POLITICAL STRUCTURE CENTRALIZATION AND SOCIAL POLICY: A CROSS-NATIONAL STUDY By Mary Vanderlaan Meyering A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1979 ABSTRACT POLITICAL STRUCTURE CENTRALIZATION AND SOCIAL POLICY: A CROSS-NATIONAL STUDY By Mary Vanderlaan Meyering This thesis tests hypotheses concerning the impact of political and economic variables on social policy. The research investigates the role of political structure centralization and economic development in affecting health, education and social welfare policies. Political structure centralization is conceptualized utilizing three dimensions: the constitutional status (federal or unitary) stipulating the number of sovereign decision-making arenas, the degree of central government control over taxation and spending -- fiscal centralization —- and the degree of informal power sharing among groups (such as political parties and interest groups) in society. Earlier studies have suggested that centralized decision-making structures lead to social policy development due to standardization of procedures. efficiency in implementation and uniformity in efforts, although there is debate whether the impact of federalism on policy is conservative or expansionary. The literature fails to differentiate among various aspects of centralization, however. Modernization theorists and others suggest economic development leads directly to advances in social policy. A generally accepted hypothesis is that economic development has an independent positive effect on policy expenditures and policy outcomes, or, Short term program results. The empirical research based upon a cross-national sample of Page 2 Mary Vanderlaan Meyering sixty-seven countries shows that federal/unitary status has no impact on such indicators as health expenditures, life expectancy, nurse availability or nutrition levels. Nor does it affect education outcomes such as gross enrollment ratios and the percent of the population en- rolled in higher education. Formal legal status has no empirical relationship to the social welfare policy area, but it does interact with economic level to affect education expenditures: unitary structures constrain the otherwise positive impact of economic development on education spending. Contrary to hypotheses in the literature, the research shows that in most cases where fiscal centralization affects policy, the impact is negative. Fiscal centralization has a direct negative effect on health and social welfare expenditures and on the size of the public sector. It has no effect, however, on health or social welfare outcomes, educa- tion expenditures or gross enrollment ratios. The effect of fiscal centralization on enrollments in higher education is dependent upon economic level. Only in underdeveloped nations does fiscal centraliza- tion have a positive effect on enrollments. The impact of informal politicalgauthority concentration (measured by levels of interest group activity and the percent of legislative seats held by the majority party) is not easily generalized: the effects of partisan dominance and of interest articulation change depending upon the policy output or outcomes being considered and the level of economic development. Moreover, in most cases where partisan dominance effects social policy, it does so in interaction with economic level. Partisan dominance and economic level constrain each others effects. For example, whereas partisan dominance has a positive impact Page 3 Mary Vanderlaan Meyering on health expenditures in underdevelOped nations, at medium and high levels of development, it has a negative impact. While limiting interest articulation has a direct positive effect on health expenditures, the impact of interest articulation on gross enrollment ratios and relative income shares of the poorest forty percent of households changes with economic level. Finally, this research demonstrates that political structure variables constrain or moderate the impact of economic develOpment on SOCial policy. In a majority of the policy situations, the effects of economic level are mediated by the political context. Thus, the research demonstrates that the structure of decision- making has implications for policy and that economic develOpment is an important contextual variable within which decision-makers must operate. Yet, the structure of government in turn affects how economic resources are translated into social development. The research, moreover, warns against unilateral statements about the effects of structure and economic level on policy. For those national leaders and policy makers who struggle to extend the grasp and horizon of every citizen, and, who consider the pur- suit of social equality a legitimate politi- cal issue. iii ACKNOWLEDGEMENTS A number of scholars and my extended family have made this work possible. I am deeply indebted to Brian D. Silver who freely gave of his time to guide me through every step of the research process, from the early conceptualizations of ideas to the final analyses. Among other things, Brian read more drafts of the chapters than, I'm sure, he cares to recount. Mbst importantly, he constantly reminded me that social science research must always take its impetus and direction from theory. The important contribution Brian made to this work by way of comments, advice, dialogues and criticisms is greatly appreciated and humbly acknowledged. Lastly, I must give credit in another important area: at several crucial junctures, Brian's spirited energy spilled over to refuel my own dwindling energy supplies. The published work of Robert W. Jackman played an important role in helping me formulate the initial ideas and questions that prompted this research. I want to thank Bob especially for his comments on early drafts of the work and particularly his suggestions on how best to utilize the income and social security data that are available. His comments led me, at several points, to consider alternative explanations and research strategies. 0. Charles Press shared with me his thoughts on federalism and its implications for intergovernmental relations. He encouraged me to think of federal/unitary status as a continuum of political form rather than iv as a polar dichotemy. Several graduate colleagues took time to discuss ideas with me. I especially want to thank Ruth Hoogland DeHoog, Gary L. Crawley and J. Pat Willerton for their thoughth1 suggestions. My extended family encouraged my efforts. And finally, my husband, Al, who often became the victim of my bad moods and who suffered through my periods of doubt and uncertainty, never failed to provide emotional support in large amounts. I cannot forget, either, the long hours Al spent helping me code and transcribe data. His unconditional support saw me through this project. TABLE OF CONTENTS Page LI ST OF TABIJES I I I I I I I I I I I I I I I I I I I I I I I I I I Xi LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . xvii LIST OF SYMBOLS . . . . . . . . . . . . . . . . . . . . . . . . .xviii CHAPTER I: PERSPECTIVE: POLITICS AND POLICY . . . . . . . . . . 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . 1 The Genesis of the Question . . . . . . . . . . . . . . . . . 2 The Impact of Political Structures on Social Policy Outputs and outcomes I I I I I I I I I I I I I I I I I I I I I I I 1 1 The Question of Distribution, . . . . . . . . . . . . . . . . 14 Education POIicy "" A Note a o o o o o o o o o o o o o o o o 16 The Impact of Various Forms of Political Structure Central- 1 mtion I I I I I I I I I I I I I I I I I I I I I I I I I 2 1 Government Centralization in the Literature . . . . . . . . . 2# Fiscal Centralization . . . . . . . . . . . . . . . . . . . . 26 Federal/UnitaryStatus..,................. 3o Federalism's Impact on Policy . . . . . . . . . . . . . . . . 35 Political Centralization . . . . . . . . . . . . . . . . . . 38 The Special Case of the Heterogeneous Society . . . . . . . . 40 Other Factors Affecting Social Policy . . . . . . . . . . . . 47 Footnotes . . . . . . . . . . . . . . . . . . . . . . . . . . 50 CHAPTER II: THE RESEARCH: HYPOTHESES, METHODS, DESIGN AND DATA . 58 Cultural Heterogeneity . . . . . . . . . . . . . . . . . . . 63 P011tical Centmlization o o o o o n o I o I o o o o I o o o 65 CHAPTER II: (Continued) Data, DesigryandMethods . . . . . . . . . . . . Measuring Government Centralization -- The Independent Variables . . . . . . . . . . . . . . . . . . Fiscal Centralization . . . . . . . . . . . . . . Constitutional Status -- Federal/Unitary . . . . Informal Political Authority Concentration . . . Commitment to Social Policy and Policy Outcomes . Education . . . . . . . . . . . . . . . . . . . . Health . . . . . . . . . . . . . . . . . . . . . SOCial welfare I I I I I I I I I I I I I I I I I Other Variables Important to the Study Social Heterogeneity . . . . . . . . . . . . . . Economic Data . . . . . . . . . . . . . . . . . . Demographic Variables . . . . . . . . . . . . . . The Sample and the Level and Methods of Study . . The Sample . . . . . . . . . . . . . . . . . . . The Problem of Missing Data . . . . . . . . . . . The Level of Study . . . . . . . . . . . . . . . Footnotes . . . . . . . . . . . . . . . . . . . . CHAPTER III: POLITICAL STRUCTURE CENTRALIZATION: ITS MEASUREWT I I I I I I I I I I I I I I SCOPE AND Federal/Unitary Status: Formal-Legal Structure Centrali- “tion I I I I I I I I I I I I I I I I I I I I FiscalCentI‘alizationo. o o o- n. o. o o o o Informal Political Authority Concentration . . . The Three Dimensions of Political Structure Centralization Footnot e S I I I I I I I I I I I I I I I I I I I I Appendix —- Some Summary Statistics . . . . . . . vii Page 66 67 67 69 70 74 75 76 77 80 80 81 82 82 82 84 85 87 9O 9O 92 99 108 111 113 Page CHAPTER IV: THE IMPACT OF GOVERNMENT CENTRALIZATION ON HEALTH POLICY I I I I I I I I I I I I I I I I I I I I I I I 1 124' Centralization and the Size of Public Spending . . . . . . . 114 Indicators of Health Policy Outputs and Outcomes . . . . . . 126 Government Centralization and Health Policy . . . . . . . . . 129 Fisca]. cerltmlimtion I I I I I I I I I I I I I I I I I I I I 1 30 Informal Political Authority Concentration . . . . . . . . . 135 The Role of Demographic Variables . . . . . . . . . . . . . . 139 The Role of Economic Development . . . . . . . . . . . . . . 140 The Joint Occurrence of Types of Centralization: NMlti— variate Analyses . . . . . . . . . . . . . . . . . . . . . 143 Other Interaction Effects . . . . . . . . . . . . . . . . . . 145 Specifying A Final Model . . . . . . . . . . . . . . . . . . 150 Outcomes of Health Policy: The Life-Nutrition—Medical Services Dimensions . . . . . . . . . . . . . . . . . . . 158 A Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 176 Footnotes . . . . . . . . . . . . . . . . . . . . . . . . . . 178 Appendix A: Summary Characteristics of the Dependent and Independent variables I I I I I I I I I I I I I I I I I I 18 1 Appendix B: Regress of HEXPOP on Centralization Measures, Int emction MOdel s I I I I I I I I I I I I I I I I I I I I 182 Appendix C: Regression Analyses: Factors Affecting Health Outcomes I I I I I I I I I I I I I I I I I I I I I I I I I 184 CHAPTER V: THE IMPACT OF GOVERNMENT CENTRALIZATION ON EDUCATION POLICY I I I I I I I I I I I I I I I I I I I I I I I I 1 94 Bivariate Analyses -- Determinants of Education Expendi- tures I I I I I I I I I I I I I I I I I I I I I I I I I I 197 Multivariate Analyses -- Interaction Effects . . . . . . . . 208 Multivariate Analyses -- Specifying a More Complete Model . . 212 The Impact of Political Structure Centralization on Educa- tion outcomes I I I I I I I I I I I I I I I I I I I I I I 216 viii CHAPTER V: (Continued) Page Multivariate Analyses -- Specifying a MOre Complete Model . . 225 A SW I I I I I I I I I I I I I I I I I I I I I I I I I I 236 Footnotes I I I I I I I I I I I I I I I I I I I I I I I I I I 2% Appendix A: Summary Characteristics of the Variables . . . . 239 Appendix B: Education Outcomes and Centralization and Economic Variables, Interaction MOdels . . . . . . . . . . 240 CHAPTER VI: POLITICAL STRUCTURE CENTRALIZATION AND SOCIAL "MARE POLICY I I I I I I I I I I I I I I I I I I I 250 Indicators of Social Security Policy Outputs and Outcomes . . 251 The Impact of Political Structure Centralization on Social Security Expenditures . . . . . . . . . . . . . . . . . . 255 Multivariate Analyses -- Interaction Relationships . . . . . 264 Multivariate Analyses -- Specifying a More Complete Model . . 270 Political Structure Centralization and Income Distribution . 275 Toward An Explanation of the Incomes Share of the Poorest . . 275 Multivariate Analyses -- Specifying a More Complete Model . . 285 The Role of SSEXPOP . . . . . . . . . . . . . . . . . . . . . 290 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 291 Footnotes . . . . . . . . . . . . . . . . . . . . . . . . . . 293 Appendix: Summary Characteristics of the Social Welfare axld Other Variables I I I I I I I I I I I I I I I I I I I 294 CHAPTER VII: POLITICAL STRUCTURE CENTRALIZATION AND SOCIAL WELL- BEING: COMMENTS AND CONCLUSIONS . . . . . . . . . . 295 Centralization and Social Policy . . . . . . . . . . . . . . 296 Federal/UnitaryStatus...................296 Fiscal Centralization . . . . . . . . . . . . . . . . . . . . 297 Informal Political Authority Concentration . . . . . . . . . 298 The Role of Economic Level . . . . . . . . . . . . . . . . . 299 ix CHAPTER VII: (Continued) Page Conclusions: Politics, Social Policy and Social Well- Being I I I I I I I I I I I I I I I I I I I I I I I I I I 303 The Research and Previous Literature . . . . . . . . . . . . 305 Footnotes I I I I I I I I I I I I I I I I I I I I I I I I I I 311 BIBLIOGRAM I I I I I I I I I I I I I I I I I I I I I I I I I I I 312 Table 3-1 3-2 3-3 3.4 3—5 3-6 3-7 3-8 3—9 4—1 4-2 4-3 4-4 u—s 4-6 LI~7 4-9 4-10 4-11 LIST OF TABLES Parson ' S r - Fiscal mta I I I I I I I I I I I I I I I Pearson's r - Fiscal Data and FCI . . . . . . . . . . . Associations Among the Informal Political Authority Con— centration Indicators - tau . . . . . . . . . . . . . Associations Among the Informal Political Authority Con- centration Indicators - gamma . . . . . . . . . . Correlations Among Indicators of Formal Structure - tau . Gamma - Indicators of Informal Political Authority . Tau - Indicators of Informal Political Authority . . Pearson's r - Correlations Among All Political Structure Centralization Indicators . . . . . . . . . . . . Linear Regression of Size of Public Sector on FU Status Linear Regression of Size of Public Sector on FCI . . Logarithmic (Curvilinear) Regression of SPS on FCI . Polynomial (Curvilinear) Regression of SPS on FCI . . Regression of SPS on FCI, Controlling for EPC . . . . Linear.Regression of SP8 on Seats Majority Party . . Logarithmic Regression of SPS on SMP . . . . . . . . Polynomial Regression of SPS on SMP . . . . . . . . . Regression of Size of Public Sector on SMP, Controlling EPC I I I I I I I I I I I I I I I I I I I I I I I Regression of SPS on Interest Articulation . . . . . Regression of SPS on IA, Controlling EPC . . . . . . xi Page 96 98 100 101 102 103 103 104 105 116 118 118 119 120 121 122 122 123 124 125 Table Page 4—12 Pearson's Correlation Coefficients Among the Health POIicy Indicators I I I I I I I I I I I I I I I I I I I 127 4-13 Regression of'HEXPOP on Federal/Unitary Status . . . . . . 131 4-14 Linear Regression of HEXPOP on FCI . . . . . . . . . . . . 132 4-15 Logarithmic Regression of HEXPOP on FCI . . . . . . . . . . 132 4-16 Polynomial Regression of HEXPOP on FCI . . . . . . . . . . 133 4-17 Linear Regression of HEXPOP on SMP . . . . . . . . . . . . 136 4-18 Logarithmic Regression of HEXPOP on SMP . . . . . . . . . . 136 4—19 Polynomial Regression of HEXPOP on SMP . . . . . . . . . . 137 4-20 Regression of HEXPOP on Interest Articulation . . . . . . . 137 4-21 Linear Regression of HEXPOP on EPC . . . . . . . . . . . . 140 4-22 Logarithmic Regression of HEXPOP on EPC . . . . . . . . . . 141 4-23 Polynomial Regression of HEXPOP on EPC . . . . . . . . . . 143 4-24 Regression of HEXPOP on Political Structure Centralization and EPC, Interaction Models . . . . . . . . . . . . . . 147 4-25 Regression of HEXPOP on Federal/Unitary Status and Hetero- geneity, Interaction MOdel . . . . . . . . . . . . . . . 150 4-26 Regression of HEXPOP on the Independent Variables . . . . . 152 4-27 The Impact of EPC on HEXPOP: Expected Values of HEXPOP . . 154 4—28 The Impact of FCI at Various Levels of EPC: Expected ValuesomePOPooo..o...ao.........155 4-29 The Impact of SMP at Various Levels of EPC: Expected values Of HEXPOP I I I I I I I I I I I I I I I I I I I I 156 4—30 Bivariate Relationships Between the Independent Variables and Life Expectancy, 1970 . . . . . . . . . . . . . . . 159 4-31 Bivariate Relationships Between the Independent Variables and the Ratio of Population to Nurses . . . . . . . . . 161 4—32 Bivariate Relationships Between the Independent Variables and Calories Supply, 1970 . . . . . . . . . . . . . . . 163 4-33 Regression of Life Expectancy on the Main Independent vanabl es I I I I I I I I I I I I I I I I I I I I I I I 1 $ xii Table 4-34 4-35 4-36 4-37 4-38 4-39 4-40 441 4—42 4-43 Regression of Life Expectancy on EPC and SMP . . . . . . Regression of Life Expectancy on EPC, SMP and HEXPOP . . Regression of P0pulation/Nurse on the Independent Vari- ab]. 6 s I I I I I I I I I I I I I I I I I I I I I I Regression of Nurses on the Logged EPC Variable . . . . . Regression of Nurses on ln EPC and HEXPOP . . . . . . . . Regression of Calorie Supply on the Independent Variables Regression of the Age Variable on EPC . . . . . . . . . . Regression of Calorie Supply on In EPC and the Age vanabl e I I I I I I I I I I I I I I I I I I I I I I I Regression of Calorie Supply on the Age Variable . . . . Regression of Calorie Supply on the Age Variable and HmoP I I I I I I I I I I I I I I I I I I I I I I I I Regression of HEXPOP on Centralization Measures, Inter- action MOdelS "" Appendix B o o o o o o c o o o o o 0 Regression of Life Expectancy on EPC and Centralization Indicators, Interaction Models -- Appendix C . . . . . Regression of Population per Nurse on EPC and Centraliza— tion Indicators, Interaction Models —- Appendix C . . Regression of Calorie Supply on EPC and Centralization Indicators, Interaction Models -- Appendix C . . . . . Regression of Life Expectancy on Centralization Indi- Regression of Population per Nurse on Centralization Indicators, Interaction Medels -- Appendix C . . . . . Regression of Calorie Supply on Centralization Indi- Pearson's r Correlation Coefficients Among Some of the Education Policy Indicators . . . . . . . . . . . . . Regression of EDEXPOP on Constitutional Status, 1970 . . Linear Regression of EDEXPOP on FCI, 1970 . . . . . . . . Logarithmic Regression of EDEXPOP on FCI . . . . . . . . xiii Page 169 170 171 172 172 174 175 175 176 177 182 184 186 187 188 190 192 195 198 200 200 Table 5-5 5-6 5-7 5-8 5-9 5-10 5—11 5-12 5—13 5-14 5-15 5—16 5-17 5-18 5-19 5-20 5-21 5-22 5-23 5-24 5-25 5-26 Polynomial Regression of EDEXPOP on FCI . . . . . . . . . Linear Regression of EDEXPOP on SMP . . . . . . . . . . . Logarithmic Regression of EDEXPOP on SMP . . . . . . . . Polynomial Regression of EDEXPOP on SMP . . . . . . . . . Regression of EDEXPOP on Interest Articulation . . . . . Linear Regression of EDEXPOP on EPC, 1970 . . . . . . . . Logarithmic Regression of EDEXPOP on EPC . . . . . . . . Polynomial Regression of EDEXPOP on EPC . . . . . . . . . Regression of EDEXPOP on EPC and Centralization, Inter- action Medels . Regression of EDEXPOP on Centralization Variables, Interaction Models . . . . . . . . . Regression of EDEXPOP on All Surviving Variables . . . Regression of EDEXPOP on EPC and FU * EPC . . . . . . . . The Impact of EPC on EDEXPOP in Federal and Unitary Nations: Expected Values of EDEXPOP . . . . . . . . . Bivariate Relationships Between the Independent Variables and GER . Bivariate Relationships Between the Independent Variables and SHE . . . Regression of GER on EPC and Centralization Indicators, Interaction Models -- Appendix . . . . . . . . . . . . Regression of GER on Centralization Measures, Interaction M0d915’-AppendiXo o o o I o I o o 0 Regression of SHE on EPC and Centralization Indicators, InteraC'tion Medds -- Appendix 0 o o o o o o o o o o 0 Regression of SHE on Centralization Indicators, Inter- action Mbdels -- Appendix Regression of GER on All Surviving Significant Variables Regression of GER on Heterogeneity, 1n EPC * IA and IA . Regression of GER on Final Surviving Variables and EDEXPOP. xiv Page 202 202 203 203 204 205 206 206 210 211 213 214 215 218 221 240 242 245 247 226 227 228 Table 5—27 5-28 5-29 5-30 5-31 6-1 6-2 6—3 6-5 6-6 6-7 6—8 6-9 6-10 6-11 6-12 6-13 6-14 6-15 6—16 Page The Impact of Interest Articulation and Extreme Ethnic Heterogeneity on GER at Various Levels of EPC, Adjusting for EmEXPOP I I I I I I I I I I I I I I I I I 230 Regression of SHE on All Surviving Significant Variables . 231 Regression of SHE on the Significant Variables From Tab]. 6 4-28 I I I I I I I I I I I I I I I I I I I I I I I 2 32 Regression of SHE on Final Surviving Variables and EDEXPOP. 233 The Impact of FCI at Various Levels of EPC: Expected vanes Of SHE I I I I I I I I I I I I I I I I I I I I I 2 314’ The Impact of SMP at Various Levels of EPC: Expected Values of SHE . . . . . . . . . . . . . . . . . . . . . 235 Pearson's Correlation Coefficients Among Some of the Education Policy Indicators . . . . . . . . . . . . . . 253 Regression of SSEXPOP on Federal/Unitary Status . . . . . . 256 Linear Regression of SSEXPOP on FCI . . . . . . . . . . . . 257 Logarithmic Regression of SSEXPOP on FCI . . . . . . . . . 257 Polynomial Regression of SSEXPOP on FCI . . . . . . . . . . 258 Linear Regression of SSEXPOP on SMP . . . . . . . . . . . . 259 Legarithmic Regression of SSEXPOP on SMP . . . . . . . . . 259 Polynomial Regression of SSEXPOP on SMP . . . . . . . . . . 260 Regression of SSEXPOP on Interest Articulation . . . . . . 260 Linear Regression of SSEXPOP on EPC . . . . . . . . . . . . 263 Logarithmic Regression of SSEXPOP on EPC . . . . . . . . . 263 Polynomial Regression of SSEXPOP on EPC . . . . . . . . . . 264 Regression of SSEXPOP on EPC and Centralization Measures, Int emotion MOde]. S I I I I I I I I I I I I I I I I I I I 2 65 Regression of SSEXPOP on Centralization Measures, I nt emction MOd 81 s I I I I I I I I I I I I I I I I I I I 2 67 Regression of SSEXPOP on All Surviving Independent van-ab]. es I I I I I I I I I I I I I I I I I I I I I I I 2 71 Regression of SSEXPOP on Final Survivors . . . . . . . . . 273 XV Table 6-17 6-18 6-19 6-20 6-21 6-22 6-23 6-24 6-25 6-26 6-27 The Impact of FCI on SSEXPOP at Various Levels of EPC: Expected Values of SSEXPOP . . . . . . . . . . . . . . . The Impact of SMP on SSEXPOP at Various Levels of EPC: Emected values Of SSEXPOP I I I I I I I I I I I I I I Bivariate Relationships Between LOW INC and the Independent vmabl es I I I I I I I I I I I I I I I I I I I I I I I Regression of LOW INC on EPC and the Centralization Measures, Interaction Models . . . . . . . . . . . . . . Regression of LOW INC on Centralization Measures, Inter- action MOd-el S I I I I I I I I I I I I I I I I I I I I I Regression of LOW INC on All Surviving Variables . . . . . Regression of LOW INC on Survivors of Table 6-22 . . . . Regression of LOW INC on EPC * IA and SMP * EPC . . . . . . Regression of LOW INC on EPC * IA, SMP * EPC and SSEXPOP . The Impact of IA at Various Levels of EPC: Expected Vflues 0f LOW INC I I I I I I I I I I I I I I I I I I I The Impact of SMP at Various Levels of EPC: Expected vanes Of Low INC I I I I I I I I I I I I I I I I I I I Page 273 274 277 280 282 286 287 287 288 289 289 LIST OF FIGURES Figure 4-1 4-2 4—3 4-4 “—5 4—6 4—7 Ingafitmic Relation Ship I I I I I I I I I I I I I I I I Polynomial Relationship . . . . . . . . . . . . . . . . Graph: Health Expenditures Per Capita, 1970 on Fiscal Centralization Score, 1970 . . . . . . . . . . . . . Graph: Health Expenditures Per Capita, 1970 on Seats Mjonty Party ’ 1970 I I I I I I I I I I I I I I I I Graph: Health Expenditures Per Capita, 1970 on Electric Power Consumption Per Capita, 1970 . . . . . . . . . Graph: Life Expectancy, 1970 on Electric Power Consump- tion Per Capita,, 1970 o I o I o o o I o o o o o o o 0 Graph: Calories Per Cap as % Daily Requirement, 1970 on Electric Power Consumption Per Capita, 1970 . . . . . Graph: Education Expenditures Per Capita, 1970 on Fiscal cthization Score, 1970 o o o o o o o o I o o o 0 Graph: Education Expenditures Per Capita, 1970 on Electric Power Consumption, Per Capita, 1970 . . . . Graph: Percent Population in Secondary and Higher Educa- tion on Electric Power Consumption Per Capita, 1970 . . Graph: Social Security Expenditures Per Capita, 1970 on Electric Power Consumption Per Capita, 1970 . . . . . . xvii 117 117 134 138 142 165 166 201 207 224 262 FU VPD PS LFV LFS IA EC SUH SMP PDP CE CC FCI SPS HEXPOP IMR LE DRS LIST OF ABBREVIATIONS Federal/Unitary Status Vertical Power Distribution Party System, Number of Parties Legislative Party Fractionalization in Votes Received Legislative Party Fractionalization in Seat Share Interest Articulation by Associational Groups Electoral Competition Selection of Upper Legislative House Percent Legislative Seats Held by Majority Party Power of Executive to Dissolve Parliament Central GoVernment Share of General Government Current Public Expenditures Share of Total Public Revenues Collected by the Central Government Central Government Share in General Public Consumption Expenditures Central Government Share in General Government Civil Con- sumption Expenditures Fiscal Centralization Index Size of the Public Sector Health Expenditures Per Capita Infant Mortality Rate Life Expectancy Population Per Doctor xviii NRS Population Per Nurse BDS Population Per Hospital Bed CAL Calorie Supply as Percent of Daily Requirement PRT Protein Supply Per Capita Per Diem EDEXPOP Education Expenditures Per Capita GER Gross Enrollment Ratios SHE Enrollment in Secondary and Higher Education 31 Number Students at Primary and First Levels of Education 82 Number Students at Second Level S3 Number Students at Third Level T1 Number Teachers Primary and First Levels of Education T2 Number’Teachers Second Level T3 Number'Teachers Third Level YSP Years Schooling Provided SSEXPOP Social Security Expenditures Per Capita LOW INC Income Share, Lowest 40% of Households MID INC Income Share of the Middle 40% of Households TOP INC Income Share of the Top 20% of Households POP 50 Millions of PopUlation with Per Cap Income Below $50, U.S. POP 75 Millions of P0pulation with Per Cap Income Below $75, U.S. % 50 % POPUlation with Per Cap Income Below $50 7% 75 35 Population with Per Cap Income Below $75 EPC Electric Power Consumption CHAPTER I PERSPECTIVE: POLITICS AND POLICY Introduction A debate goes on among students of the Third World over the appli- cability of traditional (Western) economic policy to the Third World setting and specifically, the usefulness of the conceptual separation between notions of optimum growth and distribution policies. The grow— ing °2£§§9§E§l29926w§9291ars is that aggregate growth as a social objec- tive has been deficient; increased economic growth and higher gross national products have not led to hi588?,}3Y8;$ of living or to the eradication of poverty, unemployment and concentrations of wealth in the Third World. Instead, it is argued, nations must redirect growth policies, outline redistributive goals and clearly state social objec- tives in an effort to ultimately provide qualitative along with quanti- tative growth. This debate over optimum growth versus redistribution also goes on (but in slightly different terms) among scholars who focus on the welfare state in post-industrial societies. Policy makers face realities of poverty and malnutrition amidst affluence. * * * This thesis attempts to shed some light on one issue area involved in this debate -- it investigates variations among nations in the effectiveness of national social policies (in areas of education, health, and welfare) in affecting levels of living and redistribution. Specifically, it examines the relationship between political structure centralization and social policy commitment and outcomes. While the research does not focus explicitly on economic growth policies as opposed to redistributive policies, it will allow initial answers to the questions "Have growth policies been stressed at the ex- pense of social equality and the redistribution of highly concentrated wealth?" 0r, "Has economic growth and deveIOpment in turn affected social development?" Moreover, and very importantly, the research focuses on the role of government structure centralization and political factors in affecting social equality or qualitative growth. It has been this question -- on the role of political factors in shaping policy outputs and outcomes -- that has sparked a whole new literature in political science. The Genesi§;of the Question A major feature of political science literature in the last decade and a half has been its emphasis on the study of what governments do as evidenced by discussions of governmental performance, capacity, resiliency, stability, outputs and institutionalization. These dis- cussions, which ultimately have emphasized the primacy of politics in affecting open, participatory societies, moreover, provided political science its raison d' etre. Almond and Powell, for example, have developed a paradigm around such concepts as system structures (obser- vable activities which make up the political unit), functions, capabil- ities, inputs and outputs.1 They suggest that political systems can be compared on "performance," or, how well various systems use their resources to meet the demands of their citizens. Similarly, David Easton has forwarded a systems model, arguing that all political "systems" (or sets of related elements) involve inputs, conversion structures and outputs and that political systems can be clearly dif- ferentiated from social or economic systems for purposes of analysis.2 In another vein, Samuel Huntington argues that political institutional- ization and political organizations and procedures are necessary for the establishment of viable nation states, that is, states that are stable and ozderly.3 To Huntington, the primacy of politics, including the importance of establishing "legitimate" authority so that orderly change can occur, is unquestionable. Huntington praises Leninist Communism for its emphasis on organization and its seeming ability to create political institutions and authority after revolutionary periods.“ Finally, other thinkers and scholars, among them V. 0. Key and Duane Lockard, have held that political competition and participation ensures an equitable distribution of power in society and results in a more equit— able distribution of goods among citizens, benefiting the lower strata. The implication of this literature, then, is that government structure and character colors and determines policy outcomes, affects the distri- bution of goods and opportunities, and ultimately, affects the stability and well-being of the larger society. This emphasis in the literature on governments' performance and political outputs, however, seemed somewhat ironic when, in the late 1960s and early 1970s, the findings of empirical studies on determinants of policy and program expenditure levels -- studies growing out of this earlier literature -- began to appear. The consensus of this more recent literature focusing on the American states is that politics and political structures have little or no impact on policy. In separate 'WM“—-*p“‘ "i-w-' , w..- ..rau—i.-.-.. r. .u e~- ~---v- articles, Dawson and Robinson, Dye, and Hofferbert and Sharkansky report results that show no significant independent relationships among politi- cal factors (such as party competitiveness in state legislatures and governorships, tax structures, participation and voting turnout, apportionment, divided control of state legislatures, and the character of state legislative, executive and judicial branches) and policy out- put.5 The authors did find, however, that the wealth of the state or its level of economic development and levels of education attainment consistently and independently related to what policies are adopted and how much money is allocated to resulting social programs. Maxwell and Aaronson report, for example, that with some exceptions, "For every ten percent increase in a state's per capita income, state-local expenditures increase on the average, by approximately six percent."6 Quite naturally, these findings have increased the interest in a study of governments' results or policy and political performance. They encourage political scientists to take a careful look at the impact of various political arrangements, to ask further questions about the policy process and to develop new measurement techniques. They contri- bute also to the initiation of a debate that goes on over the pre- eminence of political influence over policy making, adOption and imple- mentation. The "new" literature stemming from this debate includes cross-national studies/6s well as further studies concentrating on the American states. And findings of the literature have sometimes supported and sometimes challenged the initial American states studies. Some of the literature in the field of comparative politics and several studies focusing on other nations have supported earlier American states findings: political structure and the nature of the regime have little to do with types of public policy adopted and ultimately with conditions within society. (The study extended beyond concern over*policy expenditure levels.) Frederic Pryor, David Lane and others argue that political systems in general are converging to new "mixed" forms and that present differences between systems and politi- cal structure in general have little effect on the nature of society. Pryor finds that regime type, whether it be communist or capitalist, playsvlittlemrole in explaining public consumption expenditures.7 Lane argues that certain social forces and institutions associated with economic and socialchange or endemic to industrial structure, such as division of labor in the economic sphere (where some workers get creative tasks and others get dull tasks), have similar effects wherever they are present, whether in Western - capitalist states or in state - socialist societies. Lane finds more similarities than differences between capitalist and socialist societies in areas such as access to education and occupational prestige, and he argues that there is in- equality in both system types in control over wealth, political power and income and status.8 Other scholars argue that the level of economic development of a system, not the type of political structures present, determines the quality of life a society enjoys. Phillips Cutright found in separate works (1965, 1967) that economic development leads to greater equality in the distribution of worker income among industrial sectors and that the level of economic development is more important than regime type (in this case, democracy) in determining policy activities in social 9 welfare areas. Likewise, Robert Jackman found that economic develop- ment relates to greater policy effort in the areas of social and economic equality, whereas the degree of regime democraticness adds nothing to this relationship.10 The findings of these studies, however, including those on the American states, do not stand unchallenged and the debate over the pri- macy of political factors and their influence on policy and expenditure levels (and ultimately on the quality of life in general) continues. Several studies have found pQIitical factors to be important determinants of.American state policies. Edward Jennings' 1978 study of eight states\ suggests that in states whose electoral systems and partisan alignments 3’ are organized along lines of class cleavage, welfare policies will be t 2 more generous and welfare effort will be greater.{£ Strouse and Jones1 [/a M] and Sack anduarris13 show that federal aid and intergovernmental trans- fers -- political variables in the sense that they are elements of political control and the products of multi-level decision-making -- are increasingly important in determining amounts spent on various social policy programs in the states. And, Jack Walker demonstrates that apportionment practices (whether urban areas are granted full representation in the legislature) affect the acceptance of new policy ideas and innovative programs.14 Moreover, Cnudde and McCrone argue that the pre-eminence of political factors varies, depending on the issue at hand. They suggest that political factors such as degree of political competition, the party system and the openness of the system are important relevant variables in issues close to the have - have-not struggle.15 Likewise, Fry and Winters argue from their 1970 study that political factors (political participation, Democratic vote, interparty competition, apportionment, legislative party cohesion, and others) are more important than socio-economic factors when benefit and burden allocation patterns (and overall redistribution) rather than levels of 7 public expenditures are studied.16 Similarly, Hofferbert and Sharkansky argue that "The distribution of benefits or sanctions is perhaps the most important output dimension for political scientists...," and stress the difference between the level and distribution of benefits of social policy.17 Sharkansky, in a separate work, reasons that political factors will affect the size of the increment of change in spending more than they affect absolute levels of spending, due, in part, to a feedback effect where the results of previous policies and decisions are considered.18 Finally, Andrew Cowart shows that past experience with similar programs as well as political norms of involve- ment established in previous programs greatly affect the levels of new expenditures in the states.19 There is evidence from recent American state studies, then, that the nature of the political setting affects the level_and the impact of social policy. Those interested in cross-national research, too, have continued the study of the influence of politics on policy. Hogan suggests, after studying Mexico and Canada, that the influence of politics on policies may increase as a nation reaches a higher level of economic development. In Canada, political factors were as important as socioeconomic factors in explaining policy outputs.20 In further support of Hogan's argument, Peters finds a pattern of increasing importance of political variables over economic development indicators in France, the United Kingdom and Sweden since 1910.21 Moreover, Peters, in conjunction with Doughtie and MCCulloch finds (in a study of twenty-one developed democracies) that a nation's culture and type of decision-making structures in- fluence the types of social policy adopted.22 Cameron and Hofferbert, arguing’along similar lines that the structure and nature of intergovernmental relations should affect political outcomes, find that within-system disparities in educational expenditures vary systematically between federal and unitary states.23 And, the researchers involved in the International Study of Values in Politics found that the national political setting and macro-processes influence a government's activeness in areas of change and equality.24 In another vein, Harold Wilensky posits that several things inter- act to lessen a country's commitment to expanding welfare policies: local control and decentralization in government structures and in labor federations, social pluralism or heterogeneity and the degree of stability of the party system and in the revenue structure.25 And finally, Noel Boaden and Douglas Hibbs, in separate works, show that current national policies are reflective of the ideological bent of the party in power. Boaden gives evidence from boroughs in England and wales that party (which party is in control) affects the establishment of priorities among services; Labour councils were bigger spenders and generally more active in certain policy areas, especially those areas affecting the working and lower classes.26 Hibbs concludes from his study of Western industrial nations that "...governments pursue macro- economic policies broadly/in accordance with/the objective economic interests and subjective preferences of their class-defined core political constituencies."27 After a reading of the literature, it becomes evident that there is little clear consensus among scholars on the impact of politics on policy or on poiiWesults. Nor has the study progressed in a cumula- tive manner; different researChers use very different measures for their dependent and independent variables, and a wide range of political factors has been considered in separate studies. Moreover, some researchers study expenditure levels while others look at impact variables and service levels, or'at the distribution of benefits. Several themes do emerge from the literature, however. First, it .CMZ-LJFV'IRL seems that the issue - aLea mayldetermine whether or not political factors will play an important role in shaping policy outcome. The area of education policy typically becomes highly political and controver- sial. Hofferbert and Sharkansky note this in their later'American states studies. Moreover, welfare policies can be politically controver- sial, especially where socio-economic class or ethnic divisions are salient. Jennings, Wilensky, Hibbs and Boaden all note this phenomenon. In the United States, this kind of phenomenon became evident in the 1960's in Congress (and in the states) when welfare as well as civil rights policies were being challenged by political elites from the South Political factors may come into play in a special or more obvious way when certain issues are being addressed; education issues and issues related to the have — have-not struggle (or to the redistribution of society's goods) particularly seem to be affected by the structure of politics. Second, then, is a corollary theme, that social andlgr demggiaphic conditions -- heterogeneity, economic division, the size of the "needy" population, et cetera -- may affect the_role or impact of political} factors on policy decisions. Such demographic factors as the percent of the population that is school-aged and the health of the population, moreover, determine "need" and thus partially determine policy responses. Ecological pressures are discussed by several researchers including Sharkansky, Wilensky and Peters. 