearer to the citizen, organizes most services and the departement is "left with the residuum." Police and education are two services the State is particularly concerned with. Thus, it regulates them more closely even though they both take up large chunks in a commune's budget. In education, Ridley and Blondel tell us that: "....powers are shared between the central government and the commune. The basis of the division is broadly as follows. The central government has been responsible for teaching staff and curricula ever since Napoleon; the provision and upkeep of school buildings and the employment of domestic staff are the reSponsibility of the commune. There are exceptions to this broad distinction. Large communes often run schools of their own (generally technical schools) where they appoint the staff and whose curricula they determine. The better grammar schools, 45 on the other hand, are state institutions and fall entirely outside the concern of local authorities. For all other public schools, however, the communes are responsible for the school building programme and are therefore able to 67 influence the pace of education development in their area...." We will have more to say on the administration of education when we consider the variety of services provided by the communes. b. Social Welfare The unitary conception of the State implies that the State inter- venes actively in many areas of social and economic importance. It also implies that such intervention can occur at whatever level of government, the central administration may choose. This has been the case in the field of Social Wehare which we conceive as emcompassing health and hospitals, public assistance, and Social Security. Hospitals are generally organized at the departmental level. A minimum amount of spending and some responsibility for public assistance are mandatory. Financial costs are apportioned between central government, depart- ment and commune, with the department spending the majority of such funds. (In fact, welfare accounts for fully one-half of a department's budget.) Moreover, a recent law reorganized the service and transferred most of the administrative work to its officials. Nevertheless, as Ridley and Blondel point out, "Claims for assistance are still made in the name of the commune, decisions are taken in the 67 .Qp. cit., p. 105. 46 first instance by a committee composed of all the mayors of the canton in which the applicant's commune is located, subject to appeal at departmental level."68 While the evolution of services is undoubtedly upward, that still does not eliminate a sizeable role played by communes willing to meet a need that is far from completely fulfilled by the activities of a relatively weak Social Welfare System. France has no national health service. There is a national insurance system (under Social Security) which reimburses for medical expenses. There are public hospitals where certain categories of patients are treated free of charge. There is, moreover, a whole patchwork of arrangements which helps the poor and indigent not only in money payments but also in kind (fuel and clothing). These services emanate from the local welfare agencies called "bureaux d'aide sociale." On the other hand, public hospitals are few. Certain categories of people who are only marginally able to pay are excluded from certain categories of aid. The system then displays a number of areas where additional aid is needed to fill an important need, thus encouraging other units of government such as communes to participate more than is required by law. The Social Security System through its generalized features as well as through its allocations to large families, and a miriad of other privately operated, but nationally subsidized insurance systems, put money in the hands of the public and make the construction of communal hospitals more possible. Large governmental grants are also available for large capital outlays. 47 The voluntarism of the system is such that private charitable institutions still exist providing services parallel to those of local authorities. Such institutions may be reimbursed for services rendered to persons entitled to assistance. we conclude that welfare services can be provided by local government, the social security organs, by the mutual aid societies, by charitable institutions, or by industry particularly in the nationalized sector. The central government provides welfare services for its own employees, for school children and servicemen. Supervision of standards and coordination of health services lies with the Ministry of Public Health and Population. The Ministry of Labour and Social Security is the other central supervisory administration. It will be noticed how both ministries have two basic functions which seem to overlap and involve a sharing of responsibilities. Both are administrative umbrellas for a host of agencies and services. We conclude that in the absence of a unified system, such as England's, local authorities have a large amount of autonomy in this policy area. Local hospitals once established cannot easily be eliminated.69 They acquire the status of a public corporation and although the mayor and three councilors (as well as three prefectoral nominees) are on each hospital board, these develop a large degree of autonomy even from the parent commune. Their sources of revenue include fees paid by patients, costs reimbursed through the Social Security and Social Aid System and local 69 Ibido, pp. 285-3130 48 government grants for investment. The ”bureaux d'aide sociale" are also public corporations and although each is almost completely financed by the commune, and is administered by a committee which includes the mayor and other councilmen, the prefect exercises close supervision, through his nominees to the committee. Sources of revenue are (1) assigned taxes (part of the local entertainment tax), (2) annual grants from the municipal and departemental councils, and (3) income from property. we will come back to the local financing and operation of welfare services later when we will also consider the special arrangements available in Paris. We can, however, conclude that when it comes to sum up what would maximize the degree of local autonomy (given what we know of the mayor- prefect relationship), we can definitely say that the commune's size is likely to be a crucial variable.70 The prefect can afford to be very difficult with a small village, which might happen to be the only Communist commune in the department. He is likely to be more careful with mayors of large industrial suburbs, where even if the whole departement and its conseil general are not controlled by the Communist Party, the latter is adequately represented and has many allies. This is particularly likely to happen in the"Red Belt of Paris." Thus, by chosing to sample in Paris and among communes of over 10,000 population, we are hoping to maximize the possibilities for large variances in policy, resulting from greater local autonomy, 791212: p. 107. Ridley and Blondel concur with this distinction between large and small communes. 49 even though communities of the Paris area are also more closely supervised by the central government. In addition, the higher standard of living of the area makes for larger resources on one hand, and expectations of higher standards of performance from the citizenry. But let us look at how and why Paris is so different. C. The Administrative Center in a Stalemated Society One implication of centralized unitary control is an over- development of the political and administrative center, which eventually becomes an economic center too. The site of the political center has a special administrative status in most nations. In France this is motivated even more by the fact that all revolutions and revolts have begun in Paris and ended there too. 1. Administration and Local Government in Paris until 1964, the Paris area included three départements -- the Seine, the Seine-et-Oise the Seine et Marne, and the City of Paris proper. The latter has had a special status within the Department of Seine (see Chart II-l). Since the expenditures to be explained in this study were made in 1963, we will focus on the structure prevailing at the time and describe it as if it still existed. The administrative features that make the Paris area different are: 1. The Department of Seine has two executives: (a) The Prefect of Seine (b) The Prefect of Police 2. The City of Paris is subdivided into wards which have the name of arrondissement and 20 appointed "mayors." 50 MAP 1 REORGANIZATION OF THE DEPARTEMENTS OF THE REGION, 1964-68 fl 7‘ h 1 fit fl :- I l ‘1 I *1 fl :1 I .. _._.._ :__. .__...' .__._ —-._ I v I . l2 j I]! ' .‘ 1 .'l , I ‘i' ‘ ‘ l g u i ,, /:;=!I ' warm .1 K) ‘73,, . w, Mum 1., y g . "I! l: i ‘ " , f 5': l I w. H '1 “U W 5’3 41!: ‘l ‘ “H g" 3:33 gg'i auteur-.4 g '1'“ ‘1' '"Eu’" ! fl: (3)5 ## h i ? : i M ~ " 1 '1 Before 1964 1 Ville de Paris 5 Hauts-De-Seine ZIZZ Seine 2 Seine-St. Denis 6 Val-De-Marne m“ Seine Et Oise 3 Val-D'Oise 7 Seine-Et-Marne 4 Yvelines 8 Essonne Source: Delegation Generale au District de la Region de Paris. Adapted from Annmarie Hauck Walsh Urban Government for the Paris Region, (Praeger, 1968), p. 53. 51 mam¢3 mmHHHAou some .zoeumm memam .moneoHemHmam aaemoaHmmme H Hm Z«—————§ Y where Z, the political variable, is intervening between antecedent variable X (i.e., Resources) and policy Y (Welfare Spending); and where Z's regression coefficient b must always remain significant. The Spuriousness model 2. 2( x )Y where socioeconomic variable X is causing both the politics and the policies; and where Z's regression coefficient becomes insignificant. 22 Ezekiel and Fox, op, cit., p. 62. 3Blalock shows that this can happen when the relationships are of the form Xé— Z——9Y. Controlling for the effects of Z would yield a partial coefficient of approximately zero value, when, as we can see, spuriousness is not the case. See Blalock, _p, oit,, pp. 445,448. 24Charles F. Cnudde and Donald J. McCrone, "Party Competition and Welfare Policies in the American States," American Political Science Review, pp. 858-866. 85 Cnudde and McCrone have, however, raised the interesting possibility of an alternative to both the spuriousness and the developmental models. The hybrid relationship_of the form \\\\\5: The high plausibility that socioeconomic variables such as personal 3. X m {f—bfi income or resources may have a direct effect on policy as well as function as antecedents in a developmental sequence which includes political variables, makes it worthwhile to replicate the test that Cnudde and MCCrone use in their analysis.25 The test for a hybrid relationship consists in controlling alternatively for each of the independent variables and noticing a reduction in the regression co- efficients of each of the variables in each case. If both reductions are large enough to be significant, the hypothesis of hybrid relationship stands confirmed. Regression coefficients have one disadvantage, they do not allow for an accurate comparison of the effects of each individual variable. Beta weights, or so-called standardized regression coefficients, allow the quick comparison of the relative contribution made by each variable, because they are computed with standardized data; meaning that each variable has been divided by its respective standard deviation, thus eliminating distortions due to the use of units of measurement that differ in their variance. 2 5Idem. 26 Blalock, op, cit., pp. 450-453. 86 Thus, we see that testing a hypothesis is not simply a matter of comparing simple correlation coefficients with partials. The data cannot speak for themselves. TWO particularly interesting possibilities that one must be aware of are suppressor variables and curvilinear relationShiPS-27 While the two are entirely different phenomena, they may present the researcher with similar manifestations; mainly, initially weak zero order correlations. While the existence of a suppressor effect will become easily detectable once control variables have been introduced and the initially weak coefficient suddenly increases in value, curvilinear relationships can only be detected by trial and error methods, such as plotting each Observation on a set of axes or through the careful scrutiny of the residuals. The examination of the residuals can reveal a curvilinear relation- ship because they are assumed to represent error variance from the regression line. If we have explained all there is to explain in the dependent variable with our independent variables, then residuals should have a pattern of randomness. If, however, they seem to be related to some exOgenous variable, we can conclude that bringing that variable into our explanatory scheme will increase the value of the coefficient Of multiple determination, R2,and bring randomness to the residuals. A curvilinear relationship exists when the residuals are 27See Walter Dean Burnham and John D. Sprague, "Additive and Multiplicative Models of the VOting universe: The Case of Pennsyl- vania, 1960-1968," American Political Science Review, 64 (June, 1970), pp. 471-490. On the subject of the effects of suppressor variables and how they can be detected in survey data, see MOrris Rosenberg, The Logio of Survey Analysis (Basic Books, 1968), Ch. 4. 87 related to one or more of the independent variables28 that have already explained part of the variance. The researcher must then exercise his judgement as to what he believes may be the exact algebraic form of the relationships. Multiplicative models may be hypothesized and tested. Through successive data transformations, he may find the most appropriate curve and corresponding algebraic expression.29 Increases in the total variance explained the R2 provide indication that a better equation has been found. He can then proceed with the same methods as if he was dealing with a simple additive model. 2. Multicollinearity and Its Remedies From the above discussion, it becomes Obvious that regression analysis is a powerful tool for theory building. There are, however, some pitfalls that are all too easy to fall into, and political 28See Hubert M. Blalock, Jr., "Theory Building and the Concept of Interaction," American Political Science Review, 30 (June, 1965), pp. 374-380. 29See N. R. Draper and H. Smith, Applied Regpession Analysis (New York, 1966), Ch. 5; and Joseph B. Kruskal, "Transformations of Data" in International Encyclopedia of Social Sciences (Free Press, 1968), VOl. 16, pp. 182-193. 88 scientists have not always avoided them.30 One Of these is the problem of multicollinearity which is defined as a high correlation between two or more independent variables. This phenomenon gives rise to a considerable sampling error and makes it difficult to choose reliably between alternative models. Blalock says: "The reason why sampling errors become infinite when X and Z are perfectly related is that we can form arbitrary linear combinations of the two simultaneous equations, making the regressions coefficients indeterminate."31 Thus, another means by which we detect multicollinearity is a rise in the standard error of the regression coefficients. At the extreme (perfect correlation between the independent variables), least squares estimation yields an unlimited number of equally likely 0 3 Among recent studies that have come in for some criticism are Bruce M. Russet, "Who Pays for Defense?" American Political Science Review (June, 1969), pp. 412-429; and Jerry Hollenhorst and Gary Ault, "An Alternative Answer to: Who Pays for Defense?" American Political Science Review, 65 (September, 1971), pp. 760-763. Another one is Charles F. Cnudde and Donald J. MeCrone, "The Linkages Between Constituency Attitudes and Congressional Voting Behavior: A Causal MOdel," American Political Science Review, 60 (March, 1966), pp. 66-72; also, Donald R. Matthews and James W. Prothro, Negroes and the New Southern Politics (Harcourt, Brace and World, 1966) are critiqued by Hugh Donald Forbes and Edward R. Tufte in "A Note of Caution in Causal MOdeling," American Political Science Review, 62 (December, 1968), pp. 1258-1264. Matthews and Prothro are particularly criticized for allowing intercorrelations of as high a magnitude as (.95) between two independent variables purported to explain Negro political partici- pation. Finally, Fry and Winters, an, cit, are also criticized for multicollinearity in a "Communication," by James J. Noell, American Pglitigal_§gien§e_3eyieg, 64 (December, 1970), pp. 1249-1250 and in JOhn L. Sullivan, "A Note on Redistributive Politics," American Political Science Review, 66 (December, 1972), pp. 1301-1305. 31Hubert M. Blalock, Jr. Social Statistics (McGraw-Hill, 2nd Ed., 1972), pp. 457, 464. See also by Blalock, "Correlated Independent Variables: The Problem of Multicollinearity," Social Forces, 62 (December, 1963), pp. 233-238. 89 regression lines for the same set of data. J. Johnston explains the problem in these terms: "1. The precision of estimation falls so that it becomes very difficult, if not impossible, to disentangle the relative influences of the various X variables. This loss in precision has three aspects: specific estimates may have very large errors; these errors may be highly correlated, one with33nother; and the sampling variances will be very large." but he adds another set of consequences not mentioned elsewhere.... "2. Investigators are sometimes led to drop variables incorrectly from an analysis because their coefficients are not significantly different from zero, but the true situation may not be that a variables has no effect, but simofiy that the set of data has not enabled us to pick it up." It is interesting to note that these two sets of consequences imply different remedies according to whether one's purpose is the measure- ment Of the net effect of one or more variables or the construction of a predictive model. The scholar interested in isolating the effects of one variable may prefer to eliminate variables that may be approximations of the same concept, even if the level of explanation is reduced as a result 2 of it (lower R ). On the other hand, the econometrician may prefer (if other remedies are not possible) to include even variables that may have high levels of collinearity. But before settling on a specific approach to the problem, we must first examine some of the other signs by Which we recognize a level of 2 3.92- cit., pp. 159—164. 33Idem. 90 collinearity large enough to cast doubt on our findings. Edward Tufte describes some such signs in these terms: 1. Sizable multiple correlations for overall regression appear but with no particular regression coefficient reaching significance; and, 2. Large changes occur in the values of the regression coefficients when new variables are added. J. Johnston also adds: "....estimates of coefficients become very sensitive to particular sets of sample data, and the addition of a few more observations can sometimes produce dramatic shifts in some of the coefficients." Users of step-wise least-squares computer routines have come to recognize multicollinearity by the fact that values of their parameters may be observed to vary with the order in which they were entered. We will enter our variables in multiple mode (simultaneously), partly because our hypotheses are very specific as to which variables should be entered when.36 A crude sign of collinearity has been the level of correlations between any two variables; (.70) and above being viewed as a threshold because more than half the variance explained by these two variables 34"Improving Data Analysis in Political Science," World Politics, 21 (July, 1969), pp. 641-654. 