10 Third, it seems clear that form§£;piogram experience and norms of pgggggmLinvolvement (i.e., earlier political decisions) affect present decisions about new policies. Walker, Jackman and Cutright find this relationship or control for it in looking for further determinants of policy and the role of politics. Fourth, scholars generally acknowledge some role for economic -Jh:" development in determining policyexpenditure levelsfixfor example, al- though there is not widespread agreement about hdw far this relationship goes. Expenditure and general policy implementation require some amount of economic resource capacity, but Jackman28 and others suggest a curvi- linear relationship here, i.e., there is a point at which the rate of welfare state development begins to lag behind (continued) economic development. Others, like Hogan and Peters, posit that the importance of system wealth as a determinant of policy maries over time and that after reaching a certain level of development or growth, political factors play a more important role than economic factors in shaping ’w policy. 6M5” U ‘KJJ" 3": I : a f.' l...‘- ' Finally, it is generally recognized that a study of policyexpendi- ture levels is not the same thigg_as a study of policy impact or results.29 Sharkansky, Fry and Winters and others point out that there U Jets”: cJ(O"-¢ /o~f"‘ k may be a gap in the translation from expenditure to services or from ”mm—... expenditure to equal distribution. Peters and Heisler note that ... converting laws, rules, decisions, regulations, and even authorized expendituresinto real consequences for, or im- pacts Upon, the society or any other'target of policy (e.g., the environment or’another policy), is neither‘a routine process nor one free of politics. Indeed, one of the 11 interesting developments over time in the role of politics in public policy in highly developed societies is that it seems to have moved from_p§st;portion of the system in which inputs are convertsd_into outputs to that segment where out— ppts are converted into outcomes. (emphasis mine)30 Expenditures and actual service levels may better be considered part of policy output whereas distribution patterns better indicate policy 933- {ggps or snort-run results. The question of what role politics plays, then, must be considered in two parts. The Impact of Political Structures on Social Policy Outputs and Outcomes New research focusing on the influence of political factors on policy or policy effectiveness can benefit from the relationships uncovered by earlier researchers and go on to further specify the im- n x l‘ r‘,.,.1(,: VII'I': t‘ pact of politics. Given the various findings and assertions in the literature, we may try to specify under what conditions political variables might affect policy, rather than continuing the rather sterile debate between those who argue for and those who argue against the importance of political factors. The framework of the investigation Should allow for the question "Which factors play a role under what circumstances?" It seems reasonable to expect that depending on cir- cumstances (the policy areas under study, the level of government studied, the historical time period, the level of’national wealth, national demographics, et cetera) political factors may play a greater or lesser role in affecting policy results. ‘Peters and HOgan both seem to suggest that such a developmental model may best explain policy; that is, over time policy determinants change in importance. In this study I shall investigate the variations among countries 12 in commitment to various social policies and the effectiveness of those policies in providing for the general welgibeingrdfpcitlaens.x In particular, I will investigate what influence political structure centralization has on social policy outputs and outcomes. I shall study three policy areas -- education, general welfare and health -— that directly affect the day to day well-being and quality of life of citizens and that are the concern of all governments whatever the regime ideology or level of political "deve10pment." That is, any viable nation-state must be concerned with providing for'at least minimal levels of health, education and welfare for its populace if it is to survive as a legitimate political entity. No nation can long survive with a sick or disabled citizenry, nor can a state function politically, economically or socially if basic human needs are not being met. Economic change and growth, for example, requires healthy, literate workers. Moreover, pressure for increased activity in these policy areas (as opposed to other'areas such as national defense or transpor- tation), as well as debate over the correct role of government and the proper methods of implementation of social policies, are features of political life everywhere. It is also the case that these policy areas are ones in which we would especially expect to find a role for politics. Regarding education policy, for example, Cameron and Hofferbert state: Education policy lies at the crossroads of the economy and the polity. The education policy process involves interb action of class, religion, ethnic and linguistic conflicts. No other arena of social policy more accurately reflects patterns of social distribution which have cumulated —‘“‘__ E *me'f/I 13 through past resolution of these conflicts in favor of particular groups. And because educational outcomes structure the allocation of values within a society not only for the immediate period but also~for future genera- tions, no other arena of policy has been more politicized by those advocating or opposing fundamental social reform.31 flw-dm ‘‘‘‘‘‘‘‘‘ a In this study I will focus upon thenimpact’gflthe degree of govern- . .- - .‘N-‘v’ "’ ment centralization or authority concentration onngSi l policngutputs and outcomes and the reLationship of centralization to the extent to which various nations are committed to education, health and welfare policies. Further, I will focus on what that commitment means in terms of actual goods and services available to citizens. Is there an easy translation from outputs (and commitment) to outcomes? One way to measure national ppmhitment to social policy areas is to look.at the level of expenditures appropriated to those areas in the total system. How do nations compare, for example, on amount of money allocaEedgtofhealth or to education on a per capita or per student basis? How do they compare on proportion of national wealth devoted to social policies? These figures would indicate a country's pgligy effort- They would not indicate, however, actual short—term policy impact, services delivered or'pglggyLoutcomes. That is, money allocated may "get lost" in implementation or otherwise fail to be translated into actual goods and services; money may not reach target groups. It is necessary, then, to look at the level of actual goods and services at the citizens' disposal if we wish to draw any conclusions about the effectiveness of policy and political performance or outcome. Data on 14 such things as educational attainment, literacy, infant mortality and the distribution of material goods (such as income) must be collected. The Question of Distrigption Up to this point, the issue of the distpibution of and access to goods and services provided by such social policies for the well—being of society has not been addressed, even though the "well-being of society" implies the well-being of all groups and "quality life" implies at least minimum levels of need satisfaction for all citizens. Yet, it is generally through social security or social welfare plans that nations typically try to improve the lot of the "have-nots" or the lower income groups in the society, or try to establish a program of risk- sharing, whether or not the stated goal of the program is a redistribu- tion of the material goods society has to offer. Titmuss argues, for example, that in general, social security and service programs affect "equality" by affecting the distribution of income, command over resources and overall life chances.32 And, Jackman states: ... while it is not necessarily the case that the intent behind social security programs is to eliminate the variance across individual incomes within societies, these programs are typically designed to raise the incomes of those at the lowest levels of income by protecting them against unem- ployment, illness, and so on. If they do raise the income of such groups in this manner, these plans may make the distribution of material rewards more egalitarian.33 The implication is that with more social security spending, a nation inevitably becomes more egalitarian socially. Cutright finds similarly that the effect of large social security programs is to make 15 income distribution more equal even though such programs are not usually designed to erode the privileges that the rich enjoy and give them over to the poor. Nations with a large percentage of their gross national product committed to social security plans do tend to transfer a large amount of income to poorer groups in the system.34 It need not be the case, in other words, that social security or social welfare policies are a result of altruism on the part of policy makers.35 Such programs may be implemented for very practical reasons. But it does seem that such programs lead to some redistribution of valued goods and services among groups in society. This is not to argue that redistribution is easily accomplished or that it is a necessary corollary to economic growth. It is clear that in spite of their social welfare programs, the industrialized.and wealthy Western nations have not been successful at eliminating problems of wealth concentration and poverty. Moreover, those who study the Third World and processes of political and social change argue that traditional (Western) economic theory -- which stresses the primary importance of industrialization and steady growth of the gross national product -- has failed in its implementation to redistribute benefits of growth and thereby eliminate the disadvantages of the lower economic strata. Third World societies -- even those that have experienced greatly expanded per capita incomes -- have not experienced the ftrickle-down" of benefits from modernized to rural sectors which \‘ . . 4-.....sLH“ . 3 theorists argued would result from industrial growth. 6 Instead, as Hollis Chenery argues, it is clear now that in spite of years of growth in the Third World, Q3? wealth has been very unequally distributed and ““un“'. - major social problems have not been alleviated.37 Distribution of goods _._—‘__,_ 16 and services is an importantgguestion to be addressed by social poligy in all nations. In a study of the effects of political factors on social policy choices, level of commitment to the policies and policy outcomes, therefore, distribution patterns must be considered. The issue of social equality, or, "...the structural issue of the extent to which access to and the consumption of material goods are distributed in an egali- tarian manner within societies,"38 must be addressed. This is so especially when considering the outcome or short-term consequences of the "social welfare policy" dimension, since this dimension is generally recognized as redistributive. But we may argue that the other social policies -- health and education -- also are indicators of efforts to redistribute chances between the haves and the have-nots for attaining valued goals. -? It is clear, however, that it is not enough to look at absolute levels of program-expenditure. Rather, distribution patternsgpresent another dimension of_policy_results. They allow us to generalize further about the shgft:runtimp§§t of policy choices on society; that is, how have those policy efforts affected the overall quality of-life of all groups? (Only an historical or time-series analysis involving data covering a longer time span would allow generalizations about ultimate policy impact. This thesis investigates immediate policy outcomes, or results that are of immediate concern to policy makers.) Educatioinolicy - A Note . U To say that national health policies and programs affect citizens' quality of life and indirectly indicate efforts at a redistribution of valued goods -- such as health -- sparks little controversy. The poor, 17 who cannot afford private health care, are subsidized by the national government through various health-care programs. But given recent literature on the failure of education to affect social or occupational equality in the United States, we must justify the use of education policy in this study as an indicator of efforts at redistribution of life chances. My use of education policy in this study (as with the other pol- icies) is twofold: 1) I am interested in the influence of political structure on education expenditures and service levels and on education policy outcomes and 2) I am assuming that education policy is at least an indirect indicator of efforts at assuring a quality life to all groups within society. It is this second usage that must be justified. In his seminal study, Christopher Jencks argues that equalizing educational opportunities in the U.S. has not led to equalization of income or of occupational status between individuals -- education policy has not affected equality of opportunity or equality of results. In- stead, Jencks argues, factors such as luck and on-the—job competence -- factors which are only moderately related to schooling, family back- ground or standardized test scores -- result in unequal adult life chances. Therefore, Jencks calls for direct policy moves toward insur- ing equality of results where some insurance system can neutralize the effects of luck, redistribute income and "break the link between vocational success and living standards."39 Nilensky, in agreement with Jencks, posits that education is over- whelmingly meritocratic and VOCational, especially in relatively affluent societies, and that it is not equalizing. He argues further that when a nation develops mass higher education it lags behind in 18 welfare program development. Whereas health and welfare contribute directly to equality, according to Wilensky, education can provide no more than a peripheral contribution to equality (of results).40 Moreover, Carl Weinberg points out that educational opportunities in the U.S. have not contributed to social mobility and may, in fact, reduce mobility chances for lower groups: Education is now considered almost exclusively Egg channel for social mobility in our society. When the school dif- ferentiates students early in their educational career, it is denying to those who are allocated to inferior status the only opportunity they have for social mobility.... Mobility for the poor today is as doubtful as it was fifty years ago. Only one out of three persons improves his social position from one generation to the next and of per- sons in the lower or poverty classes, less than three out 'of 100, on the average, are able to clamrtheir way to pro- fessional status. If it were not for the random fortunes of athletic scholarships, the figure would be even smaller.“ On the contrary, others argue that the impact of education has not been slight and that education policy may have implications for redis- tribution efforts, if not for movement toward equality of results. Murray Milner writes that higher levels of education enable blacks and other*minorities to secure better jobs: in March, 1970, 63% of nonwhite men with four or more years of college were professionals or technical workers, while h% of high school graduates in the nonwhite male cate- gory had such jobs. Also, Milner notes a change between 1960 and 1970 in the differences in occupational status between groups with similar 19 levels of education. Whereas in 1966 his "index of dissimilarity" was around 38, in 1970 it was around 28. Milner concludes by pointing out: 1) that there are still differences in educational attainment even though job market discrimination seems to have decreased for those with college training and 2) that however much Americans are committed "tardily, reluctantly and hypocritically" to racial equality, they are not committed to economic or social class equality.u2 Milner ultimately argues for efforts to reduce economic inequality and to increase equality of results or benefits. Further, Becker’and Chiswick posit a direct link between schooling inequality and income inequality in U.S. states and within states, and in several other countries. They state: "Evidence ... indicates that schooling usually explains not a negligible part of the inequality in earnings within a geographical area and a much larger part of differences in inequality between areas."’+3 They also give empirical evidence to support the idea that additional expenditures on education may yield higher returns in low income areas -- education policy may be used to improve life chances of poorer groups. In further support of this view,’ Jan Tinbergen argues that even though education has been expanded in recent decades in the U.S., it has not expanded enough to meet demands of the organizers of production or of modern society in general. There- fore, he argues, income inequality remains unchanged from what it was before expansions in education. Income inequality is a function of (among other things) educational opportunity andattainment.1+4 Russell Harrison argues in his 1976 work that fiscal reform of U.S. education systems would better ensure equality of benefits, pro- mote equity and increase the efficiency of society's allocation process 20 for social, economic and political resources. (His most basic reform would entail disproportionately increasing expenditures in areas with the highest rate of returns from educational investments -- theoreti- cally, the poorer areas.) According to Harrison, education (and fiscal reform) would equalize benefits because, among other things, education can raise the average level of political participation and thereby in- crease equality of political participation.45 Finally, Lawrence Joseph criticizes Jencks' work on philosophical and logical grounds. Joseph argues that Jencks did not make a convinc- ing case (in showing conventional wisdom about the link between educa- tion and equality to be wrong) because he was not clear on his under— lying assumptions about the nature of education, and the meaning of equality and equality of opportunity. In this view, Jencks fails to discredit the equality of opportunity concept as it relates to open education systems. Joseph argues: What Jencks fails to understand, then, is that the lack of relationship between equality of opportunity and equality of results is a logical, not an empirical one. The norma- tive implications of the meritocratic model are such that we should not expect equality of Opportunity to lead to equality of reSUlts.... Thus, regardless of the validity of Jencks empirical analysis, his argument about the deter- minants of inequality in adulthood is essentially irrele- vant to his policy conclusion regarding income redistribu- tion. That is, his normative argument for equality of cir- cumstances in no way depends upon whether equality of opportunity leads, empirically, to equality of results.... 21 The key point which Jencks fails to understand, of course, is that equality of opportunity in its meritocratic form, is not meant to lead to equality of reenlts. Equality of opportunity will be compatible with equality of results only in a society which is already radically egalitarian.46 Providing education to one's citizens may be viewed as a condition of equality or’a concomitant of efforts toward equality of results (as opposed to equality of opportunity or equality as based on a merito— cratic notion). Education is something that is vitally important to any effort at providing well-being for citizens in that it can open new horizons and create new Opportunities for them. Many scholars argue that education plays a positive role in facilitating redistribution. But if education is not a concomitant of social equality, it is also not antithetical to it. For several reasons outlined earlier; and given that there are some indications that education may relate to income distributions (for example) and improving life chances, education is an important policy area for consideration. The limitations involved in considering educa- tion as an indicator of social equality are recognized. It can only be an indirect indicator of larger efforts at basic redistribution. The Impact of Various Forms of Political Structure Centralizatign The political factors under consideration in this study are patterns of intergovernmental relations and the role of national govern- ments vis—a-vis their subnational units in public policy authority and command over resources. Specifically, the focusis on_the/r01e of the . national government in public policy authority (setting priorities. chooging and writing policy, spending and taxing, implementing or /’ ” ./’ 22 enforcing policy, controlling resources) and on the degree of policy authority concentrated at the national level. Since patterns of centralization and intergovernmental relations establish a framework within which political, social and economic variables must operate, we would expect the nature of center - periphery relations to be a critical determinant of policy. The question under study here is: What is the effect of political structure centralization on social policy? Discussions in the literature on government centralization have generally distinguished several types of centralization: financial or figggl, formal — legal (federal or unitary structure), and more general political centralization, such as the degree of vertical power distri— bution. Although some refer to structural centralization in loose terms and without clearly defining the concept, other writers argue that different types of centralization exist. Cameron and Hofferbert posit that fiscalidpcentralization and constitutional federalism are not necessarily identical and give England (unitary), Norway (unitary) and Austria (federal) as examples of where the two aspects of "centralization" 47 diverge. These authors state: ... The important difference in intergovernmental relations is not simply that between defacto centralized.and decentral- ized but that between unitary and federal systems of govern- ment. Two federal (or unitary) systems may manifest differ- ent degrees of allocation of financial responsibility to the central government and yet the two may not differ in any measureable way in the structure of intergovernmental rela- tions.... On the other hand, two systems one of which is unitary and the other federal may have very different 23 structures of intergovernmental relations in spite of com- parable levels of fiscal centralization (e.g., Norway and Austria).48 Samuel Beer, similarly. conceives of federalism as a special kind #9 of decentralization, while neither Werner Pommerehne nor Paulo Vieirra find a significant relationship between constitutional federalism and fiscal centralization in their multinational studies.5O Moreover, Michael Reagan argues that even when the formal criteria of "federalism" are met, little has been learned about actual power distributions or political function distributions among levels of government. The for- mal (legal) meaning of federalism has little real significance politi- cally, and several factors must be looked at to see where power lies.51 Many other writers concur with Reagan that federal systems vary signi- ficantly in their level of more general political centralization and in their basic characteristics. Federalism in broad terms and as a fermal - legal arrangement simply guarantees some decentralization of decision making and guarantees the formal autonomy of multiple decision points.52 Divisions of functions between units, for example, differ among federal systems. Given these observations from the literature, it may be useful to study the policy impact of political structural centralization in three separate parts, fiscal, formal - legal_and politica . Such a breakdown of political structure centralization into several types, and.a consider- ation of the effects of each type on social service policies has not been evident in the literature to date. 2# Government Centraligation In The Literature Several writers conceive of national government centralization in more general terms in discussions of the effects of centralization on policy or society. I will consider these arguments about centralization first. Marion Levy argues that centralization of governmental control and authority (presumably including control over financial and all other matters) is quite necessary in changing ("developing") societies where a strong sense of policy direction is needed for the sake of stability. It is a strong, centralized political structure that has the coercive and financial capacity to handle the overwhelming problems encountered in "transitional" societies.53 Centralization, then, pro- vides policy direction and enforcement where weaker, subnational units could not. Hage and Hollingsworth forward similar arguments in their considerations of how structure might affect political performance. They hypothesize that governmental centralization (in economics and political power) leads to policy efficiency and standardization across regions within nations, promoting social equality across social classes 5“ They encourage study of such hypotheses in attempts to and groups. integrate social theory and social policy. In a study of the world's twenty-two richest nations, Wilensky uncovers some evidence for the hypothesis that centralization promotes an emphasis on welfare spending and equality. Defining "centralization" as a 99mposite of indicators of constitutional form, financial central- ization and political control by the center over local chief executives, Wilensky finds that more centralized nations have greater welfare state development or spend more money for welfare or equalization pro- grams.55 (He does not attempt, however, to refine or break down his 25 concept of centralization to consider whether different types of centralization may have varying effects on policy.) Morrison and Stevenson, on the other hand, discuss the national centralization of politics and economics in African states and conclude that such centralization has been less than effective in meeting communal demands and daily needs through social policy -- centralization has led analgesic insistTDiJnL.stir.~ and. violence.56 Their empirical findings stand in contrast to Levy's theory (which also considers the developing countries) and run counter to Wilensky's findings. (The differences between Wilensky's findings and Morrison and Stevenson's results may be a function of several factors, including level of economic development and their definitions of centralization. This study may illuminate some of these problems.) Uphoff and Esman present further evidence to support the idea that centralization may have a negative impact on social welfare deve10pment. In their studies, these scholars find a positive relationship between local control and planning gpd attainments in the nutrition, health and education areas even when controlling for national wealth. Development in the social policy or human sphere, they argue, takes place where there is local initiative and resource mobilization. Uphoff and Esman, how- ever, qualify their findings by pointing out preconditions for the success of social development at the local levels. Local organizations cannot support programs for social equity on their own; they need strong linkages with and political and resource support from the center government. Moreover, the authors find that there must be "relative equity in the ownership of assets, particularly lan " before there can be effective local organization. In other words, entrenched elites make 26 local social development impossible, and, some degree of social equality must exist prior to welfare development. (The fact that these pre- conditions are absent in many Third World nations seems to limit the applicability of Uphoff and Esman's strategy for develOpment, and, implications for policy making are unclear.) Eiscal Centralization Several other writers address the question of the policy impact of fiscal centralization. According to Frederichryor, economic centrali- zation leads to a greater proportion of national wealth being devoted) to education due to the unity of budgetary organs, greater comparability of programs within the nation and greater policy emulation across units. Pryor argues that centralization allows for hierarchical organ- ization, well-defined administrative links and standardization of tools and procedures, which in turn lead to a set of binding instructions on organs of government.58 In a more decentralized setting, on the other hand, approved budgets may only represent an "upper limit" on spending and may in fact be reduced in various circumstances by local or lower- level administrators. Furthermore, Peacock and Wiseman seem to suggest that greater financial centralization leads to greater allocations of funds for social policy for three reasons: 1) centralization is both a product and a cause of uniform standards of policy being developed, 2) centralization enables the national government to overcome problems which transcend or are external to local units and, 3) it allows for a greater reliance on more efficient or wealth elastic revenue systems.59 Given these works, then, we would expect fiscal centralization at the national level to lead to greater social services or public spending, for example, on education. 27 Pommerehne and Oates, in separate works, stress the importance of fiscal centralization in any attempts to redistribute income. Pommerehne argues that state and local governments do not have the necessary tax margins available to them to be able to enforce an aggressive redistri- ~bution policy (since national governments generally regulate the extent to which the locale can lay down tax rates) and that local units that enforce such redistribution may simply fail when mobile economic sub- jects "harmed" by such policy migrate to another state or locale more advantageous to them.60 (One generalization from Pommerehne's six- nation study is that major differences in income encourage a stronger redistribution effort by the central government, which in turn spurs greater centralization.) Oates concurs with Pommerehne when he concludes that higher degrees of mobility may make redistributive programs impossible to implement at the decentralized level and he predicts greater fiscal centralization in this policy area. It is reasonable to expect, Oates says, that a basic determinant of the proper level of government to handle the dis- tribution of a service for the public is "the geographic pattern of the effects of the output." For example, due to the mobility of families and of workers in a modern society, all groups and locales are concerned about providing quality, sometimes standardized education. The whole of society feels the effects of educational policies in any given area within the nation. Or, pressures for improved levels of services for poorer areas -- areas that are a cause of concern or embarrassment to the more affluent of a nation -- may be a centralizing force. The benefits of improved services to poor areas serves the whole nation: the poor benefit from improved services, and the consciences of the rich 28 are relieved. The effects of the output are felt by all. that Oates also elaborates on factors or problems external to locales may tend to increase the demand for fiscal centralization: ... the growing knowledge, mobility, and wealth that accom- pany economic development tend to reduce the scope for in— dependent budgetary policies by decentralized levels of government. Local public officials find over time that they become less insulated from the policies adopted in other localities.... It is also possible that technical progress over time will permit more effective centralized provision of some public services so that previously nonexistent economies of scale will come into being....61 Problems external to locales, or ones too large or "too hot" to be handled by local politics, must be handled on a higher level where financial and administrative and coercive capabilities are greater. Oates and others point this out. The fact of "externalities" and their multidimensionality becomes clear, too, in U.S. national politics and, not surprisingly, also in writings on minority populations and their K mistreatment at the hands of local politicians. For instance, on this aspect of externalities and redistribution efforts Daniel Meynihan writes: The necessity for concentrating decision making at the na- tional level will be enhanced if current trends in racial concentration persist. Between 1960 and 1966, the number of children under age fourteen in metropolitan areas increased by 3.3 million.... The average annual rate of increase of nonwhite children (2.4 percent) was three times the rate for 29 white children. Ninety-five percent of the nonwhite increase was in central cities.... According to one estimate, by 1970 Negroes will constitute 40 percent or more of the popu- lation in fourteen of the nation's major cities.... In southern communities accustomed to taking collective measures to prevent Negro accession to power, there may be movements toward metropolitan governments in order to maintain Negroes in a minority voting status; but, in general, continued and possibly heightened racial tension is likely to inhibit greatly the development of true metropolitan governments. (A fortiori the resolution of conflict between central cities and suburbs will have to occur'at the federal level save for the few states with sufficient political and fiscal resources to handle such matters at the level of state government.62 Where subnational units feel (or prove to be) inadequate to meet social demands and situations, therefore, we may expect increased pres- sures for national government involvement and centralization in fiscal matters. The question then is, Is centralization a precondition for or a spur to social equality and a more equitable distribution of society's goods? Oates does not agree with either Peacock and Wiseman or Pryor that fiscal centralization leads to greater public spending. He finds that after allowing for effects of the level of income on the size of the public sector relative to the size of the economy as a whole (something Peacock and Wiseman also consider when they discuss effects of affluence), fiscal centralization has no independent effect on the size of public spending.63 (Nor did he find, in a quick look at federal or 30 unitary status, that that status accounted for variations in the size of the public sector.) However, Oates does not at any point break the "public spending" figure down into policy areas, nor does he consider variations in policy outcomes, something that this study will consider. Oates' concern #8.Qefli¢réd‘0n the question of which level of government should perform which functions. The literature, in review, provides no consensus on what impact fiscal centralization has on public spending, nor has more detailed study been undertaken to determine the effect of fiscal centralization on various policy areas and policy results (except for Pryor's look at education and Oates' and Pommerehne's theoretical consideration of redistributive policies). Important questions remain to be answered. FederalLUnitary St§_._t_1_.1_§ The impact of federalism on policy is a subject area that has received little attention from political scientists. Yet the question does not seem unimportant or irrelevant given that almost forty percent of the world's population lives under federal governments.6u Many scholars have gone to great lengths explicating the meaning and history of federalism.65 The explications in volumes and essays have noted few common principles among federal systems, however, and, after reading the literature on the nature of federalism and considering the world's federal systems, one is capable of saying little more than that federalism refers to a structural and legal form. Federalism constitu- tionally and structurally establishes the autonomy of multiple decision points (although it does not specify divisions of functions between those points) and an actual and areal division of power between a general government and several subnational governments. It ensures that a 31 citizen will be subject to two jurisdictions.66 This is not to diminiSh the significance of a nation's claiming federal (or unitary) status. The pattern of intergovernmental rela- tions -- whether or not there will be subnational jurisdictions and how centralized structures will be -- should have important consequences for policy and for the well-being of society as a whole if only because federal/unitary status stipulates the number of decision arenas (one or several). Given the attention paid to it by scholars and national leaders, moreover, we should expect federalism to make some difference (compared to politics in unitary systems) when implemented. An initial consideration of federal systems in the world indicates that federal systems do differ in important respects, and this seems to justify a consideration of various aspects of centralization. Oates writes: As one examines the development of different countries and notes, for example, the sorts of considerations that led to the establishment of a particular federal government, it becomes clear that the unique historical experiences and circumstances of each country have led to a differentiation in governments that is difficult if not impossible to ex- plain in terms of a simple set of general principles.67 Switzerland's brand of federalism reflects the linguistic, reli— gious, and cultural differences that exist among her territorial regions. The cantons make up a loose federation and are concerned primarily with "cleavage engineering" and depoliticizing national issues. Cantonal governments handle most matters of importance to citizens and, relative to other federal situations, the national government is politically 32 weak.68 The national government really can do little more than urge that policies be adopted. For example, on the matter of fiscal equalization among regions the Swiss Constitution says: "The Confederation gpggup- agpg financial equalization among the cantons. In particular, when federal subventions are granted, the financial capacity of the canton and the mountainous regions must be considered in an appropriate fashion" (emphasis mine).69 This language is mild, relative to that used in other nations concerned with fiscal equalization. India, likewise, is an example of where a federal system has been imposed upon an expansive territory with regional, cultural, ethnic and language diversity. And, there are, in Indian politics, conflicts between the federal and state governments. National goals, duties and responsibilities of the various levels are points of contention rather than areas of national consensus. For instance, R. J. May points out that federal financial relations preceded the emergence of political federalism and that has led to disagreements over issues such as how revenues might be shared between the center and the states. While the Indian Planning Commission is concerned with broad national objectives for development, the Finance Commission is concerned with maintaining the autonomy of state governments.70 Canada and the Federal Republic of Germany may also be compared. Canada has a flexible federalism fiscally, where the provinces have asserted their rights and responsibilities vis-a-vis Ottawa (consider, for example, Rene' Levesque's Parti Quebecois and other provincial move- ments). This is so especially in the area of education where Canada out—performs many other nations: while provincial responsibilities are understood, there is strong emphasis on federal-provincial COOperation 33 on education policy and funding, and, there is at the same time, demand for financial help (toward general equalization) from the center.71 Donald Smiley writes: ... the fundamental idea that Ottawa, through unconditional subsidies, should guarantee the revenues of the provinces to a national average has been an influencial standard in federal-provincial fiscal relations from the publication of the Rowell-Sirois Report (1940) onward.72 Germany, on the other hand, is fiscally more inflexible. The responsibilities of municipalities and the Lander in funding higher education, for instance, are minuscule. A Bund-Lander Commission for Education Planning was not established until 1970; cooperation between center and periphery has been lacking.73 Note, too, the language of Article 107, paragraph 2 of the Basic Law regarding fiscal equalization among Lander. The federal government is, comparatively, powerful and deals from a position of strength vis—a-vis Lander. A federal law requiring the consent of the Bundesrat spall gpspgp a reasonable financial equalization between finan- cially strong Lander and financially weak Lander, due account being taken of the financial capacity and require- ments of communes....7u Australia represents a socially homogeneous nation where federal form was adopted due to sneer geographical size.75 It is, moreover, a case similar to Canada in that the viability of multiple decision points has been maintained. But it is also a case where standardization and uniformity among regions has occurred: Australia's experience in the sphere of federal finance has 31* generally been regarded as successful. Since the latter part of the 19303 there has been no real evidence of small states' discontent and the federation has been politically stable, with strong emergence of national standards in fis- cal and administrative fields; there has been a marked re- distribution of revenue with the apparent concurrence of the larger wealthier states. Some writers have expressed reser- vations about the centripetal tendencies in Australian federal finance, but an examination of the political forces at work provides little evidence that the bargaining power of states relative to the Commonwealth government has de- clined.76 Michael Reagan argues that the United States is an example of a new style of federalism where stress is on interdependence between national and state governments and a sharing of political and fiscal functions.77 The political game is played under conditions of mutual leverage where each level is able to exert influence and pressure on the other. Each level needs the other level. The intergovernmental pattern establiShed, then, is one of continual change and adaptation in response to social and economic forces. And the 'new federalism,' according to Reagan, stands in contrast to a more static concept of constitutional federalism. This very brief view of six nations highlights varying characteris- tics of federalism. Not only have nations adopted federal structure for different reasons, but the models of federal politics they present are different. At the same time, it may be argued that there is some- thing inherent in federal structure (that is, providing several decision 35 arenas) that differentiates federal politics from politics in unitary nations: and this difference should have some significance for policy. Federalism's Ippact On Policy Political scientists know little about the effect of federalism on policy outputs and outcomes. Nor have scholars been accustomed to thinking about federalism as a policy determinant, as something affect- ing day to day politics, or as a pattern or framework within which other political and socio-economic factors must operate. The information that can be drawn from the literature on this topic, moreover, answers few questions; there is little consensus among scholars. How does federal structure affect gggppggte spending for education or health or welfare? Cameron and Hofferbert hypothesize that the autonomy of several decision points guaranteed by federalism will have an overall conservative affect on aggregate educational spending. Unitary systems -- which have one decision making forum and which en- courage uniformity -- should spend more for education. Their data from eighteen European and North American nations, however, shows no relation- ship between federal structure and aggregate educational expenditures.78 In further support of this view, Wilensky found in his twenty-two nation study that centralization (one indicator of which was federal/ unitary status) led to increased welfare spending and interest in redistribution. The United States, Wilensky writes, has no interest in general welfare state development or in redistribution and, instead, operates by the "success ideology."79 Wilensky thus implies that federalism should have a negative impact on aggregate spending, since national elites will have lesser capacities for enforcing national pro- grams. 