5 J. Johnston, _p, cit., pp. 160. 6 The specific package of routines used for this study is the Michigan State University Agricultural Station Programs which includes step-wise deletion and addition variants of least-squares routines. 91 would be common variance.37 But there can be just as much of a problem when three or more variables intercorrelate at moderate levels (i.e., 50), but again result in as much common variance as is explained. Johnston citing Farrar and Glauber suggests the computation of an F statistic* for the R2 of the (k-l) variables in order to discover whiCh of all the variables is most severely affected by the collinearity in its ability to predict the dependent variable.38 What are the remedies to multicollinearity? None are too satis- factory. If there is little interest or theoretical reason for keeping the variables separate, the advice has always been to combine them into one through factor analysis or index construction or settle on the use of the most powerful of the variables. If there is indeed a theoretical reason for keeping the variables distinct, and determining their relative impact on the dependent variable, much will still depend on whether a spuriousness test is required. In 37 Some have suggested as much as .85 or .80, but much depends on the quality of the data and measurement. Blalock, _p, 315,, argues that multicollinearity is a problem because of the error variance existing in any of social science's indicators. See also Donald E. Farrar and Robert Glauber, "MMlticollinearity in Regression Analysis: The Problem Revisited," Review of Economics and Statistics, 49 (1967), pp. 92-107. Actually, any particular size of intercorrelation is not a sufficient protection against "harmful multicollinearity." It is more the size of the intercorrelation in relation to the total variance explained (in the dependent variable). This can be expressed as rij (Ry; Lawrence R. Klein, An Introduction to Econometrics (Prentice— \ Hall, 1962). 38 92. cit., pp. 163-164. *We will rely on the significance levels computed by the least- squares routine we are using. These are derived from a t-test performed for each of the regression coefficients. This is why it will not be necessary to show the value of the standard error of these coefficients. 92 his study a rigorous spuriouness test is necessary to confirm the effect of political variables on policy, independent of the effects of variables such as resources and personal income. In that case, the only remedy is to construct many indicators for the same concept (i.e., Communist representation, Left representation as percent of total municipal council membership), and use the one that is least correlated with other independent variables. However, a spuriousness test is not necessary in assessing the net impact of political variables in comparison to the impact of social structure (class) variables. Political theory assumes that the latter encompass and precede the former to some extent. This is why, as described in Chapter One, we will use the variables alone sequentially and determine which yields a better explanatory model. More radical remedies suggested by Blalock, Coleman and Boudon involve the creation of stratified subsamples out of the main sample, such that the multicollinearity will disappear in one or more of these subsamples.3 One objection raised here is the same usually made for all stratified samples; that it is difficult to know how other relationships have been affected by this method. But sometimes it is impossible to eliminate all the common variances, even after breaking up the sample. Finally, a more effective way of dealing with the problem is through the use of time series data. The combination of cross-sectional with time 39 James S. Coleman, Introduction to Mathematical Sociology (The Free Press, 1964), Ch. 6. See also "Evaluating the Relative Importance of Variables," American Sociological Review, Vbl. 26 (December, 1961), 93 series analysis is in fact often used in econometric studies to solve 40 such problems. 3. Sampling Design In selecting a sample for policy choice studies which use correlation and regression methods, the concern is less for strict representativeness than for adequate variance. This is because no claim is made that the sample represents anything but its own universe. While this may seem to make the preceeding description of the Stalemated Society superfluous, it is because a distinction must be made between statistical and conceptual representativeness.41 It is because only conceptual representativeness is claimed here that we have deemed it necessary to present thorough descriptive detail and statistics. If through statistical techniques (i.e., random sampling) we could claim true statistical representativeness, the descriptions would not need to be as painstaking. Actually, various studies of policy choices do not present such descriptions because their sample (i.e., the fifty American states) is coterminous with the real universe, or may come very close to approximating it. But another reason for the lack of thorough description and rigorous sampling procedures in such studies is the ability of readers to judge for themselves as to the validity of the findings. A 0For a recent application of such a combined approach in political science, see Virginia Gray, "Time Series Analysis of State Spending," paper presented at the Annual Meeting of the Midwest Political Science Association in Chicago on May 3—5, 1973. 41See Hans L. Zetterberg, On Theory and verification in Sociology (Bed- minster Press, 1965), pp. 128-130 on representativeness in samples. 94 TABLE III-l POPULATION GROWTH (1954-62) BY SIZE OF COMMUNE'S POPULATION AMONG SUBURBAN COMMUNES Population Size All French Communes Study Sample 5,000 - 10,000 30.75 (216) ----- , (not present) 10,000 - 20,000 37.94 (167) 26.30, ( 8) 20,000 - 50,000 43.95 ( 91) 36.95, (26) 50,000 - 100,000 22.17 ( 18) 16.65, (12) 100,000 - 300,000 33.20 ( 2) 14.99, ( 2) Figures compiled from a report called, "Les Finances des Communes de plus de 5,000 habitants" (Exercise, 1965); Les Communes a l'Heure de la Region; Ministére de l'Interieur; Direction Generale des Collectivites Locales, Service de Statistiques et d'Analyses Financieres, and from Annuaire Statistique de la Ville de Paris et des Communes Suburbainesde la Seine, "Annee 1965," 1968, and Institut National de la Statistique et des Etudes Economi ues; Recensement General de la Population de 1962 (Résultats du Depouillement Exhaustif), "Fascicules Departementaux;" Seine, Seine et Oise, IRhone volumes; (INSEE, 29, Quai Branly, Paris 7e, 1966~) 95 recent study of city council decision-making in the Bay Area of San Francisco was criticized specifically because the sample was hardly representative of American city councils of the Reform type (California's Reform movement having been very unique), and even less of American city councils in general. Given the relative obscurity of studies of French local government, until recently, the description here was necessary to convince the reader of the sample's conceptual representativeness. The creation of a sample for this study was subject to the combined influences of finite resources and of a concern for the generalizability of any findings. Finite resources dictated that a large portion of the sample be drawn out of the Parisian basin (where more large communes can be found), and more specifically, from the Seine Département whose archives provide easy access to the budgets of 80 large communes. Finite resources and the vagaries of research schedules made it difficult to collect adequately large samples from either the Outer Paris area or other provincial urban areas. To combine them all with those of Central Paris would have created a biased sample. Another influence, needless to say, was statistical theory, which in the case of least-squares estimation puts emphasis on a minimum sample size (i.e., N==30).43 This is in order to fulfill the basic assumption of normality of the sampling distribution. Sample size, moreover, imposes limits on the number of variables that can be entered in any one equation without 42See Robert Eyestone, The Threads of Public Policy; A Study in Policy Leadership (Bobbs-Merrill, 1971) and Ronald Loveridge, City Managers in Legislative Politics (Bobbs-Merrill, 1971). Also, Kenneth Prewitt, The Recruitment of Political Leaders; A Study of Citizen-Politicians (Bobbs- Merrill, 1971). All of these are a part of the so-called Urban Governors Series. 3 See Ezekiel and Fox, _p, cit., pp. 306—317. 96 .NmHuHmH .mma .Ho .aa ..uHo .mm.:.muomuanmn ooo.m we moan we mmnnaaoo mow mooamcfim meg: onHmo upommu m aoum woaamfioo shaman .aO%A mm Hams mm mcawfium .um mo mmoum some: wnu mmvsaoae coawmm mmnafiuoommou mnu How coaumw>mv vpmvawum mnu mum mammnuamumm :H mouawfimx Aom.~ V Amm.mav ~m.o~ m~.mH ~m.a~ mo.aa o~.qm oo.a~ No.m~ mmuauaeamaxm GHMMHOS HNHUOm Aoa.aav Aqo.eav. Ne.oq oo.om a~.wq Ho.m¢ No.mo mo.mo mo.Hm mmuaufieaoaxm HafiOfiumU—uflm mw.w- m.om~ mm.am~ wa.mma Ho.ksm no.0em oa.m- mmuauacamaxm wowumuomo annoy Amaze Amaze Aomnzv Amauzv mHm coHuMfium> mo coaumfi>mn mo aoauoa>on unmfiowmmmoo wumvnmum com: udmaowmmmoo oumwamum com: oasmaum> mumz QMHUMAMm mom ZOHH mo mHzmHUHmmmou Qz< monHmD nm Magoo A.HHIHHH ou mIHHH mmHemHv .momlaom .ma ..uwo xmm .mammm damp moafimnusnsw mmcoaaoo mow um mwpmm mp mHHH> ma on oswmumwumum mufimonc< cw madame“ mason umufim mzu ma moaa aw ouo> «Hazmo oak HN.Hm 1 mm.mm mm.mm om.mH em.am mmasaaoo HH< m~.mm om.mm .m.a om.mm om.~H om.em maaaaao mo.mm No.m~ .m.a so.om cm.mH om.o¢ scum mo.mm mm.mm .m.a NH.mm ma.m mo.Hm muo>oo ao.mm mm.om .m.a mm.mm H~.HN mm.om moosuuumumufiaamo HH.mm mm.~m mm.mm mm.mH o.~m «mum aowa mH.mm oa.mm .m.= mo.mm ma.mH mm.om oaouoaomumucammumcmamaoo Ne.am mm.em .m.a mo.mm om.mH o~.mm mmafloaom m~.mm oa.mq .m.: H~.mm mm.mH HH.oq mamm>mouuumm mo.Hm mm.mq om.mm No.5H H.om maomm “mono mm.~m oe.mm oa.~m Ho.~m mm.mH mm.mm mmafiqmmg mN.Hq HH.H¢ om.om om.mm om.ma om.mq muumuamz mo.~m om.Hm om.Hm mm.om mo.ma Hm.mm ommumaummooz mm.mm ~m.am o~.mm ~4.Hm om.HH cm.on mamanom N¢.mq mm.qm om.~m cm.mm mm.m o~.mm umaoammm mm.mm mm.os mm.om mm.m~ om.m~ om.m~ maouq< mm.eq mo.mm mm.mm Hm.~m om.mH om.am mammmHHa> mm.qm mm.mm om.Hm mo.em «m.HH om.mm maumzuusmumamomameu mm.mm oH.mm oq.mm Ne.mq oa.mH ~.~¢ Homlmaummmaono mo.mm mH.Hm o5.mm Hm.mm om.mN om.om xamammm mfi.m¢ om.Hm om.mm mm.mo Ne.m m.mm summon mm.~m om.mm mc.mq ~.mm m.- m.mm Hamumuo mo.Hm mo.mm m.Hq mm.~m «.ma w.om ameomo Hm.mm mm.mq mm.mm Ho.Hm ~4.mH m.aq muoo maumm mmmH mmmH moo> mmmao mhmflHHQQ—‘A UnuOn—HSH. *momH 0U O> umHG3§OU mHNGOHmmmw Ohm" NGH¥HO3 mmggou mmafiuflmo HQUO> MHHQNU 8 Ud—QUHOQ UCUUHNW UfiUUHQm mama zH AmzoozH q mo mmqm gon mm.om o¢.om .m.c ww.oq no.5 «.mm moon .um NH.Nw o~.cm .m.s cw.mm mm.ma n.m¢ maamnusmHHH> wa.~o wH.wm wo.mq mo.oH m.mm mmu< GONA an.ww om.qq em.wm m.mH w.mq zmmwom Hm.ww oo.qq cm.mm m.mH w.mq mfiumm umuoo Nw.m¢ «C.Nc oa.w~ No.¢o o.m m.om wumHHHH>odaou wa.no mw.mm O¢.mm ow.mc c©.HH m.mq coso .um mm.mm mm.mq oo.mm o~.ao m.m N.Nm mummaam>umna< m©.om om.mm 05.nm -.mm o.¢H «.me Nommusm oa.mq oo.am om.Hq aw.am c.5H m.am mammHmOImAIm>=oooHHfi> mw.Ho co.mm oc.mm o~.om ¢.uuomH< om.¢n o~.¢m om.qq mm.~m m.wH c.Hm souomumsu oo.wm cm.mq oo.om Nu.we om.mH| m.¢q muoo mfiumm omma mama ouo> mmmao mumwaweoz unochoe moma muo> umfiasaaoo mamaommmomoum wafixuo3 mmoaaaou moawuomo nouo> madame on unmouom unmoumm ufimouom NomH 2H AmMUMDOmmm m4m A mg ma.moa mm.mm mm.- o.q~ oo.m~ mmaaammuo> oo.mmfl Hc.on m~.q~ m.m~ oa.m~ mmammeo ma mm.HmH oa.oe H4.H~ em.m~ mm.- magma “mono mm.mm oo.mm m.mq eo.om m.mm on.¢m maomnmaomummamuaom mm.Hm mm.~m o.mm mm.m~ m.- oo.m~ manaoaoaaa> m~.mm mm.Hm m.He cm.mm m.m~ om.om mo>am> so.mm oo.He ~.mm mo.m~ ~.- ow.m~ yam: .um H~.mOH oq.mm m.om mm.~m m.- om.mm unmamao om.om oq.mm oq.mq mm.Hm ~o.- Nw.om «you mmumm mmmH mmmH muo> mmmao mHMHHfiQOZ HSOfi—JH mcmH mu O> Hagan—EEOC men—Ofimmwm Ohm wflHJHOS mmdn—EEOU mOEHufiOU Hmu O> OHHDNQ 0n— udmuumm uamuumm admonmm Nema zH AMZOUZH A Aoamnuaoo mm.wHH om.mq oo.~e me.Hm mm.¢~ oo.om mwaouuaoz mo.~oa om.me oo.m¢ mo.m~ mm.m~ so.mm maumzuusmuuamwoz «H.~ma He.om Ho.me om.e~ HH.¢~ mm.o~ «you wanna mummmo umm mmmH mHMHHHQQZ UDOEDH mcmH muo> umHfiSEn—OU mHchAmmeOHm wwwHU wGfiv—HO3 mmawuomo Houo> madame on uaooumm unwouom unmoumm monsaaoo mead zH AHZOUZH A A one mo onaa o now NHuHHH oases mom moo}: 35m mmmf mom. 954mg cow. Nm~.1 mom... H ooa. $3.1 mum}: mom. m mmof mmw. omm. Nmmf woof mmAMHmHo mwmf mom. mom. color Hoof. mam. mmmMHmHu own... mma. ooq. owquoof mow. own. mmmMqu oaof Noa. mmN. «mm;- moo.l own. omm. 5mm. mmng momf mma. oqm. omo.1¢¢o. mam. Now. non. man. mnHmHo ooo.l Homf wwmf moo. Nmaf 55.1 «mm... MNHF mao. oqa. Boo—$3. oow.l woo. mmH. moo. omo. :0. mmo. Two. one. own. mmo... mamxmoz H331 Nmo... :Hf ooo. oaof mmmf quf hoof Nmmf moof 0:11 «mot: omo< mom... com. one. omilowq... 03... H9». omm. mom. «.3. new... 01». now... go» mom... mac. Hon. mam. mom. «ma. Nmm. new. wqa. wma. «mmf moo.l moo. Hoof 5.395 33..- mmmf Noe}: :m. Hem. mam; 5mm... mmqf mmmf «ma; mom. wmmf omm. mom... Nam}: mzmo OMm 35m 9625 H m mmmmHmHo omaHMHmHU mmmmEA omens mmHmHU abozmm. MMOB own? Bow Eomo mzmn mood zH AmmHmHo mace mHm azmozmmmezH zmmstm zmmzemm monmeqmmmoommezH MI>H H.393. 136 Old cities with increasing populations have as poor a turnout as the newly sprouting bedroom communites. Fast change seems to be breaking down the Old Stalemate, with its high levels of intensity and political participation. In these old bastions of the Left or the Right, the clash produces high participation, regardless of which side has established dominance over the years. We can now appreciate why party competition means altogether something else in the French urban context. Moreover, Table IV;4 shows turnout to have very little correlation with any of the policy variables. Repeated efforts to develop more complex explanatory models with voter turnout as an independent variable have failed. As for the main political indicators, those measuring party control are all strongly intercorrelated as we would expect. we see particularly how the correlations between Communist Vote in 1959 and the Communist Representation of 1953 (which resulted out of Propor— tional Representation) is the highest of all. We also notice how the same Communist Vote is highly correlated with the two social structure variables of werkers (.788) and Professionals (.805). We thus have additional evidence of the politicization of cleavages under the Old Stalemate. Table IV-4 summarizes the testing of hypotheses of simple relation— ships between individual independent variables and each of the policy areas. The first surprise is that the Taxable Resource level is not very highly correlated with either Social Welfare or Educational Spending. This is in opposition to most findings of previous research. The predicted negative and significantly higher correlation between Personal Income and each of the three policy areas provides us with a 137 TABLE IV-4 ZERO ORDER CORRELATION COEFFICIENTS BETWEEN INDEPENDENT VARIABLES AND PER CAPITA EXPENDITURE CATEGORIES FOR 1963 (36 PARIS CORE CITIES) S.W.Sp. Ed.Sp. Tot.Sp. Demographic Measures Population Growth (GROWTH) -.050 .042 -.097 Population of School Age (YOUTH) .116 .390 .015 Socioeconomic Measures Adult Education (ADED) -.572 -.475 —.515 Illiteracy (ADUNED) .669 .545 .647 WCrkers (W) .567 .665 .607 Income (I) -.460 -.566 —.375 Taxable Resources (R) .270 .162 .501 Political Measures Voter Turnout (TRNOUT) .188 .122 .061 Left vote (LFT59) .620 .521 .523 Communist Vbte (CIST59) .661 .696 .586 Left Party Representation (LFTRP59) .685 .517 .437 Communist Representation (CISTRP59) .579 .572 .453 Communist Representation (CISTRP53) .616 .561 .490 Governmental Variables Grants for Operating Expenses (REPART) .155 .327 .062 Grants for Investment Expenses (SUBV) —.119 .165 -.097 See Table III-12 for a more detailed description of all variables. 138 clue as to the seemingly weak effect of Resources. Income as an indicator of poverty (thus of need and demand for services), acts as a suppressor variable.1 We will have better evidence of this peculiar effect when control variables are introduced. Equally surprising is the weak relationship between the Youth variable and Educational Spending. The coefficient is as large as that of Grants for Operating Expenditures. This seems to indicate how such grants are distributed. On the other hand all political variables are strongly correlated to each of the policy areas, and the coefficients are in fact larger than those of relationships with socioeconomic variables. Particularly strong is the relationship of Left Representation with Social Welfare Spending. Since LEFTRP59 includes Socialist Representation, we conclude that even though Socialist cities spend less on Social Welfare than Communist cities (Tables IV-l and IV—2), a marginal increase in Socialist strength in city councils is translated into larger corres- ponding increases on Social welfare Spending than equivalent marginal increases in Communist strength. Since Socialisusoften participate in Center coalitions, they are able to ask for higher Social Welfare Spending as one of the payoffs for their participation in the coalition. Communists must often wait until they have at least majority status in a coalition, or sheer outright control of the city council, to be able to enact large spending programs. In other words, Left Representation has more of a truly linear relationship with Welfare Spending. The two other policy areas, however, show Communist Representation as the more powerful variable. Thus, it appears that Communist 1 See Morris Rosenberg, The Logic of Survey Analysis (Basic Books, 1968). Ch. 4. 139 TABLE IV—S INTERCORRELATIONS AMONG PER CAPITA EXPENDITURE CATEGORIES FOR 1963 (36 PARIS CORE CITIES) 2 l S.W.Sp. Ed.Sp. Tot.Sp. Adm.Sp. Inv.Sp. S.W.Sp. - .558 .704 .549 .287 Ed.Sp. - - 0743 0691 0519 Tot.Sp. - - _ .786 .550 Adm. Sp. - _ - - 0470 1Investment Spending (Capital Outlays). 2Administrative Spending (Expenditures for General Control). For other details on variable labels, see Table III-12. 140 municipalities differ on all policies and that they differ particularly from Socialist administrations in their commitment to spending across- the-board. This would be proof of the complete distinctiveness of the Communist ghetto. However, such conclusions must await a more thorough test of relationships. Finally, Table IV-4 again displays a certain consistency in the relationships of independent variables to each of the dependent. The intercorrelations between budgetary areas on Tables IVe4 and IV—5 also tend to show the distinctiveness of each policy. The correlation with Investment Spending (capital outlays) seems to confirm Education as a more capital intensive policy area than Social welfare. At the bottom of Table IV—4, the two types of grants are both weakly related to the spending policy areas. It must be noted, however, that both of them are strongly related to Growth and Youth, projecting a picture of fast growing cities with many newly arrived families, with multitudes of children, and requiring more central help than other French communities. These are but preliminary generalizations and we turn now to more rigorous tests of relationship. C. The Determinants of Social Welfare Spendipg l. Spuriousness Tests The behavior of regression coefficients upon introduction of specific variables will be relied upon to provide us with evidence as to the nature of relationships in each of our spuriousness tests. As Tables IV-6 and IV¥7 show, the effects of the Left Representation variable are so strong as to reduce the regression coefficient of Income by 64%. Conversely, the Resources variable has no appreciable 141 TABLE IV-6 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND LEFT REPRESENTATION ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Resources .42 (.27) .31 .031 (.026) .011 LEFTRP59a .60 (.69) .70 .264 (.280) .0005 2 Coefficient of Multiple Determination: R = .565 Significant at .0005. TABLE IV-7 PARTIALS, BETAS, AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME AND LEFT REPRESENTATION ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Income -.23 (-.46) -.16 -.055(-.150) .190 LEFTRP59b .60 ( .69) .61 .251 (.280) .0005 2 Coefficient of Multiple Determination: R = .499 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aLEFTRP59 is Left Representation, 1959. bSame as above. 142 TABLE IV-8 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND INCOME ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963. variable Partial Beta Regression Significance Correlation weights Coefficient Level Resources .48 ( .27) .45 .045 ( .026) .003 Coefficient of Multiple Determination: .RE_5_.AOO Significant at .0005. TABLE IV-9 PARTIALS, BETAS, AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND WORKERS ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficients Level Resources .31 (.27) .25 .026 ( .026) .072 workers .58 (.56) .56 1.139 (1.139) .0005 Coefficients of Multiple Determination: Rz = .386 Significant at .0005. Figures in parenthesis are zero—order correlation coefficients for the relationship between the particular variable and the dependent variable. 143 TABLE IV-lO PARTIALS, BETAS, AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF WORKERS AND LEFT REPRESENTATION ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weights Coefficients Level workers .21 (.56) .20 .417 (1.154) .211 LEFTRP59a .51 (.69) .55 .229 ( .280) .001 2 Coefficient of Multiple Determination: R = .503 Significant at .0005. TABLE IV-ll PARTIALS, BETAS, AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF WORKERS AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) IN 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression SignificanCe Correlation Weights Coefficients Level workers .32 (.56) .33 .683 (1.154) .060 CISTRP59b .35 (.58) .37 .150 ( .236) .040 2 Coefficient of Multiple Determination: R = .404 Significant at .0005. Figures in parenthesis are zero—order correlation coefficients for the relationship between the particular variable and the dependent variable. aLEFTRP59 is Left Representation, 1959. bCISTRP59 is Communist Representation, 1959. 144 effect on Left Representation. Thus, from the outset, the independent nature of the effects of Left Representation are well established. Corroborative evidence comes also from Table IV—lO, where we find that the regression coefficient of the Workers variable has also been reduced by approximately 64%. The strength of political variables in this policy area is also demonstrated by the powerful effects of the Communism indicators. Tables IV-ll and IVe12 show the effects of Communist Representation of 1959. Although weaker than Left Representation (it accounts for only 14.9% net explained variance in Table IV914), its effects are not weakened by the introduction of Resources, Income or Workers. The effects of Income and Workers are in fact again reduced by the political variable (the regression coefficients here decline by 48%and 41% respectively.) Finally, the variable Communist Representation of 1953 is even stronger in Table IV-15. We have not deemed it necessary to present more data on CISTRP53, because its intercorrelations with Income and Workers are large enough to produce multicollinearity and the reductions of significance levels which usually accompany it. Nonetheless, the use of this variable has demonstrated very clearly the very close correspondence between social and political cleavages and their durability in societies where the Structured Model prevails. It is also an indicator of the durability of the Communist Party's hold on certain segments of the electorate, and of its impact on local policies. We find here a little of what policy analysts call the effect of incrementalism. Certain policies introduced by the Communists in 1953, when they had some influence on municipal councils, have not been eliminated by succeeding administrations . 145 In Table IV-8 we see definitive confirmation of the suppressor effect of the Income variable on Resources. The size of the regression coefficient of Resources is increased by 78%. In summation of all the spuriousness tests, it can be said that most tables seem to confirm the developmental hypothesis2 (both regression coefficients remain strong and are often increased). This means that both variables contribute to the explanation independently, but are linked in some form of causal sequence. Tables IV-7 and IV-lO seem to be excellent examples of spuriousness. Yet the alert researcher must distinguish between the causes of such a finding. While the low significance of the Income variable is due to the suppressor effect between it and Resources, the Workers variable remains insignificant because of high collinearity with LEFTRP59. This is what is revealed in Tables IV-13 and IVFl6. Thus, we cannot conclude that either Income or Workers as indicators of need and demand are not important explainers of Social Welfare policy. It is simply that their effect is difficult to detect embedded as it is in covariations with other variables. Table IVell, on the other hand, is a good example of a hybrid relationship3 (as explained in Chapter Three), where both regression coefficients decline significantly. In other words, the Social structure 2We have been guided in our distinction between spuriousness and developmental models by those made by Charles F. Cnudde and Donald J. McCrone in "Party Competition and welfare Policies in the American States," American Political Science Review (September, 1969), pp. 858-866; who in turn relay on Hubert M. Blalock, Jr., Causal Inferences in Non- Experimental Research (University of North Carolina Press, 1964). 3Cnudde and MeCrone,'gp. cit., pp. 860—862. 146 of a city determines whether the city will be communist controlled, which in turn increases Social Welfare expenditures. The latter are also, in addition, increased by the same type of social structure that brings about a Communist administration. 2. The Explanatory Model we have thus established that the basic equation explaining the determinants of Social Welfare Spending is of the form Ei = f(R,I,P) + e where, Ei = Expenditures in function i P = Party Control Variable I = Income Variable R = Resource Variable e = Error term.which can be reduced if other minor variables are included. Table IV-13 portrays this equation in the best combination of high explanatory variance (62.7%) and parsimony of variables used. Moreover, in this version of the basic equation, Left Representation accounts for a full 22% of the variance explained, independent of other parameters. Income, which had been found insignificant, now explains 6.6%. This means that even though it may appear, at first, that in Left cities the level of Poverty (Income) has no effect on spending, the conclusion is reversed when we distinguish between Left cities with high Resources (industrial strongholds) and those with low Resources (working-class bedroom communities). Tables IV-14 and IVe15 show variations of the basic equation when other indicators of party control are used. The relatively weaker effects of CISTRP59 and CISTRP53 reconfirm what we had previously surmised as to the results of marginal increases in Socialist 147 TABLE IV-12 PARTIALS, BETAS, AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Income -.25 (-.46) -.23 -.079 (-.150) .145 CISTR1>59a .46 ( .58) .46 .190 ( .236) .005 2 Coefficient of Multiple Determination: R = .378 Significant at .0005. TABLE IV-13 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, INCOME AND LEFT REPRESENTATION ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) 0F 36 PARIS CORE CITIES IN 1963. variable Partial Beta Regression Significance Variance Correlation weights Coefficients Level Explained Resources .50 ( .27) .38 .038 ( .026) .002 12.7 Income -.39 (-.46) -.31 -.105 (-.150) .023 6.6 LEFTRP59b .61 ( .69) .55 .225 ( .280) .0005 22.7 2 Coefficient of Multiple Determination: R = .627 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. bLEFTRP59 is Left Representation, 1959. 148 TABLE IV- 14 PARTIAES, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, INCOME AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Variance Correlation Weights Coefficients Level Explained Resources .51 ( .27) .42 .42 ( .026) .002 16.1% Income -042 (-046) -o 38 -0129(-0150) .012 10.2% CISTRP59a .48 ( .58) .43 .174 (.236) .004 14.0% 2 Coefficient of Multiple Determination: R = .539 Significant at .0005. TABLE IV¥15 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, INCOME AND COMMUNIST REPRESENTATION, 1953, ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Variance Correlation Weights Coefficients Level Explained Resources .49 ( .27) .41 .041 ( .026) .003 13.8% Income -.32 (-.46) -.30 —.103 (-.150) .059 5.4% CISTRP53b .49 ( .61) .47 .555 ( .720) .003 14.4% 2 Coefficient of Multiple Determination: R = .544 Significant at .0005. Figures in parenthesis are zero—order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. bCISTRP53 is Communist Representation, 1953. 149 TABLE IV-l6 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, WORKERS AND LEFT REPRESENTATION ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficients Level Resources .41 (.27) .29 .029 ( .26) .041 workers .26 (.56) .22 .443 (1.154) .131 LEFTRP59a .58 (.69) .57 .215 ( .280) .0005 2 Coefficient of Multiple Determination: R = .525 Significant at .0005. TABLE IV-l7 PARTIALS, BETAS, AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, WORKERS AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) 0F 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weights Coefficient Level Resources .38 (.27) .29 .029 ( .026) .027 WOrkers .31 (.56) .29 .606 (1.154) .079 CISTRP59b .41 (.58) .41 .170 ( .236) .016 2 Coefficient of Multiple Determination: R = .420 Significant at .0005. Figures in parentheahsare zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aLEFTRP59 is Left Representation, 1959. bCISTRP59 is Communist Representation, 1959. 150 strength on city councils as compared to the more discontinuous effects of Communist political power on the field of Social welfare. Can we draw the conclusion that the existence of many Communist administrations causes Socialisusto push for even more radical policies in an effort to compete effectively with Communists for the working-class vote? The statistical analysis seems to support such a conclusion. This is not, however, what an astute observer of the local politics of the Paris basin would conclude. The Socialists are simply in a better bargaining position in many cases. Moreover, their efforts are limited to one policy area in exchange for lesser spending (as we will soon see) in other areas. Finally, their tendency to finance services through user charges must also be considered. we thus conclude that the Socialists are simply playing the role that we would expect in a Structured Model setting which features a multi-party system. They express the basic underlying social conflict and perpetuate it organizationally, but in a slightly milder form. The tendency towards bipolarity introduced by the Two Ballot Electoral System will make them move closer to the Communists organizationally (as percent elections show), even though it can be said that their constituency has increasingly become more like that of Center Parties. Another set of variations on the main equation comes from the substitution of the Workers variable for Income in Tables Ivel6 and Ive17. In both tables we find that this indicator of social structure is not as good of an explainer as Income. Multicollinearity makes it impossible to include in one equation indicators of social structure and income, as well as some political variable (i.e., LEFTRP59). 151 Finally, questions as to the causal sequence between social structure and politics cannot be answered satisfactorily without survey data (to ascertain the true class vote) and time series data (to solve the multicollinearity problem). Various attempts were made to improve on the total variance explained through the addition of new variables, but the trade-off was in each case a reduction in the significance of these and other component variables. A different approach met with greater success, as Tables IV—18 and Iv¥l9 show. Various non-linear models were tested through variable transformations. In Table Ive18, an equation of the form Log Y = a + blx1 = b2x2 not only increased the total variance explained, but improved the significance levels of individual variables. The logic behind such a transformation was inspired by an examination of predictors and residuals of the Social Welfare Spending equation (including R, I and LEFTRP59). Since some of the predicted values turned out to be negative, it was hypothesized that a logarithmic expression would bemmre appropriate. Table Iv-l9 is an attempt to improve on the explanatory power of a model which includes the effects of Communism. Using CISTRP53 squared increases the total variance explained in Tables IV-l4 and Iv415, as well as the relative strength of the political variable. A better mathematical expression of the phenomenon of institutionalization could not be found. In 1953, the electoral system used was the P. R. System which means that Communist Representation was a much closer reflection of Communist electoral strength. Table Iv-l9 tells us that the Two Ballot Majority System, the effect of time and the coinciding cleavages have converted certain cities into Communist strongholds, not only in 152 TABLE IV-l8 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME AND RESOURCES AND LEFT REPRESENTATION, 1959, ON THE LOGARITHM 0F PER CAPITA SOCIAL WELFARE SPENDING (LOG S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weights Coefficients Level Income -.47 (-.55) -.40 -.0018 .004 Resources .43 ( .17) .31 .0004 .011 a LEFTRP59 .60 ( .71) .52 .0030 .0005 2 Coefficient of Multiple Determination: R = .642 Significant at .0005. TABLE IV- 19 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME, RESOURCES AND COMMUNIST REPRESENTATION, 1953, SQUARED ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 36 PARIS CORE CITIES IN 1963- Variable Partial Beta Regression Significance Correlation weights Coefficients Level Income -.32 (—.46) -.29 —.0097 .067 Resources .48 ( .27) .38 .0380 .004 CISTRP53.SQb .53 ( .65) .50 .0070 .001 2 Coefficient of Multiple Determination: R = .566 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aLEFTRP59 is Left Representation, 1959. bCISTRP53.SQ is Communist Representation, 1953, squared. 153 terms of representation, but also in terms of policy. We also tested multiplicative models which featured an interaction effect between politics and resources but with lesser success. D. The Determinants of Educational Spending l. Spuriousness Tests Tables IV-20 and Iv 21 seem to demonstrate that in Educational Spending, the Political variable CISTRP59 is not much weaker than in Social welfare. We chose to present the results of spuriousness tests, using CISTRP59 as the main political variable, because it was somewhat stronger than LEFTRP59 and about as strong as CISTRP53. The results of tests with these other indicators were essentially the same. In all cases, the political variable appeared at first as having an important effect on spending. It is eventually reduced to insignificance in Table IV—23 by the combined effects of Resources, Income and Youth. Interestingly, the effects of CISTRP59 seem to fare well in the initial tests against each of the variables, and even against any two variables combined (see Table Iv-22). How can this be? We will see that this is because of a double suppressor effect. But in more practical terms, the higher spending of Communist cities on education is seen as the combined product of their larger youth population, poorer families and usually strong industrial base. But let us first examine the two-variable tests. Table IV-20 shows that again, as in Social welfare, Resources has no significant effect on the regression coefficients of CISTRP59. In Table IV—21, we again have a case of declining coefficients, but with CISTRP59 appreciably less affected than Youth. Thus, it appears at first that Communist cities 154 TABLE IV-ZO PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA EDUCATIONAL SPENDING (Ed.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficients Level Resources .25 (.16) .21 .019 (.014) .139 CISTRP59a .59 (.57) .59 .217 (.211) .0005 2 Coefficient of Multiple Determination: R = .371 Significant at .0005. TABLE IV-21 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF COMMUNIST REPRESENTATION, 1959, AND YOUTH ON PER CAPITA EDUCATIONAL SPENDING (Ed.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weights Coefficient Level CISTRP59b .47 (.57) .50 .184 (.211) .004 Youth .15 (.39) .14 .472 (1.284) .381 2 Coefficient of Multiple Determination: R = .343 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. b Same as above. 155 TABLE IV—22 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, YOUTH AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA EDUCATIONAL SPENDING (Ed.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weights Coefficients Level Resources .39 (-.56) .37 9.033 (.014) .021 Youth .34 ( .39) .36 l.204(1.284) .046 a CISTRP59 .43 ( .57) .42 (.211) .010 2 Coefficient of Multiple Determination: R = .439 Significant at .0005. TABLE IV-23 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF YOUTH, RESOURCES, INCOME AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITAL EDUCATIONAL SPENDING (Ed.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Youth .33 (.39) .30 1.010 (1.284) .055 Resources .53 (.16) .47 .042 ( .014) .002 Income -.52 (-.56) -.47 -.l43 (-.l72) .002 CISTRP59b .27 (.57) .23 .083 ( .211) .119 2 Coefficient of Multiple Determination: R = .596 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. bSame as above. 156 TABLE IV-24 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND YOUTH ON PER CAPITA EDUCATIONAL SPENDING (Ed.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Resources .44 (.16) .46 .042 ( .014) .008 Youth .54 (.39) .61 2.029 (1.284) .001 2 Coefficient of Multiple Determination: R = .315 Significant at .0005. TABLE IV-25 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND INCOME ON PER CAPITA EDUCATIONAL SPENDING (Ed.