36 Samuel Beer argues in the opposite direction. He postulates that American federalism has led to the creation of a "professional bureau- cratic complex" (made up of professional administrative and technical specialists in government bureaucracy) and an "intergovernmental lobby" (made up of governors, mayors, county supervisors, city managers and other officeholders). Each of these centers of influence, although they are in contention with each other, work in the same direction toward greater government expansion and spending "...with no view to the over— all direction of the nation."80 The intergovernmental lobby, according to Beer, presses the case of its constituents before the President and Congress, while the professionals (who are fragmented into many areas of specialization) influence the same power centers for the implementa- tion of their newly developed programs. In sum, there is confusion about whether there inheres in federal- ism a conservatizing or an expansionary potential with regard to ex- penditures. How does federal structure affect socialppoligy outcomes? Again, there are various viewpoints. Ivo Duchacek argues that federalism is costly in that it produces uneven progress toward economic, educational and social develoPment among provinces.81 Cameron and Hofferbert support that idea with data from eight Western European nations (four federal and four unitary). They argue, in summary fashion, that federalism by its nature invites variation in policy performance between subnational units, that it does little to encourage redistribution within the nation, and that it fails to produce equality in distribution generally. They find that patterns of allocation of resources are affected by the structure of intergovernmental relations. Variations in 37 expenditures across units are more constrained by variations in re- sources in federal than in unitary systems, i.e., education policy out- puts in the federal systems favored the most advantaged areas, whereas the unitary systems more successfully implemented equalization formulas in educational expenditure allocation.82 Similarly, in a ten-nation study, John Echols finds that whereas revenue equalization and equalization of social expenditures across sub- national units is not a simple function of the communist - non-communist distinction, federal/unitary status does have an effect. Unitary states tend to produce greater equalization among units than do federal states.83 In contrast, Walker contendswthat American and Canadian federal structures have affected the deve10pment of national standards and expectations which pressure "lag" states and provinces to adopt new programs. Because of this, presumably, there are fewer inter-unit dis- parities in social policy. He does find evidence, though, that the well- develOped units adopt new programs more rapidly than do poorer, less industrialized units.8a While there is some evidence at hand that federal/unitary status has important consequences for within-nation distribution patterns and social policy impact, this study will examine aggregate cross-national patterns of the effects of federal (or unitary) status. That is, I may find as Cameron and Hofferbert did in their eight-nation study, that the answer to questions about the impact of federalism on policy outputs and outcomes differs depending upon the level of analyses —- national or subnational -- and the level of national resources. On the other hand, I may find that a federal system of intergovernmental relations is an important policy determinant, the impact of which is obvious at the 38 national level as well as at the local level, and in rich and poor nations. Political Centralization Reagan and others argue that federal or unitary status, although it establishes a framework within which politics must be played, tells little about the actual distribution of power or informal political authority within society. It is obvious that there are federal systems with "strong? national centers (e.g., Australia, Germany) and with politically "weak'national governments (as in Switzerland). This sug- gests that informal political authorityiboncentration may be something political power decentralization among federal systems. In considering political authority centralization as a third dimension of political structure concentration, I am focusing on the variability in vertical and horizontal power diffusion and in opinion cleavage within and among nations. These aspects of the political sys- tem are informal in that they are not specified by law, but are results of a vying for power among groups in society. I term this variation in social power configurations informal_political authority concentration and treat it as a third aspect of centralization. The degree of infor- mal political centralization may vary depending upon such things as the number of political parties in the system (given that parties are not illegal), the degree of central control inherent in the structural organization of the parties and the legislature, the opportunity for various groups to articulate their interests, party or bloc representa- tion in the legislature, et cetera. (William Riker argues, for instance, that the variation in the degree of party centralization in a 39 nation causes variation in the degree of government structural central- ization.)85 It may be argued that in systems where there is only one party (or where one party is clearly electorally dominant), informal political authority is more centralized and opportunities for presenting opposing points of view are limited. Opinion cleavage may be minimal. Rubin and Weinstein characterize single-party states in the following way: In most single-party states the chief executive is also head of the party, symbolizing the identification of party with state and nation.... Cabinet members in a one-party state are high-ranking officials in the party. The highest organ in Guinea's PDG, the bureau politique, generally has seven or eight of its twelve members in the cabinet. The national executive of Tanzania's TANU has at least half of its mem- bers in cabinet posts. It has been practically impossible in most states for a man or woman to gain a high position in the civil service without being a member of the party. It is difficult to separate the party from the government, from the legislature and the administration, in a one-party state. Party officials become administrators or para- administrators.... There is a tendency to create a monolith in which everything is absorbed: youth movements, women's movements, trade unions, cooperatives.86 (Introduction to African‘vljolitics) In such cases —- and in the late 1960s just over one-third of the 8 world's polities could be classified as one-party 7 -- it should be expected that this concentration of power will have some measureable 40 effect on social policy outputs and outcomes. (This is not to denigrate single-party systems. Many leaders of the Third World nations in the throes of political and social change have argued convincingly for the necessity of providing a structural organization united and strong enough to confront organized colonialist and imperialist forces. I postulate here, however, that such a centralization of authority should have some observable effect on health, education and social welfare policies and their implementation.) On the other hand, where two or more parties are electorally viable and have opportunity to introduce policies, alternatives and dissenting opinions, there is relatively greater power diffusion and potentially greater opinion cleavage. This results in a lesser degree of centrali- zation of command. This situation, too, should result in observable effects on the nature and level of social policies. Therefore, I will examine the viability of considering informal political authority con- centration as an independent factor of or another side of governmental centralization. The Special Case of the Heterogeneous Society In a study focusing on the impact of political structure centrali- zation on social policies, the case of social heterogeneity (i.e., ethnic, cultural, linguistic or religious fragmentation) becomes impor- tant for several reasons. First, ethnic identification and primordial attachments are important social phenomena that confront political systems everywhere. This leads one scholar to argue that homogeneous and heterogeneous policies differ from each other as much or more than ‘KN‘WW ”up“, '"”"“"““'-.~.....M .M -... ” ‘H'm ‘. -_ 7", . . do nations differentiated by levels ofyper capita GNP.88 Others argue / that ethnic identity is heightened by/a variety of things: colonialism, 41 imperialism, social mobilization, economic development, integration attempts and issues of national concern such as education and the nature of the political system.89 Joshua Fishman writes: Indeed just as newer'plant and animal species are constantly coming into being, so newer cultural formations are con- stantly being formed, some of them larger and some of them smaller than others that came before them. The existence today of Indonesians and Israelis and Palestinians and Pakistanis (as well as Chicanos and blacks and Boricuas), none of whom were massive primary cultural groups a third of a century ago, is a tribute to the human need for meaning- ful and.immediate groups of this kind. As certain groups disappear (coalesce, break up, assimilate), others rise to take their place.90 Moreover, as the expansive literature on national political inte— gration and the more recent discussions of federalism make clear, cultural-ethnic diversity operates on or within the political sphere -- it increases demands on the system.91 For example, Herbert Tingsten argues that a dangerous tension arises when a government in the plural society embarks on measures that might pit one group against another, such as in the provision of public goods and services. Each group strives for the competitive advantage.92 Socio-cultural heterogeneity in short, has implications for social policy commitment and implementa— tion. Second, the issue of the heterogeneous society comes up in dis- cussions of federalism because many see federal structure as a political solution to problems of national integration and political stability. 42 Duchacek argues, probably correctly, that unless minority groupings are guaranteed special rights and services, societal equality cannot be achieved and the groups may never emerge from a position of inferiority. He goes on to posit that the new federal formula is a means for allow- ing various groups a voice in national policies that will affect them while at the same time allowing subnational identities to exist without encroachment.93 McWhinney, Livingston and Franck, in separate works, similarly argue that federalism can be a solution to cultural diversity or subnational allegiances within societies -- federalism presents a compromise solution among groups who agree to join for limited pur- poses.9u Another group of scholars sees federalism as a means to assure expression of local interests and ameliorate communal conflict over public goods. Melson and Wolpe consider the role federalism might play in a "modernizing," ethnically heterogeneous country. The authors con- tend that "modernization" or the process of change reinforces communalism through its stress on mass participation: political competition between groups is encouraged and leads to communal conflict. It becomes necessary in such a society to allow for local, geographic or ethnic autonomy so that minorities can be represented. Policy makers must consider communal needs. For Melson and Wolpe, federalism is a solution and tool for a heterogeneous, divided society whereby various communal groups can be respected and represented at the national center. FederaliSm might preempt secessionist movements and dissolution of the state since it allows for some minimal national integration. Melson and Wolpe hypothesize that if government institutions at the national and regional level become identified with one particular communal group 43 they lose their legitimacy for the larger society. However, if a greater number of equally strong communal groups Operate to influence governmental institutions those institutions will retain their impartial- ity and legitimacy. The flaw in the Nigerian federal system, according to these authors, has been the result of not creating enough states to ensure minorities a modicum of self-determination.95 Anderson, von der Mehden and Young also address problems involved with coming to terms with diversity. These authors argue that ethnic or cultural subgroups should be recognized and given some "psychic assurance" that communal interests will be defended. They propose es- tablishing governmental institutions and structures that guarantee sub- groups some positions of authority. Moreover, they point to India's use of a federal structure to accommodate diverse groups.96 Morrison and Stevenson's studies of Sub-Saharan Africa seem to suggest a similar role for federalism. They find that centralization of authority -- measured by economic and political factors -- intensi- fies political conflict and especially communal instability (which reflects conflicts between groups holding incompatible views). This situation will continue, the authors say, as long as financial demands on national authorities exceed their responsiveness to local consti— tuencies and as long as coercive centralization is the response to political demands from local units.97 Several writers, among them W. H. Ferry, have considered the role of federalism in the United States where a black minority has definite "communal" interests. Ferry contends that black identity, self- consciousness and pride can develop only if blacks are left to organize and.administer their own affairs under a system of local political an autonomy. Such a system can work, however, only if white racism and paternalism is eradicated and replaced with tolerance and respect for cultural diversity.98 Ferry's type of federalism is close to that proposed by those concerned with diversity in the Third World. On the other hand, some theorists see not federalism but centrali- zation or unitary government as the solution to social heterogeneity and providing social services in that context. They contest the notion that federal structure is the best solution. Clifford Geertz ultimately argues for a strong, overriding national center. Although Geertz recognizes the culturally plural situations in many new states, he argues that establishing federal systems of ethnic homelands in the new states would exacerbate communal conflict. Ideally, an overriding national identity and civil government should be established within which ethnic groups can feel at home. Geertz calls for the aggregation of narrow tribal ethnic and linguistic groups into generalized ethnic blocs that can operate within a common social framework.99 Rabushka and Shepsle posit that democratic forms can never work in plural societies, which they define as societies where incompatible institutional systems coexist, where the practice of politics is almost exclusively along ethnic lines and where coercive force and instability will inevitably exist. A federal system of local autonomy, Rabushka and Shepsle argue, has been attempted in several countries and has failed. A major problem for the plural society involves the ethnici— pgpigp of collectively provided goods, i.e., difficulties arise when it comes to governmental resource and services allocation and making policy in those areas. at is, what may be good for one group may be a harm \_- -..—(pr ' for another -- a generalized public interest does not emerge in the plural 45 state so that making public policy is difficult.100 Rabushka and Shepsle forward some possible solutions for carrying on politics in the plural state: 1) deny independent decision-making authority to locales: 2) put restrictions on independent decision-making authority; 3) put restrictions on political competition: 4) restrict the scope of government; 5) create homogeneous societies and possibly new states; and 6) create a permanent external enemy which may motivate internal cooperation. Coercive centralization, they conclude, may be the only alternative fer a plural nation or at least the best form of government if stability is desired. Further, May argues that the 'federal solution' may lead in reality to weakening minority ethnic blocs and giving powerful ethnic groups protection: In some of the newer federations among former colonies, where the plurality of society has presented barriers to national integration, the acceptance of certain national norms has sometimes been included as part of the federal settlement.... In cases like this, though the professed motive is to promote the national entity, the acknowledged effect is to place limitations on groups with weak bargaining power (generally but not always minorities) in order to satisfy the interests of groups with stronger bargaining power (generally minori- ties). Moreover, such national forms often provide the basis for a nationalization of standards, and an extension of preferences to favored groups, in other fields.101 Wilensky presents another argument in favor of centralization in the case of the heterogeneous society. He posits that ethnic-linguistic 46 and religious diversities prohibit meaningful participation in less parochial voluntary associations and encourage separatist movements which make the deve10pment of a wider civic virtue impossible. Also, minority groups sometimes begin to create education and welfare services of their own, since they feel isolated and alienated from the larger society. For these reasons, and, when the heterogeneous society is governed by decentralized structures, programs committed to wider social welfare and social equality lose out. Social heterogeneity, given sharp expression by decentralization, slows down welfare state develop- ment. (Wilensky gives the U.S. as an example of this situation.) But a centralized government, according to Wilensky, can contain and dampen cleavages -- as in Belgium and the Netherlands. In such a context, the welfare state and.a commitment to equality among groups and people can be deve10ped.102 This literature, in sum, presents the_common theme that deciding upon and delivering social services is ebmplicated by socio-cultural heterogeneity; heterogeneity leads to communal conflict over and the possible ethnicization of public goods and services; The debate among writers, however, focuses upon how best to achieve political stability in the plural society. While one group feels that political stability and integration can best be served through the utilization of a federal structure, another group argues that centralized structures can create a wider civic virtue and minimize communal hostilities. But what remains clouded over is the issue of the impact of heterogeneity on H" I'd‘ MGM,“ social policy, and further, how that relationship affects the relation- ““““ 4...‘ I ship between federal/unitary status and social policy. Does the constitutional form affect social policy decisions in a unique way in 47 plural societies? Is there a "federal solution" or a "unitary solution" to problems of providing social services in plural societies? How do heterogeneity and federal/unitary status interact, if at all, to affect policy? This study may begin to clarify these issues. Other Factors Affecting Social Policy Although my interests focus on the impact of political centraliza- a? :35.) L... .. «sawmills»... tion on policy, any study attemptingEto shed light on the relationships I—W1<~ ...- 1W.“ --” ’“M ..- an! c‘-i~*‘ between these variables must consider the role of determinants other. Y" '1 i I'.V'( h. 'i 4 than political centralization. Earlier studies, alluded to above, have outlined several important social policy determinants or other factors influencing social equality. We cannot ignore the influence that the levelgof economic develop-t ment has on social policy output and outcomes. As mentioned previously, Cutright and Jackman discuss the major role that the level of economic development plays in producing social welfare and social equality or in allowing for them. Furthermore, Wilensky argues that economic growth (along with demographic and bureaucratic factors) is aproot cause of the general emergence of the welfare state. Economic development is -f--.—P-—-’"-‘I~ .- even more a factor than is the type of political system.103 But Wilensky also points out that after accounting for effects of economic growth levels on welfare state development, there is still some way to go in explaining policy outputs and outcomes. Musgrave also finds evidence that with rising per capita income, a country_becomes more interested in social security and other transfer systems. And, he argues, for a number of reasons, the Optimal level of fiscal\oentralization in a system will vary with the level of economic development.104 48 Koropeckyj and Hogan, in separate works, write that the effect of economic factors vary over time or among systems, depending on the level of economic deve10pment attained.105 Looking at socialist countries only, Koropeckyj finds that developed and underdevelOped countries show lower declines in inter-regional inequality than do countries at inter- mediate levels Of development. Hogan finds that economic variables are Ofil§§§ importance in determining policy_at middle-range levels of economic development (e.g., Canada) than they are in poor or affluent societies (e.g., Mexico and the U.S.). The authors of Values andpthe Acpive Communipy, too, find that the impact Of economic development in determining levels of social activity in the realm of social welfare varies among nations. They argue that in India, the resources of the locality have little bearing on active- ness, whereas in the U.S., economic resources play a large role in determining social activeness among elites and in policy outcomes.106 What does seem to be an important mariable to consider, they find, is the broad ideological commitment of national leaders and of citizens toward change and national development. In deciding what influence centralization has on policy then, the level of economic development must be considered.as an important 99p- textual variable. That is, level of development will_be controlled for so that I can investigate what role other variables play in affecting r 741:)».- . , policy. Social characteristics must be considered as well. As Sharkansky, / Wilensky, and Heisler and Peters make clear, social demographics such as the age structure of the pOpulation (the number of school age children, the number of people over age 65) and the geographical size of 49 the country must be taken into account. Moreover, Jackman discusses the impact of several societal characteristics such as the rate of growth of the population.107 Finally, taking into consideration the discussions of Walker, Jackman and Wilensky, social welfare gogram/pxfien’énce and political no 5 established in the past within the realm of social policy must be given recognition for the role they play. It should be expected that whatthings have been done in the past will affect policy outcomes of the present (incrementalism). * * * In this chapter I have taken care to outline and review the literature that addresses the questions of political structure central- ization, social policy, the heterogeneous society, ahdaso on. In doing so. I have pointed out the disagreement and dissensus among scholars on these issues and, more importantly, have broadly outlined areas where further study is necessary. Chapter II accomplishes two tasks. First, a summary of the hypo- theses (from the literature) to be investigated in this thesis is presented. Second, the data, research design and operationalization and measurement methods are discussed. CHAPTER I FOOTNOTES :1Almond, Gabriel and Bingham Powell, Jr., "A Developmental Approach to Political Systems: An Overview," in Jason Finkle and Richard Gable, editors, Political Development and Social Change, p. 51 ff. 2 Easton. David, "An Approach to the Analysis of Political Systems," World Politics, vol. 9 (April, 1957): 383-400. 3 See especially, Samuel Huntington, Political Order in Changing Societies, Yale University Press. New Haven, 198g. )4. 5 See Richard Dawson and James Robinson, "Inter-party Competition, Economic Variables, and Welfare Policies in the American States," Journal of Politics, vol. 2 (may, 1963); Thomas Dye, "The Independent Effect of Party Competition on Policy Outcomes in the American System," paper presented to the 1965 Annual Meeting of the American Political Science Association, Washington, D.C., September 8-11; Dye, Politics, Economics and the Public: Policy Outcomes in the American States, Rand MCNally & Co., Chicago, 1956: Ira Sharkansky and Richard Hofferbert, "Dimensions of State Politics, Economics and Public Policy," American Political Science Review (September, 1969): 867-879; Richard Hofferbert, "The Relation Between Public Policy and Some Structural and Environmental Variables in the American States," APSR (March, 1966): 73—82. 6 Maxwell, James and Richard Aaronson, Financing State and Local Governments, The Brookings Institute, Washington, D.C., 1977. p- 34- Ibid., p. 331+ ff. . 7 Pryor, Frederic, Public Expenditures in Communist and Capitalist Nations, Allen and Unwin Ltd., London, 1968. 8 Lane, David, The End of Ideology: Stratification Under Social- ism, Penguin Books, London, 1971, p. 132 ff. 9 Cutright, Phillips, "Inequality: A Cross National Analysis," APSR (August, 1967), vol. 32, p. 562; "Political Structure, Economic Development and National Social Security Programs," American Journal of Sociology (March, 1965), vol. 70, p. 537. 10 Jackman, Robert, Politics and Social Equality, John Wiley and Sons, New York, 1975; "Political Democracy and Social Equality: A Comparative Analysis," American Sociological Review (February, 1974), 50 51 vol. 39, p. 29; "On The Relation of Economic Development to Democratic Performance," American Journal of Political Science (August, 1973), V010179 P' 611. 11 Jennings, Edward Jr. ,"Party Constituency and Welfare Policy Change in American States," Paper presented at Midwest Political Science Convention, Pick-Congress Hotel, Chicago, 1978, pp. 22, 34. 12 Strouse, James and Philippe Jones, "Federal Aid: The Forgotten Variable in State Policy Research," Journal of Politics (February, 1979), vol. 36, p. 200 ff. 13 Sack, Seymour'and Robert Harris, "The Determinants of State and Local Government Expenditures and Intergovernmental Flows of Funds," National Tax Journal (March, 1964), vol. 17. P. 75. 14 Walker, Jack, "The Diffusion of Innovations Among the American States," in Sharkansky and Hofferbert,'eds., State and Urban Politics, Little, Brown and Co., Boston, 1971. pp. 386, 389, 390. 15 Cnudde, Charles and Donald McCrone, "Party Competition and Welfare Policies in the American States," APSR (September, 1969), vol. 63, pp. 858-866. ‘ 16 Fry, Brian and Richard Winters, "The Politics of Redistribu- tion," APSR (June, 1970), vol. 64, p. 508 ff. 17 Jacob, Herbert and Michael Lipsky, "Outputs, Structure and Power: An Assessment of Changes in the Study of State and Local Politics," in Sharkansky and Hofferbert, State and Urban Politics, Little, Brown and Co., Boston, 1971, p. 19. 18 Sharkansky, Ira, Spending in the American States, Rand McNally, Chicago, 1968, pp. 2,73. 19 Cowart, Andrew T., "Anti-Poverty Expenditures in the American States: A Comparative Analysis," Midwest Journal of Political Science (May. 1969). vo1 13. pp. 219-236 20 Hogan, J. B., "Social Structure and Public Policy," Comparative Politics (July, 1972), vol. u, pp. 477-509. 21 Peters, Guy, "Economic and Political Effects on the Development of Social Expenditures in France, Sweden and the United Kingdom," MJPS (May. 1972). vol 16 pp. 225-238 22 Peters, Guy, J. C. Doughtie and M. K. McCulloch, "Types of Democratic Systems and Types of Public Policy," Comparative Politics (April, 1977), vol. 9. p. 327. 23 Cameron, David and Richard Hofferbert, "The Impact of Federalism on Education Finance: A Comparative Analysis," European Journal of Political Science (September, 1974), vol. 2, pp. 228-229. 52 24 International Study of Values in Politics, Values and the Active Community, The Free Press, New York, 1971, Chapter 10. 25 Wilensky, Harold, The New Cgrporatism,_Centralization and the Welfare State, Sage Publications, Beverly Hills, 1973. 26 Boaden, Noel, Urban Policy Making: Influences on County Boroughs in England.and Wales, Cambridge University Press, Cambridge, 1971, p. 112. 27 Hibbs, Douglas, "Political Parties and Macroeconomic Policy," APSR (December, 1977), vol. 71, pp. 1467-87. 28 Jackman, Robert, Politics and Social quality, Chapter 3. 29 For further discussion on this topic, see also Ira Sharkansky, "Government Expenditures and Public Services in the American States," APSR (December, 1967), vol. 61, pp. 1066—1077, "Economic and Political Correlates of State Government Expenditures; General Tendencies and Deviant Cases," MJPS (May, 1967), vol. 11- Jacob and Lipsky, op cit.; Jackman, op. cit., ch. 2. 30 Heisler, Martin O. and B. Guy Peters, "Toward a Multidimensional Framework for the Analysis of Social Policy," The Annals (November, 1977), vol. 434, p. 67. 31 Cameron and Hofferbert, op. cit., p. 226. 32 Titmuss, Richard, Commitment to Welfare, Pantheon Books, New York, 1968, pp. 64, 65, 173. 33 Jackman, op. cit., p. 12. 34 Cutright, Phillips, "Income Redistribution: A Cross National Analysis," Social Forces (December, 1967), vol. 46, pp. 180-190; "Reply to Glenn," Social Forces (June, 1968), vol. L16, p. 539. 35 See, for example, Norval Glenn, "Social Security and Income Redistribution," Social Forces (June, 1968), vol. 116, p.- 538. 36 See, for example, various articles in Jason Finkle and Richard Gable, Political Development and Social Change, John Wiley and Sons, Inc., New York, 1971; Samuel Huntington, op. cit.; Irma Adelman and Cynt ia Morris, Economic Ggggth and_§gcial Equity in Developing Countries, Stanfbrd University Press, Stanford, 1973, ch. 4’and 5, Robert Rhodes, ed., Imperialism and Underdevelopment, Monthly Review Press, New York, 1970; Clive Y. Thomas, Dependence and Transformation, Monthly Review Press, New York, 1974. 37 Chenery, Hollis in Chenery, Ahluwalia, Bell, Duloy and Jolly, Redistribution with Growth, Oxford University Press, London, 1974, pp. xiii ff. 38 Jackman, op. cit., p. 4. 53 39 Jencks, Christopher, Ingguality, Basic Books Inc., New York, 1972. 40 Wilensky, Harold, The Nep_Copporatism..., op. cit., especially p. 10. 41 Weinberg, Carl, Education and Social Problems, The Free Press, New'York, 1971, pp. 4, 6e7. “2 Milner, Murray, The Illusion of Equality;_ The Effects of Edu- cation Opportunity, Inequalitygand Sogial Conflict, Josey-Bass, Inc., San Francisco, 1972, especially pp. 79-82. 43 Becker, Gary and Barry Chiswick, "The Economics of Education: Education and the Distribution of Earnings," American Economic Review (May. 1966). vol 56 P- 358 44 Tinbergen, Jan, "Tendencies and Determinants of Income Distri- bution in Western Countries - A Note," in Jagdish Bhagwati and R. Eckhaus, egs., Development and Planning, Allen and Unwin Ltd., London, 19729 P' 1 3' “5 Harrison, Russell, Equalipy in Public School Finance, Lexington Books, Lexington, Mass., 1976, p. 190 ff. Joseph, Lawrence B., "Normative Assumptions in Educational Policy Research, The Case of Jencks' Ingguality, " The Annals (November, 1977), vol. 434, p. 101 ff. 47 Cameron and Hofferbert, op. cit., pp. 230, 232. as ' ' ibis.- 49 Beer, Samuel, "A Political Scientist's View of Fiscal Federal- ism," in Wallace Oates, ed., The Political Economy of Fiscal Federalism, Lexington Books, Lexington, Mass., 1977, pp. 21—22. - 50 Pommerehne, Werner, "Quantitative ASpects of Federalism: A Study of Six Countries," in Wallace Oates, op. cit., p. 275 ff.; Panio Vieirra, Toward A Theopyiof Decentralization: A Cppparative View of FortyeFive Countries, unpublished dissertation, University of Southern California, 1967. Another multi-nation study of 58 countries by Oates, however, finds that there is an inverse relationship between federalism ' and fiscal centralization, i.e., federal nations are less fiscally centralized than other.nations and federalism and fiscal decentralization occur together. Findings of Pommerehne and Vieirra, therefore, do not stand unchallenged.- 51 Reagan, Michael, The New Oxford University Press, New York, 1972, p. 7. 52 For example, see Reagan, 0p. cit., pp. 7, 157; Cameron and Hofferbert, op. cit., p. 230; W. Oates, Fiscal Federalism, Harcourt, Brace, Jovanivich, Inc., 1972; D. Elazar, "Federalism," in International 54 Encyclppedia of the Social Sciences, vol. 5, New York, Macmillan, 1968, p. 353 ff.; William Riker, Federalism: Origini_0perationlpSignifi- cance, Little, Brown and Co., Boston, 1964; pp. 132, 133, 125, 129. 53 Levy, Marion, Modernization and tthStructgre of Societies, vol. II, Princeton, New Jersey, Princeton University Press, 1966, pp. 488, 766. 5“ Hage, Jerald and J. R. Hollingsworth, "The First Steps Toward the Integration of Social Theory and Social Policy," The Annals (November, 1977). vol. 434, pp. 9, 14, 18-19. 55 Wilensky, Harold, The Welfare State and Eqpality, University of California Press, Berkeley, 1975, pp. 52, 68. . - 56 Morrison, D. and H. M. Stevenson, "Integration and Instability: Patterns of African.Deve10pment," APSR (September, 1972), vol. 66, pp. 906, 920, 924. 57 Uphoff, Norman T. and Milton J. Esman, Local Opganization for Rural Develppment: Analysis of Asian Experience, The Rural Development Committee, Center for International Studies, Cornell University, Ithaca, New York, 1974, pp., x—xvii. 58 Pryor, Frederic, op. cit., pp. 44-47. 59 Peacock, Alan T. and Jack Wiseman, The Growth of Public Expen— ditures in the United_Kingdpm, Princeton University Press, 1961. 60 Pommerehne, op. cit., pp. 285, 304. 61 Oates, Fiscal Federalism, op. cit., pp. 223, 224, 240. 62 Moynihan, Daniel, "The Relationships of Federal to Local Authorities," Daedalus (Summer, 1967), vol. 96, p. 802. 63 Oates, op. cit., p. 210. 6“ Blondel, Jean, Introduction to Comparative_Politics, Weidenfeld, London, 1969, p. 294. 65 See note 51. See also: Thomas Franck, Why Federations Fail, New York University Press, New York, 1968; David Truman, "Federalism and the Party System," in A. Macmahon, ed., Federalism Mature and Emergent, Doubleday and Co., Inc., Garden City, New York, 1955; Franz Neumann, "Federalism and Freedom: A Critique," also in Macmahon; Edward McWhinney, Comparative Federalisp, University of Toronto Press, Toronto, 1976; U. K. Hicks, F. G. Carnell, J. R. Hicks, W. T. Newlyn and A. H. Birch, Federalism apgpgggnomic Growth in Underdevelpped Countries, New York, Oxford University Press, 1962; Ivo Duchacek, 99p: parative Federalism: The Territorial Dimension of Politigs, Holt, Rinehart and Winston, Inc., New York, 1970; Robert Inman, Martin McGuire, Wallace Oates, Jeffrey Pressman and Robert Reischauer, Financing the New Federalism, John HOpkins University Press, 1975; S. M. Lipset, 55 Political Man, Doubleday and Co., Garden City, New York, 1960. These are a few of the major readings on the concept. 66 See, Rufus Davis, "The 'Federal Principle' Reconsidered," in Aaron Wildavsky, Amgrican Federalism in Persppctive, Little, Brown and Co., Boston, 1967; Franz Neumann, op. cit., pp. 44-45, 54; Ivo Duchacek, op. cit., p. 194; Daniel Elazar, "Federalism," op. cit. 67 Oates, op. cit., p. 182. 68 Dunn, James A., "Consociational Democracy and Language Conflict: A Comparison of the Belgian and Swiss Experiences," Comparative Political Studies (April, 1972), vol. 5, pp. 3-40. 69 Article 42 of the Swiss Constitution, cited in Pommerehne, op. cit. 70 May, R. J. , Federalism fld Fiscal Adjustment, Oxford: The Claredon Press, 1969, pp. 114~123. 71 72 Smiley, Donald, Canada in Question: Federalism in the Seven- ties, McGraw-Hill, Ryerson, Toronto, 1976, p. 118. Cameron and Hofferbert, op. cit., p. 238. 73 Cameron and Hofferbert, op. cit. 74 cit. Article 107 (2) of Berman Basic Law, cited in Pommerehne, 0p. 75 Oates, op. cit. 76 77 78 79 Wilensky, op. cit., pp. 32-36, 52; For a similar claim concern- ing U.S. commitment to welfare or redistribution see: Frank Munger and Richard Fenno, National Politics and Federal Aid to Education, Syracuse University Press, 1962, pp. 91-967’171; Richard Wagner, The Fiscal 0r ization of Americap_Federalism, Markham Publishers, Chicago, 1971, p. 1%; International Study of Valuep in Politigp, op. cit., pp. 9—10. 80 Beer, Samuel, "Political Overload and.Federalism," Policy (Fall, 1977). vol- 10, no. 1. pp- 5. 9. 15- 81 my, Op- 0113., P0 660 Reagan, op. cit. Cameron and.Hofferbert, op. cit., pp. 233, 235-238. Duchacek, op. cit., p. 313. 82 Cameron and.Hofferbert, 0p. cit., pp. 230-234, 240; , "Social Structure Politics in Subnational Systems: A Comparison in Four'Nations," in Allan Kornberg, ed., 99mparative Legislative Systems, David McKay, 1973, p. 159. For'a discussion regarding the strain on the 56 U.S. federal system caused by interstate differences in services, etc., see: Richard Wagner, op. cit., pp. 22—23. 83 Echols, John Minor, Politicsprolicy_and Ehuality Under Communism and Democracy, unpublished dissertation, University of Michigan, 1976? , "Politics, Budgets and Regional Equality in Communist and Capitalist Systems," Comparatiye Political Studies (October, 1975), vol. 8, no. 3. 84 Walker, op. cit., pp. 383-84. 85 Riker, William, op. cit. 86 Rubin and Weinstein, title given, Praeger Publishers, New York, 1974, p. 181. 87 Blondel, Jean, op. cit., Chapter 9. 88 Fishman, Joshua, in Fishman, C. Ferguson and.J. DasGupta, eds., Langpgge Problems in Developing Nations, John Wiley and Sons, New York, 1968, p. 53 ff. 89 See, for example: Tamotsu Shibutani and Kian Kwan, Ethnic Stratification, Macmillan Company, London, 1965, p. 54 ff.; Donald Horowitz, "Cultural Movements and Ethnic Change," in The Annals (September, 1977). vol. 433, pp. 6-18; Clifford Geertz, "The Integra- tive Revolution," in Finkle and Gable, 0p. cit., pp. 655-669; D. Rothchild and M. Rogin article on Uganda in G. Carter, ed., National Unity and Regionalism, Cornell University Press, Ithaca, New York, 1966, p. 3 9 ff.; Crawford Young, The Politics of Cultural Pluralism, Univer- sity of Wisconsin Press, MadisonTi976, p. 20 ff. 9O Fishman, Joshua, Bilipgual Education, Newbury House Publications, Inc., Rowley, Mass., 1976, p. 8; see also Dunn, op. cit., pp. 32-33. 91 For a discussion of this phenomenon in one country see: Leo Depres, Cultural Pluralism and Nationalist Politics in British Guiana, Rand McNally, Chicago, 1967, especially p. 276 ff. See also, Crawford Young, op. cit., Chapter 1. 92 Tingsten, Herbert, The Problem of Democracy, The Bedminster Press, Totowa, N. J., 1965, pp. 117, 124-25. 93 Duchacek, op. cit., pp. 97, 354. 9“ McWhinney, 0p. cit., p. 406; Franck, 0p. cit., pp. x, 177, 180 ff.; Livingston, quoted in Reagan, 0p. cit., p. 25. 95 Melson, Robert and Howard Wolpe, "Modernization and the Politics of Communalism: A Theoretical Perspective," APSR (December, 1970), 96 Anderson, C., F. von der Mehden and C. Young, Issues of Political Develppmepp, Prentice-Hall, Inc., Englewood Cliffs, N.J., 1975, 57 pp. 62-66, 103-107. 97 Morrison and Stevenson, op. cit., p. 903 ff. 98 Ferry, W. H. , "The Case for a New Federalism," Saturday Review, June 15, 1968. PP- 114-17- 99 Geertz, Clifford, "The Integrative Revolution," in Geertz, ed. , Old Societies and New States, The Free Press, New York, 1963, pp. 103- 158. 100 Rabushka, A. and Kenneth Shepsle, Politics in Plural Societies, Charles E. Merrill Publishers, Columbus, Ohio, 1972, p. 213 ff. 101 May, Op. cit. , pp. 45-46. For a discussion of the fate of various groups under U.S. federalism see Inman et a1. , op. cit. , p. 18. 102 Wilensky, in Industrial Society and Social Welfare. p- xviii: , The Welfare State..., p. 53 ff.; , The New Cogporatism..., p. 35 ff., 7a 103 Wilensky, The WelfarLState... , pp. xiii, 25-29. 10“ Musgrave, Richard, Fiscal Syptgmg, Yale University Press, New Haven, Conn., 1969, pp. 155, 344. 105 Koropeckyj , I. S. , "Equalization of Regional Development in Socialist Countries: An Empirical Study," Economic Develggment and Cultural Change (October, 1972), vol. 21, p. 68 ff.; HOgan, op. cit., p. 509. 106 The International Study of Values in Politics, op. Cit- . P- 21- 107 Jackman, op. cit., pp. 47-49, Chapter 3. CHAPTER II THE RESEARCH: HYPOTHESES, METHODS, DESIGN AND DATA After reviewing the relevant literature, it becomes clear that scholars have forwarded conflicting hypotheses concerning the nature of the relationship between political structure centralization and social policy. To the extent that conflicting viers pervade the literature, political scientists have only limited knowledge of the effects of politics on policy. This research represents an attempt to test a series of popular (or*predominant) hypotheses about the impact of several types of centralization on health, education and social welfare policies using data from sixty-seven nations. In this section I will outline the major hypotheses to be tested. The first set of hypotheses pertain to the effects of various types of centralization on commitment to social welfare: H1 : The greater the central government's financial respon- sibility for social policy, the greater will be the funds expended in that policy area in the total system. Thus: H1a: Fiscal centralization leads to a larger public sector as a proportion of the whole economy. H1b: Fiscal centralization leads to greater allocations of funds to education in the total system. H10: Fiscal centralization leads to greater allocations of funds to health in the total system. H1d: Fiscal centralization leads to greater allocations of funds to social welfare in the total system. 