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Resources .43 ( .16) .37 .033 ( .014) .010 Coefficient of Multiple Determination: R2 = .442 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. 157 Spend more in relation to their younger population. This is equivalent to saying that they spend more on a per pupil basis. In Table IV423, we again find that it is Youth that is made insignificant in a four- variable test. How is it then that CISTRP59 becomes insignificant in a five-variable relationship? To understand, we must again look at suppressor effects impinging on the relationship between (Taxable) Resources and Educational Spending. Just as in Social Welfare, Income acts as a suppressor (see Table Iv-25), but in Table IV-24 we find the Youth variable having the same effect on Resources. This means that some cities with a high level of Taxable Resources spend more on Education, because their industrial base also implies a population with large families, while other cities spend less because their Resources are the result of high personal income and therefore smaller families (with private school-attending children). we see how the nature of the suppressor effect is much more complex in this policy area. 2. The Explanatory Model Thus it is that these suppressor effects are strong enough to create a four—variable equation which accounts for 56.2% of the variance (see Table IV-26). And while CISTRP59 increase the R2 to .596 in Table Iv-23, its own low level of significance (.119) compels us to eliminate this political variable from the basic equation. The political variable plays only a minor part in Education. Thus, we are again left with a four-variable equation Ei f(R,I,Y)-+ e where Ei Expenditures in function i. 158 TABLE IV—26 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF YOUTH, RESOURCES AND INCOME ON PER CAPITA EDUCATIONAL SPENDING (Ed.Sp.) OF 36 PARIS CORE CITIES IN 1963. variable Partial Beta Regression Significance Correlation weights Coefficient Level Youth .45 ( .39) .41 1.360 (1.284) .007 Resources .57 ( .16) .53 .048 ( .014) .0005 Income -.60 (-.56) -.55 -.169 (-.172) .0005 2 Coefficient of Multiple Determination: R = .562 Significant at .0005. TABLE IV-27 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME, RESOURCES AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA EDUCATIONAL SPENDING (Ed.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weights Coefficient Level CISTRP59a .42 ( .57) .36 .132 ( .211) .013 Resources .43 ( .16) .34 .031 ( .014) .010 Income -.52 (-.56) -.50 -.152 (-.l72) .001 Coefficient of Multiple Determination: 2 = .544 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. 159 Table IV—26 not only shows the importance of adequate taxable resources in Educational Spending, which is a widespread phenomenon, but it points out that (lack of) personal income as an indicator of demand is just as important a determinant. This is peculiar to French cities and can be attributed to the fact that upper income French communities will send their children to private religious schools rather than invest in a high quality public school as is often done in American upper-class suburbs. In turn, this behavior may be the result of the fact that Education is controlled by the central govern- ment which establishes nation-wide standards. Moreover, since incor- porations and the creation of exclusionary school districts are not allowed, and since, as we have seen, classes are often not geographically segregated, it will be only when a city is homogeneously lower income that there will be an unequivocal demand for more public education. In the stimulation and articulation of such a demand, the Communists will play an important role. The Socialists here exert a more centrist influence on spending levels. This then is the pattern of Educational Spending in the Paris Core area. Even though the Communism variable was eventually found insignifi- cant, we cannot say that the Structured Model of Policy-Making does not apply to this issue area. The powerful effect of the need variable, the fact that Communist cities spend more in relation to the size of central grants they receive (see Table Ive29), leads us to conclude that Educational Spending displays distinct patterns that are defined by the coinciding of social and political cleavages. Finally, Table Iv;29 shows why we have not included Grants for Operating Expenditures in our equations; its effect parallels that of 160 TABLE IV-28 PARTIALS, BETAS, AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME, YOUTH, AND THE INTERACTION OF RESOURCES AND COMMUNISM (CISTRP53xR) ON PER CAPITA EDUCATIONAL EXPENDITURES OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficients Level Income -.44 (-.56) -.41 -.1246 ' .008 Youth .36 ( .39) .29 .9690 .040 CISTRP53xR? .47 ( .67) .48 .0004 .003 Coefficient of Multiple Determination: R2 = .612 Significant at .0005. TABLE IV-29 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF GRANTS, OPERATING EXPENDITURES, (REPART), RESOURCES, INCOME AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA EDUCATIONAL EXPENDITURES (Ed.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficients Level b REPART .25 ( .33) .19 .463 .163 Resources .48 ( .16) .40 .036 .004 Income -.52 (-.56) -.49 -.l49 .002 CISTRP59c .34 ( .57) .29 .107 .049 2 Coefficient of Multiple Determination: R = .572 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP53xR is Communist Representation, 1953, multiplied by Resources. bREPART is Grants for Operating Expenditures. cCISTRP59 is Communist Representation, 1959. 161 the Youth variable (and is actually slightly weaker) as we would expect since the size of the pupil population is the allocation formula used by the central government. Since the Communism variable was seemingly close to being significant, we attempted to find an explanatory model which would include this light, but nonetheless detectable effect. As previously, our concern was again the discovery of non-linear effects and possible increases in total variance explained. Table Ive28 shows that the inclusion of an interactive effect best describes the reality. Admittedly, the Resources variable was already quite important on its own, in Table Iv-26 and IV—27, but the fact that the resulting model accounts for more variance (61.2%) explains and describes a likely variant of the process of political institutionalization; makes it all the more satisfactory. We can say that Taxable Resources is the main determinant of Educational Expenditures and that the presence of Communist administrations in 1953 has amplified its impact. This contrasts with Social Welfare where Resources is only a limiting factor, not a major determinant. E. The Determinants of Total Operating Expenditures l. Spuriousness Tests Tables IV-30, IV-31, and IV-33 show that in this policy area, the political variable (although not as strong as in Social Welfare) has a significant effect. Here, too, CISTRP59 was used to demonstrate results, which were somewhat weaker with LEFTRP59 and as good with CISTRP53. As in previously described policy areas, the nature of the Income-Politics- 162 TABLE IV-30 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA TOTAL OPERATING EXPENDITURES (Tot.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Resources .60 (.50) .54 .243 (.226) .0005 CISTR1>59a .57 (.45) .50 .915 (.834) .0005 Coefficient of Multiple Determination: R2 = .497 Significant at .0005. TABLE IV-31 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA TOTAL OPERATING EXPENDITURES (Tot.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Income -.19 (-.37) -.20 -.309 (-.571) .250 CISTRP59b .33 ( .45) .35 .653 ( .834) .049 2 Coefficient of Multiple Determination: R = .237 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. bSame as above. 163 Policy relationships was again ascertained as hybrid (regression coefficients declined 46% for Income, 22% for CISTRP59 in Table IV-33). As elsewhere, Income regains significance in the four-variable test. This policy area differs from the others in that it is strongly related to Resources to begin with (.50). The suppressor effect is detectable (see Table lve34) in this case too; the Resources correlation reaches (.76) and its regression coefficient is increased 35%. But the effect is all the stronger on Income whose regression coefficient is increased by 56%. Thus, the single dominant fact about Total Operating Expenditures is the powerful role played by Resources, accounting as it does for 39% net variance. 2. The Explanatory Model In Tables IV-35, IV-36, Iv-37, and Ive38, we observe that the Social Structure variable, Workers is even more problematic as an indicator of demand for Total Expenditures. Its high level of inter- correlation with Income and CISTRP59 creates highly unstable coefficients for any of these variables. Any additional effect it accounts for is rendered insignificant. Thus again, we chose as our best explanatory model (combining parsimony in variables and high variance explained) the same four-variable equation as in Social Welfare (Table IV-35). E1 = f(R,I,P) where this time, P = CISTRP59 The relative strengths have, however, changed. As previously mentioned, Resources is much more important. Other variables such as Youth and Growth were added but the resulting increase in variance explained was relatively insignificant. Table Ive38 shows that Density and Population were among minor variables that fared the best. Even 164 TABLE IV- 32 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND INCOME ON PER CAPITA TOTAL OPERATING EXPENDITURES (Tot.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Resources .76 ( .50) .68 .307 ( .226) .0005 Income -.64 (-.37) -.58 -.892 (-.571) .0005 Coefficient of Multiple Determination: R2 = .562 Significant at .0005. TABLE IV-33 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, INCOME AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITAL TOTAL OPERATING EXPENDITURES (Tot.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Signficance Variance Correlation Weights Coefficient Level Explained Resources .71 ( .50) .66 .297 ( .226) .0005 39.2 Income -.51 (-.37) -.43 -.663 (-.571) .002 13.2 CISTRP59a .39 ( .45) .29 .544 ( .834) .023 6.7 Coefficient of Multiple Determination: R2 = .629 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. 165 TABLE IV434 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, WORKERS, AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA TOTAL OPERATING EXPENDITURES (Tot.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Resources .63 (.50) .50 .227 ( .226) .0005 WOrkers .50 (.61) .46 4.262 (5.575) .002 CISTR1>59a .24 (.45) .20 .373 ( .834) .157 2 Coefficient of Multiple Determination: R = .626 Significant at .0005. TABLE IV-35 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, INCOME AND WORKERS ON PER CAPITA TOTAL OPERATING EXPENDITURES (Tot.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Variance Correlation Weights Coefficient Level Explained Resources .64 ( .50) .57 .259 ( .226) .0005 26.1 Income -.28 (-.37) -.28 -.423 (-.571) .107 3.2 Workers .40 ( .61) .39 3.638 (5.575) .018 7.7 Coefficient of Multiple Determination: R2 = .633 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. 166 though the WOrkers variable is not insignificant, its elimination would still leave 74.7% of the variance explained. The fact that Density and Population are negatively related seems to suggest an "Economies of Scale" effect. This means that for a given level of Resources and Need, there is likelihood of less spending in heavily p0pulation areas, since more people can be serviced out of the same facilities. In the teeming urban slums, it costs less to administer governmental services (possibly because quality suffers too). Finally interactive effects are even more evident in this policy area. In Table IV-39, the interaction of Resources and Communism adds explanation over and above that accounted for by Resources alone, and increases the amount of variance explained to 65.5%. Table Ive40 provides one final variation on the basic equation. The higher percentage of variance explained in this policy area (65% and above) was a source of puzzlement to this researcher, who assumed that functional policy areas could more easily be explained by specific variables (i.e., Youth and Education), than total spending where a host of influences can be conjured. Yet the results of the data processing seem to show that in the final analysis, most influences cancel each other out for Total Expenditures, while one influence, that of Resources, is amplified. Finally, it can be said that functional areas can fluctuate both in total magnitude and relatively to their share of the total budget. That is where much of the political give-and-take is likely to occur. Thus, the fact that the Communism variables remain significant points to an undeniable tendency among Communist cities of the Paris Red Belt 167 TABLE IV- 36 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, WORKERS, POPULATION, DENSITY, AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA TOTAL OPERATING EXPENDITURES (Tot.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Variance Correlation Weights Coefficient Level Explained Resources .780 ( .50) .72 .325 .0005 36.0 WOrkers .288 ( .61) .25 2.265 .116 2.1 P0P62a -.421 ( .90) -.28 -.001 .018 5.0 DENSb -.470 (-.26) -.32 -.0047 .008 6.6 CISTRP53C .499 ( .61) .43 2.297 .004 7.7 2 Coefficient of Multiple Determination: R = .768 Significant at .0005. TABLE IV-37 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, INCOME AND THE INTERACTION OF RESOURCES AND COMMUNISM (CISTRP53XR) ON PER CAPITA TOTAL OPERATING EXPENDITURES (Tot.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Variance Correlation Weights Coefficient Level Explained Resources .48 ( .50) .43 .193 ( .226) .004 10.2 Income -.43 (-.37) —.36 -.548 (—.571) .011 7.0 CISTRP53xR§ .46 ( .73) .41 .0016 .006 9.3 Coefficient of Multiple Determination: R2 = .655 Significant at .005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aPopulation 3126, 1962- CCOmmunist Representation, 1953. bDensity dCommunist Representation, 1953, multi- plied by Resources. 168 to spend more across-the—board. They also are deeply in debt and tax heavily. This is in contrast to Robert C. Fried's finding among Italian cities where Communist administrations had a mild tendency4 to underspend, undertax and balance their budgets. We must, however, point out that Fried's conclusion of the weakness of party control as a variable in Italian local policies is not convincing without looking at the specific policies included in Total Expenditures. To expect that poor provincial and non-industrial cities will display an across-the—board spending pattern (as opposed to emphasizing one main policy area) because of party ideology is unrealistic. We will find in Chapter Five that even the more radical French Communist Party will not display such a pattern in the French provinces, and will choose to focus on only Social Welfare Spending given the limited resources. F. _Two Other Minor Policy Areag Expenditure are Of course not the only part of a budget that are likely to reflect local policy choices. 1. The Business Tax Methods of revenue collection can equally reflect which segments of a society are to shoulder a heavier burden and which are to be spared. As pointed out in Chapters Two and Three, French communes are not free to levy whichever taxes, at whatever incidence, their 4 Robert C. Fried, "Communism, Urban Budgets and the Two Italies: A Case Study in Comparative Urban Government," Journal of Politics (July, 1971). 169 TABLE IV¥38 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, WORKERS, POPULATION AND THE INTERACTION OF RESOURCES AND COMMUNISM (CISTRP53XR) ON PER CAPITA TOTAL OPERATING EXPENDITURES (Tot.Sp.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weights Coefficient Level Resources .48 ( .50) .41 .185 ( .226) .005 Workers .52 ( .61) .40 3.733 (5.575) .003 P0P62a -.30 ( .09) -.18 -.00068 .053 CISTRP53xRb .43 ( .73) .36 .0014 .016 Coefficient of Multiple Determination: R2 = .709 Significant at .0005. TABLE IV-39 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME, RESOURCES, COMMUNIST REPRESENTATION, 1959, ON PER CAPITA BUSINESS TAXATION REVENUE (Bus. Tax.) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Variance Correlation Weights Coefficients Level Explained Income -.45 (-.27) -.28 -.040 .007 6.0 Resources .85 ( .62) .74 .032 .0005 40.0 CISTRP59c .66 ( .54) .46 .082 .0005 17.5 2 Coefficient Of Multiple Determination: R = .791 Significant at .0005. Figures in parenthesis are zero—order correlation coefficients for the relationship between the particular variable and the dependent variable. aPopulation Size, 1962. bCommunist Representation, 1953, multiplied by Resources. cCommunist Representation, 1959. 170 administration might choose. For the most part, they must rely on the Centimes and the Taxe Locale, which can vary only in magnitude and only within centrally defined limits. The Taxe Locale rates, in fact, vary very little. There are, however, some minor taxation areas that are left to the community's option and discretion. One of these is called Taxe sur les Professions and is usually collected in the form of a license fee from those operating a business or more properly, those we might call self-employed. This is a tax that is very much recurred to by Communist municipalities. To a large extent this can be considered a symbolic policy area, since not more than 10 Francs per capita are ever raised in this fashion. Its almost punitive aspect symbolizes the dominance of a proletariat over an embattled middle-class. It, above all, reflects the community's political ideology within the Left-Right dimensions of French politics and society. This is indeed what we find in Tables IV-39 and IV#40. The equation of Table IV-39 was also attempted with LEFTRP59, but this yielded aenmiler R2. In other words, this particular policy area is characteristic of Communist municipalities exclusively. Socialist administrations use this tax only in so much as their resource level (and as explained below, their spending programs) warrant it. There is, moreover, an unusual effect imparted here by Resources. It is not the lack of Taxable Resources that causes a city to levy this additional tax, but exactly the opposite. This means that it is to the extent that a city is capable of spending, that it will raise such a tax. Table IV—39 proves the purely optional character of this tax. 171 TABLE IV—4O PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME, RESOURCES AND THE INTERACTION OF RESOURCES AND COMMUNISM (CISTRP53xR) ON PER CAPITA BUSINESS TAXATION REVENUE (Bus.Tax) 0F 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Variance Correlation Weights Coefficients Level Explained Income -.30 (-a27) -.14 -.020 .084 1.0 Resources .61 ( .62) .36 .015 .0005 7.1 a CISTRP53xR .83 ( .90) .69 .000257 .0005 25.5 2 Coefficient of Multiple Determination: R = .882 Significant at .0005. TABLE IV-4l PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF YOUTH, GROWTH AND LEFT REPRESENTATION, 1959, ON PER CAPITA OPERATING EXPENDITURE (Op.Exp.), GRANTS (REPART) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficients Level Youth .36 (.68) .34 .470 .041 Growth .47 (.76) .46 .0986 .006 b LEFTRP59 .33 (.35) .25 .0365 .047 2 Coefficient of Multiple Determination: R = .669 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP53xR is Communist Representation, 1959. bLEFTRP59 is Left Representation, 1959. 