58 59 These hypotheses grow out of the discussions of Levy, Hage and Hollingsworth, WilenSky, Pryor, Peacock and Wiseman and Pommerehne, all of whom consider TISCal centralization directly or indirectly. Their works make up the bulk of the literature on this aspect of centraliza- tion. suggesting that centreyzetion 1%.d.art0.-.stwdamimtiw._of.-pro- cedures, efficiency in implementation, uniformity of policy efforts throughout a system, and, ultimately, greater redistributive efforts. (Oates, of course, is agdighenting voice here in that he argues central- ization has no effect of the size of public spending.) It may be added further, that fiscalflcentralization might logically lead to the integration of advocacy groups (political parties, interest groups), national policy makers and bureaucrats and promote policy effort and commitment, whereas decentralization might lead to effort fragmentation and local resistance. The second set of hypotheses follows from the first: H2 : The greater the central government's financial respon- sibility for social policy, the greater the policy outcomes. Thus: H2a: Fiscal centralization leads to greater education policy impact (outcomes). H2b: Fiscal centralization leads to greater health policy impact. H2c: Fiscal centralization leads to greater social welfare policy impact. Here it is postulated that centralization, besides affecting policy outputs (expenditure levels, for example) affects the outcomes or the short - run impact of policy. As noted earlier, Jacob and Lipsky argue ,.— ‘a policy outcomes)~than on actual expenditure levels, which may simply 6O reflect social-economic factors. In considering the link between centralization and the outcomes of education, health and welfare policy I anticipate being able to draw some conclusions about the general effects of centralization on social well-being and social equality (Chapter VII). Iraniassuming that the outcomes of education, health and social welfare policies haveiimpiica- tions fgpmsogial well-being and social equality. The short-term impact of these policies affects social equality in that policy may either improve or harm the life chances of lower income groups. I Eontendywith Titmuss, Jackman and Cutright, moreover, that social welfare policies should_directly affect social equality. Another'area of concern is the effect of formal - legal centrali- zation (federal/unitary status) on policy outputs and outcomes. The literature suggests that federal or'unitary status has an independent effect on policy even after considering and controlling for other fac- tors that affect policy, such as economic development. The literature provides no clear consensus on just what effect federalism has on policy. Beer and Wilensky present opposing arguments on federalism and expendi- ture levels while many others who discuss centralization and policy imply that federalism has a conserwatizing nature. Therefore: H3 : Federal systems, by guaranteeing the existence of mul- tiple decision points with formal autonomy, allocate less money to social services in the total system than do unitary systems. H3a: Federalism leads to a smaller public sector as a pro- portion of the whole economy. H3b: Federalism leads to a smaller allocation of funds to education in the total system. H30: Federalism leads to a smaller allocation of funds to health in the total system. 61 H3d: Federalism leads to a smaller allocation of funds to social welfare in the total system. Also: H4 : Federal structure, because it invites policy variation among units, leads to uneven policy results (or short— run impact) and greater concentrations of policy out- puts in society. Federalism has a negative impact upon social equality. H4a: Federalism leads to a lesser impact of education policy on society. H4b: Federalism leads to a lesser impact of health policy on society. c dfiér‘f ' 3‘" H40: Federalism leads to a lesser impact of social welfare policy on society. "’ ‘*~.~a-ac Duchacek, Echols and Cameron and Hofferbert argue that federalism has direct negative consequences for distribution patterns and equali- zation attempts among subnational units. I will suggest here that there may be inherent in federalism a trade-off between local control and social equality and that local control arguments often are raised by groups interested in maintaining differentials and their own economic advantage. That is, equality may not be the most preferred goal of decision-makers or of those who influence them. Several writers have spenfilated on related ideas. For example, May states: If a simple generalization can be made about financial rela- tions in a federation it is that federations contain rela- tively poor units and relatively rich units and that the poor'units favour’a redistributive system of federal finance, and in order to achieve this generally support a central government with strong fiscal powers or’a centralized scheme of revenue allocation subject to periodic independent review, while the rich units favour fiscal decentralization and 62 revenue transfers, if they are necessary, based on units' relative contributions to the federal revenue.1 Jerome Rothenberg, in considering the best unit size for optimal governmental performance, writes that a maximization of redistributive efforts calls for centralized government, while concerns for "home rule" generate units too small for efficient public output production.2 Wilensky infers that local control stands in the way of social welfare program development by allowing for the expression of opinion of so many conflicting communal groups.3 Duchacek suggests that local control and equality are a "pair of desireables" which in reality are often conflicting.“ And Richard Hill points out that in urban and metro- politan governments, decentralization leads to inequality by maintain- ing and perpetuating class and status privilege, by yielding differential access to income and economic goods and services, and by reducing the financial resources of central cities.5 This literature, then, SUpports the notion that local control inherent in federalism diminishes efforts at promoting social equality through a redistribution of social goods, services and income. Moreover, Grant McConnell discusses the difference in results that may be obtained, given the same distribution of interests, depending upon whether the political context is centralized or decentralized with local control.6 To McConnell, local autonomy is a conservative prin- ciple that protects the power of the status elites and ignores needs of local minorities. The already weak are disadvantaged and unrepresented in policy choices in the decentralized setting. Moreover, decentrali- zation leads to a multiplicity of government agencies (with many becom- ing obscure) and a lack of accountability. In this situation, organized 63 interests and not the general public influence policy outcomes. There- fore, according to McConnell, centralization better'allows for meeting the need of the broad public: Policies generally adhering to maintenance of the status quo and favoring the concrete interests of existing elites will tend to be associated with organizations based on small units; alternatively, large units will more probably produce policies favoring change directed to the general, diffuse and widely shared interest of a broad segment of the popu— lation.7 Other scholars like Grodzins, Reagan and Weisbrod refute arguments typically made in defense of local autonomy -- that decentralization encourages creativity and innovation, that local government is closer to the people it rules and that decentralization reflects variations in preferences among units within a nation.8 Local control may not produce the benefits often ascribed to it in theory, and it may, in fact, be harmful to the "good of all" in the sense that it detracts from social equality. Therefore, given arguments in the literature we should ex- pect that: H5 : There is a trade-off relationship between federalism (and local control) and social equality. figural Het elfieflgl-fl Another set of questions revolve around issues of cultural hetero- geneity and consequences for social policy. Theorists tend to agree that heterogeneity leads to greater demands for social services and that demands spiral with communal competition, often outstripping govern- ment capability. But they argue over which is the most efficient 64 constitutional form in the situation where commitment to citizen well- being is complicated by ethnic jealousies. The hypothesis from the literature that can be tested is: H6 : Social heterogeneity has an independent negative effect ~99 social policy outputs and outcomes. What is not clear from the literature is how ethnic pluralism affects the relationship between political structure centralization and social policy. Again, there is little consensus of opinion expressed in the literature. Watts contends that in the multicultural federalism there will be pressures for equalization policies and redistribution, since the existence of differentials in the range and quality of public services sparks tensions between groups and regions.9 It has been argued, on the other'hand, that federalism invites policy variation since local groups rule themselves and have varying levels of resources at their'disposals, and that national standards and national programs are harder to implement in these circumstances, especially considering cleavages within society and the isolation of groups from one another. Moreover, some have argued that federalism may protect the privilege and status of richer, more powerful groups. Cameron and Hofferbert found that in federal systems social policy reinforces social dispari- ties (see earlier discussion). And.it seems reasonable that where heterogeneity has led to battles or animosities among communal groups, the various groups will be unwilling to give up some of their local autonomy to a national center which could then enforce equalization policies. Local ethnic groups may want to determine their own education, health and social welfare procedures and priorities, shunning attempts at uniformity. It may well be the case that pressures for equalization among groups 65 (or regions) will be greatest in systems where the various cultural groupings are not given autonomy or local control and where each group must influence the central government for social services and public goods. In this situation where groups do not have their own (local) governments, they must compete with everyone else for attention from the one main decision-making forum -- the central government. Moreover, the centralized government may more easily impose universal standards and enforce uniform policy procedures. Nor can it justify huge dis- parities in social services among groups or regions. Heterogeneity in the unitary system, then, may increase pressures for social services due to competition among groups. These opposing arguments will be examined in the analytical chapters to follow. Political Centralization Finally, I have argued that political centralization (among other things, the degree of vertical power distribution) might constitute a separate dimension of government centralization (along with financial and formal - legal centralization). A concentration of informal poli- tical power at the national center could logically lead to uniformity of opinion, or at least a minimum of opinion cleavage, and to greater policy consensus among decision makers. National commitments to social policy may be strengthened by the fact of elite c00peration. And, the literature suggests that where there is a centralization of authority, implementation and standardization are enhanced. Local authorities cannot successfully oppose the national policy or squandor resources. These arguments will be considered by testing the following hypotheses: H7 : In systems where informal political authority is con- centrated at the center, more funds are allocated to social policy in the total system. 66 Thus: H7a: Informal political authority concentration leads to a larger public sector as a proportion of the whole economy. H7b: Informal political authority concentration leads to greater education policy outputs. H70: Informal political authority concentration leads to greater health policy outputs. H7d: Informal political authority concentration leads to greater social welfare policy outputs. Also: H8 : Informal political authority concentration leads to more even policy results and lesser concentrations of social policy outputs in society. Social equal- ity is enhanced. H8a: Informal political authority concentration leads to greater education policy impact in the short run. H8b: Informal political authority concentration leads to greater health policy outcomes in the short run. H8c: Informal political authority concentration leads to greater social welfare policy outcomes in the short run. Tests of the aforegoing hypotheses are presented in Chapters IV to VII. Data, Design and Methods In order to proceed to test hypotheses, the concepts and variables to be measured empirically must be operationalized. This is no simple matter. The data—gathering stage of this study has been arduous, but illuminating. Yet, although in some areas I am limited by data avail- ability (such as in measures of fiscal centralization, and figures on governmental expenditures) there are great possibilities for the study of social policy and policy impact. In this section I will discuss the data sources utilized in the study. 67 Meaggring Government Centralization - The Independent Variables The causal variables of major importance to this study are fiscal centralization, federal/unitary status and what I shall term informal political authority concentration. I have argued that each of these represents a different aspect or dimension of government authority and the degree of concentration of that authority at the national center. Arguments from the literature that suggest and justify this thrgge dimensional consideration of government authority_and its concentration have been presented earlier. In Chapter III I give a more detailed discussion of the scope and measurement of centralization. Here I will briefly discuss what data are available for use in operationalizing the independent variables. Fiscal Centralization To measure the degree of fiscal centralization of a nation informa- tion on revenues and expenditures must be relied upon, since that is all that is available for a large number of countries. These data for 1968, 1969, or 1970 are available from the International Bank for Reconstruction and Development (IBRD) in its publication, World Tables 122; (revenue data) and World.Tables 1276 (expenditure data). For pur- poses of this study, I need a measure of the amount of independent fiscal decision-making power in the granting of public services at various levels of government within the nation. A number of scholars have argued that the greater the central government's financial responsibility for social policy, the greater the funds expended in that policy area in the total system. As Wallace Oates and others interested in financial centralization have argued, there are four different ways to operationalize fiscal 68 centralization given the data available from international, comparable sources such as the IBRD.10 The following may be considered: 1. The percentage of total public revenues collected by the cen- tral government (these data are available for 59 nations for 1968 or 1969): and 2. The percentage share of the central government in current public expenditures (these data are available for 63 nations for 1970; the figure attributes grant funds to the share of decentralized levels of government but includes expenditures by all central government departments and agencies). One possible problem with these two figures is that, for some nations, in their computation, contributions and expenditures for social security programs were excluded from the share of the central government while they were included in totals for the public sector. Therefore, these figures may actually underestimate the degree of fiscal central- ization. Further, the annual Yearbook of National Accounts Statistics pub- lished by the United Nations allows a consideration of two other pOSsible indicators for fiscal centralization: 3. Central government consumption expenditures, or, the percent share of the central government in all public consumption expenditures (available for 41 nations and includes wages of public employees and current purchases of goods and services from private enterprise, includ- ing defense expenditures); and #. The percent share of the central government in civil consump- tion expenditures (all expenditures for non-defense purposes; available for 29 nations). 69 This last indicator seems most appropriate to this research, given my focus on health, education and welfare policies. Moreover, defense figures are a real potential source of internation variation in expendi- tures, in that some nations spend much more on defense than others. However, these data are available for only twenty-nine nations and therefore may limit the consideration of fiscal centralization. The first indicator may not completely reflect fiscal centralization in that, in some nations, revenues are regularly collected by the center and then given over to subnational or administrative units to be spent for purposes the latter decide upon. Therefore, in an attempt to establish the relative significance of these four measures as indicators of fiscal centralization and to clarify how the data may best be used, I shall proceed by considering the interbrelationships among these indicators. This is done in Chapter III. It is interesting to note here that Oates found similar results in measuring fiscal centralization using each of the four indicators (singly) for a sample of fifty nations, while Break found little difference in results when using the central government share in total revenues and the central government share in total ex- penditures.11 Constitutional Status - FederalZUnitary I have argued in Chapter I that we should expect constitutional status -- federal or unitary -- and resulting intergovernmental rela- tions to have an effect on social policy. This is a second dimension of government authority concentration. The variable, obviously, can have only one of two "values" and in order to rank a nation as federal or unitary national constitutions can be consulted. Here I use Ivo Duchacek's discussion and classification of nations.12 Of the 7O sixty-seven nations in this study, Duchacek lists the following as federal: Argentina, Australia, Austria, Brazil, Burma, Cameroon, Canada, West Germany, India, Libya, Malaysia, Nigeria, Pakistan, Switzerland, the United States and Venezuela. Of these Duchacek indi- cates uncertainty about the status of the federal structure in Burma and Libya, given political turmoil in these states in the 1960's. After reading case studies and other country source materials, moreover, I decided not to classify these nations as federal for the purpose of this study. Although the Burmese post-independence constitution es- tablished a federal nation (where states had the right to secede), a military coup in 1962 effectively replaced federalism with dictatorial powers in the hands of a Revolutionary Council under General Ne Win.13 Similarly, in Libya the federal divisions of power (between three pro- vincial units and the central government) were abolished in 1963 by means of’a constitutional change made by the Cabinet. This move to stabilize a badly fragmented political system failed, however, and in 1969 Muammar Qaddafy and his cohorts seized power, cancelled the con- stitution, and established a highly centralized military regime.” Neither Burma nor Libya, then, were federal states after 1962 or 1963. A more thorough examination of federal structure is given in Chapter III. ‘Infgrmal Political Authority Concentration To operationalize the concept informal political authority con- centration, indicators that demonstrate the degree of informal power distribution and opinion cleavage within a system must be found. At the same time it seems prudent to avoid stickier problems such as debates over definitions and measurements of the term "power." One simpler way 71 to indicate the degree of informal political power centralization in a system would be to ask questions about its party and legislative systems, since these institutions presumably function to aggregate interests, represent various blocs of opinions, choose among alternate policies, play'rule-making roles, and vary in their degree of centralization or hierarchical structure. Moreover, by looking at the number of viable parties in a system, party competition and the nature of legislative representation, it may be possible to "measure" the extent to which power to rule or initiate is informally shared among groups and thereby gain some insight into the process of opinion representation. In this context it would seem_aggfggxiaieato,gather the following kinds of data: party nomination procedures and powers, the working organization of the parties, power relationships and influence between central and peripheral party organization, the timing of elections (whether local and central elections are held concurrently), and on party ideology. These kinds of data, however, are not available for more than a few countries. Data are available, though, on several other aspects of party and legislative structure that point to the degree of informal political authority centralization. I have collected data on several aspects of national party and legislative structures. Using Jean Blondel's classifications, I have noted (for 1968) the party system and party gtrength (based upon the percentage of votes received) for each of sixty-seven nations.15 A country may have no parties, a full single—party system, a forced coalition of several parties under the aegis of one dominant part, a dominant single-party system, a two party system (where two parties get over 90% of the vote), a two and a half party system (where two parties 72 get 75-80% of the votes), a multiparty system with a dominant party, or a multiparty system without a dominant party. Characterizing party systems by the number of parties (to yield discrete categories) simplifies the process of studying parties and in- dicating degrees of informal political centralization and hierarchical control. Theoretically, political phenomena are continuous in nature. That is, differences among party systems are numerous, ranging from the nature of their support (whether they are mass parties, imposed organ- izations, traditional parties, etc.) to internal structure (variables such as internal democracy, leadership, centralization) to party ideology.16 Yet, the number of parties in the system is a variable commonly used.and it does seem to broadly reflect the degree of party centralization and strength. For preliminary purposes, I have noted the number of parties in each system according to Blondel's classifications. For'purposes of this research, I will consider three classifications based on intuitive judgements about degree of party centralization and ability to impose the will of the party upon the system. I will consider no~party and single-party systems, two and two and a half party systems, and multiparty systems. Second, I have computed the percentage of seats held by the major party in the lower house of the legislature. This figure is based upon election and legislative data recorded in The Europa Yearbook 1971: A Worldggurvey, volumes I and II for one national election held usually between 1968 and 1970.17 These data reflect the degree of opinion cleavage present in the rule-making body, but also indicate how concen- trated legislative power is, i.e., if one party holds 80% of the seats in the body, it may be expected that any dissenting opinion can be 73 over-ruled and that agendas are controlled and priorities are established by the major’party. If, on the other hand, the major party holds about 50% of the seats, fewer possibilities exist for easy predominance. The major party will have to negotiate and persuade. Third, I have used Blondel's 1968 data on the right of dissolution of the parliament by the executive. Blondel reports, for'a large sample of nations, whether the executive has all powers, limited powers or no power to dissolve the chamber. This information indicates the degree of power concentration at the center -- in this instance, the degree of power-sharing between executive and chamber. Several other kinds of data have been collected as well. I have recorded for each nation the method of selection of persons to the upper legislative house: whether they are appointed, inherit the post, or are elected by constituencies (Blondel, 1968). This information tells us something about the concentration of power: if members are elected by national constituencies and are thus accountable to them, power is more diffused in the system than if members are appointed at the center. Finally, I have included some data from Textor and Banks' A Cross-Policy Survey reflecting subjective judgements on the part of the scholars on the degree of electoralpgompetition and the degree of interest articula- tign by associational groups (or organized interest groups) all for 1963.18 Ideally, it would be preferable to avoid "subjective" data of this nature and instead rely on "hard" data -- things that can be counted or directly observed. However, at this time the nature of avail- able measures and the unavailability of information prevent the gather- ing of "hard" data. At times, honest judgements of trusted scholars must be relied upon instead of postponing study until such "hard" data 74 is within reach. (See Chapter III for a discussion of how the foregoing data are finally employed.) Commitment to Social Policy;agd_Poligngutcomgs The dependent variables in this study are the levels of policy out- puts and outcomes in the policy areas of education, health and welfare. A few remarks about the terms outputs and outcomes are appropriate at this point. Following the usage established by Levy, Meltsner and Nildavsky,19 I use outputs to refer to what governments do and the goods and services produced by government organizations. Examples of outputs are expenditures, taxation, regulation, court decisions, actual goods and services given, et cetera. Outcomes, on the other hand, refer to s..- the effects/6f;policiesfinthe intended audience in the short run. WM'W‘W ' W Outcomes are what citizens see and they therefore inVOIVe evaluation according such things as citizen's normative judgements, and, often involve questions about the distribution of benefits. What this re- search cannot address is pglicy result, since results refer to ultimate, long range policy impact and final consequences. A study of policy results would require data coverning a long time span and would involve time series analyses. Policygcommitment as operationalized in this study, then, is indi- cated by expenditure levels and by actual service levels. In the educa- tion sphere, expenditures and such things as numbers of students served indicate commitment or output. Outcomes are indicated by literacy rates and educational attainment. Outcomes of social welfare policies, on the other hand, are indicated by the distribution of income and material goods within society. The consideration of outcomes of social policies is important in 75 that it allows initial answers to questions about the effects of govern- ment structure centralization on policy impact and short—term results, daily levels of living of citizens and social equality. Moreover, the research should shed light on the process of translation from policy output to policy outcomes. Below, I will discuss the data used to study policy output and outcomes for each social policy area. Education The data on expenditures for education are usually for 1970 and were collected from the United Nations Statistical Yearbook 1973.20 According to the source, data on public expenditures on education include recurring expenditures, teachers salaries, capital expenditures, and, where applicable, subsidized private education. The data reflect monies expended for education at every level of government. Although several publications give figures on education expenditures, the Statistical Yearbook presents them in the most consistent manner for the largest sample of nations. Several other variables indicate the way the allocated money was spent and to what effect. First, I have collected data on the number of students and of teachers at each level of education -- pre-primary and primary, second and third levels -- for 1970 so that I may compute student-to-teacher ratios over the whole school-goer population and for each level of education. They also allow me to compute what percen- tages of the population are at the various levels of instruction. These figures were collected from UNESCO's Statigtical Yearbookg197g (not the same publication as above).21 According to the compilers, the figures are those reported by each nation in response to UNESCO inquiries. Second, using the same source, I have recorded the gross enrollment 76 ratio for each nation. These data, expressed as a percentage, give the total enrollment of all ages divided by the population of the specific age group which corresponds to the ages of primary and secondary schooling. That is, they indicate the percentage of school-age children in school, taking into account the varying national school systems and age requirements. The data indicate how effectively educa- tion policy outputs have been in providing an education to young citi- zens. Third, data on the average number of years of schooling offered at the first and second levels for 1970 were collected from the IBRD‘s World Tables 1926.22 This figure varies among countries because duration of schooling varies from country to country and according to the level of instruction. And finally, the World Tables were also a source for the 1970 adult literacy rate. In the few cases where the Tables had no figures for a country, I used Taylor and Hudson's World Handbook of Political and Social Indicators figures for 1965.23 "Adult" here gener- ally refers to persons 15 years of age and older. Both years of school- ing offered and literacy rate indicate results obtained from money expended in the area of education. dearth Figures on health expenditures are not as readily available as those for education. The source of these data is the World Tables 1976 from which I collected current government health expenditure data for fifty-eight nations.24 Compilers of the Egblgg define current health expenditures as including purchases of goods and services by the central government for health, hospitals, population control and the like. Although indicators of the general health of the population, such 77 as deaths from communicable diseases, are not uniformly available for a large sample of nations, we can get a picture of the outcomes of health policy outputs by referring to other indicators. Using the WorldfiTgbles 1976 and the UNESCO Statistical Yearbook, I have collected data for 1970 on the infant mortality rate per thousand live births, life expectancy at birth, the number of citizens for every doctor, nurse, and hospital bed in the nation, and, data on nutrition. Nutrition data consist of two figures: calorie supply per capita, given as a percentage of daily requirements, and protein supply (total grams) per capita per diem and indicate the accessibility of good nutrition sources to the population. Social Welfare I am interested in studying countries' efforts to improve the lot of the "have-nots;" the disadvantaged, the unemployed, the sick or disabled and, generally, lower income groups. By considering social security program expenditures -- which typically establish a system of risk-sharing, provide for the needy and affect the distribution of resources with society -- I hope to be able to indicate national commit- ment to providing for the well- -being of all strata of society. The International Labour Office publication, The Cost of Social Securit , provides data on social security expenditures.25 These figures include monies for medical care and benefits in cash and in kind, such as family allowances, aid to the disabled, and pensions. These data, ‘usually for 1970, are available for fifty-two countries. I have also recorded social security expenditures per capita and as a percentage of gross domestic product in purchaser's value, 1970, from this source. These figures reflect a nation's level of commitment to providing an 78 adequate level of living for all citizens. Next, indicators of social welfare policy outcomes are needed. As I have argued in Chapter I, social security programs -— whether or not they were intended to -- do affect the distribution of income and, therefore, the command over resources within society. Income_is_trans- ferred_throughsocial_welfare programs to poorer groups in a system.§éj' H_/ (Social welfare policy, then, also contributes to the level of social equality, or, the distribution of the consumption of and access to material goods in society. This will be more fully discussed in]. Chapter VII.) In order to indicate social welfare policy outcomes (or, the immediate impact of welfare policy on the intended audience) I will use data reflecting income distribution. The data -- on the income shares of the lowest forty percent, the middle forty percent, and the top twenty percent of households -- indicate relative inequality by com- paring income shares of groups of individuals (households) to their population share. The data allow an investigation of the immediate effectiveness of social welfare policies aimed, basically, at poorer, disadvantaged groups in society. Does centralization affect the income share, for example, of the lowest forty percent of households? The income data come from Chenery, et al., and are available for forty-four of the nations included in this study.27 (They indicate the distribution of income before taxes.) These data on income shares are sensitive to inequality in particular ranges and therefore allow a consideration of income shares at the lower income range. As Ahluwalia points out, this may be of special interest to policy makers who must consider varying needs of different policy target groups.28 79 (Other indices of inequality, on the other hand, such as the Gini coefficient, summarize measures of inequality over entire populations.) Distribution of income patterns within societies, moreover, reveal much about social equality. That is, income determines one's command over resources and one's ability to attain a certain level of living. Due to the nature of income data in general, however, the process of generalizing and drawing conclusions about the impact of political structure on welfare policy outcomes, and later, on social equality will be limited. This is so because national income data are not strictly comparable across systems.29 Nations differ in aggregating and report- ing procedures, i.e., they may gather information on individuals, households or economic sectors and they may report figures reflecting distributions before or after taxes. Further, whereas some nations may report only money or wage income, others may report wages and income from other sources such as investments, property, transfers and other goods and services received. Where only wage income is reported, the relative position of those in entreprenurial and upper classes may be underestimated, thereby reducing the observed level of inequality. Finally, the data used here, aggregated by households wherever p0ssible, reflects the assumption that income within a household unit is dis- tributed equally, and it does not allow the researcher to control for household size and age structure, two variables that affect the distri- butional results of income. Yet, the available data may be useful in highlighting trends in distribution patterns and in levels of equality of access to goods. Income data, such as they are, present social scientists with some in- sight into problems of unmet needs and policy effectiveness. 80 Other Variables Ipportant to the Research Social Heterogeneity In Chapter I I discussed the importance of considering the effect of social heterogeneity or ethnic pluralism (as another independent variable) on relationships between types of centralization and social policy commitment and outcomes. However, since there are many different definitions of "ethnic groups" and given that social scientists do not generally agree upon which criteria to use in defining ethnic groups, operationalizing and measuring ethnic pluralism is not a straightforward task.30 Again, I am also limited by data availability. I will measure social heterogeneity using two types of information. First, Marie Haug has constructed a "pluralism" index based on Textor and Banks' data indicating language, race and religious heterogeneity, sectionalism feelings among groups within the nation and interest articulation by non-associational groups (informal ethnic, religious or similar groups).31 This index uses a simple arithmetic summation of scores on these variables, allowing the researcher to estimate national pluralism as negligible, moderate, marked or extreme. According to Haug, the language, race and religion variables reflect cultural dif- ferences among groups, while the sectionalism variable may reflect regionalism or communalism. The interest articulation item, on the other hand, indicates the political saliency of ethnic (or other) groups. These data were recorded for sixty-seven nations. Second, I have used The World Factbook 1974 to record the percen- tage of the population in the largest ethnic group.32 These data indicate the presence or absence of a dominant ethnic group (i.e., a low score on this variable indicates an ethnically plural nation, or, 81 the absence of a dominant ethnic unit). In a recent study of African nations, Jackman argues that ethnic dominance rather than gphpig heterggeneity may be the factor to consider when studying the effects of ethnic politics.33 When one ethnic group has numerical strength enough to impose its will, its politics may become tyrannical and elicit destabiliZing reactions from other groups. Conflicts over allocations of public goods may increase. Competing power centers in an ethnically plural society, on the other hand, may produce more moderate politics and an atmosphere of compromise. Economic Dagtg Besides the data outlined thus far, several other types of data will be employed. First, various economic data will be used to charac- terize the nature of the systems included in this research. In order to measure the "size of the public sector" I will use figures on tax revenue as a percent of national income. These data for 1970 are listed for most countries in the IBRD World Tables 1976, referred to earlier. The Igpl§§_also report gross domestic product for 1970 in United States dollars. To indicate a country's level of economic development, industrial- ization and national affluence, another important characteristic of nations, either of two types of data may be used: gross national pro- duct per capita (I have data from the World Tables 1921 and 1276 for GNP per capita, 1969 and 1973) and electric power consumption per capita, 1970 (World Tables, 1976). These two indicators have been shown in previous studies to be highly intercorrelated and empirically equal. Finally, since all expenditure figures are given in national currency units, I have used the World Tables 1976 to record the exchange 82 rates. This figure allows me to translate all expenditure figures into 1970 United States dollars for easy comparison. (Where this source gave no exchange rate, I used the National Basic Intelligence Factbook, published by the Central Intelligence Agency.34 This source gives ex- change rates for 1974 and 1975 and was used for ten nations out of sixty-seven in this study.) Demographic Variables Information about national populations, such as population in millions, population density, population growth rates and population age structure (ages 0-14, 15-64, and 65 and over) were collected from the Tables 1221 and 1226. These data become relevant at several points in the research. The Sample and the Level and Methods of Study The Sample An attempt has been made to include as many nations as possible in this research, but this effort has been limited by data availability. Three groups of data, particularly, have limited the number of cases to be considered here: data on fiscal centralization, data on social security expenditures and data on income distribution and income shares. For a majority of the variables, however, I have data for sixty-seven nations. These countries include ten Southeast Asian nations, eighteen Latin American nations, nineteen African and Middle Eastern nations, and twenty Atlantic area or Western European nations. These countries are listed by geographical area on the following page. The sample, then, represents nations from all parts of the globe and at various levels of * * §t§t§t§§ Sh Sh‘h St §t§h *- COUNTRIES INCLUDED IN THE STUDY, BY GEOGRAPHICAL AREA AFRICA AND MIDDLE EAST Botswana Cameroon Chad Israel Ivory Coast Libya Malagasy Republic Mali Mauritius Nigeria Rhodesia Senegal Sierre Leone South Africa Sudan Swaziland Tunisia Turkey Zambia LATIN AMERICA Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Republic Ecuador Guatemala Honduras Jamaica Nicaragua Panama Paraguay Peru Trinidad & Tobago Uruguay Venezuela * federal nations Stqt §h¥h§h§t§t ** ATLANTIC AREA Australia Austria Belgium Canada Finland France Germany, F. R. Greece Iceland Ireland Italy Luxembourg Netherlands New Zealand Norway Portugal Sweden Suitzerland United Kingdom United States SOUTH AND EAST ASIA Burma India Japan Korea , Republic Malaysia Pakistan Philippines Sri Lanka Thailand South Vietnam # marked or extreme pluralism (Haug's Index) 84 social and political change. It includes many regime-types. There are obvious gaps here, however. The Communist nations are not represented in this research, for several reasons. First, much of the data I have collected for sixty-seven nations are not available in comparable form for the Communist states. Second, accounting practices vary between Communist and other nations. Often, annual expenditure data are not available for the Communist states and expenditure cate- gories used by these nations are significantly different than those used by the non-Communist states. Figures from Five and Ten Year Plans which set goals to be reached, moreover, cannot be sutstituted for expenditure data. Also, it would have been desirable to include a greater.number of Third World nations in this research, given the data, so that the sample could better reflect the great variations in national structure and political experience that exist among world's nations. This research, and generalizations made from it, therefore, are recog- nizably somewhat limited in applicability at this time. The Problem of Missing Data In choosing indicators for the variables and concepts of this research I followed two criteria: indicators shOUld be theoretically relevant to the concept, and, should be available for a large sample of nations. In spite of this attempt to gather data for a large sample of nations, however, data are limited to some extent in areas of policy outputs and outcomes. For example, although for most variables data are available for all of the sixty-seven nations in the study,35 health expenditure data were available for only fifty-eight nations, education expenditure data were available for sixty-one nations, social security expenditure data existed for fifty-two nations and income distribution 85 data were available for forty-four nations. Gurr suggests several remedies for missing data: missing cases can be discarded, the mean value of the indicator can be used where that figure is missing for a country, interpolation can be used especially in time-series analyses, or values may be estimated using contextual information, information from relevant literature or cases similar to the missing case.36 Rather than inserting means, interpolating or estimating values (which may introduce more error into the sample), however, I have chosen to simply vary the number of cases (N) in the analysis depending upon the immediate topic of research. In other words, analyses involving social security expenditures will be based upon a sample of fifty-two nations, whereas analyses of the impact of structure on income distri- bution patterns will be based upon data from forty-four.nations (through use of the pair-wise deletion options). This method allows for a con- sideration of the hypotheses with the available data (even the smallest N, 44, represents a reasonable sample of the world's nations) and does not introduce greater error into the research. The Level of Study This research poses questions at the aggregate or systems level. It relies on national-level data. I have hypothesized, moreover, that variations among nations in degree of governmental centralization will have an important, measureable impact upon public policy. And, I expect that relationships will be observable at the system level. The nation- state is an important social unit about which we can still learn a great deal. Several scholars, however, have outlined possible problems inherent 86 in national level studies using aggregate data: the accuracy of aggre- gate data may be limited or faulty, the degree of comparability of measurements may be diminished -- especially when units provide infor— mation by processes of self-enumeration, and, aggregate data may not indicate much about real properties of the collective unit.37 For purposes of this research, I have attempted to minimize these problems. Most of the data, for instance, have been taken from United Nations' sources which have been carefully compiled in an attempt to maximize comparability. Moreover, the intent of the research is to identify national level trends in the data rather than to characterize or repre- sent properties of the collective based upon aggregate data or averages. It may well be that social change and policy change can more clearly be seen at the intra-national level, as Merritt and Rokkan suggest.38 Mbreover, there are problems involved in generalizing about countries which are internally vast and diverse. Ideally, social units studied should coincide with units perceived as "real" by inhabitants.39 However, subnational data are simply not available at this time. Nations and data-gathering organizations such as the U.N. or IBRD have not gone far beyond making national-level data available. * * *- The hypotheses presented earlier will be tested using regression analysis and least squares estimation techniques. These analyses and their results will be presented in Chapters IV, V and VI. Chapter III presents a discussion of the principal independent variables and their measurement. CHAPTER II FOOTNOTES 1 May, R. J., Federalism and Fiscal Adjustment, Oxford, The Clare- don Press, 1969, pp. 161-62. 2 Rothenberg, Jerome, "Local Decentralization and the Theory of Optimal Government," in Julius Margolis, ed., The Analysis of Public Output, National Bureau of Economic Research, New York, Columbia University Press, 1970, pp. 33-35. 3 Wilensky, Harold and C. N. Lebeaux, Industrial Societyiand Social Welfare, The Free Press, Glencoe, Illinois, 1965, p. xxii. 4 Duchacek, Ivo, Comparative Federalism, Holt, Rinehart and Winston, New York, 1970, p. 348. 5 Hill, Richard, "Separate and Unequal: Governmental Inequality," APSR (December, 1974), p. 155? ff. 6 McConnell, Grant, Private Power and American Democragy, Vintage Books, New York, 1966. 7 Ibid., p. 114. For arguments in defense of local control see: Oates, Fiscal Federalism, Harcourt, Brace, Jovanovich, Inc., 1972, pp. 11, 13; Richard M. Nixon quoted in Robert Inman, et. al., Financing the New Federalism, John H0pkins University Press, 1975; Jeffrey Pressman in Inman, et al., above; Walter Hiller, Neg_Dimensions of Political Economy, Harvard University Press, 1966, p. 168. For a refutation of these argu- ments see: Michael Reagan, The New Federalism, Oxford University Press, New York, 1972, pp. 118-119; Morton Grodzins in Daniel Elazar, ed., The American System, Rand MCNally, Chicago, 1966, pp. 210-211; Burton Weisbrod, External Benefits of Public Education, Princeton, N.J., 1964, p. 117, argues that most important issues are of nation-wide concern and local decision making becomes less isolated as nations unify. 9 Watts, R., Multicultural Societies and Federalism, Studies of the Royal Commission on Bilingualism and Biculturalism, no. 8, Ottawa: The Queen's Printer, 1970, pp. 40-46. 10 Oates, Wallace, Fiscal Federalism, op. cit., p. 196. 87 88 11 Break, George, Intergovernmental Fiscal Relations in the United States, Brookings Institute, 1967. 12 Duchacek, Ivo, op. cit. 13 See Josef Silverstein, Burma: Militdpy Rule and the Politics of Stdgnation, Cornell University Press, Ithaca, 1977. 14 See Omar El Fathaly, MOnte Palmer and Richard Chakerian, Political DevelOpment and Bureaucracydin Libyd, D. C. Heath and Co., Lexington, Mass., 1977: Colin Legum and John Drysdale, African Contem- mrarLRecord, Africa Research Limited, Exeter, United Kingdom, 1970. 15 Blondel, Jean, An Introduction to Comparative Government, Weidenfeld and Nicolson, London, 1969. 16 Ibid., chapters 9 and 10. 17 The Europa Yearbook 1971: A World Survey, Europa Publications Limited, London, 1971. 18 Textor, Robert and Arthur Banks, A Cross—Policy Survey, Cambridge, Mass., Institute of Technology, 1963. i9 Levy, Frank 8., Arnold J- Meltsner and Aaron Wildavsky, Urban Outcomes, University of California Press, Berkeley, 1974, pp. 1-23. 20 United Nations, Statistical Yearbook 1973, New York, 1974. 21 United Nations U.N.E.S.C.O., Statistical Yearbook d972, Louvain, Belgium, 1974. 22 International Bank for Reconstruction and Development, World Tables 1976. 23 Taylor, Charles and Michael Hudson, World Handbook of Political and Social Indicators, Second Edition, Yale University Press, New Haven, 1972- 24 World poles,1976, op. cit., Economic Data Sheet 2. 25 International Labour Office, The Cost of Social Security, Geneva, 1972, 1976. 26 See Chapter One. 27 Chenery, Hollis, Montek Ahluwalia, C. L. C. Bell, John Duloy and Richard Jolly, Redistribution with Growth, Oxford University Press, London, 1974. 28 Ibid., p. 7. 29 See discussions about the problems of income data in Malcolm Sawyer, "Income Distribution in OECD Countries," OECD Occasional 89 Studies, July, 1976, pp. 3-36; Chenery, et al., 0p. cit., chapters 1, 2. 30 See the discussions of this topic in Robert W. Jackman, "The Predictability of Coups D' Etat, A Model with African Data," American Political Science Review (December, 1978), no. u, p. 1262. 31 Haug, Marie, "Social and Cultural Pluralism as a Concept in Social System Analysis," The American Journal of Sociology (November, 1967), pp. 294-3ou. 32 The World Factbook, Publishing Sciences Group, Inc., Acton, Mass., 1974. 33 Jackman, Op. cit. 3“ Central Intelligence Agency, National Basic Intelligence Fact- book, Office of Geographic and Cartographic Research. 35 The number and variety of nations to be included in the research was initially dictated by the availability of data on fiscal centrali- zation. 36 Gurr, Ted R., Politimetrics, Prentice-Hall, Inc., Englewood Cliffs, N.J., 1972. 37 See Erwin Scheuch, "Cross National Comparisons Using Aggregate Data: Some Substantive and Methodological Problems," in Richard Merritt and Stein Rokkan, eds., Comparing Nations, Yale University Press, New Haven, 1966. 38 Merritt and Rokkan, op. cit., pp. 268-271. 39 Linz de Miguel, in Merritt and Rokkan, op. cit., p. 275. CHAPTER III POLITICAL STRUCTURE CENTRALIZATION: ITS SCOPE AND MEASUREMENT A major assumption underlying this research is that there are several separate dimensions making up what is more generally referred to as political structure centralization or government authority con- centration, and further, that the various dimensions or types of government centralization specified in Chapters I and II may have differing effects on social policy. Are there in fact empirical grounds for dissecting the concept government centralization into several dis- tinct components? Or, as is often assumed in the political science literature, do constitutional, economic and political aspects of cen- tralization or concentration coincide? In this chapter, the claim that there are several separate aspects to the degree of regime centralization is tested ‘using data from sixty-seven nations. At the same time, I deve10p measures for my prin- cipal independent variables: fiscal centralization, formal-legal status, and informal political authority concentration. Fedemd/Unitary Status: Formal-Legal Structure Centralization In Chapter II, I have discussed the classification procedures Used to designate the nations included in this research as federal or unitary. The "measurement" of this aspect of political structure centralization is straightforward. That is, following Duchacek and Elazar, we may rely on constitutional stipulations of the structure of intergovernmental 9O 91 relations (when the constitution is, in fact, in force). The following nations in my sample are federal: Argentina, Australia, Austria, Brazil, Cameroon, Canada, West Germany, India, Malaysia, Nigeria, Pakistan, Switzerland, the United States and Venezuela. According to Duchacek, some basic characteristics and operating principles of federal systems are: 1) the constitutional stipulation of federal structure, 2) a clear division of significant powers among substantially self-sustaining centers, each with independent rule-making power, and 3) an areal division of power.1 Moreover, Duchacek and Elazar distinguish between federalism or non-centralization and decen- traligatiomfdmrundtary states.2 Whereas decentralization refers to a conditional diffusion of specific powers to subordinate local govern- ments by a central government (these powers being, theoretically at least, subject to recall by a unilateral decision), federal constitu- tions stipulate sovereign powers for both the federal and provincial governments. As Beer points out, the unitary government often uses the local governments as agents for carrying out central government policies (decentralization), or it may use local offices to implement policies directly.3 This kind of central government dominance and direction is usually absent in the federal system. The nature of center—periphery or intergovernmental relations varies between unitary and federal systems. This aspect of political structure centralization is captured by ranking states as federal or unitary. However, even a brief consideration of the fourteen federal states in this study points out what my be significant differences in center-periphery relations among federal states. That is, it may prove useful to further classify federal systems into those with strong and 92 those with weak central governments. Drawing from Duchacek's and Elazar's discussions of federal systems, several criteria for characterizing federal systems can be established: Can the center abolish states or otherwise enforce its will upon the states? Can the states destroy the union or secede? Can the center unilaterally change the constitutional division of powers between the center and subnational units? Can states unilaterally change the divi- sion of power? Do center and state governments each have significant sovereign powers? To classify the federal systems according to these criteria, I have relied upon the advice of several area specialists and information from national case studies.4 The results of this effort are depicted in the figure on the following page. In sum, while the main objective in this research is to consider differences between unitary and federal systems in their impact on policy, it may be useful to examine the effects of different types of federal systems . Fiscal Centralization A second type of government centralization is fiscal centraliza- tion. How much control does the central government have over national spending and taxation? I have argued that formal-legal structure does not prescribe or predict either the level of fiscal centralization or of informal political authority concentration. The formal-legal status of a nation is but one dimension of its general level of gOVernment cen- tralization. This phenomenon is supported by the data and will be commented on a little later. Varying levels of fiscal centralization among national economies 93 .mooHsom Mom 3 090: com ** msosmapsm * *nmvmfixmm cooaosmo maamnpms< *maosuoso> manpms< Haumnm asmsnoo who; mfiHosz " mamhmamz " massesomwr " mascH “ .m.s menses ocmawoseesm _ . . . . . . . _ _ . . pomssmc>om “ u u u u “ pcmssho>ow . _ . _ _ _ _ _ . _ . _ Hmwoooe as a r a r muwli n in m. m a awesome . mcohwm “ Mme: . _ *nmezmzzmmeoo smmmmemma ass amazes zmmrmmr monmH>Hm mmzom ho mmDB bhomopmo N m opca cocooow ma Ho>oH Hm>HovcH oonooamcoo one om one mm .Ho>oH Ho. one no agreemacwam one monswdm HH< .II! i. .0 O .- rm.u mm. mm. om.- om.- we. as.. mm. mm. mm.. .1 am. . am.u . mm. on. mm.. mm.- mm. mm. mo... mmf No. no. no. em... mmf emf mm... 1 ® a me... so. 2. S... e of. an... em... 7- «H mzm Um mag >mq mm Hum 00 mo mo Qm> 4H mZm Om mad >mq mm H0& 00 mo mo Qm> D\a *mHomeHocH coapmuaamwpsoo onseoshpm Hmoapdaom HH< moos< mcoapmaoynoo I H m.:omnmom munimapes 106 soapmflsoapns pmoHcch sense seasons: memom coavapomsoo Hmnopooam meson .soaemuaamcoaeomwa o>aemamamoq moeo> .coaemnaamsoaeomwe oaaemamemoa sopmam hpnmm xcocH coapmuaamwesoo doomed mcHSpHoccmxm coapmsdmcoo pccscno>ou mosso>om chasm accesso>oo Hespsoc mouspdoschm onmnm psoscwc>oo Hmwvsoo soapspanpmam Hoxom Heoapwo> seesaw assess: .Hmwoeoa 4H mzm om mag mm Hum 00 mo Ho Qm> Db .sH caramesmoc w a .rH oemuoeoe no esmoaeasmam u o scoaemasoaewm someones mo mao>oa onswooos we ecsoaeacwan $35268 mun canes 107 and the interest articulation variable. The party variables measure several things: the number of parties in the system (PS), the degree of electoral competition, EC, and the degree of opinion dominance in the legislature, LFV, LFS, and SMP. For my purposes. PS and SMP (% seats majority party) are m0st interesting theoretically. These variables indicate partisan strength within society and within the formal politi- cal arena within which they operate. LFV and LFS, on the other hand, differ from SMP in that they indicate the likelihood that two randomly selected legislators will be from the same party. SMP gives the percent of seats in the legislature held by the major party and gets at legis- lative fractionalization from another direction. Moreover, SMP repre- sents 1969 to 1970 figures whereas the other two indicators reflect 1960 to 1965 party strengths. Of the two preferred measures of partisan control, SMP on average correlates more highly with the other party variables than does PS (.735, as opposed to .652). Moreover, SMP reflects partisan strength in a much clearer manner than PS (or the type of party system). SMP indicates party dominance or concentration in the legislature -- the law making and policy setting body. SMP should provide the clearer picture of the possible impact of informal political authority (i.e., how power has been divided in society and outside of the formal structures of government) on government outputs. The interest articulation variable, IA, indicates the presence or absence of regular, active representation of interests within or from the larger society. (The variable is drawn from Banks and Textor who categorize interest group activity as significant, moderate or negli- gible. I have recoded the variable to create a dichotemous variable where 1 signifies negligible activity and where, presumably, there is a 108 concentration of interest activity in the hands of a few.) It points to the way in which influence is shared among groups in society. If there are substantial or moderate levels of interest articulation, we may assume that many different groups are vying for power or influence. IA also reflects existing opportunities to divide power and influence government policy decisions. Therefore, informal political authority concentration will be indicated in later analyses by both SMP and IA. The Three Dimensions of Political Structure Centralization The measures of association presented in the tables above seem to justify the claim that the political structure centralization concept reflects at least three distinct dimensions of centralization or authority concentration. They provide the basis for a study of the impact of different types of centralization on social policy. It may well be the case, in other words, that the several types of centraliza- tion work to influence social policy in different ways. This will be the topic of the following chapters. Here I will briefly discuss the relationships among the three types of centralization. That F/U status is something quite different from fiscal centrali- zation or informal political authority concentration becomes clear with inspections of Tables 3~3 and 3-9. In Table 3-3, F/U and VPD correlate very weakly with all other political characteristics (except, of course, with each other). In Table 3-9, again, it is evident that F/U status does not correlate at medium or high levels with FCI (common variance is 14%), with SMP (0.6% common variance) or with IA (3% common variance). That is, relative to the high levels of r among the variables within each dimension (highlighted in Table 3-9), correlations between dimen- §dgm§_are low. 109 The highest levels of association between dimensions of centrali- zation are those between fiscal centralization and informal political authority concentration. Still, there is only 12% common variance between FCI and SMP. It is safe to argue that FCI and SMP do not lie along the same dimension of political structure centralization, and, in fact, represent different things. The amount of common variance between FCI and IA, however, is somewhat higher at 33%. In spite of this low to medium relationship between FCI and one aspect of informal political authority, IA, it still seems theoretically interesting to consider their separate impacts on social policy. Certainly, they are not close to representing the same dimension of centralization. It is interesting that IA correlates with the fiscal variables in Table 3-9 at higher levels than do the other indicators of informal political authority and that IA correlates slightly higher with the fis- cal variables than it does with the other informal political authority variables I have considered. This suggests that the degree of fiscal centralization and the level of interest articulation are related to each other to some extent, even though there is not a similar relation- ship between FCI and SMP (or the other party variables). Presumably, when interest articulation from competing groups is high and when each group clamors for its share of goods from government or its share of the national financial pie, this leads to 9; reinforces fiscal decen- tralization. When interest articulation is high, there is incentive for fiscal decentralization, and vice versa, i.e., with fiscal decen- tralization comes an increase in lobbying access points and greater Opportunity for more groups to organize for a share of the goods. The exact nature and strength of this relationship, in sum, is not clear. 110 In all, the consideration of the various indicators of political structure centralization has laid the groundwork for the analysis of the impact of several types of centralization on social policy. CHAPTER III FOOTNOTES 1 Duchacek, Ivo, Comparative Federalism: The Territorial Dimension of Politics, Holt, Rinehart and Winston, Inc., New York, 1970, Chapter 5; Daniel Elazar, "Federalism," International Eneyclgpedia of the Social Sciences, Vol. 5, MacMillan, New York, 1968. 2 Ibid. 3 Beer, Samuel, The British Political System, Random House, New York, 1974, pp. 60, 61. 4 I am grateful for the help of Professors O. C. Press and M. Bratten, of the Department of Political Science, and D. Bailey of the Department of History, Michigan State University; also see, James Dunn, "Consociational Democracy and Language Conflict: A Comparison of the Belgian and Swiss Experiences," ComErative Political Studies (April, 1972), Vol. 5, pp. 3~40; Donald Smiley, Canada in Question: Federalism in the Seventies, McCraw-Hill, Ryerson, Toronto, 1976; Michael Reagan, The New Federalism, Oxford University Press, New York, 1972; R. J. May, Federalism and Fiscal Adjustment, Oxford: The Claredon Press, 1969; Jean Holmes and Campbell Sharman, The Australian FederaldSystem, Allen and Unwin, Sydney, 1977; L. A. Sheridan and Harry Groves, The Constitu- tion of Malaysia, Oceana Publishers, Inc., Dobbs Ferry, New York, 1967; Mushtaq Ahmad, Government and Politics in Pakistan, Space Publishers, Karachi, Pakistan, 1970; Herbert Feldman, Revolution in Pakistan, Oxford University Press, 1967. 5 Computed from World Survey of Education, volume 5, UNESCO publi- cation, New York, 1971; International Bank for Reconstruction and Development, World Tables 1971, Table 2; United Nations, Statistical Yearbook 1973, New York, 1974. See among others, James Buchanan, "Federalism and Fiscal Equity," American Economde Review, Vol. 40 (September, 1950), p. 583; Harold Wilensky, The Welfare State and Equality, University of California Press, Berkeley, 1975. 7 Rawkins, Philip, "Outsiders as Insiders: The Implications of Minority Nationalism in Scotland and Wales," Comparative Politics, Vol. 10 (July, 1978), p. 519 ff. 8 In order to check for'possible biases created by the fact that there was some missing data in the calculation of FCI in 29 cases, or, 111 112 to check for possible differences in results when using the 38 cases where I had all the data for the FC index and the other 29 cases, I will regularly consider the separate effects of fiscal centralization using only the 38 cases where all pieces of information were available. 9 Using the tau figures to calculate a variable's average correla- tion with all other variables in Table 3—3, we can again distinguish these two sets of characteristics: F/U .138 PS . 347 VPD .097 LFV .197 SUH . 174 LFS .290 PDP .122 IA .280 EC .343 SMP .218 Variable CE CR GC FCI FU-X- VPD SUH PDP IA* SMP LFV LFS PS EC * dichotemous variables CHAPTER III APPENDIX Some Summary Statistics peep 68.70 73.14 69.06 71.96 .21 2.64 2-73 1.8 2.20 60.64 .592 .510 1-91 1.28 113 Standard Deviation 21 20. 18. 21 -34 18. 92 65 01 .41 .68 .44 .935 .898 .27 .212 .244 ~753 I2 61 59 41 67 67 62 60 6O 56 67 67 CHAPTER IV THE IMPACT OF GOVERNMENT CENTRALIZATION ON HEALTH POLICY In this chapter I will investigate the relationships between the three aspects of political structure centralization and health policy outputs and outcomes. In general, previous research posits that cen- tralization has a positive effect on social policy outputs and outcomes. In Chapters V and VI I will extend the analysis of the impact of cen- tralization on social policy by considering, in turn, education and social welfare outputs and outcomes. Centralization and the Size of Public Spending Before focusing on the separate social policy areas, there is a more general hypothesis that should be tested, given our concern with the effects of political structure on the size of public policy spending:and commitment: that political structure centralization, and financial centralization especially, has a significant, positive effect on the size of the public sector. This hypothesis was forwarded by several scholars whose arguments were reviewed in Chapter I. For example, Peacock and Wiseman argue that fiscal centralization leads to a larger public sec- tor, or greater social policy spending, because centralization leads to, among other things, a standardization of policy implementation.1 Pryor makes a similar argument for education spending.2 Before considering separate policy areas, then, we consider the larger picture of political structure and the size of social policy spending generally. Does 114 115 political structure centralization affect the size of the public sector, or, overall national commitment to social programs? There is some disagreement over the direction of the relationship, however. While Cameron and Hofferbert argue that federalism has a conservatizing effect on social policy allocation decisions, Beer con- tends that in federalism inheres an expansionary potential,3 and Oates .3 * ‘ reports results that show that fiscal centralization has no independent * - ' 1 “r“, as edged: on the size of public spending after controllingffo; national affluence.4 (See Chapter I for further discussion.) The hypothesis that centralization has a positive effect on the size of the public sector can be tested with regression analyses by plotting the size of the public sector (hereafter referred to as SPS) against the three aspects of centralization: FU status (federal/unitary), FCI (fiscal centralization index) and informal political authority con- centration (SMP, percentage of legislative seats held by the majority party and IA, the degree of interest group articulation). The impact of EU status on SPS can be tested using the following regression equation: Y = a1+ blx1 + e where Y = SPS, size of the public sector, 1970. (Following Oates, I use tax revenue as a percent of national income.) X = FU status, 1970. (A dummy variable is used where 0 = federal, 1 = unitary.) e = stochastic disturbance or error term. Results are shown in Table 4—1. After considering data from 65 nations, we must reject the hypothesis that federalism has an independent or important effect on the total size of public spending for social 116 policy. The R2 is extremely small and the parameter estimate does not pass the simple test for inclusion as a causal variable (that is, it is less than twice the size of its standard error). National level data, then, support neither Cameron and Hofferbert's nor Beer's arguments. Federalism has neither a conservatizing nor an expansionary effect on the size of the public sector; empirically, FU status has no effect on SPS. Table 4—1 Linear Regression of Size of the Public Sector on FU Status Independent Parameter Standard Standardized Variable Estimate E559; Estimate F/U status -2.11 “aha -.068 constant 26.91 3.98 2 -2 m :11 P12! .005 .000 .220 The next step is to investigate the hypothesis that fiscal central- ization has a positive effect on SPS. It is not clear from the litera- ture what form the relationship between fiscal centralization and SPS takes; should we expect a linear relationship? Or, is it possible that a curvilinear relationship exists (although the possibility is not raised in the earlier literature)? A curvilinear relationship in this context would suggest that as FCI increases, SPS increases and then, at some point, levels off or even begins to decrease. Figures 4—1 and 4—2 depict these curves: 117 SPS SPS FCI FCI Figure 4—1 Figure 4—2 Logarithmic Relationship Polynomial Relationship The impact of fiscal centralization can be tested using the follow- ing equations: Y = a-+ b1X2 + e Y = a + b1X3 + e (logarithmic form) Y = a + b1X1 - b2X4-+ e (polynomial form) where X = 1n FCI 3 X4 = FCI, squared. Results are shown in Tables 4-2, 4-3 and 4—4. It appears that the linear equation best fits the data from 65 nations -- although the loga- rithmic form is close to the linear in fit -- and it is the simplest model. (The polynomial parameter estimates do not pass the test of being at least twice the size of their standard errors.) But what is most interesting is that these results show that the hypothesized effect of FCI on SP8 is in error. In fact, these results show that FCI has a 118 linear negetive effect on SPS. As fiscal centralization increases, the size of the public sector decreases. Peacock, Wiseman and Pryor's arguments do not stand when tested empirically. Table 4—2 Linear Regression of Size of Public Sector on FCI (N = 65) Independent Parameter Standard Standardized Variable Estimate Edger Estimate FCI, 1970 -.384* .085 -.550 constant 52.87 6.34 R2 32 F .300 .285 20.17 * Starred estimates are at least twice the size of their standard errors. Table 4—3 Logarithmic Regression of SPS on FCI (N = 65) Independent Parameter Standard Standardized Variable Estimate EEEQE Estimate ln FCI, 1970 -24.46* 5.46 -.547 constant 128.96 23.19 R2 B2 F .299 .280 20.10 * Starred estimates are at least twice the size of their standard errors. 119 Table 4—4 Polynomial Regression of SPS on FCI (N z 65) Independent Parameter Standard Standardized Variable Estimate Errer Estimate FCI, 1970 -.650 .659 -.926 FCI, squared .002 .005 .381 constant 61.29 21.76 R2 B2 F .300 .270 9.99 Recall that Oates contends fiscal centralization has no effect on the size of the public sector after the level of national wealth is controlled for. This further hypothesis was tested using 1970 data for 65 nations based on the following equation: Y = 1-+ b X2 + b2X5 + e where 1 X = electric power consumption, EPC, per capita, 1970. 5 Results are reported in Table 4—5. Once again, the findings do not support the arguments in the fiscal centralization literature. Even after controlling for levels of national wealth, FCI has a significant negative impact on the size of the public sector. (As would be expected, the level of economic development has a positive effect on the size of public spending. Richer nations can afford to spend more on social policies than poorer nations. While there must be a national commitment to social policy spending to improve the day to day life of citizens and meet basic needs of citizens, there must also be resources to allo- cate before social well-being can be achieved.) 120 Table 4-5 Regression of SPS on FCI, Controlling for EPC (N = 63) Independent Parameter Standard Standardized Variable Estimate Edge; Estimate FCI, 1970 —.181* .082 -.259 EPC, 1970 .0026* .0005 .574 constant 33.66 6.57 32 82 f. .546 .523 26.50 * Starred estimates are at least twice the size of their standard errors. To test the effects of a third aspect of government centralization -- informal political authority concentration -- on SPS we consider two independent variables: the percentage of parliamentary seats held by the majority party (SMP) and a dummy variable denoting the degree of centralized control over interest articulation by associational groups (0 = significant or moderate interest articulation, 1 = negligible in- terest articulation).5 In the case of the relationship between SMP and SPS, again, we are not sure what form of relationship to expect -— linear or curvilinear. The equations to be tested are of the same form as those discussed earlier (page 117). Tables 4-6, 467 and 4—8 present results. The linear and logarith- mic equations appear to fit the data from 56 nations equally well, i.e., 121 the logarithmic equation does not improve upon our ability to predict SPS.6 (The parameter estimates of the polynomial form are not signi- ficantly larger than their standard errors.) Again, the unexpected result is that SMP has a negative effect on SPS -- as the percentage of seats held by the majority party increases, the size of the public sec— tor decreases (although the rate of that decrease may level off at high levels of SMP). Table 4—6 Linear Regression of SPS on Seats Majority Party (N = 56) Independent Parameter Standard Standardized Variable Estimate E5223 . Estimate SMP, 1970 -.22* .080 -.371 constant 38.59 5.16 R2 B2 F .14 .12 7.5 * Starred parameter estimates are at least twice the size of their standard errors. 122 Table 4-2 Logarithmic Regression of SPS on SMP (N = 56) Independent Parameter Standard Standardized Variable Estimate §E£2£ Estimate ln SMP —12.94* 4.69 -.374 constant 77.54 18.98 R2 B2 F .140 .120 7.65 * Starred estimates are at least twice the size of their standard errors. Table 4—8 Polynomial Regression of SPS on SMP (N = 56) Independent Parameter Standard Standardized Variable Estimate .EEEQE Estimate SMP --562 .453 -.947 SMP, squared .0026 .0034 .585 constant 48.56 14.01 R2 fig F .148 .110 4.0 It is prudent here to consider whether or not this relationship based on a bivariate equation, is causal or if it is only a spurious re- sult of the level of economic development or other features. Table 4-9 Shows results obtained when regressing SPS on SMP while controlling for 123 economic deve10pment (Y = a-+ b1X6-+ b2X5 + e). When we control for national wealth, the effect of SMP on SPS becomes insignificant. That is, the earlier finding of an inverse relationship between SMP and SPS may well be due to the effects of the level of economic deve10pment on SMP. The percentage of legislative seats held by the majority party, then, as an aspect of informal political centralization, has no causal impact on the size of public spending. Table 4—9 Regression of Size Public Sector on SMP, Controlling EPC (N = 56) Independent Parameter Standard Standardized Variable Estimate Egrer Estimate SMP, 1970 -.O80 .067 -.135 EPC, 1970 .003* .0005 .656 constant 24.84 4.65 R2 B2 F .510 .490 23.12 * Starred estimates are at least twice the size of their standard errors. Finally, I shall consider the effect of the degree of interest ar- ticulation (IA) on the size of the public sector. The hypotheses con- cerning the effects of centralization lead us to expect that as IA increases and informal political authority is centralized, the size of the public sector will increase. IA is represented here by a dummy variable where 0 = significant or moderate IA (signifying decentralized 124 informal political power) and 1 = negligible IA (centralized informal power). The equation is: Y = a«+ le9 + e where X = dummy variable IA, 1963. 9 Results in Table 4-10 Show that there is a strong negative relation- ship between IA and SPS. Where informal political authority is central- ized (where independent social groups do not have opportunity to articulate interests and otherwise attempt to influence politics) the 'size of the public sector tends to be small. Again, decentralization, not centralization, leads to increased spending. Table 4—10 Regression of SPS on Interest Articulation (N = 60) Independent Parameter Standard Standardized Variable Estimate E3292 Estimate IA, dummy, 1963 -18.17* 2.51 —.725 constant 34.62 1.80 R2 B2 F .526 .516 52.25 * Starred estimates are at least twice the size of their standard errors. This finding of a significant relationship between the degree of interest articulation and SP8 is further supported by the results shown in Table 4—11 where the level of economic development is controlled for. Even after adjusting for the varying levels of national wealth among 125 nations, the impact of low levels of IA (where informal political power is centralized) remains. Whatever the level of economic development of a nation, therefore, where a large number of social groups are allowed to articulate their interest and influence decision—makers, the amounts of money allocated to the public sector and social policy (among which is health, education and social welfare policy) increases. Table 4-11 Regression of SPS on IA, Controlling EPC (N = 60) Independent Parameter Standard Standardized Variable Estimate Egrpm, Estimate IA, 1963 -11.75* 2.92 -.470 EPC, 1970 .002* .0005 .414 constant 27-98 2-49 R2 B2 F -632 .615 37.80 * Starred estimates are at least twice the size of their standard errors. In summary, while FU status does not significantly affect the size of a nation's public sector (SPS), FCI has been shown to have a negative and linear effect on SPS, even after considering the level of national affluence. The results of the impact of informal political authority concentration on SPS are mixed; whereas the legislative strength of the majority party (SMP) has no independent effect on SPS after controlling for levels of economic deve10pment, the lack of open 126 interest group articulation (the presence of only negligible levels of interest group activity) has been shown to have a negative impact on public spending. A number of arguments in the literature have thus been shown empirically to be incorrect. Indicators of Health Policnyutputs and Outcomes Let us now investigate the impact of government centralization on health policy. We begin with a consideration of various indicators of health policy outputs and outcomes. (For a discussion of the health data and their sources, see Chapter II.) I concur with Levy, Meltsner and Wildavsky that several separate results of policy decisions must be 7 Because of imperfect implementation considered: outputs and outcomes. of outputs (for example, resources allocated) and because of such things as unanticipated consequences, there may not be a strong rela- tionship between outputs and outcomes, or, immediate results and distri- butions of outputs. It seems justifiable to consider both health policy outputs (here, expenditure allocations) and outcomes, even though earlier studies have not separated the two dimensions. Table 4—12 is presented to test the argument that outputs and out- comes are theoretically and empirically separate. Two kinds of data are reflected in the table. First, I have collected data on health expenditures (total expenditures for health, expenditures as a percent of the Gross Domestic Product, and expenditures per capita) to indicate outputs and.levels of commitment to the health policy area. Of all possible output indicators, expenditure figures are the most readily available for a large sample of nations. Second, I have collected data on infant mortality rates, life expectancy, nutrition and numbers of doctors, nurses and hospital beds in proportion to the population to 127 IMR DRS NRS BDS CAL PRT HEX HEXGDP HEXPOP IMR LE DRS NRS BDS CAL PRT Table 4-12 Pearson's Correlation Coefficients Among the Health Policy Indicators -.38 -.25 “050 -031 -0LI'2 J HEX .45 HEXGDP . 58 m HEXPOP Infant Mortality Rate per 000 live births, 1970 Life Expectancy, 1970 Population Per Doctor, 1970 Population Per Nurse, 1970 POPUlation Per Hospital Bed, 1970 Calorie Supply/Cap as % of daily requirement, 1970 Protein Supply/Cap/Diem, 1970 128 Table 4—12 (continued) HEX Health Expenditures in U.S. dollars, 1970 HEXGDP Health Expenditures as 70 of GDP, 1970 HEXPOP Health Expenditures Per Capita, 1970 represent the immediate impact (outcomes) of health expenditures (and commitments). That is, how are life chances, daily nutrition and medical facilities affected by centralization? (See the Appendix to the chap- ter for a summary of the various characteristics of these data.) Table 4—12 reflects the fact that two dimensions of health policy results are in fact inherent in these data. Whereas the average corre- lation among the output (expenditure) variables is .63.and the average correlation among the outcome variables is .57, the average correlation of the output variables with the outcome variables is only .39. That is, variables within each group correlate more highly with each other than with variables of the other group. To simplify the analysis of the health policy area I have chosen to use one expenditure variable (per capita expenditures) and three outcome variables (life expectancy, population per nurse and calories per capita as a percent of daily requirement). Health expenditures per capita will be used both because it is theoretically the most interest- ing of the three expenditure variables, given my concern with the quality of life and well-being of every citizen in the society, and because it has the highest average correlation (.725) with the other output variables. The outcome variables reflect three areas of health policy -- what I will term Life—Nutrition-Medical Services. Each of these aspects is represented by the indicator correlating most highly 129 with all the other outcome variables within each of the three concern areas. For example, calories per capita correlates on average at .60 with all other outcome variables; protein supply correlates at .55. Therefore, to indicate the nutrition aspect of health policy outcomes, I use calories per capita.8 Government Centralization and Health Policy The cross-national analysis presented here represents the first scholarly attempt to study determinants of national health policies on a cross-system basis, and specifically, to consider the impact of political structure centralization on health outputs and outcomes. More- over, the centralization and social policy literature has not focused on the health policy field. (That literature does investigate, to some extent at least, the eduation and social welfare areas.) Hypotheses from the literature do focus generally on social policy, however, of which health policy is a part. To that extent, we can test these hypo— theses using health policy data in an attempt to increase our knowledge of health policy determinants (here, government centralization) and health policy outputs and outcomes. I will begin this analysis by considering bivariate relationships and "partial theories" of the impact of several types of political structure centralization on commitment to and outcomes of health poli- cies. Next, I will consider multivariate and noneadditive relationships, thereby testing for spuriousness while specifying the more complex relationships between types of centralization and health policies. In general, the literature suggests that political structure centraliza- tion leads to greater efforts in social policy areas. We should expect unitary government, fiscal Centralization and informal political power 130 concentration each to lead to greater health expenditures and higher levels of health policy outcomes. The regression of health expenditures per capita (HEXPOP) on federal/unitary status (FU) allows us to address the question of federalism's impact on allocations for social policy -- whether it presents a conservative or an expansionary influence. There is little consensus in the literature on this point, and there is no empirical evidence supporting either argument. The equation I consider here is: Y a + b1X1 + e where Y = health expenditures per capita, 1970 X1 = FU status, 1970. Results in Table 4—13 appear to challenge both arguments. Using data from 57 nations, we find that the EU status of a nation -- its formal-legal character determining the number of sovereign units within the nation -- has no independent effect on a nation's level of health expenditures. It has neither an expansionary effect nor a conservative impact, and, instead appears to play a negligible role as policy deter- minant. This aspect of centralization, in short, carries little signi— ficance for per capita spending on health care, as evidenced by these data from 57 nations. Fiscal Centralization We may logically expect, on the other hand, that the level of fis- cal centralization of a nation -- the degree to which the central government has responsibility over taxation and spending -- will have significant impact upon the level of health expenditures. It has been argued, for example, that fiscal centralization leads to the standardi- zation of procedures, to greater efficiency in implementation and to 131 Table 4—13 Regression of HEXPOP on Federal Unitary Status (N = 57) Independent Parameter Standard Standardized Variable Estimate E2223 Estimate F/U status, 1970 -2.21 14.02 -.04 constant 21.59 12.46 R2 fiz F .000 .000 .248 uniformity in policy efforts. For these reasons, Levy, Hage and Hollingsworth, Pommerehne, and Peacock and Wiseman hypothesize a posi- tive effect of fiscal centralization on social policy expenditure levels.9 To determine empirically the nature of the relationship between FCI and.HEXPOP, I test the linear, logarithmic and polynomial models. The results of the regression analyses are presented in Tables 4—14, 4-15 and 4-16. It is immediately clear that FCI has a significant negar tive impact on per capita health expenditures and that the hypothesized positive relationship is empirically invalid. Once again, arguments forwarded in the literature are found to be untenable when tested em- pirically with cross—national data. What is not so clear is the £92m of the relationship -- whether it be linear, logarithmic or polynomial. That is, does the effect of FCI level off or even decrease at some level? Figure 4—3 depicts the graph of these relationships and the actual data points. From the graph it appears as though a few outlying data points may be distorting the true relationship between FCI and HEXPOP. In fact, when the Netherlands and 132 Table 4-14 Linear Regression of HEXPOP on FCI (N = 57) Independent Parameter Standard Standardized Variable Estimate Errgr Estimate FCI, 1970 -1.02* .276 -.502 constant 93.61 20.45 R2 R2 F .252 .233 13.81 * Starred estimates are at least twice the size of their standard errors. Table 4—15 Logarithmic Regression of HEXPOP on FCI (N = 57) Independent Parameter Standard Standardized Variable Estimate Error Estimate 1n FCI, 1970 -71.33* 17.03 -.547 constant 322.32 72.37 R2 R2 F .299 .282 17.54 * Starred estimates are at least twice the size of their standard errors. 