172 TABLE IV—42 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF YOUTH AND GROWTH ON PER CAPITA INVESTMENT EXPENDITURE GRANTS (SUBV) OF 36 PARIS CORE CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficients Level Youth .29 (.58) .32 2.164 .072 Growth .39 (.62) .41 .429 .023 2 Coefficient of Multiple Determination: R = .442 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. 173 Table IV—40 puts the argument even more succinctly: the inter- active term summarizes the spending capability of high Resources with the spending commitment of Communist municipalities. The 25.5% variance explained by the multiplicative variable is not truly reflective of its true strength because of the occurrence of common variances. If we compare with Table IV-39 and observe that the R2 has been increased from 0.791 to 0.882, we can conclude that this one variable must account for approximately all but the 6% of variance explained by Income. 2. Central Grants—In-Aid Tables IV441 and IV-42 demonstrate the contrast between the two types of grants-in-aid. Grants for Operating Expenditures can be more adequately explained (R2 = .669) and the political variable is a sizable part of this explanation. These tables substantiate the statements we made on the basis of the averages reported in Table IV-l and IV-2. Table IVF43 shows that REPART is a policy area, which has been shaped by previous invest- ment programs and acquired a greater predictability. Left cities are those that have developed such programs over time and have thus been able to qualify for the grants.5 Above all, Table IV-4l demonstrates that the size of such grants is determined by the size of the Youth population. This is as we would expect, since a good chunk of REPART grants are available for the financing of primary school expenditures. _— 5 See Mark Kesselman, The Ambiguous Consensus: A Study of Local Government in France (A. Knopf, 1967). 174 The LEFTRP59 variable was best at explaining than the other political variables. This shows that this is one policy area where, again, Socialists are even more effective (in relation to their control of councils) than Communists in making their cities qualify for every franc and centime that can be extracted from central coffers. Table IV-42 demonstrates that Youth and Growth are different facets of the same phenomenon and are the main determinants of the size of grants that are aimed mainly for capital outlays. G. Conclusions This chapter has demonstrated that Social Welfare Spending, Business Taxation and Grants for Operating Expenditures are the three policy areas which can be explained in terms of a Left to Right continuum. In Educational Spending, Communist cities stand out in the magnitude of their outlays, but as we have seen, this to a large extent is the product of their peculiar combination of characteristics. This much must, however, be said -- Communist cities spend as much and slightly more than their resources allow, even when we consider their larger youth populations and higher poverty. This, therefore, can be taken as evidence that party control has an effect that operates within the context of many other correlated factors. The same to some extent can be said of Total Operating Expenditures. We have discovered that there is indeed a greater commitment to spending across-the-board by Communist cities, but it is, again, an evidence that must be put in the context of the other operating variables. Resources emerges as an important variable in all policy areas, even after we have made the crucial distinction between it and personal wealth. 175 we can see, however, how a research design which omits that distinction and uses blunt measures such as Economic Development Indices6 and the like, would have difficulty in discovering any independent effect by political variables. It is important to note at this stage that the only other study that was able to detect the impact of party control was conducted among English and Welsh County Boroughs with socioeconomic variables that resemble ours in that they were based on taxable property values.7 James Alt was able to prove conclusively (his cross—sectional analysis was repeated for ten consecutive years) that Labor Party control is a powerful explainer of Educational, Housing and Taxation policies (higher spending and lower tax rates). He, too, found party competition to have no significant effect (with or without controls) on policy. Alt did not differentiate between need and resources, but his wealth variable seems to be a closer approximation of our Income variable (it is highly correlated with socioprofessional categories) than of our Taxable Resources. Thus, it is no surprise to find that Alt's variable is also negatively related to policies. 6 See Thomas R. Dye, Politics, Economics and the Public: Policy Outcomes in the American States (Rand-MCNally, 1966). 7 James Alt, "Some Social and Political Correlates of County Borough Expenditures," British Journal of Political Science (January, 1971). 176 Finally, I believe that the reason why his findings are so similar to those we have just outlined is because English County Boroughs are also heavily industrialized. Mere importantly, their industrialization took place under the English equivalent of class polarization. Such heavy industrialization produces both the mobilized working- class and the industrial tax base that its representatives can tax to finance adequate welfare programs. Findings such as Fried's in Austria, that Socialist rule explains only housing policy,8 must again be considered within the sampling frame chosen (which includes cities with as low a population as 1,000). we turn now to our second subsample to see exactly what happens to our hypotheses in a more impoverished and less industrial setting. 8Robert C. Fried, "Politics, Economics and Federalism: Some Aspects of Urban Government in Middle—Europe," unpublished paper delivered at the Comparative Administration Group, American Society of Public Administration, university of Indiana, October, 1970. CHAPTER FIVE FINDINGS (PART TWO) In this chapter we address ourselves to the question of the generalizability of the hypotheses tested in Chapter Four. For this purpose we must first establish what the second sample represents in relation to French urban patterns, old and new. We must also specify some of the methodological limitations inherent in the small (18 cities) size of the subsample. we can then proceed to the data analysis in the same fashion as in Chapter Fbur. Again, we will first look at means and averages for each of the policies concerned, and at simple correlations, so as to establish the overall patterns of relationships and test some of the basic two-variable hypotheses. Since the sample size limits the number of variables that can be included in the development of an explanatory model, we will use one stage to test the basic hypothesis and identify some of the major determinants of each policy area. As in Chapter Four, these will consist of the major ones of Social Welfare, Education, Total Operating Expenditures and the two minor ones of Business Taxation and Grants-in-Aid. 177 178 A. The Validity of the Second Sample l. The Sample and French Urbanization In Chapter Two we specified that French urbanization does not present the same features everywhere. We made a distinction between a new and an old urbanization. It is, however, useful in the process of identifying a city as belonging in one or the other category of the dichotomy; to start with, a non-mutually exclusive typology of French urban societies. These are: 1. The Industrial—Urban 2. The Parisian 3. The Provincial-Commercial 4. The Suburban-Low Density 5. The New and Fast Growing Towns Each of these categories constitutes a particular variety of the urban phenomenon worthy of study in isolation. It can, however, be seen that the first two categories have some of the features of what we have previously called the Stalemate or the Structured Policy. The other categories are mixtures: of old and new urban patterns, of urban and rural cultures, of Stalemated conflict and American-style growth inspired consensus. The sample of 18 communes from Lyons and the Outer Paris urban ring represents these three categories. It is a mixture of nine cities of the old non-industrial, provincial urbanization and nine cities of the new urbanization or suburbanization. Their political behavior, however, presents similar features. Relatively less mobilized electorates and less active administrations are witnessed in both cases. In the case of 179 the Lyons communes, a relative economic stagnation had dried up resources for any major spending programs. In 1963 when we observe these communes, economic renewal was just barely starting. In fact, such a stagnation remains typical of most other provincial areas until the late 1960's. Today this trend has completely been reversed. The long stagnation was particularly aggravated by a relative neglect on the part of the central government's policies, again until the recent efforts at deconcentration and regional planning. Needless to say, the outer ring communes received even less attention than the provincial suburbs of major metropolises. The political climate of such areas has been characterized by Mark Kesselman as much more consensual than that of Paris.1 While the cleavages are still more pronounced than in rural areas, it can be said that the center of the political spectrum is much larger than in Paris. Socialism in the provinces is a respectable middle-of-the-road umbrella for politicians of various extractions. The remnants of the old anti-clericalism still play a role as common denominator for worker, farmer and small shopkeeper, alike. Thus, we find that Socialists and Gaullists collaborate in coalition more often than in Paris to keep the Communists out of power. In this they are more successful: St. Fons in the Lyons area, and Houilles in the Paris suburban ring are both cases where the Communists lose in the second ballot after having dominated the first. They lose to a Gaullist-Socialist alliance. 1The Ambiguous Consensus (A. Knopf, 1967). 180 We see that in politics, the two urban areas of the second sample display striking similarities. If a difference must be drawn, it is that the Gaullists thrive with greater ease in the quickly growing suburban areas where other parties have not had time to institutionalize their hold on parts of the electorate. They make deep inroads in newly growing cities, such as Poissy, where they lure enough of a newly arrived working-class to win. 2. Methodological Limitations We have just shown that despite the distinct character of the two urban areas from which we drew the second sample, the similarities are such as to allow for a meaningful analysis. Comparisons with our first sample may thus prove valuable. But this is not the only problem. A sample size of 18 is bound to lower the significance levels of the structural equations and even more, that of the individual explanatory variables, below what we have so far taken as acceptable levels. Stated differently, the problem is that for a given significance level, the number of variables that can be added to an explanatory model is bound to be even more limited than in our previous sample. Similarly, our ability to distinguish between the effects of each variable and correctly gauge the magnitude of its explanatory power is also hindered. Mere potentially harmful to the meaningfulness of our tests and results would be a lack of sample homogeneity, and variables that seem to have two distinct sampling distributions. While the assumption of normality of the variables is all too often taken for granted, there is nonetheless little doubt that some of the practical implications of this assumption must be checked. Such checks include plotting the data and 181 assessing the size and nature of the variances, standard deviations and coefficients of variation. (See Table III-3.) Such scrutiny in our case (see Chapter Three) seems to confirm the assumptions of homogeneity that we already have supported in terms of the French urban context. Tables III-2 through III-12 in Chapter Three provide ample evidence. In each of the income categories the cities of the second sample display a greater poverty and lower political mobilization as well as lesser Communist penetration; regard- less of whether they are located in the Paris Outer Ring or in the Lyons area. On the other hand, however, we find that as previously stated, growth levels are still not substantially different from those of the Paris Core. This growth statistic, however, presents lesser deviations from its mean than for the Paris sample. we found, in fact, that most of our variables in this sample had smaller standard deviations than for the Paris sample. Smaller standard deviations improve the reliability of measurements. These are then surmountable limitations. MOreover, this small sample is not being used in isolation. Since the task here is one of corroborating previously tested hypotheses, we can tolerate lower significance levels for those relationships that are in the same direction as those of the initial sample. On the other hand, if some relationships are reversed or radically changed in magnitude, we must have at least as high a significance level as for the initial hypothesis testing. we will have more to say on the validity of the discovered relationships (effects on the dependent variable) when we examine the patterns of independent variable intercorrelations. But let us now look at the policy areas through some sample means. 182 B. Expenditure Patterns Even though Table Vel presents even fewer consistent patterns than Tables IV-l and IV-2, it nevertheless makes it possible to describe the nature of some expected unambiguous relationships. Despite the small size of each category of party control, one unequivocal trend emerges: Social Welfare Spending continues to follow a Left to Right continuum. The second generalization to be drawn is that whereas the effect of party control on the other spending areas was minor and disappeared after appropriate controls in the Paris sample, it is non-existent in Table V-l from the outset. Similarly, Gaullist cities show an even greater propensity to spending in Education and total spending, and continue to be the recipients of large central grants for investment purposes, while drawing few, as yet, for Operating Expenditures. Inter— estingly, this latter area continues to display a slight Left to Right pattern as we demonstrated in the previous chapter. Finally, Table V-l demonstrates that if there is one major difference between the two samples, it is that regardless of party control, the suburban and provincial cities consider Social Welfare a costly luxury, while Education seems to warrant more of their efforts. Also, expenditures associated with other policy areas than those just mentioned (i.e., roads), seemed to take up more of their share of the budget. This may be due to the fact that in the outer suburban areas and in the provinces, local governments have to concern themselves with matters (i.e., country roads or cultural facilities) which are amply and centrally supplied in the Paris Core. Needless to say, most of these generalizations require further testing through the introduction of appropriate controls. But let us first turn to the zero—order correlations. 183 TABLE V-1 SELECTED PER CAPITA POLICY AREAS OF 18 PROVINCIAL FRENCH CITIES, GROUPED BY TYPE OF PARTISAN CONTROL OF MUNICIPAL COUNCILS IN 1963. Number of Partisan S.W.Sp.a Ed.Sp.b Tot.Sp.e REPARTd SUBVe Cities Control 3 Communist 27.47 54.18 266.49 19.22 30.26 5 Socialist 20.39 41.30 269.25 10.68 24.24 3 Center Left 19.77 43.89 259.37 8.94 13.06 3 Gaullist 16.02 62.88 408.28 9.93 30.24 4 Conservative 18.74 37.88 217.87 8.92 31.83 7 Right 17.50 48.14 299.43 9.30 31.15 18 A11 Cities 23.18 46.76 279.11 11.30 26.22 See Table III-12 for variable labels and their meanings. a b Social welfare Spending. Educational Spending. cTotal Operating Expenditures. dRepartitions, Grants for Operating Expenditures. eSubvention, Grants for Investment Expenditures. NOTE: The Right category is made up of Gaullist and Conservative cities combined. The Center Left Category describes cities whose municipal council is controlled by a coalition of Socialists, Radical Socialists, Christian Democrats and Left Gaullists, where no one party affiliation prevails. Various combinations are possible. 184 C. Zero-Order Relationships A quick glance at Table V-2 is all that is needed to remove any lingering doubts as to the validity of this sample. The socioeconomic variables are even more closely intercorrelated than in the Paris area. Correlation between Adult Education and measures of the social structure (Workers, Professionals) are of the order of .85 and above, while they did not exceed .74 in the Paris sample. Similarly, the Income indicator is more closely related to the Workers variable and less related to the Professionals variable. What does this tell us about the socioeconomic setting of the second sample? In the provinces, the chances for a worker to have an education above the primary level are remote, while in Paris the better educational system (endowed with major national lycees and technical colleges), to say nothing of cultural institutions such as libraries, museums, etc., makes it more feasible for him to break out of his proletarian ghetto. Conversely, while in Paris professionals are among many with a better education (.74), in the provinces they are practflxily the only ones (.900). Similarly, while they constitute the majority of those with higher incomes in Paris (.866), they are among many others (i.e., landowners and wealthy merchants) in the provinces (.807). Finally, the relationship between Personal Income and Taxable Resources is even.weaker (-.113) than in the previous sample (+.309). In the provinces, as well as in the suburban fringe, the industrial base of some communities does not come even close to compensating for the abject poverty of the population, which may include many semi-rural families. On the other hand, the rich communities are exclusively residential with little of the commercial base or the dense structures of the Inner Paris. .aoaumaaaom onu aH mamcoammmmoum ucoo pom .5 .mmusufiuaomxm ucoaumo>oH pom mucouu .moowuao>n=mw mu .mwuauwpaomxm maaumuomo pom muamuu .mcowuHuHMQOMp .omoa .ooo> shone .omoa .oso> noessaaouo .ooaumospm oasp pom NHIHHH magma mom omo. >mbm qwm.| mam. HmmbmpHm HzmnzmmMQZH ZMMBNHm OZQZ< mZOHH m4mthe Communist Representation component within it. Both indicators of Communist strength fare well when Resources are controlled. Table V414, in fact, shows that CISTRP53 is the most powerful variable (r = .60) in the equation that yields the highest proportion of explained variance. Thus, again we see how important the election of 1953 was in determining 2 Joseph Duplouy, Le Credit aux Collectivites Locales (Berger-Levrault, 1967). 191 TABLE V—5 PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND LEFT REPRESENTATION ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Resources .57 (.54) .54 .026 (.026) .016 LEFTRP59a .41 (.35) .34 .066 (.068) .103 Coefficient of Multiple Determination: R2 = .412 Significant at .019. TABLE V—6 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weights Coefficients Level Resources .57 (.54) .56 .027 (.026) .006 b CISTRP59 .63 (.46) .48 .100 (.094) .015 2 Coefficient of Multiple Determination: R = .521 Significant at .004. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aLEFTRP59 is Left Representation, 1959. bCISTRP59 is Communist Representation, 1959. 192 TABLE V-7 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND COMMUNIST REPRESENTATION, 1953, ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) 0F 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Resources .44 (.54) .56 .017 (.026) .073 CISTRP53a .60 (.66) .54 .291 (.354) .011 2 Coefficient of Multiple Determination: R = .547 Significant at .003. TABLE V-8 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, INCOME AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weights Coefficients Level Resources .63 (.54) .57 .027 ( .026) .008 Income .08(—.21) .06 .014 (-.043) .751 CISTRP59b .56 (.46) .51 .105 ( .094) .023 2 Coefficient of Multiple Determination: R = .531 Significant at .012. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP53 is Communist Representation, 1953. bCISTRP59 is Communist Representation, 1959. 193 spending patterns as observed ten years later. The Communists have left an even clearer mark here, on the communities where they once held policy-making power. The proof that these patterns are mostly the result of that once existant power and not as much the result of today's Communist electorate, is in the lesser strength of the variable CIST59 (.52) (see Table V-3). Needless to say, comparisons between the effects of highly inter- correlated variables are risky. The WOrkers variable, nonetheless, appears throughout most of these spuriousness tests as playing a more powerful role in the determination of welfare Spending than it did in the Paris area (compare Tables V—8 and V-lO). This is partly because of the lesser validity of Income as a demand or need indicator in the provinces where fluctuations in propertied wealth are not as sharp. This is why it fares so poorly in Table V-7. We are thus in a position where no one explanatory design provides a completely satisfactory causal structure. In Table ve10, CISTRP59 is rendered insignificant. Mbreover, any design where CISTRP53 and Workers appear together runs into very strong collinearity (+.877). we can, nonetheless, conclude that Communism past and present does seem to have detectable effects on Social Welfare Spending. The size of the working- class is an equally good explainer. This means that Gaullist and Socialist controlled councils spend more if their constituencies are dominated by WOrkers. Finally, the level of Resources plays its important delimiting effect, but not as powerfully as in other policy areas (i.e., Total Spending) as we shall soon see. Thus, many similarities to the Paris Core area can still be observed. 194 TABLE V-9 PARTIALS, BETAS, AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES AND WORKERS ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 18 PROVINCIAL CITIES IN 1963. Variable Partial Beta Regression Significiance Correlation Weights Coefficients Level Resources .36 (.54) .32 .016 (.026) .156 Workers .49 (.62) .46 .381 (.506) .047 2 Coefficient of Multiple Determination: R = .461 Significant at .010. TABLE V-10 PARTIALS, BETAS, AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, WORKERS AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA SOCIAL WELFARE SPENDING (S.W.Sp.) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weights Coefficient Level Resources .40 (.54) 1.08 .052 (.026) .006 Workers .28 (.62) .26 .215 (.506) .067 CISTRP59a .13 (.46) .10 .021 (.094) .100 2 Coefficient of Multiple Determination: R = .556 Significant at .011. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. 195 TABLE V-ll PARTIALS, BETAS, REGRESSION COEFFICIENTS FOR THE EFFECTS OF YOUTH, RESOURCES, INCOME AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA EDUCATIONAL SPENDING (Ed.Sp.) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weights Coefficient Level Youth .47 ( .55) .46 2.966 (3.533) .073 Resources .31 ( .42) .27 .034 ( .028) .253 Income 3 -.05 (-.l6) -.O4 —.025 {-.085) .847 CISTRP59 .Ol ( .14) .00 .002 ( .075) .989 2 Coefficient of Multiple Determination: R = .384 Significant at .150. TABLE V-12 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF GROWTH (OF POPULATION) AND YOUTH ON PER CAPITA EDUCATIONAL SPENDING (Ed.Sp.) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weight Coefficient Level Growth .35 (.56) .36 .618 .170 Youth .33 (.55) .34 2.161 .194 2 Coefficient of Multiple Determination: R = .394 Significant at .023. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. 196 TABLE V—13 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF GROWTH (OF POPULATION) AND RESOURCES ON PER CAPITA EDUCATIONAL SPENDING (Ed.Sp.) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weight Coefficient Level Growth .65 (.56) .59 1.008 .005 Resources .54 (.42) .45 .055 .024 2 Coefficient of Multiple Determination: R = .521 Significant at .004. TABLE V-14 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF RESOURCES, INCOME AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA TOTAL OPERATING EXPENDITURES (Tot.Sp.) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Correlation weight Coefficient Level Resources .75 ( .75) .75 .632 ( .635) .001 Income -.04 (—.11) -.O3 -.095 (-.400) .894 CISTRP59a .00 (-.03) .00 .008 (-.097) .990 2 Coefficient of Multiple Determination: R = .572 Significant at .006. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. 197 E. The Determinants of Educational Spending This policy area displays a non-partisan pattern. None of the party control variables yielded any significant explanation; in fact, even their simple correlation coefficients are weak (see Table Ve3). Table V-15 summarizes the results of a series of spuriousness tests, permutations and other explanatory models. As we can see in Table V412, the highly intercorrelated variables of Youth and Growth explain a larger proportion of the variance than the four of Table Vell. The optimal explanatory model includes two variables that we can call environmental: Growth and Taxable Resources. The growth variable is descriptive of the new patterns of French urbanization which resemble more the American suburban phenomenon with its non-partisan commitment to public education. This is in contrast to the old urbanization of the Paris inner ring, where private (parochial) education plays a larger role and separates classes and political constituencies, and where it was the older communities that spend more. F. The Determinants of Total Operating Expenditures If the influence of Resources was quite powerful in the first sub- sample, it becomes the only determinant of Total Expenditures in this subsample. Table Vel4 shows that it alone accounts for 56% of the variance, and all other indicators do not add anything of significance. The fact that cities of the provinces receive lesser attention from the central government, and are required to remain closer to official indebtedness ceilings than those of the old Seine Departement, has already been mentioned as a reason for the stronger impact of Resources. This is corroborated by various observers and studies3 even though it 3Idem. This is also to be found in the report called "Les Finances des...." (See Chapter Two, Table II-l n.) 198 would seem at first that Inner Paris communes would be more closely supervised and therefore required to Stay closer to official ceilings. The fact is that the supervision is closer mostly in matters such as police and communications (roads). The financial supervision has more often meant encouragement in following certain policies of a central government keen on reducing the blight of certain areas of its capital city. G. Other Policy Areas 1. Grants for Operatinngxpenditures However weak the influence of party control variables on Education (the main recipient of such grants), we nonetheless discover that the diligence of Communist administrations to qualify for every franc and centime dispensed by central authorities is still as important in the provinces as in the Paris inner ring (see Table V-15). The only difference here is that the Socialists are not as diligent. 2. Business Taxation This small and symbolic policy area again displays the impact of party control variables at its highest level. And, again, it is not the lack of resources that explains the need for such revenue but the opposite. The results of Tables V-l6 and V417, with the very high variance explained (91%) by the same three independent variables we used to explain Social Welfare Spending in the Paris inner ring, seem to confirm that cities displaying a combination of those characteristics constitute a distinct form of urbanization and policy-making. As a 199 TABLE V-15 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF YOUTH, RESOURCES AND COMMUNIST REPRESENTATION, 1959, ON PER CAPITA OPERATING EXPENDITURE GRANTS (REPART) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Correlation Weight Coefficient Level Youth .39 ( .44) .33 .574 .149 Resources -.15 (-.05) -.12 -.004 .582 a CISTRP59 .57 ( .64) .56 .081 .027 2 Coefficient of Multiple Determination: R = .497 Significant at .043. TABLE V-16 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME, RESOURCES, COMMUNIST REPRESENTATION, 1959, ON PER CAPITA BUSINESS TAXATION REVENUE (Bus.Tax.) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Variance Correlation Weights Coefficients Level Explained Resources .92 ( .83) .81 .032 .0005 64.8 CISTRP59b .56 ( .32) .25 .042 .024 5.5 2 Coefficient of Multiple Determination: R = .882 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP59 is Communist Representation, 1959. Same as above. 200 TABLE V-17 PARTIALS, BETAS AND REGRESSION COEFFICIENTS FOR THE EFFECTS OF INCOME, RESOURCES AND THE INTERACTION OF RESOURCES AND COMMUNISM (CISTRP53xR) ON PER CAPITA BUSINESS TAXATION REVENUE (Bus.Tax.) OF 18 PROVINCIAL FRENCH CITIES IN 1963. Variable Partial Beta Regression Significance Variance Correlation Weights Coefficients Level Explained Income -.57 (-.46) —.23 -l.96l .022 4.5 Resources .93 ( .83) .78 .030 .0005 60.5 CISTRP53xRa .68 ( .47) .32 .00016 .004 7.5 2 Coefficient of Multiple Determination: R = .907 Significant at .0005. Figures in parenthesis are zero-order correlation coefficients for the relationship between the particular variable and the dependent variable. aCISTRP53xR is Communist Representation, l953,multiplied by Resources. 201 social phenomenon, they display striking consistencies through time (i.e., CISTRP53) and space (Inner and Outer Paris). This minor field of Business Taxation shows that when symbolic gestures are needed, provincial Communism is not different from Parisian Communism. H. Conclusions We have shown in this chapter that despite the limitations on thorough analysis imposed by sample size, it is still possible to discern specific spending patterns. In summary, we can say that the second sample does not constitute a completely distinct universe. Provincial city budgets differ from Parisian cities to the extent of their lesser resources and industrialization, but these differences are compounded by resulting in less mobilized workers, less well organized Communist organization and more consensual Socialists. The confirmation of some hypotheses relative to minor policy areas (Grants and Business Taxation) leads us to conclude that the Structured Policy Mudel is not a special case of Parisian urbanization. It can be said that it is found in its ideal form in Paris, but has existed in varying purity throughout all French urban areas of the pre-World War II era (i.e., Le Havre and the Departement of Nord). Its features (i.e., class warfare through the use of the budget) were somewhat moderated by the more consensual culture of the provinces. Today, this trend is accelerated by the new urbanization. CHAPTER SIX CONCLUSIONS In concluding this study, we can ask -- what was accomplished here? We began by deriving from the research and literature of policy analysis, one major hypothesis pertaining tothe impact of party control as a political variable on budgetary policies. we hypothesized that in France, at the local level, changes in the party in power are likely to result in detectable (non-spurious) changes in the levels of local social welfare expenditures. In conceptualizing why political variables should have an impact onpxiicy in France, we constructed two models of policy-making, the Fluid and the Structured. The Structured Policy-Making Model describes a number of political systems such as Austria, Belgium, Netherlands, Sweden and Britain, with the following features: 1. Crystallization of sociaf cleavage hostility into corresponding permanent political and parapolitical structures. 2. Political parties make policy by seeking to maximize benefits only to their limited and captive constituencies. Political variables can also be expected to affect Social Welfare expenditures because of the redistributive nature of that policy area. To test this hypothesis, we collected budgetary data for 1963, from French municipalities in the Paris and Lyon urban areas. We formed two 202 203 samples, one of 36 cities from the Paris Core area and one of 18 cities from the outlying Paris suburbs and the Lyon urban area in order to avoid a biased sample. Budgetary data was available from the archives of French local governments; socioeconomic data was made available through French census publications, even though some crucial variables (Taxable Resources and Personal Income) were also found in the budgets (assessed valuations) of local governments. Finally, the political variables, operationalized as percent of council controlled by Left parties (and more specifically by the Communist party), were developed out of electoral results in Le Mbnde for 1953 and 1959. The newspaper was particularly useful in that it identifies the party affiliations of electoral lists, which often carry misleading labels. The method of least-squares was chosen to test the main hypothesis and its corollaries. This entailed the use of computer routines which allow for the simultaneous controlling of the effects of many variables. we were also fortunate that our particular program included the automatic calculation of significance levels for individual variables through a t-test performed on the standard error of the regression coefficients. This was particularly useful in alerting us to cases of multicollinearity, also known as high intercorrelation between independent variables (such as income and social structure). In coping with the problem we followed two strategies. When testing for spuriousness (as when controlling for resource levels), alternative indicators were used. Those that were not as highly correlated with other independent variables allowed us to perform a conclusive test. 204 In cases where no such tests were called for, because political theory did not indicate such a possibility, we chose to introduce each ' of the correlated variables in alternative sequences. we were thus able to compare the relative impact of each (in terms of total variance explained) on the dependent variable. The results of the testing were quite conclusive. In the Paris Core area, the effect of party control was very strong in determining Social Welfare expenditures, but played an intervening role in Total expenditures and to some (although less detectable) extent, seemed to increase Educational expenditures. In the 18 provincial cities, party control had an effect only on Social Welfare and was not even associated with any pattern in Educational and Total expenditures, where the pervading influence of taxable resource levels was all too powerful. The divergent findings justified our sampling strategy and allowed us to discover the powerful effect exercised by sharp disparities in resources and industrialization. Thus, we can add one more basic assumption to the Structured Model of Policy-Making —- the necessity for a minimum overall level of industrialization and taxable resources. Thus, while the Paris sample exemplified the ideal of the Structured Mbdel, the second was a variant which finds all cities spending less and the Communists expressing their relative isolation amid a more consensual setting by making the expected choice to overspend only in Social Welfare. In comparing our two sets of findings to those of Fried in Italy and 1 Austria on one hand, and to those of James Alt in England on the other, See footnotes 7 and 8 in Chapter Four. 205 we find confirmation to the conclusion that for political variables to have an effect on policy, a certain threshold of industrialization must have been achieved and with it an adequate resource base. In a systemic perspective, we can say that the relationship between socioeconomic development and policy with politics as an intervening variable is not continuous or linear over time. It is only after industrialization has sufficiently mobilized an electorate (along well defined cleavages) that political parties will begin to add to policies an imprint that can be attributed only to them. Such discontinuities should be a warning signal to students of expenditure determination that great attention should be given to their settings and to the composition of their samples. Also in comparing our results to those of James Alt's2 in Britain, we can at least minimize the plausibility that the effect of party control in France can be explained by the special status of the French Communist Party (whose political strategies could be extra-systemic in their inspiration). The British Labor Party finds as much difficulty in turning out the working-class vote in the countryside as the French Communist Party in holding on to such a vote on the runoff elections in the provinces. Among the more interesting corollary findings was the tendency for Left and Communist administrations, in both settings, to qualify for more grants-in-aid than their socioeconomic situation warranted. Grants- in-aid in turn were found to have little impact on spending levels in either samples, except in Education where they expectedly paralleled the effect of the size of the pupil population. See footnote 7 in Chapter Four. 206 We were also able to demonstrate (in a minor policy area, Business Taxation) how political variables can be found to have a powerful effect regardless of setting when the policy is largely symbolic in nature. Equally interesting was the pattern of Educational Spending in the Paris Core area where the negative impact of the Personal Income variable seemed to indicate a preference of peOple in rich communities in France for private education. we haveshown in this study how electoral systems, such as Propor- tional Representation, can institutionalize a set of voting patterns and policy-making trends that can be detected ten years later. On the other hand, we have also seen how the Two Ballot Majority System had resulted as of 1959 in divergent electoral and policy-making strategies for the Socialists. In Paris, Socialist leaders would act like Communists (even when not in coalition with them) in increasing Social welfare expenditures. Socialist leaders in the provinces acted more like Center politicians in keeping expenditures within range of the community's resource base. In coalition with Center politicians, they routed the Communists out of communes where the latter had a plurality. Finally, voter turnout was found to have no relationship to policy or party control. The same could be said also of the degree of party competition. Even though we did not specifically test this hypothesis, it was not difficult to observe that the least competitive cities often had the highest expenditures. In summarizing the more meaningful contributions that this study can be said to have made to political science, we can categorize these 207 into theoretical clarifications, substantive arguments and use of concepts and methods. Needless to say, whatever was accomplished here was possible only because of previous efforts by ingenious scholars in many fields and disciplines. The theoretical clarifications pertain to our use of the concept of institutionalization in explaining why and how some political systems display a higher politicization. In substantive matters, we hope to have presented a more accurate and up-to-date description of a political system (the French) which has often seemed to be the subject of cliches and caricatures. This applies all the more to the study of French local government. In our efforts at concept formation, we hope to have convinced fellow researchers of the necessity of differentiating between taxable resources and personal income. The powerful suppressor effects dis- covered in the form of highly industrialized communities with an impoverished population have applicability in all urban expenditure studies. we have pointed out that in order to be able to detect such a suppressor effect, special care must be given to the composition of a sample. What are the ranges in life styles that the sample represents -- industrial, rural, commercial or neo-industrial (clean industries and services)? Too much of a mix will cloud subtle relationships. But the conceptual distinction will allow the researcher more unequivocal conclusions as to the limiting influence of resource levels on governmental spending. 