133 Table 4-16 Polynomial Regression of HEXPOP on FCI (N = 57) Independent Parameter Standard Standardized Variable Estimate [Error Estimate FCI, 1970 -5-73* 1.99 -2.81 FCI, squared .034* .014 2.32 constant 243.32 65.98 R2 R2 F -344 .311 10.50 * Starred estimates are at least twice their standard errors. Sweden are taken out of the regression analyses, the parameter estimates (the negative slopes) and the Hz's are reduced by half, although the sign of the parameter estimates stays the same and the linear and loga- rithmic estimates are still at least twice the size of their standard errors. (The significance of the polynomial regression disappears, while the logarithmic form continues to explain more of the variance in HEXPOP than does the linear form.) The strength of the relationship decreases, however, as FCI explains only a small portion of the variance in HEXPOP. Furthermore, when the eight Anglo-European countries that lie on the upper half of the graph (quite apart from the other data points) are removed from the analyses in an attempt to determine if outliers (exceptional cases) are unduly influencing the true relation- 10 ship, FCI no longer has an independent impact on HEXPOP. The indepen- dent effects of FCI are unclear at this point and will be further Health Expenditures Per Cap, 1970 131+ Fiscal Centralization Index, 1970 Figure 4-3 (N = 57) 135 investigated using multivariate regression techniques later in the chapter. But FCI does not, at this point, appear to have the positive relationship often predicted in the literature. Informal Political Authority Concentration Regression analyses testing equations of the forms discussed earlier facilitate an examination of the effects of what I term informal political authority concentration on per capita health expenditures. The linear, logarithmic and polynomial models allow a test of whether SMP and IA each has a significant independent effect on HEXPOP. The effect of the percentage of seats held by the majority party on HEXPOP is reflected in Tables 4—17, 4-18 and 4-19. As SMP increases, HEXPOP decreases and then, at high levels of SMP, levels off. That is, a logarithmic curvilinear form best fits data from L16 countries. (See also, Figure 4—4. Note that the fit of the logarithmic curve to the data points improves when the two outliers -- the Netherlands and Sweden -- are excluded from the analyses.) Using simple bivariate equations, then, we find that SMP has a negative effect on HEXPOP.11 Further aspects of this relationship remain to be explored later in the chapter. It remains to be seen, for example, if SMP influences HEXPOP once economic development is controlled. When levels of interest articulation are negligible (signifying a concentration of informal political authority at the center) health expenditure levels decrease (see Table 4—20). As is the case with the size of the public sector, when interest articulation is limited, health care allocations are minimal. 136 Table 4-17 Linear Regression of HEXPOP on SMP (N = 46) Independent Parameter Standard Standardized variable Estimate EEEEE Estimate SMP, 1970 -.806* .239 -.466 constant 68.74 15.34 R2 R2 F .217 .198 11.38 * Starred estimates are at least twice the size of their standard errors. Table 4—18 Logarithmic Regression of HEXPOP on SMP (N = 46) Independent Parameter Standard Standardized Variable Estimate Error Estimate ln SMP -51.24* 13.59 -.508 constant 226.96 55.03 R2 R2 F .258 .240 14.27 * Starred estimates are at least twice the size of their standard errors. 137 Independent Variable SMP SMP, squared constant Table 4e19 Polynomial Regression of HEXPOP on SMP (N = 46) Parameter Standard Estimate grrgr -2.98* 1.31 .0166 .0098 132.26 40.58 R2 R2 F .269 .232 7.37 Standardized Estimate “10.72 1.28 * Starred estimates are at least twice the size of their’standard errors. Regression of HEXPOP on Interest Articulation Independent Variable IA, dummy, 1963 constant Table 4-20 (N = 46) Parameter Standard Estimate Errgr -39-66* 9-57 40.33 6.87 R2 R2 F .295 .278 17.17 Standardized Estimate «543 * Starred estimates are at least twice the size of their standard errors. Health Expenditures Per Cap, 1970 fang--- “ultra.-."r, --.. {...-.....1..T‘...- - .. -. a..- ..--..rv.......--;..... ...I..-:......f.;-. '.--."._:..3 ...-.I. .. -....-”1’.!-. In-.. . H ' - ’. » ~ ,___ .- .--., . .,. .- x .NkTHGRLfi . . ...... . .-....._5- ...“, ,..—.t-...t ...... l b-ao-oi . 3}: v. -- n c , . a v / ROCRQI‘THM16“*“;+{€§7 new Swine»— .. our 1 v- yc~>v< - 446.... .1:- .--=-;'~'M- -~§‘;.'; . m».— - *1-..- 2.-—— ..“E; .... _ % Legislative Seats Held By Majority 1331'th 1970 “igure 4-4 N = 46) Health Expenditures Per Cap, 1970 -.—-. ... -. .--.,.-s. a... - .. -.. .:°.::.. ....f- 7r .. .. .. . _. . . A .....--- ... .... -..--. . . ...- t-.- ..-. ...... -. >00» ”0 a... ... --- 2-...l_.--. ... .5..--,- -_,.--..-. ...-.. .,. ..t- -. ---.'...- 9 .. . . .-.-- 5.2-..- _ .. .. a. . . . , a . . - , .-. .. Ail—"".-.- M . . - H . V . n. .... ..i ...... . .. . ..-...— ...ti-..’. ...-..a..-.’....—s . ...," -..-.--,v--..-...---_ . .-.—.-._. .... a . .. . ,r-v.. . --...-n- -- -----..-_-o.a--~ ...- ..-- ..- o— -—-_.._.-...-.- -‘..—_.4 l...‘ . ----§—- -. .-.—.- p -- ,..-_’4 ..~ .»|~.. ...-’uc ~ ......~ K ---.'.... . -...-._.--_.-....-.---.....1.... . .-.—.... - . .-.. tuv-Q-.—.'-c - .-:.... . ...-Y.—-—y~.-- ...--.——.——.o~.-...---.._—.-_...---...--u ~.....--. -- :... - 4-.-" .-.. ..--. ._... . .....-t. , 4 .. .... ... If ~v9¢ .. 1 t 4 1 t no-w -... .. - v 'v . ,,. ....- ..l. .. ... . . L . , . , .. . . .. . , - -.,. .o.. .- I... ...... §.i.,.-- ..< . .. ...._ a-- A -..- . n . . .. .. - . . -...... . .....t... . ... . .. . . . ..--_ . -_ . .- - .. ..-.. .. . - a , t . .-.. . . .. ...,. , . . - H -. .. . .. . , . .-- - .. . V 1 _Y ..--..__— V ‘ --.—.-_..'-“- .I.. . ... . u -..o. . ...- .. -7 .. .. . .. . ,.. . .... , . , . , .... . r .__. . .-. -..-- . .-. .-- a- . . -.. . D. .... . . I..- ,.-. .- 5.-- . I... p . ._ -... -...-- . .i... ..... , . . ., ,V. ... . ,, ... -- .. . ..L.. . -.. . . . r— 7 c--——.——.—-.- ----..—----—.. —-m--—-;—— 00-. - . . a . u , I v-? .1 v» -|----~....L ..-.-.... . .--.. ..-..._....-..o.. >...b-—..;..--. ' ...; .. | . -.. . ..--a V -..- - .. v. . .. .- f"""' ., .....-. .... . y . . , ’ 'o‘4 - w, .q.~ ..-.‘-. ‘......---....- ..._. . .l.-. A.- ..t... .’ ..--.--».f. -o.....'.. -'-.- .... .---.-- \ ...--. 1-. --...a 5 -‘-v C-al. ‘ i ..-. .. ...- p... . .-.. (.-.. .--. .. -........’-- ..-—...! Legislative Seats Held By Majority Pam fly, 1970 Figure 4—4 (N = 46) 139 The Role of Demographic Variaylgs We would logically expect that social characteristics such as population growth rate, the percent of population over age 65 and the degree of socio-cultural heterogeneity affect health care allocation patterns. Where populations are growing at a rapid pace (characteristic of many Third World nations) health care needs may be surpassing government capabilities to provide health care services. On the other hand, where a relatively large percentage of the population is over 65 (not often the case in Third World countries), governments may perceive and respond to greater health care needs. Moreover, it was hypothesized in Chapter II that social heterogeneity has a negative impact on social policy outputs. Regression analyses indicate that demographic factors do in fact affect HEXPOP. The population growth rate and the percent of the popula- tion over age 65 both show significant relationships (the first a nega~ tive and the second a positive relationship) to HEXPOP when tested with simple bivariate equations (R2 = .240 and .431, respectively). Moreover, while ethnic dominance appears to have no impact on HEXPOP (parameter estimates are not significant and the R2 is negli- gible), marked or errreme heterogeneity has, as hypothesized, a signi- ficant and negative impact on HEXPOP when tested using a simple bivar- iate equation. In order to explore these relationships more fully, multivariate analyses must be done, since it is quite plausible to suppose that the most highly fragmented societies are also the poorer ones. Put another way, the effects of heterogeneity may be a function of economic level. 140 The_Role of Economic Development Earlier research has suggested the importance of the level of economic development in positively affecting social policy. Again, using regression analyses, I consider the bivariate effects of level of economic development on HEXPOP. Results are shown in Tables 4—21, 4—22 and 4-23 and in Figure 4—5. Table 4-21 Linear Regression of HEXPOP on EPC (N = 54) Independent Parameter Standard Standardized Variable Estimate Errgr Estimate EPC, 1970 .008* .0017 .596 constant 5.87 5.425 R2 R2 F -355 .339 22.58 with Norway and the U.S. taken out of the regression analysis: EPC .015 .0014 .844 constant -2.21 3-55 R2 R2 F .712 .706 116.27 * Starred estimates are at least twice the size of their standard errors. 141 Table 4-22 Logarithmic Regression of HEXPOP on EPC (N = 54) Independent Parameter Standard Standardized variable Estimate Errgr Estimate 1n EPC 13.31* 2.41 .652 constant -62.51 15.55 R2 R2 F .425 .411 30.34 with Norway and the U.S. taken out of the regression analysis: ln EPC 15.12* 2.23 .702 constant -72.24 14.14 R2 R2 F .493 .482 45-77 * Starred estimates are at least twice the size of their standard errors. As becomes obvious from the tables and.g:aph, two outliers -- Norway and the United States -- seem to be determining the form of the relation- ship between EPC and HEXPOP. When the outliers are removed from the analysis, the linear model best fits the data (see Figure 4-5). As was to be expected, the level of economic development has a strong positive effect on the level of health expenditures per capita (R2 = .340, all cases; R2 = .705, Norway and the United States excluded). Given what is known from this analysis and earlier studies, then, EPC will be treated here as an important contextual variable that must be Health Expenditures Per Cap, 1970 142 ' ._ ~s—;:-;<._-.:a:-i«.:: 'E‘ 4.1."... 1'” _ Electric Power Consumption, 1970 (Kilowatt hours per capita, in thousands) Figure 4—5 (N = 5“) 143 Table 4-23 Polynomial Regression of HEXPOP on EPC (N = Independent Parameter Standard Variable Ebtimate ggrpr EPC .0215* .0033 EPC, squared —.000001* .0000003 constant -3.97 5.00 R2 R2 F -569 .548 26.46 54) Standardized Estimate 1.60 -1010 with Norway and the U.S. taken out of the regression analysis: EPC .0226* .0036 EPC, squared -.000001* .0000005 constant -5.91 3.82 R2 "R2 F .739 .727 65.08 1.23 -4.24 * Starred estimates are at least twice the size of their standard errors. taken into consideration when investigating the effects of the political variables. The Joint Occurrence of Types of Centralization: Multivariate Analyses In order to more toward a fuller specification of the relationship between the independent variables and commitment to health care we must move beyond bivariate analyses to consider multivariate effects. I 144 begin by considering the possibility of multiplicative or joint effects among the centralization variables. That is, it seems logical to expect that where several types of centralization occur together or jointly we will observe results that go beyond the explanations rendered by considering types of centralization independently or in an additive model. For example, although both fiscal centralization(FCI) and low levels of interest articulation (IA, where the dummy variable = 1) have a negative impact on HEXPOP in bivariate analyses, it may be that when they occur together they interact so as to effect greater health expend— itures (by the fact that together they present a greater push or influence toward national standardization of expenditures and implementation of policy). Or, the percentage of legislative seats held by the majority party (SMP) may interact with IA to affect HEXPOP in a positive direc- tion even though SMP, too, has an independent negative impact on HEXPOP. We might find similar trends in the joint occurrence of the other centralization types. In order to test for the presence of interaction effects I create new variables by multiplying two centralization variables together. For instance, to test for multiplicative effects between SMP and IA, I multiply the logged SMP variable (the form that best fit the SMP and HEXPOP data) by IA (which is a dummy variable). Likewise, to test for joint effects of federal/unitary status and.majority party seats, I multiply ln SMP by FU (also a dummy variable). (I will also test for the presence of joint effects between FU status and other centraliza- tion measures in spite of the fact that FU status displayed no bivariate relationship to HEXPOP. It may be that when present with other types of centralization, FU status affects health policy outputs.) 145 That several types of centralization may work jointly to affect health expenditures is discounted by multivariate regression analyses where economic level is held constant. (See Appendix B for a presenta— tion of results.) None of the interaction pairs among the four central- ization measures is significant at the .05 level. The discovery that there are no interaction effects among the centralization measures seems to bolster further the argument underlying this research that political structure centralization is best conceived of as separate dimensions. It appears as though each type of central- ization has its own impact on policy that is not tied in with effects of other types of centralization. This suggests, moreover, that the conceptualizations of centralization in the literature may be mislead- ing. Wilensky, Levy, Hage and Hollingsworth, and Morrison and Stevenson all conceive of or present centralization as an undifferentiated composite of political and economic variables. Other Interaction Effects It is also important to consider the manner in which level of economic development interacts with the other independent variables of the study. For example, those who write about federalism often note that such a decentralized structure is costly to implement because of such things as duplication of offices and multiple agencies, legisla- tures, and other governing structures. For this reason, federal struc- tures are more often found in wealthier nations which can afford to try to accommodate diversity.12 It may also be that only where EPC is high does federal status affect SOCial spending. Some scholars, such as Marion Levy, argue that centralization -- economic and political, presumably -— is most efficient for new and 146 poorer nations that need to aggregate power enough to command the situation.13 For purposes of this research it is essential to consider the effect of the presence of both low levels of economic development and centralized structures on social policies and social well-being. In other words, we go beyond questions of political stability to stress questions of basic human needs and levels of living. Moreover, there is evidence to suggest that levels of interest articulation increase with social mobilization and industrialization; EPC may interact with IA in such a way to affect IA's impact on policy. For these reasons, I test for interaction effects between EPC and the centralization measures. We expect that levels of economic deve10pment affect the relationship between political structure centralization and HEXPOP and.that the effect is more than additive -- it is multiplicative. Table 4-24 shows the test of these relationships. EPC displays special interaction effects with two types of political structure cen- tralization -- fiscal centralization and informal political power con- centration (as indicated by SMP). However, the interaction effect differs depending upon the type of centralization present. First, where high levels of EPC and SMP occur together, the multiplicative effect on HEXPOP is negative. That is, although SMP becomes insignificant when controlling for EPC in an additive model, the joint effect of SMP and EPC on HEXPOP is negative. This suggests that where one-party dominance in the legislature exists in a country that is economically developed, commitment to health expenditures is moderated or lessened. Conversely, increasing party competition would increase health expenditures in the wealthier countries. Second, where high levels of fiscal centralization and EPC occur 147 no.3. aqua Admuzv poopmcoo ans. men. ems. sna.a cam * Hos as men. hrs. amoo. mmoo. cams .oam rmm.u ca.aa *mo.oan cams .Hoa as as as Ere 23.8. mom. amo. name. ooao. can a «H mam. mam.u Hmoo. momoo. cams .oam was. rm.ma Na.mmu moms .sH ra.om Hm.om Ammnzv asosnsoo nan. am.o- soc. *HNO.- cam * mam ca one. Hm.o omo. ammo. cams .oam mao.n mm.aa mn.aa cams .mam as .mm ornsasna oonaownocoem gowns ownosopm oemeaonm woeosmwam cacoawca ssoosoaoasm maoooz 53036ch Jim was 2033339200 owseoapm Hmogaaom so momxmm mo coamwonwom 3N1: panama 148 3mm. mmm. H: E. as. so. concerns oonaowoocnsm .mHoHHo osmocmpm adore mo onam one ooaxp enema pm.ono moodsfivmo oosnmvm * oo.N wmoo. moo. Ha.mfi Henna Undocopm :m.m Admnzv pompmcoo saoo. cam * pa *aoo. cams .oas oc.m cams .nseosn s\a consaunm Horosmwom oapmmwm> scoocoauaaH Aconcassoov amt: oases 149 together, the joint effect is to increase health expenditures in spite of the fact that the independent effect of FCI on HEXPOP is negative and that the effect of FCI controlling for EPC is negative. In other words, when high levels of fiscal centralization exist in an affluent society, commitment to health care expands, as indicated by greater levels of health dollars per person. This finding seems to support hypotheses in the literature, and to controvert our earlier observation of a negative relationship between fiscal centralization and health expenditures. But, it would be improper to consider these findings as more than tentative until the relationships have been considered within the con- text of a more complete model of the relationships between the independent variables and HEXPOP, since some of the relationships that have been uncovered may be spurious. Another type of interaction may be significant as a causal factor. The literature on federalism postulates that there is a "federal solu- tion" for socio-cultural diversity and the problems that heterogeneity poses for social policy implementation. Federal structure, it is argued, can ameliorate communal conflict over social goods and services by allowing local autonomy and control. Where federal structure and hetero- geneity occur together, the otherwise negative effects of heterogeneity should disappear. In contrast, where heterogeneity is high and govern- ment (at the center) is controlled by ethnic interests -- assuming that politics is ethnically based in the plural society as Rabushka and Shepsle and others argue -- we would expect the effect on HEXPOP to be negative. The argument is that where there is an ethnicization of goods and services, the ruling ethnic groups are not concerned with sharing benefits of health care with other groups. 150 It is clear that data from forty nations severely question the argu- ments in the literature: federal/unitary status does not interact with heterogeneity to affect the impact of socio-cultural heterogeneity on health policy. At the aggregate level, arguments about a federal or unitary "solution" are empirically invalid or unsupported. (See Table 4-25.) Table 4:25 Regression of HEXPOP on Federal/Unitary Status and Heterogeneity, Interaction Model (N = 40) Independent Parameter Standard Standardized Variable Estimate Egrgr "Estimate Heterogeneity, dummy var. -23.08 23.74 -.312 Federal/Unitary -4. 78 20.23 -. 053 Hetero. * FU -5.47 27.12 -.O67 constant 34.88 18.66 R‘2 R2 F .133 .063 1.90 Specifying a Final Model At this point it is unclear what roles the independent variables play, if any, in explaining the level of health expenditures. That is, I have considered bivariate equations and interaction models separately; it is not clear from the analyses whether the variables that have been found to be significant in zero-order and interaction tests survive in multivariate tests. 151 In order to move closer to a final specification of the causal relationships between the independent variables and HEXPOP, I have regressed HEXPOP on the variables that survived earlier tests. When theoretically relevant independent variables are entered together in this manner, it should become apparent which relationships are spurious (or are functions of other variables) and become insignificant, and which stand the test of a more complex specification. The multivariate regression analysis demonstrates that when con- sidered in context with other important variables, socio-cultural heterogeneity, population growth rate and the interaction term of FCI and EPC are no longer significant at the .05 level. That is, they add little or nothing to an explanation of HEXPOP when other political, social and economic variables are considered. The presence of high levels of ethnic, linguistic and religious fragmentation has no direct effect on HEXPOP, nor, as shown earlier, does it interact with economic or political variables to affect HEXPOP. The rate of pOpulation growth, also, must be excluded from the final equation if we hold strictly to the .05 significance level: population growth rate is significant at the .08 level; it has a marginal impact on HEXPOP that is not as impor- tant a determinant as the other variables. But its presence adds only .003 to the overall R2. It is interesting that the ln FCI * EPC interaction variable adds nothing to the explanation of HEXPOP when the main effects of In FCI and EPC as well as the other variables are present. Apparently the interaction effects discovered earlier (Y = a - b1(ln FCI) + b2(ECPC) + b3(ln FCI * EPC)) are spurious when several political variables are included in the regression. (The role of the ln SMP * EPC interaction 152 variable, on the other hand, proves to be very important to HEXPOP.) Table 4—26 displays the regression of HEXPOP on the variables that survive multivariate analyses. The results confirm some arguments from the literature and discount others. It is clear, in any case, that the existing literature has not given an accurate account of relationships between politics and policy. Table 4—26 Regression of HEXPOP on the Independent Variables (N = 40) Independent Parameter Standard Standardized Variable Estimate Errgr Estimate Pct POP 65 + 3.28* -593 -339 In SMP 73.57* 6.44 .729 EPC .240* .012 17.80 In SMP * EPC —.057* .003 -17.48 1n FCI -132.11* 8.18 -1.01 IA 31.95* 4.25 .437 constant 269.71 37.08 R2 R2 F .963 .957 148.67 * Starred estimates are at least twice their standard errors. The notion that citizens' needs and demographic characteristics influence expenditure decisions is supported by this research. In the case of health expenditure levels, for example, as the prOportion of 153 people aged 65 or older increases, allocations for health care likewise increase, all other factors being equal. Also, as expected, the re- search demonstrates that the level of economic development or national wealth is an important determinant of expenditure levels. It deter- mines the level of resources at decision-makers‘ disposal and therefore constitutes an important contextual variable determining the environment within which political factors will Operate. But the results also indi- cate that not only do political factors have independent effects on health expenditures, they also moderate the impact of economic level of health-care commitment. In other words, while economic level influences politics, political variables also affect the relationship between economic level and HEXPOP. This is discussed further below. Some of the research results challenge or discount theories in the literature. First, it is clear that the present study questions claims from scholars of American state politics that the nature of politics has no effect on social policy, and specifically, on expenditure levels. In the case of health care, the level of interest group activity, the degree of major party dominance in the legislature and the location of fiscal decision making -- center or locale -- all play a role in deter- mining spending. (The nature of their impact is discussed below.) This role is independent of (but moderated to some extent by) the economic environment. Second, these research results refute several hypotheses conCerning the nature of the impact of political centralization on policy. While Peacock and Wiseman, Wilensky, Hage and Hollingsworth, and others argue that fiscal centralization leads to greater social spending, and while Oates argues that fiscal centralization has no impact on social spending, 154 this research demonstrates that fiscal centralization has a negative irpggr on health expenditures, other factors being equal. Table 4—28 demonstrates that at each level of economic deve10pment considered, as the degree of fiscal centralization increases, the per capita health expenditure decreases. (All other significant variables are held con— stant at their means.) Regardless of the level of national affluence, then, centralization of fiscal activities -- spending and taxation -- will have a negative impact on commitment to health care- Table 4—27 The Impact of EPC on HEXPOP: Expected Values of HEXPOP* (adjusting for the effects of the other variables by inserting mean values) Level EPC Expected HEXPOP (mean for Third World Nations) 329 23.58 (mean for total sample) 1742 37.18 (mean of Anglo-American, developed nations) 4557 64-52 * in U.S. dollars Moreover, the relationship between informal political authority concentration and HEXPOP is more complicated than would have been ex- pected given the centralization literature. While a concentration of interest articulation in the hands of a few has a positive effect on the level of health expenditures (all other things considered) the posi- tive impact of majority party dominance in the lower house of 155 Table 4—28 The Impact of FCI at Various Levels of EPC: Expected Values of HEXPOP* (adjusting for effects of other variables by inserting mean values)@ r39 13ng E Third World E Total Sample E Developed Nations 329 1742 4557 54 61.58 75.33 102.67 72 21-58 35-33 62-67 90 -10.92 2.83 30.17 _— _ L _ * in U.S. dollars ** The FCI values represent the mean value of FCI for the sample (72) and values at one standard deviation above and below the mean. I calculate HEXPOP figures using the log natural of the FCI value since the logged FCI variable has been used in regression analyses. @ Cell entries are the predicted values of health expenditures per capita, 1970, at specified values of economic deve10pment and fis- cal centralization, holding the effects of SMP, IA and the percent population aged 65 or over constant at their means. —— ; ‘ the legislature is moderated by economic level. (This is clear from Table 4-26. The final equation is: HEXPOP = 269.70-+ 3.28 (% popula- tion age 65+) + 73.57 (ln SMP) + .240 (EPC) - .057 (In SMP -x- EPC) — 132.11 (ln FCI) + 31.95 (IA, dichotomous)). That is, although both EPC and In SMP have an independent positive impact on HEXPOP, they interact to moderate each others' effects. The interaction variable (In SMP * EPC) has a negative impact on HEXPOP. When ppjpp SMP and EPC are high, HEXPOP is lower than when, for example, EPC is high and SMP is 156 low. (See Table 4-29.) In other words, the effects of SMP differ depending upon the level of economic deve10pment. When EPC is 329 (the mean level of electric power consumption for Third World nations) higher levels of majority party control lead to higher levels of health expend- iture. At middle and high levels of EPC, however, increased majority party control leads to decreased expenditures. Table 4329 The Impact of SMP at Various Levels of EPC: Expected Values of HEXPOP* (adjusting for the effects of other variables)@ rag grin-g“ E Third World E Total Sample 3? Developed Nations 329 1742 4557 39. 38 4. 39 46. 33 129.81 60.64 26.42 36.26 55.44 81.90 41.98 28.26 1.14 * in U.S. dollars ** The SMP values represent the mean value of SMP for the sample (60.64) and values at one standard deviation above and below the mean. I calculate HEXPOP figures using the log natural of the SMP values since the logged SMP variable has been used in regression analyses. (0 Cell entries are the predicted values of health expenditures per capita, 1970, at specified values of economic development and majority party strength in the legislature, holding the effects of fiscal centralization, interest articulation and the percent of population aged 65 or older constant at their means. We may also consider this relationship in terms of the effect of EPC on HEXPOP -— the positive effect is moderated by the political situation. When the major party clearly dominates the legislature, 157 increased economic deve10pment and national wealth does not lead to pro- portionately greater health spending. Health spending actually decreases. This trend is evident, too, at the mean level of SMP for the nations in the sample (60.6%). Although the level of expenditures increases al- most threefold where SMP is near 40% and.EPC jumps from 1742 to 4557 (the first line of Table 4—29), expenditures increase only one and a half times where SMP is 60% and EPC increases from 1742 to 4557 (the second line of the table). In sum, the final model enlightens us as to the impact of political structure centralization on HEXPOP: 1) federal/unitary status has no impact on health spending, 2) fiscal centralization has a negative impact on expenditures, 3) informal political authority concentration has a more complicated impact on HEXPOP -- while low levels of interest articu- lation positively influences commitment to health care, the impact of party dominance depends upon national economic level. It is clear that the nature of political organization and the structure of'politics -- whether control is centralized or decentralized —- has importance to health policy. But it is also clear that the nature of that impact differs depending upon what aspect of political structure centralization is being considered. Government centralization is not a unidimensional concept in its structure or its results. The final model also confirms the role of economic development; although in general it has a positive effect on HEXPOP, those effects may be moderated by the political situation. Both economic and political factors must be considered in an explanation of commitment to health care. Finally, whereas one demographic variable -- the percent of the 158 population aged 65+ -- proved to have an impact on HEXPOP, other demo- graphic characteristics such as socio-cultural heterogeneity added nothing to the explanation. This array of political, economic and social variables (whose inclusions were justified by theory), then, appears to explain 96% of the variance in HEXPOP among nations. Qrtcomes of Health Poligy: The Life-Nutrition-Medical Services Dimensions In this section I consider the effects of political structure centralization on health policy outcomes. Does centralization affect the actual outcomes of health policy decisions as experienced by citi- zens? As noted earlier, I operationalize "outcomes" by referring to life expectancy, calorie supply as a percentage of requirements and population per nurse- The tables on the following pages show the bivariate relationships between the independent variables and the Life, Nutrition and Medical Services dimensions. Although several of the centralization measures appear to be significant in explaining the health outcomes, it becomes apparent that the level of economic development plays a relatively larger role in explaining the variance. In each case, a logarithmic form best fits the EPC and outcomes data. That is, as level of national wealth increases, life expectancy and calorie supply increases and the ratio of population to nurses decreases. But, this effect levels off as EPC increases. (See Figures 4—6 and 4—7.) EPC has a curvilinear impact on the Life-Nutrition-Medical Services dimension. The next step involves testing for joint effects among the central- ization indicators and between EPC and the several aspects of political structure centralization. That is, we might expect that the centrali- zation indicators will work jointly to affect health outcomes or that 159 NH .m Row 28.38 or: one. are. Ema Em as or . a 8 . we entrance mmm . o9. .1 8:0. man or. E or .m ma . me pontoon Nam. ems .. more. 3% .- saw no .0 Hm .om 9:88:00 mar. one... came. *aarm... Hos wm.m 0:16 pgmnoo coo. are... amoé mama- museum in mm headsets gowns noenfiwnm noSoflna 838853 onscreen senescence annotate: so u a on? .nnocoeoomxm ems one moHQofiHg egocomoocH one sooxpom wmdnmcodpoaom tsunami/Hm onto canoe 160 .mHoHHo owmocmpm Remap mo omen one oofizv pmmoH we can movmsapmo ooHHmpm * mm.m No.mm pompmqoo arm. arm.- are. *ac.cu near rescue .aom mm.fi wm.nw pompmcoo mam. rmm.u eo.m *mc.mH- asaosomowouor .mm mmmmammwm wowwm mmmmmmmmm. mmmmmammm oouaowuocoem oncogene. woeoaflnm ssoosoaoosH Goddesses onus oases 161 so.amo a.ooam romances ems. moo.u ma.mma mam.aoou can re arson ooomo sentence Rm. mom. modem *3;on .3. mfi.omm mm.nmm pompmcoo ooo. arm. ma.oa mom.am mam mm.womfi mm.mfim pampmcoo ooo. one. mo.ma ao.oa Hoe “H .me S :3me PGSMGOO ooo. ooo. mo.mme oo.mma barman o\a mm abscesses wowwm mmmmammmm moanoawn> congressman onscreen woeosnwnm stocoaoofi MO carer one can moansawoa esoosoaoocH one soossom naasncoaenaor Gm u a momnsz op coavoasmom abs oases corneas 162 .wHoHHo onmocmvm adore mo omen one coax» pmmoa as one moposapmo oohwmvm * ma.rea ms.oaoa essensoo mom. mam.- ma.mo *om.aamu +mo owe .eom .noa mm.mom mfi.fifil pcdvmcoo mos. o3. oroom mammooa ores rescue .aoa mfi.mmm om.nmfifi pompmcoo wow. ohm. ofi.mdm *mfi.¢dmm apaosomonopom .mm horseshoe wowwm nounsarnm mammmmmmm oouaosmocspm onmocmvm Hoposmwmm pcoocomoocH Acossassoov Home canoe 163 mum omre c.6328 ooo. ooo. ooo. *meé can 5 o .m R a: Rambo o3. are .. mom too? 3 mm.m wH.HNH pcmvmsoo o3. moor moo. mommf arm at: .e ommma 2328 :3. m3... ooo. *mom .. Hos 0H3: mwomOH chmcoo ooo. o3... 3.: ore- master {a .mwm noedfiunm wowwm noonfieom anaconda oouaownoefim Encoder senescence rcoosoaoofi so u so Nit canoe on? .aammdm odwodmu one moapmflfla voodoomoocH one. coospom mmEmcoafiwHom opoflgam 164 .mHoHHo osmocmpm Moose mo mean one oofizp yucca as one moposfipmo ooswmpm * om.m mm.oo esopnsoo oam. «an. mom. mom.m rmo ems .aoa .soa mm.m m©.NHa pcmemcoo mom. mom.- mo.a *om.su over rezone .aom am.m Ha.aaa osnpmsoo sea. ooo.- aa.m *aa.mau auaosomowouor mm mopmsdpmm HoHHm mopmsdpmm mmammwmmw Shenandoah onscreen wososowom esoosoaoocH Aoossarsoov mmla canoe Life Expectancy. 1970 165 Electric Power Consumption, 1070 (Kilowatt hours Per capita) Figure 4-6 (N = 63) 166 L: 23.12.31 2:... ... .-. __ a... it Marinara-r: . E ii; 1:2,: .2. 'le «(fl-1.41.1.1“. 1...“."”.Mm._”hlum.. am...“i .. .._.mrr+L.»n : W .1: .w ...-34.14.14Jul-5.1....”iaa..1r.m-M-..._.M.lai ..." i i: J: .1 m.1 .“l; .1. 7%.. ...?H.4|.1.Ll.ht.t. . ... . “...-.4 lit ..-a...._-a-:.1.1.__1_E :11333 .44 .. . _ 3. a Zita-n.1,“ .21-fa.-_.-h..-.-._.r.. am. If Lr. . _ .T Z. . .... . .... .. ffiib 11 hm : 41.144.411.74444 WWWMLI arm-”914 H_. H“ *Wm m*”_ .— —W jg” W“ r L11: . _ “__ . _ T? 4 ...-”114.141-...fii 1». LL11... 17,171.»... . a . . . T i- w l J i ”a 4:1,: .3? - .1. 11.: _ _ a“ i w is _ :1. : : a ._ . .....- 1......- 4 .12. 417. .161. 11.4 a. a; 1 . ._ .1 H 1 ._ .__ . . ... .. .. . . 2. artist...”- .-.. a. ii i 1 _ a l. _ .. .51.. .:. :1; :17... ...W £51: .. ”1 AT... L 4+ >..PIL... H_ _ . *4 .a .a _ W . 81¢: _ 4.3.1.: 1213.... 1...; : _ i Z ... . 1.1:: 5.: 2:: a . :51: .4.- .. .: :. .1 a :3: i :1 -_ a .5 fiigritaflwrwivflifi m a : ”FIL- WWFHIbMIL .Hfir: ”lb m A_ m a l .43 3L1 ...s a... at m :_ “1:: “1.11.4.4 .a : aft-C" m . ”a - #44. 11.- __;__f n _ .‘_ _ .4 45 ..M F .C. *4. . r+4-4a.1_.._-1_-w-xiii--1. :_ _ ; .E :- C as 1.1.... .__.1.::; E: :;-.....:. far. ., _ 3 amp 4.1.1.3?...-”£11....E if?” r: 1“ 1 _ :r r. 1:: arr-Er. .:_; “1:;31: _ 1.4;..- ...-31:4 m: - . Z a; 1.1:. r-.aM-..i+..._:+;. _ 1:. 1; 1 :4. 1. . . ...1--.....1. _ _ 117...: i ...-1; . if xi-t.-..-_.._. . a; : :- . agar.“ ww4+ N. gm. L: frwinummi_:._Umw;:m: .W 1.11 . . .. - 1.1.. _ . :71 1 .: 2:11:3.1211::431..-:Ira-1:-1: .. .1-- 651: other -.........-:-.r.m.a..._. :2 £115.: 7.1.4.... . ... ... . a .a... - ...” .0. “1r.i+1.»¢1r .I .u M. .. .1 .- ...-1:1..eErlar ..-_......1.._i.r§::e.a .... ...m .. ... is :4 W ..flrflm.» mw._m.._.dlnw4p.n.v.l.ml.fi.m «1.1m ia.m T“. “T “I ..LW ”#44 £1:C.Ew 7:. ”01:01...1t.... H... ..e trig: - . ._ ...... ass-er...sealer-......“.... ....- . 1...... C121. . . 131. . . a L f.. a. ...-.8832 a ... ...... 13;. a; 1:4 HHS .H00 mm Ammo .Hm mmmfiHOHMU o O 7 mm pi MP in t We SP nS 0r CU rm 9 wt ...-1 mm tn C/.\ e m Figure 4-7 (N = 62) 167 EPC may interact with centralization dimensions to affect outcomes under special circumstances. I consider the possibility that the impact of centralization will vary with economic level (a known, important causal variable). It is interesting that none of the interaction equations tested for the three outcome variables was significant at the .05 level (see Appendix c). The effects of the main independent variables -- central- ization indicators - seem to be direct and are not joint functions of economic level or other types of centralization. What remains to be done, then, is to determine the role of the main independent variables in explaining health outcomes. Table 4—33 shows the results of regressing life expectancy on the variables that proved to be significant in bivariate analyses. That is, I consider the im- portance of each independent variable when other independent variables are also present. Given the presence of several centralization indica- tors, social demographic indicators and economic indicators, two of the main independent variables seem to be significant in explaining life expectancy: the level of economic development and the degree of major party dominance in the legislature. The effects of the other variables become statistically insignificant and appear to have been a function of EPC or SMP. Just over 73% of the variance among nations in life expectancy is explained by EPC and SMP. As expected, as level of economic development increases, so does national life expectancy. Resources are available, in other words, to implement programs and pro- cesses aimed at or resulting in prolonging life. (EPC alone explains over 7r% of the variance in life expectancy.) The impact of major'party dominance and concentrated power in the legislature, however. is 168 Table 4—33 Regression of Life Expectancy on the Main Independent Variables (N = 50) Independent Parameter Standard Standardized variable Estimate Error ' Estimate FCI .027 .061 .042 SMP -.134 .049 -.251 IA -1.38 3.03 -.O61 ln EPC 3.32* .848 .527 Heterogeneity -3.71 2.03 -.162 Pop. Growth Rt. —1.42 1.24 -.117 constant 50.57 8.85 R2 152 F .794 .763 25.64 * Starred estimates are at least twice their standard errors. negative. As the percentage of seats held by the major party increases (and after adjusting for EPC), life expectancy tends to decrease. In this case, the concentration of informal political authority in the hands of a dominant political party has negative consequences for the national average length of life. This suggests that legislatures con- trolled by a predominant ruling group tend not to establish programs or pass laws designed to make life more comfortable or less hazardous. They are less willing, presumably, to commit resources to deve10pment in this area. This also suggests that centralized political control may 169 not be the best development strategy for Third World countries (as Levy suggests it is) if one priority is to increase average life expectancy. Instead, party competition and increased opinion representation could better be encouraged. Eable “:34 Regression of Life Expectancy on EPC and SMP (N = 54) Independent Parameter Standard Standardized Variable Estimate E322; Estimate ln EPC 4.53* -555 .720 SMP —.128* .047 -.240 constant 39.34 5.72 R2 fiz F .763 .752 69.20 * Starred estimates are at least twice their*standard errors. The next step involves a consideration of the role of health expenditures (HEXPOP) in explaining life expectancy. Do health expend- iture levels have an impact upon health outcomes? I argued earlier that there may g9: be a strong link between commitment (as measured by expenditure levels) and actual outcomes as experienced by citizens. Money; for instance, may never reach target groups or programs. Table 4-35 shows what happens when HEXPOP is entered as an indepen- dent variable into the regression of life expectancy on EPC and SMP (the two variables found to be significant in multivariate analyses). HEXPOP is not significant in explaining life expectancy when the other 170 variables are present. This supports the earlier contention that since there is not necessarily a strong link between outputs and outcomes, the two must be considered separately when studying the impact of politics on policy. Table 4:35 Regression of Life Expectancy on EPC, SM? and HEXPOP (N = 50) Independent Parameter Standard Standardized Variable Estimate Error Estimate 1n EPC 4.70* .665 .746 SMP -.132* .048 -.247 HEXPOP -.0144 .031 -.047 constant 38.84 5.74 R2 E?