208 1. Theoretical Considerations If there is one area where we have relied more heavily on the efforts of others, it must be in the field of cleavage theory. The works of Dahl, Rokkan, Lorwin, Liephart, Powell and Stiefbold have already been acknowledged in inspiring us.3 Their value has been in providing answers imbedded in the historical context on one hand, and organizational theory on the other. Their focus on the system of conflict resolution adopted by nations and more specifically democracies, allows to often uncover unsuspected reasons for the stability and maintenance of democracies. Their work on fragmented societies shows that high levels of consensus are not the most important underpining in the functioning of democracies. Of greater importance seems to be the patterns of reward and payoff distribution among competing elites. The behavior of Communists in bringing down the weimar Republic must be contrasted to that of the French Communists during the crisis of 1968, and points to one major difference. In the France of 1968, the Communists had something to lose (namely their control of certain institutions, like municipal councils studied here). In other words, cleavage theory provides us with answers to questions of systemic importance. It has also began tidying up the definitional looseness that had so far pervaded the study of ideology, conflict, hostilities, and participation. These are now defined in more concise 3 See footnote 18 in Chapter One and also footnote 1 in Chapter Two. 209 and behavioral terms (i.e., ideology must exhibit constraint). Many of them are tied to specific and observable entities -- organizations. One particular type of organization —- the political party, will affect the persistance of political attitudes and ideologies. We have, however, pointed out that while political parties usually have an interest in politicizing most social issues, some party systems can develop which may minimize such a politicization, particularly as it may be reflected in expenditures. This is why we have deemed it important to make the distinction between a Fluid and a Structured policy-making system. Finally, cleavage theory reminds us of the importance of the stakes and terms of the conflict in bringing about polarization or consensus. This is why, since V. 0. Key, political scientists have expected to find higher politicization reflected in issues that were called redistributive (i.e., Social welfare Spending). It must be remembered, however, that the prevailing state of the economy will determine to a large extent which issues can truly be regarded as redistributive (zero-sum) conflicts. Expanding economies can allow even the most faithful followers of Herbert Spencer to establish large Social welfare Programs, and turn most governmental spending into a giant pork barrel with a bit for everybody. Conversely a stagnating economy (which has often been the case of pre-war European democracies) may make it impossible to define any issue in other terms than "us versus them." The end of idology argument has in fact been made under the assumption of a prevailing prosperity that was supposed to be here to stay. 210 Today, when most industrial democracies are beset with shortages, production cutbacks and conservation problems, zero-sum issues are again more numerous. Whether a new politicization emerges out of new conflicts and resultant cleavages will, however, still depend on the nature and depth of past institutionalization patterns. It is interesting to speculate what changes such an increase in zero-sum issues will bring about to the American party system and the Fluid policy-making system. 2. substantive Arguments In our inquiry into the workings of French local government, we discovered that generalizations made on the premise that the French state is unitary and centralized in structure, can hide a much more complex picture of the reality. Beyond the usual question as to how rigorously central decisions are enforced, we discovered that a strong local party system can often overcome the restrictions on local autonomy imposed by constitutions and legislation, and display a surprising spirit of local initiative. We can conclude that in the burgeoning field of cross-national comparative local government, the study of local party systems and more specifically that of the local roots of national party systems, must not be overlooked. 4 This is what is advocated also by Mark Kesselman in his "Changes in the French Party System," Comparative Politics, 4 (January, 1972), 211 The relative atrophy of local party systems among American communities, often means that local interests are helpless against the power of national executive agencies and nation-wide corporations. Paradoxically, federalism can often mean more restrictive federal and state guidelines in the granting of loans and subsidies than in a system where the central state does not have to worry over the definition of its formal powers. And the actions of various federal agencies in leapfrogging local structures, and operating through their own system of service delivery can often result in lesser local con- sultation than what usually happens in the French prefectoral system. In a trend paralleling the increase in power of the national Presidency, responses to the urban crisis have also featured increases in local administrative personnel and executive (mayoral) power.5 Such a strategy has often weakened even further any form of local political will. Rarely do we see working-class towns and cities able to mount their own welfare and poverty rehabilitation schemes as we have witnessed in the Paris Red Belt. All too often American working-class communities (e.g., Toledo) are still governed by middle-class "managers" able to predominate over a party system weakened by the blows of the "reform movement." It is no wonder that we have witnessed many periodic uprisings by elements of the working-class over the power of what they call In manager-council type cities, it is the manager who has assumed powers beyond the definition of his office; in cities like New York, the mayoral office was enhanced; more recently, Atlanta and Detroit have revised their charters to grant more power to the mayoral office. 212 "pseudo—intellectuals" attempting to impose reforms without any consultation.6 In the meantime, the multiplicity of administrative arrangements and these uprisings help perpetuate the myth of local autonomy. This observer's recommendation is that at the national, as well as at the local level, a revitalizing restructuring of the party system is needed. It is more likely to improve the quality of our democracies than the continuing obsession with administrative tinkering. Why is it that the same innovative spirit cannot be applied to input structures? 3. Research Implications. For the social scientist, the administrative complexity of local government poses immense challenges to any attempt at evaluating the performance of such governments; the same problems of comparability remain.7 Some highly qualified users of multivariate techniques throw their hands up in despair and seriously advise a return to case studies. This may indeed be useful and necessary if we are to understand the actual functioning of administrative interrelationships and their economic-budgetary implications.9 But on the wider and more theoretic 6The feelings of supporters of George Wallace in the busing controversy, and also the response of some communities to the fluoridation problem, are to be remembered. See Robert L. Crain, Elihu Katz and Donald B. Rosenthal, The Politics of Communipy Conflict, the Fluoridation Decision (Bobbs-Merrill Company, 1969). See, for example, Terry N. Clark's compendium of articles in his Communi£y_Decision-Making (Chandler Press, 1966). 8 See Roy Bahl, who wrote Metropolitan City Expenditures (university of Kentucky Press, 1968), and recently expressed such a feeling in a conversation. 9 See footnote 2 in Chapter One. 213 plane, the functioning of more homogeneous local government systems such as those of France, England, or Austria, may constitute a more fruitful field10 for the systematic testing of hypotheses, as we have done in this study. The application of multivariate techniques in the realm of political phenomena is hardly a new development, yet even if the conceptual problems are improved, some problems remain in the limitations imposed by the data and format of simple cross-sectional analysis. Multicollinearity is one such problem. The simultaneous use of time series and cross-sectional analysis has recently been attempted.11 Path analysis and other econometric techniques are also resulting in interesting studies; and while it is true that technique alone cannot solve all our problems, it can still be of help. When a good research design has been found, tested in a microcosm, and when interesting and meaningful results have indeed emerged, it is valuable to pursue it further, enlarge the scope, improve the data, expand the number of variables to include those that are more difficult to construct, and bring to bear more powerful statistical techniques. 0 See footnote 43 in Chapter One. 11 See Virginia Gray, "Time Series Analysis of State Spending," paper presented at the Annual Meeting of the Midwest Political Science Association in Chicago on May 3-5, 1973. 214 4. Improvements and Future Research The possible improvements to this study are numerous. First and foremost would be an expansion in the sample size to include more cities from the Outer Paris Area, and newly growing urban areas similar to Lyon's. Conversely it would be useful to include the cities of the Lille urban area, which is the oldest industrial area of France. This would allow us to ascertain the uniqueness of the Paris Core Area, and whether the Stalemate as we have described it was indeed part of the Old Urbanization. We suspect that results would not differ. By and large, the addition of cities from the Southern areas would present only marginal benefits, each of the latter (i.e., Toulouse, Bordeaux, Marseilles) being relatively distinct from each other administratively, as well as politically. A different research strategy would be to focus data gathering at the level of the large metropolis of 100,000 population and above. This would be a study of central cities. The second most obvious improvement in the quality of the data, would result from the use of more than one time period for the budgetary data. Using an average of two or three consecutive years of expenditures would improve significance levels and allow some analysis of capital outlay patterns. Change patterns can be studied by duplicating the same data for 1955 (two years after the 1953 municipal elections, which we have found to have set a trend) and 1968 (three years after the 1965 elections which represent the first efforts by Gaullists at institutionalizing their rule). Studies of French cities after 1970 might run into some problems of comparability, given the changes brought about in tax collections and 215 assessment practices. The interesting problem of metropolitan consolidation would, however, be the main focus of any post-1970 study. This would present an excellent opportunity for the testing of various hypotheses pertaining to the effect of structural changes. Finally, the relationship between national and local government could be better explored through a study of the allocations of the national budget in each policy area. In other words, it would be useful to know the size of the local effort in relation to national expenditures in each policy area (i.e., education). This would allow a better measurement of the effectiveness of certain central directives aimed at inducing specific local policies. We could then have a some— what more substantiated base of comparison between a centralized and a federal system. As far as improvement in the operationalization of specific variables, we can point to one possible fruitful area. Political variables, measuring the strength of partisan structures and secondary organizations (i.e., party membership lists or finances, unionization and church attendance data), would allow us to have a better operational- ization of what we have called political institutionalization. We could then ascertain to what extent such institutionalization is based on control of governmental bureaucracies and budgets, and how much of it is based on participatory structures. Finally, improvements in technique can go beyond the use of multiplicative models attempted in this study. The political scientist stands to gain from a scrutiny of the more recent trends in the econometric 216 analysis of expenditures. Policy analysis was born out of the initial efforts of economists such as Fabricant,Fisher and Sachs. Today other economists are experimenting with new and more sophisticated models12 (i.e., wage and employment functions for expenditure determination) that merit our attention. 12See Richard Gustely, The Process of City Government Determination, unpublished doctoral dissertation (Syracuse University, 1973). See also Roger Schmenner, "On the Determinants of Municipal Employee Wages," Review of Economics and Statistics (February, 1973), pp. 83-90. APPENDIX APPENDIX A Data Sources Annuaire Statistique de la Ville de Paris et des Communes Suburbaines de la Seine, "Anneel965;“ Paris, 1968. Institut National de la Statistique et des Etudes Economiques (INSEE), Recensement General de la Population de 1962; Resultats du Depouillement Exhaustif, "Fascicules Départementaux" (VOlumes: Seine, Seine-et-Oise, Rhone). Ministére de l'Interieur, Direction Generale des Collectivites Locales, Service de Statistique et d'Analyses Financieres; Les Communes a l'heure de la Region; "Les Finances des Communes de plus de 5,000 habitants" (Exercise de 1965). Le Munde (March 27, 1959). Le MOnde (March 15, 1953). Compte Administratif de la Ville de Alfortville n H " " " d' Antony H " " " " d' Asnieres H " " " " d' Aubervilliers H n H H II de Bagneux H n I! H H de Bagnolet N " " " " de Bobigny n n n u H de Boulogne-Billancourt H " " " " de Bourg La Reine n H " " " de Bron cc " " " " de Cachan N " " " " de Caluire et Cuire H " " " " de Charenton Le Pont " " " " " de Choisy 1e Roi H H H II II de Clamart 217 6. 218 Compte Administratif de la Ville de H de de de de de de de de de de de de de de de de de de de de d' de de de de de de de de de de de de Clichy Clermont-Ferrand Conflans Sainte Honorine Courbevoie Creteil Drancy Genevilliers Givors Houilles Fontenay Sous Bois La Garenne Colombes Le Chesnay Le Vesinet Les Lillas Lyon Nanterre Neuilly Noisy Le Sec Nogent Sur Marne Maisons Lafitte Mbntrouge Oullins Poissy Puteaux Sartrouville Saint ans Saint Mande Saint Maur des Fosses Saint Ouen Vanves Venissieux Versailles Villeneuve La Garenne Villeurbanne BIBLIOGRAPHY BIBLIOGRAPHY Abramson, Paul. "Why the Democrats are No Longer the Majority Party." Paper prepared for delivery at the 1973 American Political Science Association Convention, New Orleans, September 4-8. ------ . "Generational Change in American Electoral Behavior," American Political Science Review, 68 (March, 1974). Alford, Robert R. Party and Society: The Anglo-American Democracies (Rand—McNally, 1963). Allardt, E. and Rokkan, Stein (eds.). Mass Politics (Free Press, 1969). Almond, Gabriel. The Appeals of Communism (Princeton University Press, 1954). Althauser, Robert P. "Multicollinearity and Non-Additive Regression Mbdels," in H. M. Blalock, Jr., Causal MOdels in the Social Sciences (Aldine Publishing Company, 1971). Alt, James. "Some Social and Political Correlates of County Borough Expenditures," British Journal of Political Science, 1 (January, 1971). Anderson, Robert T. and Anderson, Barbara G. Bus Stop for Paris: The Transformation of a French Village (Doubleday, 1965). Aubin, Roger. Communes et Democratie; Taches et MOyens de la Commune (Paris: Les Editions Ouvrieres, 1965). Backstrom, Charles H. and Hursch, Gerald. Survey Analysis (Northwestern University Press, 1966). Bahl, Roy. Metropolitan City Expenditures (University of Kentucky Press, 1968). Barnes, Samuel H. "Ideology and the Organization of Conflict: On the Relationship between Political Thought and Behavior," Journal of Politics, Vol. 28 (1966). Bauer, Raymond; Dexter, Lewis; and de Sola Pool, Ithiel. American Business and Trade Policy (Atherton Press, 1964). Bay, Christian. The Structure of Freedom (Stanford University Press, 1958). 