“ F .764 .747 45.36 * Starred estimates are at least twice their standard errors. The level of life expectancy, in sum, can be expressed as a function of EPC and SMP. The final equation from Table 4-34 is: Life EXpectancy = 39.34-+ 4.53 (ln EPC) - .128 (SMP). A second health outcome indicator is the ratio of citizens to nurses. This represents an attempt to tap the availability of medical services to the population. Table 4—36 shows results when the indepen- dent variables that survived bivariate analyses are entered into a multivariate regression. None of the political variables is empirically 171 Table “#36 Regression of Population/Nurse on the Independent Variables (N = 52) Independent Parameter Standard Standardized Variables Estimate Egggr Estimate 1n EPC -737.8* 282.42 -.605 IA -374.21 926.64 -.086 SMP —8.32 14.54 -.081 Pet. Pop. 65+ 14.38 176.02 .024 Pop. Growth Rt. 223.24 539.30 .095 Heterogeneity 1050.29 606.07 .238 constant 6374.56 2865.54 .095 R2 fiz F .488 .411 6.36 * Starred estimates are at least twice their standard errors. significant in explaining the availability of nurses to the population. ‘Nor are the social demographic variables significant. Instead, economic level alone explains over 42% of the variance in nurse availability among nations. (See Table 4—37.) As economic development increases, the ratio of people to nurses decreases. Medical services become more accessible. When one considers the role of HEXPOP in accounting for nurse availability, HEXPOP shows no empirical relationship to the level of medical services as indicated by numbers of nurses. Table 4—38 shows 172 Table 4137 Regression of Nurses on the logged ECP Variable Independent variable 1n EPC constant (N=60) Parameter Estimate ~807.54* 7106.41 R2 R2 .439 .426 Standard Standardized @3323} Estimate 136.08 -.662 876.50 F 35.21 * Starred estimates are at least twice their standard errors. Independent Variable 1n EPC HEXPOP constant Table 4-38 (N = 57) Parameter Estimate ~821.62* 1.06 7172.51 32 82 .439 .413 Regression of Nurses on ln EPC and HEXPOP Standard Standardized E3325 Estimate 183.61 -.674 8.99 .018 1058.36 I F 16.84 * Starred estimates are at least twice their standard errors. 173 results of entering HEXPOP along with 1n EPC: the overall R2 is reduced and the HEXPOP parameter estimate is much lower than its standard error. Thus, health expenditures do not translate in any direct way to an improvement of medical service availability. The final equation from Table 4—37, then, is: P0pulation/nurse = 7106.41 - 807.54 (ln EPC). The last health outcome considered in this research is level of nutrition, or calorie supply as a percent of the daily requirement as established by the United Nations. Table 4—39 presents results of multivariate analyses where calorie supply is regressed on the indepen- dent variables that survived bivariate analyses. It appears from the table that the only significant variable is the demographic variable giving the percent of the population over age 65. All other variables -- social, economic and political -- drop out. As the number of senior citizens increases. so does the calorie supply available to citizens. This relationship holds up, moreover, after controlling for level of national wealth (ln EPC is not significant). Although economic level is not significant to calorie supply in the multivariate regression, it is interesting to note its relationship to the demographic age variable. Regression analysis shows that 48% of the variance in the age indicator is accounted for by level of economic deve10pment (see Table 4—40). In other words, economic growth and national affluence increase the life span of citizens. Healthier nations will have more people living to age 65 and older. Peor nations will have relatively low life expectancies. Economic level, then, does influence calorie supply indirectly through its effect on the age variable. We may specify two relevant equations: 174 .— —— —— Table 4:39 Regression of Calorie Supply on the Independent Variables (N = 50) Independent Parameter Standard Standardized Variable Estimate Egrgr Estimate ln EPC .173 1.51 .024 FCI .072 .102 .098 IA -3.81 5.23 -.146 SMP -.043 .083 -.069 Pct. Pop. 65+ 2.73 .986 .793 Pop. Growth Rt. 2.51 3.02 .180 constant 82.32 16.07 R2 "fiz F -586 .519 8.72 * Starred estimates are at least twice their standard errors. Percent P0pulation Age 65+ = 4244.+ .0009 (EPC) and Calorie Supply = 90.33 + 2.56 (% population 65+). The equations illuminate the developmental effects: EPC has a posi- tive impact on life expectancy, which in turn has a positive impact on calorie supply. 175 Table 4—40 Regression of the Age Variable on EPC (N = 63) Independent Parameter Standard Standardized Variable Estimate ‘Erggr Estimate EPC .0009* .00016 .702 constant 4.44 2.04 R2 '82 F .493 .480 38.00 * Starred estimates are at least twice their standard errors. Table 4—41 Regression of Calorie Supply on ln EPC and the Age Variable (N = 60) Independent Parameter Standard Standardized Variable Estimate E332; Estimate ln EPC 1.20 1.23 .165 Pet. 65+ 2.11* .581 .612 constant 85.65 5.42 R2 E?“ F .562 .540 26.31 * Starred estimates are at least twice their standard errors. 176 Table 4—42 Regression of Calorie Supply on the Age Variable (N = 60) Independent Parameter Standard Standardized Variable Estimate .Egggr Estimate Pct. Pop. 65+ 2.56* .369 .742 constant 90.33 2.66 82 F .540 48.03 * Starred estimates are at least twice their standard errors. As was the case with the other two outcome variables, health expend- itures play no direct role in influencing nutrition levels (see Table 4-43). Again, it appears that expenditures do not easily or directly translate into improving actual levels and distributions of health outcomes. A Summary Although the political centralization measures significantly affect levels of health §pending, they are largely irrelevant to health gut: ggmgg (SMP does appear to have a slight negative impact on life expec- tancy, however). Instead outcomes are best understood as functions of economic development. There seems to be little, if any, relationship between health expenditures and health outcomes as measured by life expectancy, avail- ability of nurses to the population and calorie supply. Monies spent 177 Table 4—43 Regression of Calorie Supply on the Age Variable and HEXPOP (N = 60) Independent Parameter Standard Standardized Variable Estimate ‘Errgr Estimate Pct. 65+ 2.71* .499 .787 HEXPOP —.024 .052 -.067 constant 89.86 2.87 R2 '82 F ~554 .531 23.64 * Starred estimates are at least twice their standard errors. for health are not reflected in these areas. Rather, allocations may "get los " in the process of implementation. It may be, however, that this research has not tapped some important outcome or other dimension that would explain this phenomenon. It is not clear from this research, for example, where health expenditures go. The findings of this chapter, moreover, demonstrate the difficulty in prescribing development strategies for'Third World nations. Not only are the effects of centralized control complex, but there is also a significant gap between health spending and health outcomes. Finally, the effects of economic level differ according to the political situa- tion and between health outputs and outcomes. With some qualifications, however, with economic development comes health policy development. CHAPTER IV FOOTNOTES 1 Peacock, Alan and Jack Wiseman, The Growth of Public Expenditures in the United Kingdom, Princeton University Press. 1961. 2 Pryor, Frederic, Public Expenditures in Communist and Capitalist Nations, Allen and Unwin, Ltd., London, 1968. 3 See David Cameron and Richard Hofferbert, "The Impact of Federalism on Education Finance," in European Journal of Political Science (September, 1974), vol. 2; Samuel Beer, "Political Overload and Federalism," Policy (Fall. 1977). vol. 10, no. 1. 4 Oates, Wallace, Fiscal Federalism, Harcourt, Brace, Jovanovich, Inc., 1972. 5 The interest articulation variable was collapsed in this way to highlight the presence or absence of centralized control of interest articulation. Where the level of interest articulation by associational groups is negligible and the power or right to voice interests is not widely shared among social groups, we assume that the political elites at the center alone, have control over determining what "interests" or needs will be addressed. This aspect of politics, then. is centralized. 6 Data on the percentage of seats in the legislature held by the majority party are available for 56 nations only. Here and in the fellowing tables, results are based on all available data for the specific analysis being done. Therefore, N's vary from table to table. 7 Levy, Frank, Arnold Meltsner and Aaron Wildavsky, Urban Outcomes, University of California Press, Berkeley, 1974, pp. 1—23. 8 The average (absolute) correlations of the outcome variables with each other are: IMR .64 IMR and LE correlate with each other LE .68 at -.87. DRS . 47 NRS .53 BDS .53 CAL .60 CAL and PRT correlate with each other PRT -55 at .79. 178 179 The average (absolute) correlations of the output variables with each other are: HEX .515 HEXGDP .66 HEXGDP and HEXPOP correlate at HEXPOP .725 .87. 9 These arguments were reviewed in Chapter I. See, for example, Marion Levy, Modernization and the Structure of Societies, vol. II, Princeton, N.J., 1966: Jerald Hage and J. R. Hollingsworth in The Annals (November, 1977). vol. 434, p. 9 ff. 1° The estimated regression equations when all data are included are as follows: Y = 93.6 - 1.025 (FCI st. e. (.276) Y = 322.32 - 71.34 (ln FCI) (17.03) Y = 243.30 - 5.73 (FCI) + .0343 (FCI)2 (1.99) (-0145) When the Netherlands and Sweden are removed from the analyses the follow- ing equations are estimated: Y = 53-33 - -530 (FCI) (.186) Y = 169.67 - 36.41 (ln FCI) (11.77) Y = 110.89 - 2.32 (FCI) + .013 (FCI)2 (1.38) (.010) When the Netherlands, Sweden, New Zealand, Canada, Finland, the U.K., Austria and Australia are removed from analyses the equations are: Y = 8.28 - .023 (FCI) (-070) Y = 14.30 - 1.81 (ln FCI) (4.51) Y = 17.55 - .310 (FCI) + .002 (FCI)2 (.517) (-0037) 11 When the Netherlands and Sweden are excluded from the analyses, the estimated regression equations are: Y = 53.57 - .567 (SMP) F2== .145 (.197) 180 Y = 168.30 - 36.73 (ln SMP) R = .157 (12.23) Y = 85.40 - 1.58 (SMP) + .007 (SMP)2 1:52 = .140 (1.24) (.0089) 12 See Martin 0. Heisler in Dialogue on Comparative Federalism, edited by Ellis Katz and B. Schuster, Center for the Study of Federalism, Temple University, pp. 14—15. 13 Levy, Marion, Modernization and the Structure of Societies, vol. II, Princeton, N.J., Princeton University Press, 1966? CHAPTER IV APPENDIX A Summary CharacteristiCS of the Dependent and Independent Variables Variable Infant Mortality Rate Life Expectancy Pop./Dr. P0p./Nurse Pop./Hos. Bed Calorie Supply Protein Supply Health E ends. (million:) Health E ends. (% of GDP Health Expends./ Capita Fiscal Centrali- zation Index Seats Majority Party Elec. Power Cons. per cap. Interest Articulation Fmaflmmmw Status Mean 68.2774 59.6493 5769.8955 2108.4127 430.2388 106.0323 71.9677 306.1790 .0154 19.8446 71-9597 60.6393 1741.7302 .5167 .7910 Standard Deviation 53.8209 11.3522 9963.5666 2197.4885 439-2370 13.1136 18.7214 798-3979 .0143 36.7866 18.0137 21.2662 2733-2725 -5039 .4096 Cases 62 67 67 63 67 62 62 62 54 57 67 56 63 60 67 CHAPTER IV APPENDIX B Regression of HEXPOP on Centralization Measures, Interaction Medels (N = 40) Independent Parameter Standard Standardized Variable Estimate Ergo; Estimate EPC .006* .002 .427 FU 111.46 152.28 1.24 1n FCI -40.66 31.68 -.312 1n FCI * FU -22.18 36.87 -1.07 constant 169.5 129.65 EPC .004 .002 .286 ln FCI -65.29 25.04 -.501 IA -338.06 171.39 —4.63 ln FCI * IA 75.98 39.92 4.55 constant 293.22 104.96 EPC .005 .002 .388 ln SMP -27.91 13.64 -.277 1n FCI -29.68 19.26 -.227 constant ** 249.44 83.48 ** ln FCI * 1n SMP does not enter the equation due to insufficient tolerance 182 183 Independent Parameter Standard Standardized Variable Estimate 3395 Estimate EPC .0067* .002 .498 FU -20.84 133.52 —.232 In SMP -40.88 29.60 -.405 ln SMP * FU 7.15 32.93 .333 constant 167.09 120.54 EPC .005* .002 .410 1n SMP -25.29 21.16 -.251 IA 28.27 119.03 .387 1n SMP * IA *9.69 29.46 -.558 constant 116.09 89.31 EPC .0058* .002 .435 FU 10.59 15.63 .117 IA -19.29 22.69 —.264 IA * FU -2.62 23.55 -.035 constant 12.36 16.11 * Starred estimates are at least twice their standard errors 0 CHAPTER IV APPENDIX C REGRESSION ANALYSES: FACTORS AFFECTING HEALTH OUTCOMES Regression of Life EXpectancy_on EPC and Centralization Indicators, Interaction Medels Independent Parameter Standard Standardized Variable Estimate Ergo; Estimate FCI —.140 .226 —.222 ln EPC 3.97 2.37 .630 FCI * ln EPC .018 .034 .202 constant 37.21 17.34 IA -2.74 11.01 -.121 In EPC 4.62* 1.28 .734 IA * 1n EPC -.132 1.61 -.030 constant 32.77 9-87 F/U Status 1.95 8.57 .070 In EPC 5.44* 1.07 .860 F/U * 1n EPC .036 1.21 .009 constant 24.41 7.79 Independent Variable SMP ln EPC SMP * ln EPC constant Parameter Estimate -.3019: 2.46 -037 51.52 185 Standard Error .158 1-55 .026 10.22 Standardized Estimate -.644 .391 ~374 * Starred estimates are at least twice the size of their standard GITOI'S . 186 Regression of Population Per Nurse on EPC and Centralization Indicators, Interaction Models Independent Parameter Standard Standardized Variable Estimate Egggr Estimate FCI -109.32 58.12 -.896 ln EPC -1770.48* 607.68 -1.45 FCI * ln EPC 11.24 8.44 .632 constant 16127.29 4442.09 IA 1075.01 3068.33 .246 ln EPC -747.41 358.05 —.613 IA * ln EPC -210.13 449.77 -.247 constant 6713.30 2750.40 SMP 16.91 47.85 .163 ln EPC -613.28 469.42 -.503 SMP * ln EPC -4.61 7.84 -.240 constant 6586.59 3088.72 FU Status -3323.76 2322.80 -.619 1n EPC -1150.87* 290.65 -.944 FU * 1n EPC 415.97 330.33 .560 constant 9916.01 2111.34 * Starred estimates are at least twice the size of their standard errors . 187 Regression of Calorie Supply on EPC and Centralization Indicators, Interaction Models Independent Parameter Standard Standardized Variable Estimate [Egrgg Estimate FCI .276 .373 .380 In EPC 8.03* 3.89 1.10 FCI * 1n EPC -.053 .054 -.502 constant 59.27 28.75 IA 12.22 17.55 .469 ln EPC 4.63 2.04 .637 IA * 1n EPC —3.29 2.57 -.649 constant 79.42 15.73 SMP -.O38 .292 -.O62 1n EPC 4.51 2.86 .620 SMP * on EPC -.002 .048 -.014 constant 81.05 18.84 FU Status 13.58 14.25 .424 1n EPC 6.34 1.78 .871 FU * ln EPC -2.14 2.02 -.483 constant 66.06 12.95 * Starred estimates are at least twice their standard errors. 188 Regression of Life Expectancy on Centralization Indicators, Interaction Models Independent Parameter Standard Standardized Variable Estimate E359; Estimate ln EPC 5.16* .673 .819 FCI -.231 .149 -.367 SMP -.482 .206 -.903 FCI * SMP .005 .003 .960 constant 51.46 10.47 In EPC 5.26* .587 .836 F/U Status 7.39 8.28 .267 FCI .015 .115 .024 EU * FCI -.074 .126 -.223 constant 24.52 9.01 ln EPC 4.24* .832 .673 FCI -.098 .086 -.155 IA -19.36* 9.61 -.859 FCI * IA .210 .123 .774 constant 41.72 9.26 In EPC 4.10* .807 .652 SMP -.154 .084 -.288 IA -5.00 6.19 -.222 SMP‘* IA .051 .106 .168 constant 44.57 7.78 189 Independent Parameter Standard Standardized Variable Estimate Eggg; Estimate ln EPC 4.64 .563 .737 EU -.273 6.69 -.009 SMP -.161 .102 -.301 FU * SMP .041 .109 .115 constant 38.95 7-89 In EPC 4.52* .819 .718 FU .393 2.86 .014 IA -7.57 4.63 -.336 FU * IA 4.56 4.32 .200 constant 33.29 6.97 * Starred estimates are at least twice their standard errors. 190 Regression of Population/Nurse on Centralization Indicators. Interaction Models Independent Parameter Standard Standardized Variable Estimate Egggr Estimate 1n EPC -978.96* 196.08 -.803 FCI -46.82 43.62 -.384 SMP -23.43 60.27 -.226 FCI * SMP .265 .819 .268 constant 11762.70 3050.82 ln EPC -982.10* 154.43 -.806 FU 1855.08 2178.22 .346 FCI -10.96 30.21 -.089 FCI * FU -30.42 33.24 -.472 constant 9325.08 2370.94 ln EPC -872.43* 223.93 -.716 FCI -17.44 23.20 -.143 IA 3134.64 2588.39 .718 FCI * IA -38.24 33.12 -.725 constant 8728.29 2493.71 ln EPC —929.08 240.89 -.762 SMP -.145 25.03 -.001 IA 524.27 1849.70 .120 SMP * IA -14.13 31.68 -.243 constant 8037.43 2323.06 191 Independent Parameter Standard Standardized Variable Estimate ‘Egpgr Estimate 1n PEC -905.26 162.80 -.742 EU 2385.07 1936.64 .444 SMP 31.14 29.55 .301 SMP * FU -49.62 31.55 -.724 constant 6337.88 2283.44 ln EPC —876.38 232.44 -.719 EU -203.99 811.87 -.038 IA 289.26 1314.74 .066 FU * IA -677.25 1224.76 -.154 constant 7837.86 1976.19 * Starred estimates are at least twice their standard errors . 192 Regression of Calorie Supply on Centralization Indicators, Interaction Models Independent Parameter Standard Standardized Variable Estimate E2393 Estimate 1n EPC 4.06* 1.24 .558 FCI -.O44 .277 -.060 SMP .002 .383 .003 FCI * SMP -.0005 .005 -.090 constant 86.31 19.40 ln EPC 4.27* .975 -588 FU -15.72 13.75 -.491 FCI -.271 .191 -.372 FCI * FU .252 .2099 .656 constant 96.47 14.97 In EPC 2.056 1.34 .282 FCI -.197 .139 -.271 IA -35.45* 15.57 -1.36 FCI * IA .355 .199 1.13 constant 111.08 14.99 ln EPC 2.59 1.41 .355 SMP -.051 .146 -.082 IA -11.65 10.81 -.448 SMP * IA .047 .185 .135 constant 97.66 13.58 193 Independent Parameter Standard Standardized Variable Estimate £2223 Estimate 1n EPC 4.42* 1.01 .608 EU -12.72 12.04 -.397 SMP -2.15 .184 -.349 SMP * FU .201 .196 .492 constant 92.06 14.19 1n EPC 2.444 1.22 .336 EU -3.94 4.27 -.123 IA -15.94 6.91 -.612 FU * IA 7.83 6.44 -.298 constant 98.86 * Starred estimates are at least twice their standard errors. CHAPTER V THE IMPACT OF GOVERNMENT CENTRALIZATION ON EDUCATION POLICY Let us turn now to a consideration of the impact of political structure centralization on national education outputs and outcomes. Given that education historically has been a controversial issue in many nations (because, among other things, people tend to associate education with their own social advancement and opportunities) we would expect the structure of politics to have significance for education policies.1 Decisions about commitment and access to education, we expect, will be highly controversial and political. Moreover, a predominant view in the literature is that government centralization leads to greater education policy outputs and outcomes.2 First I shall consider indicators of education policy development. As in the case of health policy, one must distinguish between outputs or level of commitment and outcomes, or, short-run results of the policy outputs as experienced by citizens. Table 5-1 shows relationships among the education variables that I consider potential indicators. Given the Pearson's r correlations among these indicators, it is clear that two dimensions of education policy are represented -- an expenditure dimension (which I will use to indicate outputs) and an "impact" or "result" dimension including such things as gross enrollment ratios, literacy rate and the percent of the population in higher education. The average correlation among the expenditure variables is .98 and the 194 195 8888 ma. 3. 8.. S. 3. 3... ms. 2. Beam 8. mo. 3. no... no: 9.- no: 68 3.- am. NH. mo. 2. 9. am» so.- so. mm. so. do. 2. or. 8. a. me. me am. om. mm. a. .R. mm. mm am. an a whopmofiocH hoafiom coapmosom one We osom mcos< mesoaoammmou coapmaoHHoo m.uomnmom Hum manna a“. so.-. no. no. 2.. arm ma. mm. mmo no. as. «N. am. on. 00.- no... mo. 3. mm. mo. ma. NN. No. 3. mm. momma macaw xmnm mm» 2 NH. 3. mm mm B firm Emu 196 Table 5-1 (continued) LR GER SHE Si S2 SB T1 T2 T3 YSP EDEX EDGDP EDEXPOP Literacy Rate, 1970 Gross Enrollment Ratio, 1970 % Population Enrolled in Higher Education, 1970 Number Students, Primary and First Levels of Education, 1970 Number of Students, Second Level, 1970 Number of Students, Third Level, 1970 Number Teachers, Primary and First Levels of Education, 1970 Number of Teachers, Second Level, 1970 Number of Teachers, Third Level, 1970 Years Schooling Provided, 1970 Education Expenditures in Millions U.S. Dollars, 1970 Education Expenditures as a % of GDP, 1970 Education Expenditures Per Capita, 1970 average correlation for the outcome variable is .46. The average correlation between the two groups of indicators, however, is only .25. To simplify the analyses, I have chosen to use education expendi- tures per capita (EDEXPOP) to indicate outputs. This seems justified given my concern with the social well-being and opportunity structure for individuals in society (rather than, for example, being concerned with additional increments to the percent of GDP spent for education). Choosing from among the outcome variables is somewhat more difficult. But several indicators seem to be more interesting theoretically than the others. The percent of thegpopulation enrolled in second and third levels of education (SHE), for example, represents a cumulation of educational achievements and successes over the several levels of 197 schooling. It indicates opportunities for higher education. .§£E§§ enrollment ratios (GER) on the other hand, indicate how well educational policies are implemented and whether outputs reach target groups. (The figure is the percent of children enrolled in first and second levels, given the total number of school-aged children.) These are important figures for characterizing national achievements in education. Moreover, literacy rate and the gross enrollment ratio correlate at .82. By using GER we are indirectly tapping this other important education feature. In sum, to indicate education outcomes, I use SHE and GER. Bivariate Analyses -- Determinants of Education Expenditures Following the strategy used in Chapter IV, I begin this analysis by considering bivariate relationships and "partial theories." Based on the writings of Pryor, Peacock and Wiseman, Cameron and Hofferbert and others, we expect that, in general, government centralization will en- courage greater commitment to education and greater education outcomes. Yet, given our earlier findings, we might suspect that the relationship between centralization and education policy will be more complex than earlier literature has led us to expect. The impact of centralization may differ depending upon the aspect of structural centralization or of education policy under consideration. First, I consider the role of constitutional status in determining education expenditures. The regression equation is: Y = a + b X -+ e 1 1 F/U status, 1970 1: 13" (D H (D >4 ll stochastic disturbance. (D II Using sub-national, regional data, Cameron and Hofferbert find that federalism ultimately reduces education spending. Similarly, Pryor 198 compared nation states and concluded that centralized structure has a positive impact on educational spending. On the other hand, Beer argues that federal structure has an expansionary_effect on national spending. (These arguments were reviewed earlier.) Our findings are presented in Table 5-2. The FU variable is very close to the .05 level of signifi— cance and explains about 5% of the variance in EDEXPOP. That is, in the bivariate case, FU is marginally important to EDEXPOP. Specifically, unitary structure appears to have a slight negative impact on education expenditure levels (EU is a dichotemous variable where 0 = federal and 1 = unitary.) In other words, for the bivariate case, the Beer hypo- thesis is not clearly confirmed, but the Peacock and Wiseman, Pryor and Cameron and Hofferbert hypotheses are discounted. A clearer picture may emerge with multivariate analyses. Table 5-2 Regression of EDEXPOP on Constitutional Status, 1970 W=5m Independent Parameter Standard Standardized Variable Estimate ‘Egrgr Estimate F/U Status -52.26* 26. 73 -.274 constant 97.87 23.76 R2 H2 F .075 .055 3.82 * Parameter estimate is significant at the .056 level. 199 Next, I consider the relationship between fiscal centralization and EDEXPOP. The equations are: Y = a + blx2 + e (linear form) Y = a + b1X3 + e (logarithmic form) Y = a + b1X2 + b2X4-+ e (polynomial form) where X2 = fiscal centralization index (FCI), 1970 X = 1n FCI 3 2 X4.= (FCI) . The equations allow us to determine not only the direction of the rela- tionship and the impact of FCI, but also the form of the relationship, i.e. , linear, logarithmic, or polynomial. Tables 5—3 to 5—5 and Figure 5-1 show results of these regressions. Fiscal centralization, too, has a negative impact on EDEXPOP in the bi- variate case. But, as becomes clear from Figure 5-1, the impact of FCI on EDEXPOP levels off at high levels of FCI. That is, the form of the relationship appears to be logarithmic. Once again, earlier hypotheses are not confirmed by bivariate analyses. To test for the impact of informal political authority concentra- tion on EDEXPOP, equations of the form employed above are considered. Once again, the literature leads us to expect that the more centralized the political control, the greater will be commitment to education. Results from this research are reported in Tables 5-6 to 5-9. Bi- variate analyses demonstrate that SMP has a slight negative impact on EDEXPOP and that the relationship is a linear one. (The curvilinear forms do not improve on the R2 of the linear form and the quadratic form is not significant.) Moreover, when interest articulation (IA) is 200 Table 5-2 Linear Regression of EDEXPOP on FCI, 1970 (N = 59) Independent Parameter Standard Standardized variable Estimate Eggp; Estimate FCI, 1970 -2.45* .521 -.565 constant 232.85 38.64 R2 HZ F -320 .305 22.01 * Starred estimate is at least twice its standard error. Table 5—4 Logarithmic Regression of EDEXPOP on FCI (N = 59) Independent Parameter Standard Standardized Variable Estimate Eire; Estimate ln FCI —160.89* 32.80 -.582 constant 738.72 139.38 R2 H2 F .338 .324 24.06 * Starred estimate is at least twice its standard error. 201 7 Fiscal Centralization Index, 1970 if" (.49 ' , 1 1 (N = 59) o . l gure 5-1 IT I I l . 1.. .. . ~. ... .. .1. . 1. ... — m ... .... « * 00.. . . v- - M. . A. . LN _.1.1.—”.._ wn.. .u-a _.”.—.-. .nn ...H. ._M. —”m .1__m M.‘— _—W* o n? .W 0... rh-lt-IOL .01.. H1”- .0.-..n» . .. .v.. . ._..i. . ”....._n _*.....1.“ .. ... 1*.1 ... ..hhalmrvitrh..--LLIHTIHWATILJ? .-.H.+ .1.4. . .. . ..4. _. V -:.. .:.:..: ..:.:qu:z.1--11::.t-tl.lttxt:n,: 1;.:r.:.r..:1 : :+ 1 1 .1.1 Iltlrrtrll...‘ trips titty-tut). 0. . . c . .1..+p..1... a.” .... “a u p... ... 1.”. 1.“ .... ...J .... o .. ow.. . . . . 9 . \ ... ... . .. . ... e . _~. .. ._. .1... . . ... .m, .... _~4. .. ... .. . .1 .0 _ o .... 0» . . .... .. . o .. —.~ . .r.. H. ] ._.. .... ... .... u .- . _.1. w. . . .—.1 r '0 0.). Obit IT}.- .II..«\ .- . ... .... . . .. o L. 0.0. ..v _ L ... ~.. m.. __.— .. . . .H . .HF? EVOLIoA . . .\ 0 o 8.0. ... . .. . . a 9.. 1 .. 1. . .....- .[0L-bvvlib 0&7. 0W fi- .. .e . .. o . . n n 0 o - . . - o . . . . V ..b ...; il—rrf .oL.v- 5.5 .40910.- 1.....916 .ILI. o ..H .0”. . L.. . . ... ‘- —. .. .1 e w . o u . .o e c . . ) titrito- 0.0.-r1t'01.u.0.-..0Y01.r.. 1.1? 1... ... p..... ... ,.. .... ... ._ ._ . .... .... ... r. . .p. . .... -. to O . ...—... . .. .... .h .....U»””. ......__..1.. . r. 1“.” 1.. ....m............ .... .... ....1— .... . .. . ... .1 .HH. .. . .... .. . .... .“1._, ....J 1.“ 1.. .... .1». ..H ..w._...r11..1 ......-...rI -ot..-1.-i.._... . ' .... H... .... . . . .. . . . .. .u‘ . . .. . .h.» bib; 1111...; .olr . .Plil..s1|7.xr.0¢9.lr.t’ ...-..vl..—’-:q.u. . .. .1 . . . ..... u - ._-. -11-:1::1r ..-._ .i4n111. - .1t 1...; 2.1: .;..:1..v.1: : .1 1 \\ 1 . i Li. 1 :1; H. . 1.... . ...w 1. g .11 .1 11.11.111.11. 1... . 1 . 1.13-..- . . n . w , o . . . . . . . e 1 o . . . _ . A. . . . . . . .L . e . y . . . . . .H. “1.0I‘Oi .....0 1. 1.0.0... 0 . . - 4 Z. . ,..\ . . . ...; .. .. ... .... 1. i . “1....- --.1=..r. 21:111.}. 1.4.11: ...-.7. ....1 .... .11 .e 1C. c L . . . r ....l.’ .OI'D .-10altl‘n .111 r1! 0... ‘I‘. int-0.6. I” q. .. . . . a . . .. u . . . . m w .m . m*. o —. . . . . L . _ a . . .* .Illv. -I‘... ..., . . . ...1. .. ... . . . ...._ . ... ..u. .. . _ ...“ _.. .11. u \\n 1 O .C ....... .... .... . ... .. .... .a . ...H .. ... . . .__1._.u...._ .. ..u M". 1. " ~_»_ ..... 1 1 . ... \.... . ... .... . . . . o. . ... 1 _ .. . . 1 ...m.. .h.» .... ......“—. ....g. . u . 1P. .. -0... ..v. a! .0-_ ..‘\. .E 1. C. 1. . . . . . . a. 1. . .f. . .h -..-...... (1.. C.,...4133 291...“-.. 7... 11.11 .11. . .3 . ..x. _ _ . .1. . .-- 11.22.1111! ......12. .. .....:.:._1.: 1.1.11 1-..... a .. 1... 1. . fin nu‘ 0 ..Afi ,. ...v. .... o.“ H. e . .... . ._ .... .... ...1... . ... .1 .. . . .m.. N... .. »\ .. . h. 1” . . .... ... . .. u. . . ... ..... .1 H. . 1.. .. .”.11*......... r'! -\ \.a..-- --....1.1-.Y?- . . .. ... . 11 ... .... .. 1 . 1. . 11...;,-.-LL1$-.-.1.._. .1... z... 1 .1 . . . o . .. .1. . . ... .. .1 AI... 00.. .0I. ...YVQV..F o l. '. ....»I...0:. M. . .... . a .1.. _ .. 1 s\‘.‘ . . . .. .. . .(n: ..--.0100... -10..-. .-.t- r $.16.- n-.1.. . -. . .. 1 g. ... ...“. . . _. a 1. .g. ... . . \.. . . ... . . . . ‘51.—-.-0-0 .1 .. .. . ... 1. . . .... ... . _ . .1 .. .1 ..w ~.11_.»~.1m"1.r 1. .1 . . .. O . . .. M.. .... ... .. .-.. .... .. ..u 1. 1.. 1 .1. . H 1. .. .. 11.1.1.1. 4-- .........DD.. -§I.~.--lwi.o . .... 1.1 . .... . ._.... 1i. ..u“ 1.“.....P... 1.1% . .1. w_rl_ q_>” . .1.» .VI. >...1>TP...¢LL0 ”A . . ~ ..u.. 0‘ .. 1.... .... .... ... .. . 1 III. I o .7-I.A .00 at. . ...... .. .... .. 1.. . . .. o . Y . P . . . . . I: .01710-—P-PA.III .10.?51..'l. . .1... Li. _.A. 0.115. 0. . 1 m ..u . 1 . . "~_ .... ._ . .. — - . u w H U _ . a“ —. M. . . f. . .. e 1 . u . e n _ a. . w 9 I'd-I .0..IV bin-D ... . . . .... 1. ... “L... p .n.‘ .. .... .a. .. 1... ._. 1.1~.L.. .- .. N .. . o? . — . .. _. ... ... _.. ... .“u ... ”i.. . 1.” .u . .H._ “.1. w“. H. . .. ,. [Vb-Witnvtlo ...Olt -.l.!.YI-.O . u a . .. ... .. . .. .H.. . ......A ... ...—g. ... . ... .P ...F; i . .. 0&5. . In. . . ... ... .1 . . I . . 1 _ - . . . —.-r-0 fettlhrt...fu1ebcr1..fe.+ ..Y-o.-.0rI-H.,.l...”~“ 1v. . w." .. 1 _ Q . . .. .... .. .... ... . L00 .0. oonJwivl -|-00.O.5F.tl10 ”W”... ... .. . ... ~.. Mn". ..u. . k. .wm ..w ...» — .. -. . ... . .a ....—.. n . .. ~. .. .... .. . . . . . . .. .1 1 1 . 1. . . . 1.”. .. 1. . . . . . . .... h * .. .. . ... m .. ...“ ... ”m.“ .—_m .4”. ”h..;#.... “1». .m_w ”N.J.fi.r—n ~ _~.~ r..-‘ 501lm11.ql o—vl . 1 . ... . . .. . .. .... .. . 1 . L 1.0.. . . . 1. v . . .. . . ”H . . p . o . .— 1 . . . . . . Io! 1.1.07.0 01 oft-r610 LilT-A-g-O 19.-.0f1fi0104t . .r .. . . . . . .. . . fl . . . . . . . . Lg-.. - . . .41—.-.-- ......1._+.-..1 :r. 1 l1. 1: “42:" :1; x”. l “1.. . rut... .1... L. - ...-r -1- . .. . ... ...1 .. .1 .... . . s 1 1,11 . . .1. .. ... . ,., . . . - . . . 1 _ . . _ 1 . 1 . . a . .. , 1 1 . . . . 1 . 1 Tel. ~ . .. 1 . 1 u . . . \ . b... .I.- -... '0. I¢.l. .... b| 0.. .-PI .. . ... . .... .1” m _.*.n”“ H... ... ...u . .. a. ..\ u... ...(LFAHHUTM . £14.13... 0.109 Hirvqonlwm... ” . U .w Ir ...H p .H .. .. 'LHYHLHLPWOWHVof ......fi- . a. ...-0.0.9.. ...1MLI1A1a.h.v. .....014.... m.ma :N. a... ...h ”M.. . ..." a.. . . . . 1O I.lf.--..-.-vvv-Iitoc-0fi..06..-46... a.... .. .._.~..... ... ..m. .4. H T {7.2.1. ..1.. 1...... w... ".7 1.1.1: :1: . . ..fi. .. .. L .. ... ... .. ...” .... .11m.i m .n. ..__ —... 1.. . “”1. 11¢. ..._ .Lu .9. 1. ._ _.. .n.. .1 .. . ._ . . O . c .11.... s ...4' . .. .u. ._ g .1 ...“ a —. .“ ... ..p” u._. _... I. . ...“ kn.1,._. ... 1 .... 0ft :0. Q0051-‘-V|.I'V§b.--s.v0r.ll ...P|.. 1| . .. ... - . . . .1... ... ... ..1 ...a .. .... r»....9f.. .0. 141»; 1 . . . .... ..1 .. _. . ...w u . . . . a. _ .1.v|...|...'-Df 00.+ ff»- Volt-0. FvLJ-L..T..Jk'& . "17.. 1.». ._ ....‘ .... ...m. ..u 1 . . c .. ., .... .XIO.‘. 0!? I .I It.’ I...0-.I ”“10. .4 .— . e . . . c . . . ..i. “H.” ”Hum “M . we . n ... .nu. .. . ““1. .... . . . . 1 . . . . . .. . .. . n . ... o .. u. . .. . . . . 1.... 1.. 1. “ii-U. 11.1; . 11 1 .1 _ 31.13... ’ in , u . . . . . . . a. . . . . .._1 ... ..n . _ .. . n... ”.1 u-. . . ... p. ..104 I ... I.-. 001.91.09.03! 0 n- * a . ..O/o . . . . _. .1” . . . ... .1. . . . . .... - .1 .. 11......” .11.? liq 1 .. . . J . _ . _ . ,1 . 0 . 0. . . 1. . . . . .v1..0t. 1P... 7.10.0.1 .....- 4-00 .I. . ... . . . . .. . . w J . a I, vIIII. II. In .... . 0| 0 ”it . _ . _ . L . . 1 . . . . r _ V” . a . . « .~ “1 . n . . . . . . . VOI- vl“..l ill... ol" '4 . a . . ... ... .. . _.. L.. ...1».. .— ... . 1 . . 4.. . . : :1. ._ _ .1. i1: ..1 . . . . . 1...... 1. .1... _._. . .1 --..l..+!-I.. v. a . .. .P . . . .. ...o . 1...... .. ...‘L LV~ . u n- 6 . .. . . . _ a .—. .. ... _ .F—.1 F V I.».'.p..\. 4....- ‘luv . v — .. ... 1 .rlblotl'.rl V21»? -.IVHAvtlebif 7.0 -0 0. . .r . \» ._*. — '1. ...-0.17.. r .31-{41.211014 .1 1......1h11fi.11 ... 1 1.1 1., ...1..-\\...H ... . . 1. 01.1. , . . ... a 1. — .. ... ~.. 1.1.. .. .... .... F \ ...g. . .~ .. .... . . ..-. . . . . . e. .. .. . L1. ... ... \. ... .0“ b 0? ..l 1. . . . . u . ... .... .... .__1. k“ M _. ._ ...\. ... «4v mo. .. bu . .1 a. _ 11.. _ ..." .m.. . "H .\ L...» .0 Y&.t... . ,. . 1. .. . .. .1117 .1 -. ...-..n... -111... .... 2......1. 11. ... . .. . . 1 . . .. .... .1. fr.- L...1 .-t 4 1 mu . ‘1. 1. . . 1 1 _ 5.. 0‘. ...-1190.0! [-0- c. t 53-..: 0.91.1'... 1. . .. .i. .L. .... .L. 1. P‘ __1. __.\L.-m _... . . . 1 I. . . . . 1 y. . .. . — 1. . . ...—. h . . .... . .- ...u H. ”h. n. M. m .... o.» h. H ad.” —_.i .u. .~ .1 . . . . H . .. . ... ..l." . ... ... .. . . .. ._ .. . 1 . 1” 1:11.. i. 1 .u. _ \YW 1TH .... 1 .1 . . . _. p. . or... .--. h: I... (I ...-...-u‘lt.-[1+.-i.1ir+1.fu\-_ ”....“ 1.1. .. M . ----£313.31... .1. .1. 1. .1._. . . _ ..1 . It- o!.v-.0.0-.i..l..1 . _ . . .. . .... .. m .. . .v u m... . . ”luqnflu .fil . u . . v. .. ... .. _ .1 . .. . ... _ «n a... :1 . ... .t..... ...W . .---.1..- . 1 .. a... .. .. . . .... ......1rl....»1.L.F1h.i 1.4. “.... r . o. . ... o. a. . 7.} 1LI0 # 11.0 I . . . 0......01 ...hlv.?,..i 1.17..-.1011va'1-0r.(.. .Hv. .0. r. 0.....11.i.. ..u.. .... .M.“ h; .uu “h”. _mfim , l — 11.61.11.111“. 1 ... ... ... 1 .. . _ “...”.1..._...... ... . 1 . ,w .... ... .....— . .1 . . . .. . ... ._ .H 11 Wm. ...WM ..u” .M . ...__ q»_m H.1HMUN“ ._..r ....tivst- 1r 0 . . u .. . n . . . . .. .. .. 1 1. ._ to..- . e. . . .... ..—. .. —. w. . .1 — a . .Hntadrw.101-.HIH-0'Prt.7-HL rt. og.>§§1b ...-5.00.01? .1“*”,.7 “h.” 1 ’ m. It... :1.- ..1.... ....-. -..-..-.-.ts- - .... 1.. I... .2 _. . .11. .1_. ..1.. 1: . 1 .- ..-.. .fi .i. ... . . 1. ..H _. .1131... ... .1”. 1. .1 :M 1 m ”.11. ”.1. ._... .. . . . ... . ...n . 1, 1. ...... 1.1112111171. ....l _ I ... .-.- .11.. - to .M _. . o . .1. . ... -.. ..." ..... . o. . . t. . *be I‘d ... - n 1- . . r. .. ..IL.‘ ....111-03016 .‘Vlroru 1.01.141 PL MY...“ .4 .44 ...a. . . . “...-...!!! -12-1....-.a-1- -... - .l.--.-u-..r.-.L.1.m..,1 .. _. 1 1.. 1 .1 1:. _ . . 1* a . .: 1.... .. m . 1 _ . : 1 . _ .1: 1 11. .1: . — _. .. n 2.1... ...1" ~F». . .. .W. ..Ob-hl . w.. u. . h. I eru-TII.... “1|... . 1 . ,. ,. . . . 1.5.. .... “"4“ ”up . ~.amh “ w. m“~. . u . .... . . 1 , 1 .H. "a. ..M.~ m“ .m. . . w. .u.. I .-..v-ietzs .a...11 ..L . H- . if.- h P 1 .1 .---4...? ..a ..1. 11 ... . . ....1 1.1... 1 1.: . . *.« ”m y .o .p .h.“. “una . . ... . . . 1: .11 11 .1. 1-1;... B., 11.17.. 1|L (JILL. Ir F. 1.11 ...L .. 1. ... ..l... r 2w. x .T w. *5 4.." WODWwWo . .0 w“. awn”! .» ...fil... 1.. :. _.HHM»1_ ..fihfl monspfiosomxm soaps sea 0 (x 0\ fl 9. (d U s... a)! 04 202 Table 5-5 Polynomial Regression of EDEXPOP on FCI (N = 59) Independent Parameter Standard Standardized Variable Estimate E2293 Estimate FCI -8.68* 3.92 -2.00 FCI, squared .005 .028 1.05 constant 431.17 129.40 R2 §2 F -355 -327 12.70 * Starred estimate is at least twice its standard error. Table 5-6 Linear Regression of EDEXPOP on SMP (N = 59) Independent Parameter Standard Standardized Variable Estimate E332; Estimate SMP -1.25* .503 -.30h constant 132.22 32.31 R2 fig F .115 .097 6.15 * Starred estimate is at least twice its standard error. 203 Table 5-2 Logarithmic Regression of EDEXPOP on SMP (N = 59) Independent Parameter Standard Standardized Variable Estimate Ezggg Estimate 1n SMP -69-54* 29.51 --325 constant 337-63 119.75 R2 fig F . 105 . 087 5- 55 * Starred estimate is at least twice its standard error. Table 5-8 Polynomial Regression of EDEXPOP on SMP m=5m Independent Parameter Standard Standardized Variable Estimate Egggr Estimate SMP -.962 2.85 -.262 SMP, squared -.002 .021 -.O79 constant 123.88 88.31 R2 ER F .115 .077 3.00 204 Table 5-2 Regression of EDEXPOP on Interest Articulation (N = 59) Independent Parameter Standard Standardized Variable Estimate Egggr Estimate IA, dummy -90.68* 18.31 -.585 constant 103.37 13.15 R2 fie F .343 .329 24.51 * Starred estimate is at least twice its standard error. negligible and informal political control is centralized, education expenditures decrease. IA appears to account for about 33% of the variance in EDEXPOP among nations. These results shed further doubt on the earlier hypotheses. Based upon these analyses. we must argue that when informal political authority (of political parties and of interest groups) is concentrated at the center, commitment to education is less— ened. Bivariate analyses thus suggest that all types of political structure centralization tend to lessen education policy outputs. Relationships in each case are negative, although the strength of the relationships varies by the type of centralization under consideration. These findings are tentative, however, and must be tested in the light of other considerations. It is also important to consider the roles of economic, social and 205 demographic variables in determining education policy. For instance, because a nation must have resources and capacity to support education programs, we expect that the level of economic development will be an important determinant of EDEXPOP as well as a contextual factor affect- ing the impact of political structure on education policy. Tables 5-10 to 5—12 and Figure 5-2 illuminate the bivariate relationship. 4— Table 5—10 Linear Regression of EDEXPOP on EPC, 1970 m=5m Independent Parameter Standard Standardized Variable Estimate E332; Estimate EPC .024* .002 .858 constant 13.8h 8.88 R2 fig F .736 .730 131-35 * Starred estimate is at least twice its standard error. In the linear form, EPC explains seventy-three percent of the variance in EDEXPOP. The more wealthy a nation, the more it spends per capita for education. In other words, we may consider EPC an important variable defining the environment within which the political variables operate and policy decisions are made. (The logarithmic form does considerably worse than the linear form in fitting the data and in explaining variance. It appears, moreover, that one case -- Norway -- is strongly affecting the R2 and fit of the 206 Table_5-11 Logarithmic Regression of EDEXPOP on EPC (N = 59) Independent Parameter Standard Standardized Variable Estimate E3223, Estimate 1n EPC 33.18* 4.05 .766 constant -1N8.86 26.12 R2 fig F ~58? .579 66.98 * Starred estimate is at least twice its standard error. Table 5-12 Polynomial Regression of EDEXPOP on EPC (N = 59) Independent Parameter Standard Standardized variable Estimate Egggr Estimate EPC .0#3* .004 1.52 EPC, squared -.000002* .000000u -.726 constant .11# 6.2h R2 E?“ F .830 .821 111.57 * Starred estimates are at least twice their standard errors. 207 __. . II |o.. Q . . II .OI II . ..«IIwJ I 'OII O l I I I I I I I . . I . .——.—.—.----g,-—..-.—o—.— .——._.—.._ --.... ——9 , l . ; I r- — ...-..— -.......-- .———__- . . . . . . . —.... .-...d—Q _'c _-.—--- 4-. .1 . . . a u I v n I, I -D ‘1 Ilv III In .0 I I . ... --IIA 1%.) .11 11-1'1. . - .1 :59 ‘ L~4~~ . .. --’--.1. 4- r 13’. $35 1330 b VQ "--- 5‘.- n . s d . 1”. 1 . . mm.4_ .I.. .4. ... .... . 1.... 1. — .. 0 .fl - I I 0'. ... ... o ..t c. 1 o. . u 1 .. . .... ... ... ... ....fi . . . o . o 0‘ ~.- .... ... ,. . .. 1 ¢ . ~ . “9.: it tirl" Ill‘lv._n 1 .II' '10 II VIII.-.“I. 1-04IVIIIID .v ,I’.’ IVII.lIro.IO . po\ .. —I. . .. ... .. . . . 1 3.; 1. 14...: . .... 1 1 _ . ..o O.. c .1—.. —. . u _ . s . 1 ..‘s IIOIIIVTI'YIILIIII. .....- YIIII'.II‘¢III1I‘.IIO.I 0 I117: II .IthI .I 9 .VIIII I .l .I 1. .. . w .. . _ . . 1 . . , 1 _ _ 1 . .1 xxx?” _ . , 1 1 . 1 1 . ...1 ‘III...‘ quI -..IIIIbIIYIIDIII...I.I . I~ I ...-II; I I A.“ a. .. . . _ m _ . 1 ( .. 1.. .1 .. .M _ . 1 u . 1 \. ... .. . 1 u _ . .5. T’IQI IIIJIoI oIIII~ICII¢IIIIIII*.II~I III—I IIoI I In . . 1.. a.. g _ 1 1 1 1 H .2 . 1F. 1 A . . _ . 1 T, v|.-.I ‘4 Quit. at 9.. I' I II'I‘ . I . .I. n ’3 I I In: C. u I n ' -.~ . ‘1 u 1. . . .Ar .1 / . (I .1. . . 1 .u I . . {I‘IFMI ..’I TI .lrIIII III. IIII..IIII +II . 1. . - . . 13 . . .... 1. - .. . * . . I . VI . n . . . _ _ .4.- 1. ... .1 1 .. 1. . I I" III. EA. |.Vh IP. .I I . I IIII IA. IIIIIOL ..II‘II .IIIIIOI III III. II .I II . It. I . .. . . .1 _ .1 . 1 . ., .1 1 . 1 . 9. . .1 . . .. o » . 1 1 . 1.. .m 1 o 1. q - _ .. 1 . . 1 1 . 1 1. . . 1 .. 1 . u! o. 1 1 . N 1?. .n ...11 —. .. 1 11.. 1 ~ ~ “ h “.1 III I PI I. I. [a I OI I I.II . . . ‘0‘ ..u. I... I I on I I........,. I . I o: I .o . 1. 1 . ... .1. n y ... . 1 . ... .. . .4 . 1 . 1 . . . 1‘ 1 . ... o 1 .1 L 1 1— - .... . . . .1. .. ... —_.. .. ._ . . . . o w 1 . _ I... .. . ..., .. .... ..JI'I. _ 1p . 1. .1 .1 1... .. u . ... «,1 ..1 . . * M . . 1 N. .. . . ... 1 ‘_. [..1 ....J‘.» .1 .o . .1 .K. . .. u. .1 1.1 . . 1 . _ o‘ DIIIII- IIIIL I‘..LY . I on. 50..- IJAI I or... th.n.."? .1...r I I4 AIIIA I.I III—v.10. «I 54 OI .10! IIIII1 I-| Il AYIII I—III. +IIII.5l 1? O. I I..II I ..z . . .. a.. ..h .1... . 11". . ..I‘. . u .1 ./ . 1 1 ... ... . ~ _ i p .- .. .. 1 1 14.1. 1 1. . 1./.. ..~ .r.9 . _ .. 1. 1.1 .1. 1 1 _ ”a. .. 1. . .1 ....... .11. .1/...1.1.¢I. .1 .1” . . ..u11 . _ a . .. u. . ..1 . . . . .1 .11. .. .. — . I 1 . 