219 220 Bell, Daniel. The End of Ideology: On the Exhaustion of Political Ideas of the Fifties, new ed. rev. (Collier Books, 1962). Birdsall, William C. "A Study of the Demand for Public Goods," in Richard A. Musgrave (ed.) Essays in Fiscal Federalism (Brookings, 1965). Birnbaum, Norman. "Great Britain: The Reactive Revolt," in Merton A. Kaplan (ed.) Revolution in WOrld Politics (J. w. and Sons, 1962). Blalock, Hubert M. "Correlated Independent Variables: The Problem of Multicollinearity," Social Forces, 62 (December,l963). ------ . Causal Inferences in Non-Experimental Research (University of North Carolina Press, 1964). -----. "Theory Building and the Concept of Interaction," American Sociological Review, 30 (June, 1965). ------ . "Causal Inferences, Closed Populations and Measures of Association," American Political Science Review, 61 (March, 1967). ------ . "Correlated Independent Variables: The Problem of Multi- collinearity," in Edward R. Tufte, The Quantitative Analysis of Social Problems (Addison-Wesley Company, 1970). ------ . Social Statistics, 2nd Ed. (McGraw-Hill, 1971). Boaden, Noel T. and Alford, Robert R. "Sources of Diversity in English Local Government Decisions," Public Administration, 47 (Summer, 1969). Bon, Frederic, et. al. Le Communisme en France (Cahiers de la Fondation Nationale des Sciences Politiques, #175, Colin, 1970). Brogan, D. w. The Development of Mudern France (Revised edition, Harper and Row, 1966). Buron, Robert. Le Plus Beau des Metiers (Plon, 1963). Burnham, Walter Dean and Sprague, John D. "Additive and Multiplicative Models of the VOting universe: The Case of Pennsylvania: 1960-1968, American Political Science Review, 64 (June, 1970). Campbell, Angus; Converse, Phillip E.; Miller, Warren E.; and Stokes, Donald E. The American Voter (John Wiley and Sons, 1960). ------ . Elections and the Political Order (John Wiley and Sons, 1966). Cathlineau, Jean. Le ContrOle des Finances Communales en France (Librairies Generales du Droit et de Jurisprudence, 1963). 221 Chapman, Brian. Introduction to French Local Government (George Allen and Unwin, 1953). ------ . The Prefects and Provincial France (George Allen and Unwin, 1955). Charlot, Jean. Etude du Pouvoir au sein d'un parti politique (Colin, 1967). ------ . The Gaullist Phenomenon (Praeger, 1971). Clark, Terry N. Community Decision-Making (Chandler Press, 1966). Cnudde, Charles F. and MCCrone, Donald J. "The Linkage Between Constituency Attitudes and Congressional Voting Behavior: A Causal MOdel," American Politipal Science Review, 60 (March, 1966). ------ . "Party Competition and Welfare Policies in the American States," American Political Science Review, V61. 63 (September, 1969). Coleman, James S. "Evaluating the Relative Importance of Variables," American Sociolpgical Review, V61. 26 (December, 1961). ------ . Introduction to Mathematical Sociology (The Free Press, 1964). Colvaz, Jean-Yves. "L'idée de Dépolitisation: Jugement de Valeur," in Georges Vedel (ed.), La Dépolitisation (Caliers de la Fondation Nationale Des Sciences Politiques, N"120: Paris, Armand Colin, 1962). Controt, Aline, et Francois Dreyfus. Les Forces Religieuses Dans La Societe Francaise (Colin, 1965). Converse, Philip and Dupeux, George. "Politicization of the Electorate in France and the United States, Public Opinion Quarterly, 26 (Spring, 1962). ------ . "The Nature of Belief Systems in Mass Publics," in Ideology and Discontent, ed. David Apter (Free Press, 1964). Crecine, John P. Governmental Problem Solving (Rand-McNally, 1965). Curtius, Ernst R. The Civilization of France (Vintage Books, 1962). Cyert, Richard M. and March, James. A Behavioral Theogy of the Firm (Prentice-Hall, 1964). Dahl, Robert. "Patterns of Oppositions," in Robert Dahl's Political Oppositions in Western Democracies (Yale University Press, 1966). Dahrendorf, Ralf. "Recent Changes in the Class Structure of European Societies," Daedalus (93, L, 1964). 222 Dansette, Adrien. Religious History of MOdern France, Vol. II: under the Third Republic (Herder and Herder, Inc., 1961). Davis, Otto A. and Haines, George H. "A Political Approach to a Theory of Public Expenditures: The Case of Municipalities," National Tax Journal, Vol. 19 (September, 1966). Dawson, Richard E. and Robinson, James A. "Inter-Party Competition, Economic variables and welfare Policies in the American States," Journal of Politics, V61. 25 (May, 1963). De Savigny, Jean. L'Etat Contre Les Communes? (Seuil, 1970). Deutsch, Karl. Nationalism and Social Communication (M.I.T.Press, 1953). ------ . "Social Mubilization and Political Development," American Political Science Review, 54 (September, 1961). Dogan, Mattei. "Le V6te Ouvrier en France, Analyse Ecologique des Elections de 1962," Revue Francaise de Sociologie, 6 (December, 1965). ------ . "Political Cleavage and Social Stratification in France and Italy," in Seymour M. Lipset and Stein Rokkan (eds.) Party Systems and VOter Alignment (The Free Press, 1967). Downs, Anthony. An Economic Theory_of Democracy (Harper and Row, 1957). ------ . Inside Bureaucracy (Little and Brown, 1966). Duclos, Jacques. "Les Municipalites Communistes au Service des Populations," Nouvelle Revue Internationale, 9 (May, 1966). Due, John F. Government Finance: Economics of the Public Sector, 4th Ed. (Wirzin, 1968). Duverger, Maurice. Political Parties (2nd Ed., translated by Barbara and Robert North, London, Metheun and Company, 1959). Duplouy, Joseph. Le Credit Aux Collectivities Locales (Berger-Levrault, Paris, 1967). Dye, Thomas R. Politics, Economics and the Public: Policy Outcomes in the American States (Rand-McNally, 1966). Dye, Thomas R. and Zeigler, Harmon. The Irony of Democracy: An Uncommon Introduction to American Politics (Wadsworth Publishing Company, 1970). Easton, David. The Political System (Alfred A. Knopf, 1953). Edinger, Lewis (ed.). Political Leadership in Industrialized Societies (Wiley and Sons, 1967). 223 Ehrman, Henry w. Politics in France (Little, Brown and Company, 1968). Eisenstadt, S. N. "Initial Institutional Patterns of Political MOdernization," Civilisations, 12 (1962). ----- . "Institutionalization and Change," American Sociological Review, 24 (April, 1964). Engelman, Frederick. "Austria, The Pooling of Opposition," in Robert Dahl, (ed.), Political Oppositions in Western Democracies (Yale University Press, 1966). Eyestone, Robert. The Threads of Public Policy: A Study in Polipy Leadership (Bobbs-Merrill, 1971). Ezekiel, MOrdecai and Fox, Karl A. Methods of Correlation and Regression Analysis (John Wiley and Sons, 1967). Fabricant, Solomon. The Trend in Government Activity in the united States Since 1900 (National Bureau of Economic Research, 1952). Farrar, Donald E. and Glauber, Robert. "Multicollinearity in Regression Analysis: The Problem Revisited," Review of Economics and Statistics, 49 (1967). Fauvet, Jacques and Mendras, Henri. Les Paysans et la Politique Dans la France Contemporaine (FOndation Nationale des Sciences Politiques, Cahier #94, Colin, 1958). Fisher, Glenn w. "Determinants of State and Local Government Expenditures: A Preliminary Analysis," National Tax Journal (December, 1961). ------ . "Interstate Variations in State and Local Government Expenditures," National Tax Journal (March, 1964). —————— . Taxes and Politics: A Study of Illinois Finance (university of Illinois Press, 1969). Fried, Robert C. "Communism, urban Budgets and the Two Italies: A Case Study in Comparative Urban Government," Journal of Politics (July, 1971). ----- . "Politics, Economics and Federalism: Some Aspects of Urban Government in Mittle—Europa" unpublished paper delivered at the conference on "Problems of European Bureaucracy," of the Europe Committee of the Comparative Administration Group, American Society of Public Administration, University of Indiana, October, 1970. Froman, Lewis, Jr. "An Analysis of Public Policies in Cities," Journal of Politics, 29 (1967). 224 Fry, Brian R. and Winters, Richard. "The Politics of Redistribution," American Political Science Review, 64 (June, 1970). George, Pierre, et. a1. Etudes sur la Banlieu de Paris (Cahiers de Fondation Nationale des Sciences Politiques, Colin, 1950). Gravier, F. J. Paris et 1e desert Francais (Paris, Le Portulan, 1947). Gray, Virginia. "Time Series Analysis of State Spending," paper presented at the Annual Meeting of the Midwest Political Science Association in Chicago on May 3-5, 1973. Greene, Thomas H. "The Communist Parties of Italy and France: A Study in Comparative Communism," World Politics, 21 (October, 1968). Goblot, Edmond. La Barriere et le Niveau (Paris, 1925). Gordon, Robert A. "Issues in Multiple Regression," American Journal of Sociology (1973) . Gustely, Richard. The Process of Citngovernment Determination, unpub- lished doctoral dissertation (Syracuse University, 1973). Hamilton, Richard. Affluence and the French Worker in the Fourth Republic (Princeton University Press, 1967). Hansen, Niles M. French Regional Planning (Indiana University Press, 1968). Haveman, Robert H. and Margolis, Julius (eds.). Public Expenditures and Policy Analysis (Markham, 1970). Hays, William L. Statistics for Psychologists (Holt, Rinehart and Winston, 1963). Hitch, Charles J. and McKean, Roland N. The Economics of Defense in the Nuclear Age (Harvard University Press, 1960). Hofferbert, Richard I. "The Relation Between Public Policy and Some Structural and Environmental Variables in the American States," American Political Science Review, 60 (March, 1966). Hoffman, Stanley. "Paradoxes of the French Political Community," in Stanley Hoffman, et. al. In Search of France (Harvard University Press, 1963). Hollenhorst, Jerry and Ault, Gary. "An Alternative Answer to: Who Pays for Defense?" American Political Science Review, 65 (September, 1971). Huntington, Samuel P. Political Order in Changing Societies (Yale University Press, 1968). 225 Jennings, M. Kent and Zeigler, Harmon. "The Salience of American State Politics," American Political Science Review, 64 (June, 1970). Johnston, J. Econometric Methods, 2nd Ed. (McGraw—Hill, 1972). Kessel, John H. "Governmental Structure and Political Environment: A Statistical Note about American Cities," American Political Science Review, 56 (September, 1962). Kesselman, Mark. The Ambiguous Consensus: A Study of Local Government in France (Alfred A. Knopf, 1967). -----. "Over-Institutionalization and Political Constraint: The Case of France," Comparative Politics, 3 (October, 1971). ------ . "Changes in the French Party System," in Comparative Politics, 4 (January, 1972). ------ . "Recruitment of Rival Party Activists in France: Party Cleavages and Cultural Differentiation," Journal of Politics, Vol. 35 #1 (February, 1973). Kesselman, Mark and Rosenthal, Donald B. "Local Power and Comparative Politics: Notes Toward the Study of Comparative Local Politics," in Rodney P. Stiefbold (ed.) Frontiers in Urban Research (University of Miami Press, forthcoming). Key, V. 0., Jr. Southern Politics (Vintage Books, 1949). ------ . Public_gpinion and American Democreey (Alfred A. Knopf, 1961). Kindleberger, Charles P. Economic Growth in France and Britain 1851—1950 (Harvard University Press, 1965). Kirchheimer, Otto. "The Waving of Opposition in Parliamentary Regimes," in Robert Dahl (ed.) Political Oppositions in Western Democracies (Yale University Press, 1966). Kriegel, Annie. Les Communistes Francais: Essai d'Ethnographie Politique, 2nd Ed. (Seuil, 1970). Klein, Lawrence R. An Introduction to Econometrics (Prentice-Hall, 1962). Lainville, Robert. Le Budget Communal (8th Ed., Sirev, 1959). Lambin, Paul. "Projet de Defense de 1'autonomie des communes rurales," in Departents et Communes (July-August, 1960). Lavau, Georges. Partis pplitiques et realites sociales (Paris, Colin, 1953). Le Colloc'h, Bernard. La Revolution Silencieuse: du Gaullisme au pouvoir (Didier, 1971). 226 Lijphart, Arendt. The Politics of Accommodation: Pluralism and Democracy in the Netherlands (University of California Press, 1963)- Lipset, Seymour M. and Rokkan, Stein (ed.). Party Systems and Voter Alignments (The Free Press, 1967). Lipset, Seymour M. "The Changing Class Structure and Contemporary European Politics," in Mattei Dogan and Richard Rose (eds.) European Politics (Little and Brown, 1971). Lorwin, Val. "Segmented Pluralism: Ideological Cleavages and Political Cohesion in the Small European Democracies," Comparative Politics (January, 1971). Loveridge, Ronald. City_Managers in Legislative Politcs (Bobbs-Merrill, 1971). Lyden, Fremont J. and Miller, Ernest G. (eds.). Planning, Programmipg, Budgeting (Harvard University Press, 1965). MacKae, Duncan. Parliaments, Parties and Society in France, 1946-58 (St. Martin's, 1967). Manheim, Karl. Ideology and Utopia, translated by Louis Wirth and Edward Shils (Harvest, 1936). Margolis, Julius. "The Demand for Urban Public Services," in Harvey S. Perloff and Lowdon Wingo, Jr. (eds.) Issues in Urban Economics (Johns Hopkins, 1968). Maspetiol, Roland and Laroque, Pierre. Lat Tutelle Administrative (Sirey, 1930). Matthews, Donald P. and Prothro, James W. Negroes and the New Southern Politics (Harcourt, Brace and World, 1966). McCloskey, Herbert, et. a1. "Issue and Fbllowers," American Political Science Review, 54 (June, 1969). McHale, Vincent. "Religion and Electoral Politics in France: Some Recent Observations," Canadian Journal of Political Science, 2 (September, 1969). Meynaud, Jean and Lancelot, Alain. Les Attitudes Politiques (Paris, Presses Universitaires, 1964). Micaud, Charles. The French Right and Nazi Germapy (Durbana, N.C., 1943). Miller, Warren and Stokes, Donald. "Constituency Influence in Congress," American Political Science Review, 57 (1963). NOonan, Lowell G. France: The Politics of Continuity and Change_(Holt, Rinehart and Winston). 227 Oliver, F. R. and Stanyer, J. "Some Social Aspects of the Financial Behavior of County Boroughs," Public Administration, 47 (Spring, 1969). Parsons, Talcott. Essays in Sociological Theory (Rev. Ed., Free Press, 1954). Peters, B. Guy. "Economic and Political Effects on the Development of Social Expenditures in France, Sweden, and the united Kingdom," Midwest Journal of Political Science (May, 1972). Pierce, John G. "Party Identification and the Changing Role of Ideology in American Politics," Midwest Journal of Political Science, 14 (February, 1970). Pinner, Frank. "On the Structure of Organizations and Beliefs: Lagerdenken in Austria," a paper delivered at the annual meeting of the American Political Science Association, Chicago, September 5—9, 1967. Powell, Bingham. Social Fregmentation and Political Hostility (Stanford University Press, 1971). Prewitt, Kenneth. The Recruitment of Political Leaders: A Study of Citizen—Politicians (Bobbs-Merrill, 1971). Putnam, Robert D. "Studying Elite Political Culture: The Case of Ideology," American Political Science Review, 65 (September, 1971). Quade, E. E. (ed.). Analysis for Militapy Decisions (Rand-MeNally, 1964). Ranney, Austin (ed.). Political Science and Public Poliey (Markham, 1968). Rémond, Rene, et. a1. Forces Religieuses et attitudesepolitiques dans la France contemporaine (Paris, Colin, 1965). Ridley, F. and Blondel, J. Public Administration in France (Barnes and Noble Company, 1969). Rokkan, Stein. Citizens, Elections and Parties (David McKay Company, 1970). Rose, Richard. "Class and Party Divisions: Britain as a Test Case," in Robert Dahl's (ed.) Political Oppositions in Western Democracies (Yale University Press, 1966). Rosenberg, Mbrris. The Legic of Survey Analyeis (Basic Books, 1968). Rosenthal, Howard. "The Electoral Politics of Gaullists in the Fourth French Republic: Ideology or Constituency Interest?" American Political Science Review, 63, #2 (June, 1969). 228 ------ . "Voting and Coalition Mbdels in Election Simulation," in William Coplin (ed.) Simulation and the Study of Politics (Markham, 1968). ------ . Commentary on Duncan MacRae, Jr.'s "Parliament, Parties, and Society in France, 1946-58," American Political Science Review, 63 (September, 1969). Runciman, W. C. Relative Deppivation and Social Justice (University of California Press, 1966). Russet, Bruce M. "Who Pays for Defense?" American Political Science Review (June, 1969). Sachs, Seymour; Harris, Robert; and Carroll, John T. The State and Local Government....The Role of State Aid, Comptroller's Studies in Local Finance, #3 (State Department of Audit and Control, 1963). Salisbury, Robert and Heinz, John. "A Theory of Policy Analysis and Some Preliminary Applications," in Ira Sharkansky (ed.) Policy Analysis in Political Science (Markham, 1970). Sartori, Giovanni. "Politics, Ideology and Belief Systems," American Political Science Review, 63 (June, 1969). Schmenner, Roger. "On the Determinants of Municipal Employee Wages," Review of Economics and Statistics (February, 1973). Schmidt, Charles. Le Maire de la Commune Rurale (Berger—Levrault, 1959). Selznick, Philip. The Organizational weapon (McGraw-Hill, 1952). Sharkansky, Ira. Spending in the American States (Rand-McNally, 1966). ----- . "Government Expenditures and Public Services in the American States, American Political Science Review (December, 1967). Sharkansky, Ira and Hofferbert, Richard. "Dimensions of State Politics, Economics and Public Policy," American Political Science Review, Vol. 63 (September, 1969). Sharkansky, Ira. The Routines of Politics (Van Nostrand, 1970). ------ . The Politics of Taxing and Spending’(Bobbs-Merrill, 1970). ------ . "Economic Theories of Public Policy: Resource-Policy and Need- Policy Linkages Between Income and Welfare Benefits," Midwest Journal of Political Science (June, 1972). Singer, Jean. Le Maire et Ses Pouvoirs de Police (Berger-Levrault, 1959). Smith, H. and Draper, N. R. Applied Regression Analysis, Ch. 5 (New York, 1966). 229 Stiefbold, Rodney P. "Segmented Pluralism and Consociational Democracy in Austria, Problems of Stability and Change," in Martin O. Heisler et. al. (eds.), Politics in Europe, Structure and Processes (David McKay, 1973). Stinchcombe, Arthur L. .Constructing_Social Theories (Harcourt, Brace and WOrld, 1967). Sullivan, John L. "A Note on Redistributive Politics," American Political Science Review, 66 (December, 1972). Swartzenberger, Roger-Gerard. La Guerre de Succession (Presses Universitaires, 1973). Tarrow, Sidney G. "Political Dualism and Italian Communism," American Political Science Review, 61 (March, 1967). ------ . Peasant Communism in Southern Italy (Yale University Press, ------ . "The Urban-Rural Cleavage in Political Involvement: The Case of France," American Political Science Review, Vol. 65, #2 (June, 1971). Thibaud, Paul. "Le Communisme Municipal," Esprit, 34 (October, 1966). Truman, David. The Governmental Process (Knopf, 1951). Tufte, Edward R. and Fbrbes, Hugh Donald. "A Note of Caution in Causal MOdeling," American Political Science Review, 62 (December, 1968). ------ . "Improving Data Analysis in Political Science," World Politics 21 (July, 1969). Urwin, Derek and Rose, Richard. "Social Cohesion and Political Strains in Regimes," Comparative Political Studies, 2 #1 (April, 1969). Walsh, Annmarie Hauck. Urban Government for the Paris Region (Preger, 1968). Wantzenriether, Pierre. "Le Maire Rural et Son Sous-Prefet," Etudes des Problemes Municipaux, #6 (June, 1965). Waterman, Harvey. Political Change in Contemporatprrance (Charles E. Merrill Publishing Company, 1969). Waxman, Chaim I. (ed.). The End of Ideology Debate (Funk and Wagnalls, 1968). Williams, Philip. "The French Presidential Elections of 1965," Parliamentary Affairs, XIX:l (1965—66). Wold, H. and Jureen, L. _Demand Analysis (Wiley and Sons, 1953). Wright, Gordon. "Four Red Villages in France," Yale Review, L (March, 1952). ------ . Rural Revolution in France (Stanford University Press, 1964). 230 Wylie, Lawrence. Village in the Vaucluse (Harper and Row, 1964). ----- -. Chanzeaux: A Village in Anjou (Harvard University Press, 1966). Zetteberg, Hans L. On Theory and Verification in Sociology (Bedminster Press, 1965). 3 4 B 6 6 9 1 3 0 3 9 2 1 3 # I1 I“ H“ H H H I" “ E "in A ll 1|" I“ u Illl