1 .. .. r. . . I -0]- Qt ueol o. ”v”0.usob .u”. 0..." s ”.4“ 1f11‘." . I147... o OI we QI—I ..I XI no.1. III I In”... . I .... H I “ .... . —.Q . .1 . . .... . .a .... 11._ p.00. .11. 1.1 .1 1. . 1 . . . _.1. _ - . _ . 1 .. . . ... ..1 . q _~. 11... 1.. ...oI ..1 11 . . I . .1 ’4. . 1 5 . 1. . . w . 1 a 1 n . . 1. . . . 11. 1 1 ... . 1 .11, .1. 1. 1 . . .1 . . * . . . . 1 ..d .1. n”. . .1” 1.. 1— q... ..u ... ..f1 .1 1.. l7. / 1. . _.. 1 . m . — n 1 . ‘7 II?Lo<'.I.IA.v L v.'.b1I6Aoo h a {0.5 w—. o a on.» .r.. 'I w VI J: c..I ._oI . C.-. I III.- II a 0... 1. 4. . .I . .. . . . a . .1_.. ..1. ., . ... 11.... . . (.1 . . J ..... c . . n a . e . . . . . 1._ . ... .. .. .... ..u 1’1» . .. .1 o’fi .I’. a . Q . * c 1 . . . . 1 . 1. 1. .. 1. 1. 1......Ik11 1L 1.. . . . .7 1 r _ 1 11 1 . 1 1 _ .. .1 .. . ....1 7.. . ... 1.0.0! . .b . . . .1 . J 1.». III: IIOIIY.III ‘I. I iii-.1 Ib.‘ I oIfi II. Y’IrIVIa1 III *II ra I ”II 91‘ ..Iv..l ... II’IWII IIII.I..| II.‘ .IIIIJ. .0 lo I n . IQ I e . . 1. . . .... . . . 1. .. . _. .. . ... . . . . . 1. 1. 1 _ 11.1”.._ “.... 1.. ....1. 1. _. . m . _ 1 — m . 1. . . 1. 11.. .U... . 1. .. .i. TI f 1o .1 _ . . o. .. 1 1 H.. ..1— . .1 .1 —. r ..1 1 1 . 4’ .1— . . PI . o . n .‘ IIIII'I.‘ IIII4I.I. IIII ..IIY..II4I IYIr ..ILIIrIL .IIIII0-.—II v.9..IIoI .. ..II.I yo.l_.61. IIIII . II...IIIw. ... . .1 . . ... . . u # 1.1._. . .... 1 ... e o I u e . — ... /. . o c . _ .. ... .. . ... ...u:_1. ...”.4 ...m 1. ._1”._.1 ... . _/ ... 1. . o I. . . .. -. ."11o. . . ..1 . 1 .o ..11 o 1 . “[17‘ l I _ . I * I 1 .1 . . 1. 1... ... . w.. . . . . 1 .— . l- O .0 _ . . . IIIIIIIIOII. I I..oI .I I I . u. c. ..I..o. not» 1>I‘ I--.» .v .21 . .....Ir .-.: . I... .II.... . II. III III I 1.II . . . ... . .4. 1.1”...” ..w“111. n.. .o.. 11_11. s... A 7.!“ ~19 M I4. 1 . a at . . .11.. .. ”.....m . .u. ... .... ‘. ..u.. . .r .1. , o I’ll. JI.‘ 111 o o . a . . . .. i121 .... 11.1 .. . ... 1.31 .. Il1.I.1 1 a .4 I 1 . .... 1. ~....... ... I... .... .... ... . . l. o ‘ 'r .o. O O . so Iolo. I‘I’..-I I. 4.. _...v. ......OIo19I1 ctr. 41*- :I..—F uO‘II ..1 ~14. ..v. I I .ftI.I@II.Iu .. . I’IUV... ..I O I.- o O II u~ I .. .u. .u . .1. .. ... -. m ... .... .. . _. ... 1 o. . . I.. .001 1 .1 7.1..71.131111131111111.1:1:1,: N11 1 1.1.III1...L_.I.- p. .7./11. ...1 . ... . .. . > .n n .1 . . V . . u . 1 I. o‘....'..0 I I . II . IIIIIP . .Iw Ir L.I.I . ti. 1.7!.7‘L113 . . «179“...24 14A «II1. m.$4.‘52 IIIII1. II .tlilIIlifu I 1IPII:14‘MIIJAIIOIII0I.IJI.J‘. I . ...JI.-.II.I\!1UI\.I }?~II ..lIl Q. .. 1 1 . v . .o . 1.. \- ) "\J K . 1 . \n 1 r I: . .. . \ . I :rp...m~.¢ _ 01.1.1 ..fl 111M $1..-.“ t... ...... .. 1. .v. 0 no 9 o m o c > b I . . 1. . . ~ . . _ _ . . . I. . . . O . 1 In. . 1 . . 0 . . I L I” 1 . (.x . . O a . .V.~‘ rad . O a i . ad. .\ .1A. u. 1451 I1... ...1. a. ...1fi. 4.1.5. I _ .. 1 1 . a 7 . 1. 1 1 1, T- 117...? .1 .1. 1 .1 .... 1 1 . .1. --.I.1 _ -.. . Consumption, 1970 (Kilowatt hours per capita) Electric'Power 582 (N: igure 5-2 q I'.‘ 208 quadratic model to the data. A linear model seems to best fit the EPC and EDEXPOP data.) Next, let us consider bivariate relationships between social- demographic variables and EDEXPOP. That is, social characteristics such as ethno-cultural heterogeneity and pOpulation needs may well affect education policy decisions. Marked or extreme heterogeneity, for example, may affect allocations to education for reasons pointed to earlier. And the age structure of the population (the number of school aged children) as well as the population growth rate may affect EDEXPOP. Bivariate regression analyses show that both population growth rate and the percent of the population age 14 or’under have a negative effect on EDEXPOP (fiz's are .2h8 and .430, respectively). As the population and the number of school-aged children increases, education expenditures per capita decrease. In the bivariate case, then, the expanded need for educational services brought on by an increase in the population does not appear to increase allocations for education. Mbreover, social heterogeneity has no impact on levels of education expenditure. At the aggregate level, extreme cultural heterogeneity does not, on average, enter into or exacerbate the controversial politics of education decision-making. Again, the literature would have led us to different conclusions. Multivariate Analyses -- Interaction Effects There is reason to expect that the economic level may interact with the centralization measures to affect EDEXPOP. First, Cameron and Hofferbert found that economic deve10pment constrains the effects of federal status on education spending.3 Moreover, the role of political structure centralization generally may vary with level of economic 209 deve10pment. (We know that economic level is an important contextual variable.) The impact of SMP, for example, may depend on the level of national affluence, as is the case in the health area. Table 5-13 shows that two such interactions are significant at the .05 level. Both FU and FCI have joint effects with EPC on EDEXPOP. The effect of FU * EPC is to moderate the otherwise positive and strong relationship EPC has to EDEXPOP. Similarly, when both EPC and FCI are high, EDEXPOP decreases. What remains to be seen is whether or not these relationships are significant when a more complex model is tested, and, how these interaction effects change at various levels of economic development. One may also test for joint effects among the centralization measures. It may be that special configurations of centralization types create different policy-making environments. For example, whereas fiscal centralization has an independent negative effect on EDEXPOP, it may be that when there is also a high degree of majority party control in the legislature, the effect of FCI may be moderated or changed. Table 5-14 shows results of these analyses. Again, two interaction relationships prove to be significant. When both FCI and SMP are high, EDEXPOP decreases. But, when FCI is high and there is negligible interest articulation, EDEXPOP increases. The impact of FCI seems to be complicated. What these findings mean in the context of the larger model of determinants of EDEXPOP remains to be learned, however. These findings, like earlier ones, are tentative and possibly spurious since we have not considered a more fully specified model. This is the next task. 210 Table_5r13 Regression of EDEXPOP on EPC and Centralization, Interaction Models Independent Parameter Standard Standardized _2 Variable Estimate _E_rr__o_r Estimate _13_ y FU 17.78 16.52 .093 .808 59 EPC .036* .004 1.27 FU * EPC -.017* .004 -.523 constant --951 15-13 FCI .342 .356 .079 .853 59 EPC .064 .008 2.24 FCI * EPC -.0007* .00012 -1.40 constant -8.71 27.82 SMP -.310 .373 -.084 .723 56 EPC .013 .012 .476 SMP * EPC .0002 .0002 .362 constant 32.52 24.71 IA -11.35 18.0 -.073 .722 59 EPC .023* .003 .806 IA * EPC -.009 .035 -.02L» constant 23.33 13-55 * Starred estimates are at least twice their standard errors. 211 lglfle 5-14 Regression of EDEXPOP on Centralization Variables, Interaction Mbdels Independent Parameter Standard Standardized _2 Variable Estimate Egg; Estimate B_ 11 EPC .022* .022 .760 .749 59 FU 1.71 187.68 .008 In FCI -44.5 39.05 -.161 ln FCI * FU -3.55 45.45 -.080 constant 218.10 159.78 EPC .023* .002 .824 .731 56 FU -44.15 46.04 -.231 SMP -.424 .694 -.116 SMP * FU .357 .748 .147 constant 58.92 43.42 EPC .022* .003 .778 .738 ‘59 FU -37.64 19.96 -.197 IA -40.65 28.97 -.262 FU * IA 35.75 30.07 .228 constant 53.12 20.57 EPC .021* .002 .759 .779 56 SMP 3.20* 1.26 .873 1n FCI 148.87 84.12 .538 1n FCI * SMP -189.88* 74.41 -1.36 constant 764.50 224.39 212 Table 5-14 (continued) Independent Parameter Standard Standardized Variable Estimate @2133 Estimate i 11 EPC .020* .003 .706 .773 59 ln FCI -100.76* 32.36 -.364 IA -514.26* 221.51 -3.32 ln FCI * IA 119-74* 51.60 3.38 constant 444.16 135.66 EPC .023* .003 .811 .720 56 SMP .471 .636 .128 IA . 29.03 47.77 .187 SMP * IA -.752 .796 -.364 constant -3.95 38.14 * Starred estimates are at least twice their’standard errors. __ Multivariate Analyses -- SpecifyingLa More Complete Model Thus far I have tested partial explanations for the level of edu- cation spending across nations. A number of variables -- political, economic and socio-demographic -- have been found to influence EDEXPOP in bivariate and interaction tests. Here I move beyond partial theoretical considerations to specify more fully the determinants of education policy commitment or outputs. To that end, I regress EDEXPOP on the fUIl range of theoretically important variables that have been found to be significant in earlier analyses. These results are given 213 Table15r15 Regression of EDEXPOP on All Surviving Variables (N = 45) Independent Parameter Standard Standardized Variable Estimate Error Estimate EPC .028* .005 1.02 EU 25.76 18.56 .135 SMP .665 1.45 .145 Pop. Grow. Rt. 9.29 13.16 .106 IA -67.84 294.29 -.396 ln FCI 3.23 96.11 .011 EU * EPC -.017* .005 -.530 In FCI * EPC 2.14 2.11 .152 % P0p. 14 or under -2.41 2.13 -.259 In FCI * SMP -33.84 78.50 -.210 In FCI * IA 18.17 66.98 .466 constant 237.95 249.03 R2 E2 E .859 .811 17.75 * Starred estimates are at least twice their standard errors. It is interesting that after the level of economic development and the interaction term of PU and EPC are considered, all of the other variables fail to add significantly to the explanation of EDEXPOP. (see also Table 5—16.) Although population characteristics and the 214 centralization variables were significant in bivariate analyses, when a more fully specified model is tested with data from forty-five countries, these variables drop out of significance. In other words, their signi- ficance in the earlier tests was due to their relationship to and dependence upon EPC (or FU * EPC). Table45r16 Regression of EDEXPOP on EPC and FU * EPC** (N = 59) Independent Parameter Standard Standardized variable Estimate Error Estimate EPC .034* .003 1.18 FU * EPC -.014* .003 -.435 constant 13.94 5.99 R2 E2 E .817 .810 95.98 * Starred estimates are at least twice their standard errors. ** See note 4. When the model is correctly specified (EDEXPOP = 13.94 + .034 (EPC) - .014 (EU * EPC)), R2 is .810. Clearly, the main explanatory variable is EPC (R2 = .730). But the interaction term of FU * EPC adds .08 to R2 to bring the overall R2 to .810. The effect of FU * EPC is to moderate the main effect of EPC on EDEXPOP. That is, in unitary political systems, the otherwise strong, positive effect of EPC on EDEXPOP is moderated. This is depicted in Table 5-17. 215 Tab1e75:1Z The Impact of EPC on EDEXPOP* in Federal and Unitary Nations: Expected Values of EDEXPOP** (N = 59) EPC Constitutional E Third world 3E Total Sample E Developed Nations Status 329 1742 4557 Unitary 20.52 48.78 105.11 Federal 25.13 73.17 168.88 * In U.S. dollars, 1970. ** Cell entries are the predicted values of EDEXPOP in unitary and federal systems at various levels of economic development. Several hypotheses from the literature have thus been shown to be empirically unfounded. First, it is clear that the degree of fiscal centralization of a nation is not important to an understanding of EDEXPOP. Nor is the degree of informal power concentration in the legislature or in interest articulation important in explaining educa- tion policy commitments. Cameron and Hofferbert's findings, moreover, are opposed to our findings. Our’aggregate level data show that unitary status lessens the commitment to education spending and that this relationship is strongest at high levels of EPC. The negative im- pact becomes increasingly evident as EPC increases. Cameron and Hofferbert, on the other hand, found gg_relationship on the aggregate level using data from eighteen developed nations. When they analyzed subnational data, however, they found that federalism leads to reduced 216 expenditures. Thus, Beer's argument that federalism has an expansionary potential is partially supported in the area of education policy. But our study suggests that the main determinant is national affluence and that FU's impact is contingent on the level of economic development. Finally, our research gives no support to arguments concerning the role of population needs and of social heterogeneity in determining education policy expenditures. Although we expected that the education policy area would reflect political controversies, one potential major and salient source of controversy -- extreme ethnic heterogeneity -- bore no consistent relationship to education spending. The Impagt of Poligical Structure Centralizationgon Education Outcomes Policy outcomes indicate actual results from policy decisions that reach or are experienced by citizens and specific target groups. When studying the impact of politics on policy, then, it is vital that the research move beyond a consideration of expenditure levels to a consider- ation of what things result from policy and how policy affects citizens' daily lives and opportunities. The question I address in this section is "Does political structure have implications for policy outcomes or results?" Aside from affecting levels of expenditures, in other words, does the degree of political structure centralization affect the way in which services are distributed? To indicate education policy outcomes, I refer to gross enrollment ratios (GER, the percent of school-aged children in school) and the percent of the population enrolled in second and third levels of educa- tion (i.e. . high school and college or technical school). 8101:. What role does political structure centralization (as well as 217 several other theoretically important variables) play in affecting education policy outcomes? Tables 5-18 and 5—19 display results of bivariate regression analyses. As becomes clear from the analyses, all of the political centralization variables except federal/unitary status are significant at the .05 level in affecting GER and SHE. In each case, moreover, the effects are negative: as political structure centralization and authority concentration increase, GER and.SHE de- crease. Decentralization and local control, not centralization,appears to expand education Opportunities. As expected, economic deve10pment plays a large positive role in explaining GER (R2 = .779) and an important role in determining SHE ("E2 = .444). In both cases a logarithmic model best fits the data. (See also Figure 5-3.) That is, EPC has a positive impact on GER and SHE, but, at high levels of EPC, the impact levels off. At high levels of EPC, further increases in economic development add progressively smaller increments to educational opportunities. Finally, social-demographic factors help explain enrollments. Extreme ethnic-cultural heterogeneity, a rapidly growing population and the presence of large numbers of children age 14 or below all negatively affect gross enrollment ratios and enrollment in higher education in bivariate analyses. Below, we consider these relationships in multi- variate analyses. Mnltivagiate Analys§§_e- Interaction Models The results from bivariate analyses are partial, incomplete and possibly spurious. Several questions remain. First, does the level of economic growth affect the impact of the several types of centralization so that the effects of centralization change by economic level? Or do 218 coo moor eedeneoo oo omm . oi. .1 mo .: rods? .2 oo .9 R . mma financed own . moo. moo. scanned . Hora oo m3. ooo.- mm; no.7 Hora or on scram ssdensoo oo own. one... oo .3 tram- Hoe ca Ha.ma mo.ofia renowned oo mos. own... at . moon .1 Hoe on 4. mm . E esflnsoo oo ooo. oaof rim 3 .11 B m Mm. confirms scuba domains carcass, oonaesnoefim Rosanna Headaches ecoocoaoeaH mam $13.33. was agenda, pcoocomoocH one coozpom mmanmcofloaom casual/Hm 219 on . m an .n- refinance oo was. now. now. i.e.: one. ca om.m oorm ordnance oo omm . 3m . aoo . *moo . omm NH . mm mm. or 933980 was. T ooo. moo... wounded .86 on o2. one. :5. mos. ago am.mm mm.mmfi vqmvmsoo om m2. Rm... own imam- saw 5 was oo .8 ordnance om s3. ems: SH. 6?: arm m mm endeavor nouns m3 carcass, oosaosmdsmpm shedsmpm Hoposmnmm psoosomoocH $32383 2.83969 220 ll .mHoHHo onmocwvm Hdonv coax» enema we run mopwEavmo denudem * Norma ecceneoc Hoods oo NS. o8... mom. .34. so i .aom .rom no .N. on .mo esflocoo oo 3N. m3... mm.m LEN? .em rezone .aom mm.m mm.ow pumpmcoo oo mam . mmo . u on . n *ma . om- aeaosomohcror So no.9 Essence ma: Noooooo. *aoooooo... scanned 6% oo o3. ms; moo. *mao. omm m Mm orgasms henna cannons flmflfllfl 8328:an underarm statesmen 6.868888 Aooscdpcoov mat manna 221 moo. oso. respncoo oo Nos. ooo.- moo. mano.u «H coo. Has. redeneoc moa.- aoooo. mooooo.- eonsson.Hoa oo mos. mam.u moo. oooo.u Hos ooo. mom. scornsoc oo mom. ooo.- oao. momo.- Hos ca mao. awn. essences oo new. oor.- «ooo. soooo.- Hoe aao. Hoo. essencoo oo ooo. Hoa.- Nao. ooo.- as m m crosses scans damages cannon, ocnaohnosnem grandam attached esoosoaoocH mauw,oapme saw one nearnasn> rcooeoaoocH one cooarom nearmcoaanaom orsaas>am 222 anao. amo.» ecnancoo or non. map. moo. mao. oar ca msoo. omo. psdencoo oo ems. sob. Hooooo. moooooo. use one . mo 2.. . 9:3...“ coo moo.- oooooo. moooooo. abandon .azm on amm. omm.- Hoo. mooo.- mam moo. mam. remanded on mmm. ooo.- mao. *mso.- man as mao. mos. essences on mam. ooo.- mooo. smooo.- are m 1mm organs henna broadens carcass, Esaondocdem ohdosfim nopoamamm E38865 Aconcaesocv $1106.33. 223 .mnoaao vumoqmpm Hams» moans enema am one mopmsapmo ooanmpm * ‘3 Homo. mos. rcnrncoo Hood: oo are. mao.u oooo. *moo.u no as .aom .pom oao. Hoe. essences or arm. ooo.: moo. *mmo.n .pm rezone .aom ooo. moo. remansoc oo ems. mam.u moo. *omo.u speedomosopom soo. amo. essencoo oo.au mooooooooo. *aoooooooo.u sounded .omm oo ooo. no.6 mooooo. *moooo. ode .m mm demeanor Henna cesarean cannann> endangers. ehdoeflm abscesses ecoocoaoefi Aconcapcoov mfilw marsh % Population in Second and Third Levels of Education, 1970 224 0400-- O-‘.-.- A“ ~L o-c .777: . ”2'1 51'! :f ' .....- r. 1.. -, "”TX{ILJLI.JH22M r-u-O-O"~>->C -.. -- .. -..... ....L.-.., O o—-——-- o ......u - I, \ n‘%. ::T -..-... f . T""m"~ “or—... .t.--._. +--.... ...} “’”“"‘1 ..-.f -- . , .- o .4.—- c—f-O- —--:..-.... 1. . —.—. _n ‘— ; a“ :T- ”...—:r-o ...-‘n‘.- ...“..i ..-... I . .--—-—v .- - ’ +4-." up- c—c-t-o —.-o .-- ‘ .-...-rfi-a-.—.-.- - .-. . . ..-? - ‘- - - _. ...._._. 5..- . . . ‘- -' - . - -..? . -. . L.- . ... .-...... .— -o.‘ ----o a- ---- -*- t ¢——-J pod-..- ...”, ... ...-o—oo l -4. . ..y- .....o...{ -" — ...i_. ""v ... .-- ...-o- "‘l“‘ . :y24. +- ...... -.....1'2 ...—..1. _-. ‘. .-.-“:ba-o--—' -g--.-: . I ..1 -. —.:--’—.- ..-- O ... ..-. ~--.- 0.... ......o 1.-.. ....i uo—Q - -.- «up-+oo ...- +o‘-on: 07-) ~o< .- 1,“... "...—J. «...—o. ‘ w. ’0 ...-.... 1 -- ..-¢-.....-§ “...-o... ‘ l p—u—nob-o oo— ‘n-q—o- , o~ .. M-Dw ...-:4».- -I-n.>—-L~o-—b-L—-—---4 ...-... .. --- --.—.----. F.- -..-I o-ou—o-tr ——.—-)—“¢- -..J- .A— :4 +- ...-s.-- p...» -h. .. --..--—.mu—-¢4—o—J _~.1 t Y F —T .4 m ~a—n—O.-9~.Jn-‘o-a‘-J— -..... -Ftu.‘ '1- - o-- b».— ton-4 .F—n-— -—.—- “mt—*4 -—~--- '-- o-o....--oq ~-.—— .... *CJ‘M ......IP. ~:I--._—1-- . v.1) g.“ ..-..-.+_._...‘ -..... “-.-.-y.-- . . --..L- ’-.‘}--°--' -o ~A. . .-.- .-. ..-.. .--. ..-. . . .. . . ‘.. . - no. ! ‘n- ..& P~.;—;. ‘-!{3-:-.- :--o~4.~-' - “.---8 ...-:4; Jlaob-oop .31“. ...“ -'n 0 .IS b -o —o [ST—n... —-o-- -0- -oo.— -094 - 9'” -vo o—o- a-Wari o. 0.. n . c-b ~o —-~ i-o-o-u. hobo Ail-{1+ Hie-:-+f--.-: ‘ ‘ 1%“ WA .F‘ ~[:r—.--n.‘. +-;J_, f‘t’i‘ - d. I? ’h- a. -~_o+n‘-?—wE--—c—:—o'r c”.- Electric Power Consumption, 1970 (Kilowatt hours per cap) Figure 5-3 (N = 60) 225 the centralization measures interact in special ways to affect policy? For example, does the impact of SMP depend upon the degree of fiscal centralization (or, vice versa)? Bivariate analyses do not answer these questions. Tables 5-20 to 5-23 report regression results. (See Appendix B.) A number of interaction terms appear to be empirically significant in explaining GER and SHE in these analyses. Taken alone, however, the implications of these findings are unclear. It does seem to be the case that the interaction terms often represent moderating influences on the main effects of the centralization and economic measures. That is, they qualify or add another dimension to the relationships between political and economic variables and education policy outcomes. An interaction term also may negate or change the nature of the relationship. For example, in the case of SHE, the interaction term ln EPC * IA (which carries a negative sign) overrides the otherwise main positive effects of economic level (adjusting for the effects of’all other variables). Whereas both IA and 1n EPC have independent positive effects on SHE, when high levels of EPC exist where IA is negligible, the effect of EPC is negated and IA's effect is moderated. In order to interpret these results correctly, they must be placed within the context of a more complete model. I regress the enrollment indicators, then, on the full range of variables that have survived bivariate and partial analyses. Multivariate Analyses -- Specifying a More Complete Model Tables 5-24 to 5-27 bring us closer to an understanding of the effects of the independent variables on gross enrollment ratios. That is, we test for the significance of each variable in the presence of 226 Tableg5524 Regression of GER on All Surviving Significant Variables (N = 40) Independent Parameter Standard Standardized Variable Estimate E329; Estimate FCI 1.175 .815 .916 Heterogeneity -9.88* 4.11 -.208 SMP .949* .414 .834 In EPC * IA 9.786* 4.41 1.05 In EPC * FCI -.085 .086 -.401 Pop. Grow. Rt. -1.10 3.80 —.O46 Pct. Pop. 14'+ .106 .641 .042 SMP * IA .291 .2596 .464 In EPC 10.24 5.94 .773 FCI * SMP -.0134 .007 -1.19 IA -86.05* 34.12 -1.80 constant -30.25 58.64 R2 '82 E .874 .824 17.61 * Starred estimates are at least twice their standard errors. __ _ several political, economic and social variables. Three variables together explain over eighty percent of the variance among nations in GER: ethnic heterogeneity, level of interest articulation and the inter- action term ln EPC * IA. The equation from Table 5-25 is: 227 ———— Table_5-25 Regression of GER on Heterogeneity, ln EPC * IA and IA** Independent Parameter Standard Standardized Variable Estimate E5323 Estimate Heterogeneity -12.37* 3.47 -.259 In EPC * IA 11.7* 1.79 1.28 IA -86.13* 9.59 -1.83 constant 87.79 2.26 R2 'fi2 E .830 .818 68.54 * Starred estimates are at least twice their standard errors. ** When GER was regressed on the four variables that were significant in Table 5-24, (Heterogeneity, IA, 1n EPC * IA , and SMP, SMP was no longer significant at the .05 level. __ GER = 87.79 - 12.37 (Heterogeneity) + 11.72 (ln EPC * IA) - 86.13 (IA). The presence of all the other'variables included earlier'adds nothing to an explanation of GER; their earlier significance was spurious. There is one final relationship to be investigated - that between education spending and education outcomes. Table 5—26 shows what happens when EDEXPOP is entered into the equation with the other vari- ables. The‘R'2 increases to .872 and the parameter estimate for EDEXPOP easily passes the test of significance. In this case, there appears to be a link between education expenditures and commitment ggg education policy impact. EDEXPOP is one predictor of GER along with the economic, 228 Table 5—26 Regression of GER on Final Surviving Variables and EDEXPOP (N = 45) Independent Parameter Standard Standardized variable Estimate 2239; Estimate Heterogeneity ~12.89* 2.92 -.270 EDEXPOP .0865* .020 .284 1n EPC * IA 11.40* 1.50 1.24 IA -76.56* 8.34 -1.62 constant 78.95 2.79 R2 fig E .883 .872 77.60 * Starred estimates are at least twice their standard errors. social and political structure variables. Our final, revised equation is: GER = 78.95 - 12.89 (Heterogeneity) + .0865 (EDEXPOP) + 11.40 (In EPC * IA) - 76.56 (IA); (1?2 = .872). The interesting result is that several types of variables are important to an explanation of gross enrollment ratios -- political (structure of interest articulation), SOCial (heterogeneity) and economic (in interaction with the political structure indicator). Heterogeneity has the hypothesized effect on policy: where there is extreme or marked ethno-cultural diversity, policy levels decrease, all other things being equal. Although a centralization of political 229 authority in the hands of a few (indicated by negligible levels of interest articulation, IA = 1) has a negative impact on GER, that impact is moderated by economic level. Hypotheses suggesting that centralized control leads to greater policy levels must be qualified in this case. Finally, adjusting for the effects of the other variables, as EDEXPOP increases, there is an increase in the number of school aged children who attend school. Table 5-27 allows a consideration of several different tendencies resulting from the interplay of levels of economic growth, interest articulation and ethnic pluralism. First, it is apparent that at low levels of economic development (comparable to levels in most countries in the Third World) GER will be greatest if there is social homogeneity and moderate or significant levels of interest group activity and representation (IA = O). No matter what the status of interest arti- culation, moreover, gross enrollment ratios decrease where there is extreme social heterogeneity. At low levels of EPC, then, the impact of IA is limited. Second, at middle levels of EPC, GER is greatest where political authority is concentrated.and society is more homogeneous. Again, heterogeneity decreases GER no matter what the level of IA. Finally, at high levels of EPC, GER is greatest where informal political authority is concentrated (IA = 1) reggrdless of the degree of social pluralism. The.impact of IA on this aspect of education policy outcome changes by level of national affluence. The final equation for the gggggd outcome variable is: SHE = —.033-+ .004 (FCI) + .0005 (In EPC * SMP) - .0004 (ln ECP * FCI) - .013 (pop. Growth Rt.) + .0001 (EDEXPOP) - .00002 (FCI * SMP) - .0002 (SMP); "152 = .926). 230 Table_5-27 The Impact of Interest Articulation and Extreme Ethnic Heterogeneity on GER at Various Levels of EPC, Adjusting for EDEXPOP* (N = 45) EPC SE Third World E Total Sample I Developed Nations 329 1742 4557 IA = 1 Heterogeneity = 1 60.46 79.52 90.47 IA = o Heterogeneity = 1 70.96 70.96 70.96 IA = 1 Heterogeneity = 0 73.36 92.41 103.36 IA = o Heterogeneity = 0 83.85 83.85 83.85 * Cell entries are the predicted values of gross enrollment ratios at specified values of IA, ethnic heterogeneity, and EPC, holding con- stant the effects of EDEXPOP at its mean value. Again, several types of variables are important in explaining educa- tional opportunities (enrollment in higher education): political, social and economic. Moreover, once again it is clear that the politi— cal and economic indicators interact to influence policy levels. The economic and political situations influence each other. Finally, the analysis shows that there is a link between expenditures and policy im- pact in this issue area. (Recall that no such link appeared in the health area.) EDEXPOP is one predictor of SHE. Mbst interesting to this research is the impact of political 231 Table 5-28 Regression of SHE on All Surviving Significant Variables** Independent variable FCI ln EPC * SMP ln EPC * FCI Pop. Grow. Rt. ln EPC * IA Pct. Pop. 14-+ FCI * SMP SMP constant (N = 46) Parameter Estimate .004* .0006* ”0011 -.0003* -.00002* "' o 002* —.031 R 2 -915 fiQ ~89? Standard Standardized E3223 Estimate .0004 2.25 .0000? 1.91 .00006 -1.44 .004 -.305 .001 -.115 .0006 -.078 .000006 -1.64 .0007 -1.16 . 027 F 51.05 * Starred estimates are at least twice their standard errors. ** IA, SMP * IA, and ln EPC did not enter the regression due to in- sufficient tolerance. They appear to be highly correlated with other variables in the equation and thus add nothing more to the explanation of SHE. 232 Table 5-29 Regression of SHE on the Significant Variables From Table 4-28 (N = 49) Independent Parameter Standard Standardized Variable Estimate Egpgp Estimate FCI .004* .0004 2.27 ln EPC * SMP .0006* .00006 1.92 ln EPC * FCI -.OOO4* .00005 -1.41 P0p. Grow. Rt. -.015* .0023 -.402 FCI * SMP -.00003* .000005 -1.81 SMP -.002* .0006 -1.07 constant -.0417 .0206 R2 E2 .907 .894 * Starred estimates are at least twice their standard errors. 233 Table 5:30 Regression of SHE on Final Surviving Variables and EDEXPOP (N = 49) Independent Parameter Standard. Standardized variable Estimate ggggr Estimate FCI .004* .0003 2.07 In EPC * SMP .0005* .00005 1.72 ln EPC * FCI -.OOO4* .00004 -1.34 Pop. Grow. Rt. -.013* .002 -.358 EDEXPOP .0001* .00003 .265 FCI * SMP -.00002* .000005 -1.40 SMP -.002* .0005 -1.158 constant -.033 .017 R2 82 F .937 .926 89.01 * Starred estimates are at least twice their standard errors. structure centralization on SHE. (See Tables 5—31 and 5-32.) First, hypotheses suggesting that fiscal centralization has a positive impact on social policy levels must be qualified. Only at low levels of economic deve10pment does fiscal centralization have such an effect. At middle and high levels of national wealth, FCI has a negative impact on opportunities for higher education. From our final equation we know that the effects of FCI are constrained by both SMP and EPC. That is, its effects are not straightfbrward and they change depending on economic 234 level and the degree of majority party control in the legislature. Table45531 The Impact of FCI at Various Levels of EPC: Expected Values of SHE** (adjusting for the effects of other variables)@ m=um EPC 32' Third world 2 Total Sample I Developed Nations FCI* 329 1742 4557 54 .026 .040 .048 72 .034 .036 .038 90 .043 .032 .027 * values of FCI represent the mean of FCI for the entire sample and 1 standard deviation above and below the mean. ** In percentages. @ Cell entries are the predicted values of enrollments in higher edu- cation (percent of the population enrolled) at specified values of fiscal centralization and economic deve10pment, holding population growth rate, EDEXPOP, SMP and the interaction term 1n EPC * SMP constant at their mean values. The values of the other interaction terms change with changes in FCI and EPC. Not only does EPC constr§;g_Fg;, it is also constrained.by FCI. At low and middle levels of FCI, as EPC increases, so does SHE. The interaction terms, on other words, present important qualifications to earlier hypotheses about the impact of both FCI and EPC on policy. The impact of informal political authority concentration (as indicated by SMP) on SHE differs from the impact of fiscal centralization 235 on SHE. Whereas at low levels of EPC, SM? is associated with a decline in SHE, at middle and high levels of EPC, as majority party control in the legislature increases, so does SHE. (This trend is most pronounced at high levels of EPC.) SMP and national wealth interact to affect SHE in varying ways. Table35232 The Impact of SMP at Various Levels of EPC: Expected Values of SHE** (adjusting for the effects of other variables)@ (N = 49) EPC SMP* E Third World SE Total Sample 32 Developed Nations 39.38 .048 .033 .025 60.64 .035 .037 .039 81.90 .022 .042 .055 * values of SMP represent the mean of SMP for the entire sample and values 1 standard deviation above and below the mean. ** In percentages. @ Cell entries are the predicted values of enrollments in higher education (percent of the population enrolled) at specified values of SMP and EPC, holdinngCI, population growth rate, EDEXPOP, and on EPC * FCI constant at their mean values. The values of the interaction terms change with changes in SMP and EPC. SMP also constrains the impact of EPC on SHE. At low levels of majority party control, as EPC increases, SHE decreases. At middle and high levels of majority party control, on the other hand, increases in 236 SHE follow economic growth. This finding supports the notion that in Third.World and poor nations, advances in social policy do not neces- sarily follow from increases in GNP and industrialization. The "trickle- down" theory is challenged in this case. In both cases (in the impact of FCI and of SMP on SHE) it becomes apparent that the effects of politics (of political structures) on policy differ depending upon the economic status of the country. What is a "beneficial" form of political structure in Third World nations may be a "harmful" political situation in more advanced countries, and vice versa. This suggests, too, that development strategies cannot be universal, but must be chosen to fit (accommodate) the national situation. £a§EEE§£I Results of this investigation of education policy differ from results reported for the health policy area (Chapter IV). The impor— tance and role of political structure centralization and of economic level change by issue area and by focus of concern on policy outputs or policy outcomes. Education spending per capita is largely a function of economic level. But, the formal-legal status of the nation —- federal or unitary -- qualifies that relationship: unitary structure moderates the otherwise positive effect of EPC on EDEXPOP. Education policy 92:- 99mg, on the other hand, appear to be more a function of political and _§ggi§l indicators. The role of EPC is contingent; it influences the percent of the population enrolled in secondary and higher education and gross enrollment ratios jointly with EDEXPOP, and, more importantly, through its interaction with political centralization indicators. 237 Finally, the nature of the relationship between centralization and education policy is much more complex than the earlier literature suggests. Moreover, our research in this policy area reinforces the early contention of this study: that political structure centralization is a multifaceted concept. Whereas unitary status (centralized structure) leads to a lesser commitment to education spending (through its impact with economic level), the impact of other kinds of structural centralization on education outcomes changes with economic level. In affluent nations, informal political authority concentration (IA) leads to higher gross enrollment ratios (adjusting for all other variables). Similarly, in wealthy countries, as the majority party control in the legislature increases, so does enrollment in higher education. In this situation, however, fiscal centralization has a negative impact on SHE. In poor nations, these relationships change. At low levels of EPC, fiscal centralization has a positive impact on SHE, and SMP has a negative impact (all other things being equal). The effect of interest articulation on gross enrollment ratios, moreover, is diminished in poor nations and, in this situation, the presence or absence of extreme cultural diversity seems to be a more important determinant of GER. Heterogeneity (at low levels of EPC) leads to lower school enrollments. Results from Chapters IV and V differ in another area as well. In the area of education policy there proves to be an empirical link between policy outputs and policy outcomes. In general, as EDEXPOP increases, so does GER and SHE (controlling for other causal factors). In the health policy area, however, such a link was absent. Apparently, it is more difficult in some issue areas than in others to turn ex- penditures into actual results and opportunities for citizens. CHAPTER V FOOTNOTES 1 Cameron, David and Richard Hofferbert, "The Impact of Federalism on Education Finance: A Comparative Analysis," European Journal of Political Science (September, 1974), vol. 2. 2 Ibid., and Frederic Pryor, Public Expenditures in Communist and Capitalist Nations. Allen and Unwin Ltd., London, 1968. 3 Cameron and Hofferbert, op. cit. 4 Federal/unitary status has no main or direct effect on EDEXPOP. When considered along with EPC and FU * EPC, FU is not significant. 238 CHAPTER V APPENDIX A Summary Characteristics of the Variables Variable GER SHE FCI ln FCI sq FCI EPC ln EPC sq EPC Heterogeneity Pop. Grow. Rt. Pct. POP. 65 + Pct. Pop. 14 - SMP ln SMP sq SMP IA Mean 67.90 .0540 71.96 4.24 5497-85 1741.73 6.189 10385818.6 .4194 2.10 6.14 37-50 60.64 4.04 4121.30 .5167 239 Standard Deviation 23.76 ~0352 18.01 .2823 2488.75 2733-27 1.80 30863382 - 9 .4975 - 9388 3.811 9.056 21.27 .3650 2830.56 ~5039 Cases 60 65 67 67 67 63 63 63 8? 64 56 56 56 6O CHAPTER V APPENDIX B 240 mo. am we .a renounce oam. omoo. memo. mam * oam as non. om.m som.o omm as on one. moa.u omm. mmm.- mam aa.oa oa.ma- renounce amo.u am.m mam.mu om * cam cm are. oo.m mos.ma one as oo one. one. oo.oa ma.oa om fi:.mm no.0m esopmcoo ooo. oaoo. mamma. Hos * one ca ooom. ma.n oa.m oaa as oo ooa. ooo.- ems. ooo.- Hos .m mm mmmmmmmm. moans mmmmmmmm. mmmmmmmw oouaommosmam oanosnpm succeeded reoocoaoocH mHoooz someomaoch .maopHOHecH coapmmaamnpsoo one saw so moapom pomsaaonsm moose mo commmonwom own“ ofinmtm. 241 .3988 28:95. Hams... oofifi ammo.“ on one mopmsapmo 8.53m * zl emoa soon pantheon ooa. oa.m smo.m «H * ode ca mos. om.m mom.o ode ca mom. moéu 36m *nmonu «H mm mmmsimmmim momma. gm. a commonsense Emocnem horosmhmm racecoaoefi $33282 omN manna. 242 maa.u omm. ooa.u Hoe ooo.- mm.oa om.mu on no ooo. o3 no.2 one 5 oo.aa ofi.H pemewcoo omo. mom. 38. «H * Hoe one... made ooo- ...H mmo. our. memo. Hoa on no. om; *oaoa one 5 om.mfi afi.m©1 pompmcoo no.7 Roo. Lmo... new i. Hora ms; in. an; ism more. aam. *ao.a Hoe oo mma. Hm; room one em .2. mm flog spasm flotsam mammal, Euaoanocuem 883m Stamens ecoesoaooeH mHoooz cofipomsmecH .mohsmmoz coapmmaamnpsou co moapmm ecosHHohsm macho Mo scammonwom H NI 33. we 243 oao. ma.o am.a om on man. mam. mo.a moa.oa oam ca Ho.oa Hm.om- announce amo.u mam. moam.n oa * mam oom. oom. mos. mam oar. ma.ma *aoo.mm as oo Hoe. mmo. ma.a mom.ma oam ca om.:fi :«.Hmn panamsoo mao.au mos. mono.- mq dHHsB cam uncomm 2a deHoHcm soapwdsmom pcoonom Mo scammonwmm leqm Momma. 248 afiao. 000. vampmcoo mom. mmooo. mooo. pnmdcmmvccH Aomsqdoqoov mmuw odpma CHAPTER VI POLITICAL STRUCTURE CENTRALIZATION AND SOCIAL WELFARE POLICY I shall now examine a third area of social policy: social security or social welfare. Social welfare policies are generally recognized as ultimately redistributive: such programs are designed to improve the lot of the "have-mots" whether out of altruistic concern or practical considerations. The main focus of this chapter is the impact of structural centralization on social welfare policy outputs and outcomes. The central question is: "Does political centralization affect levels of policy commitment and outcomes, and if so, what is the nature of the relationship?" Wilensky's 1975 study leads us to expect that fiscal, formal-legal and political centralization has a positive impact on social welfare program deve10pment. (Recall from Chapter I that Wilensky uses a composite index to measure degree of centralization.) He finds, using data for 1966 from the 22 richest nations of the world, that "The greater the authority of the central government vis~a-vis regional and local units, the higher the welfare state spending and the greater the program emphasis on equality."1 Similarly, Echols. Duchacek. and Cameron and Hofferbert argue that federalism lessens commitments to social policy, while Peacock and Hiseman, and Hage and Hollingsworth argue that centralized structure leads to an expansion of commitments.2 In sum, the literature brings us to the conclusion that centralization, 250 251 not local control, enhances the deve10pment of the welfare state. There is some evidence presented by other scholars, moreover, that the greater the social welfare expenditures, the more egalitarian the society (in terms of income distributions). These arguments will be considered in the second half of the chapter when policy results as experienced by citizens, or, outcomes are considered. Indicators of Social Security Policy_0utputs and Outcomes In Chapter II I discussed the social security and income data that are available at this time. First, several expenditure figures for 1970 reflecting monies for medical care and benefits in cash and in kind are available for 52 countries and indicate commitment to social welfare programs. Again, the per capita figure stresses our concern for the well-being of igdividuals_within society. Second, to indicate social security outcomes, I refer to data on income distribution and income levels. These data are used for several reasons. First, more specific data on the impact of social security programs -- such as figures on numbers of people served as a proportion of those in need, figures on unemployment and yearly changes in employment, numbers of "rehabilitated" citizens, etc. -- are simply not published. Second, it is through social security programs that countries attempt to improve the position of the lower socio-economic groups in society. Social security programs allow for risk-sharing among groups in society. In this way, the pro- grams provide movement toward affecting social equality by adjusting the distribution of income, and thereby, the command over resources. Titmuss, Jackman and Cutright, moreover, each argue that, in effect, social security programs and expenditures lead to some redistribution of valued goods and services within society. We expect, then, that social 252 security expenditures per capita (hereafter referred to as SSEXPOP) should have a positive impact on the income shares of lower groups in society. There should be an empirical link between SSEXPOP and income distribution figures. The availability of income data will limit the research somewhat. Data for 1970-1972 on income shares of the top, middle and low income groups are available for only ## of the 67 nations included in this study. Data on absolute income levels (numbers of citizens earning less than $50 and $75 per year, for example) are even more limited -- such data for 1969 are available for only 28 nations. These data indi- cate, to some extent, poverty levels and the extent to which basic human needs, such as housing, food, and clothing, are being met in each nation. Because employing the latter figures would limit the sample too severely. and because I am particularly interested in the welfare of the lowest income groups in society, how social security affects the plight of the poorer citizens, and relative income shares, I will use one figure to indicate outcomes -- the income share of the poorest 40% of households in each nation.3 The figure allows us to be sensitive to inequality in particular ranges. It also indicates the distribution of resources and therefore, often, of opportunities to satisfy needs and wants. This income group, moreover, can be considered a primary target group for policy makers concerned with improving the lot of the poorest. (Table 6-1 reports Pearson's correlation coefficients among all the data in this policy area. It is difficult to choose one "best" income indi- cator from these figures. Therefore, I am guided by theoretical interests and data availability alone.) 253 Table 6-1 Pearson's Correlation Coefficients Among the Social Welfare Indicators socsmczx socsmcax SSEXGDP .16 SSEXGDP ssmor .116 .61 SSEXPOP LII .32 .28 .31 LII 1.61 -.39 -.75 -.7u -.1l+ LGI Low INC .32 .35 .35 .91 -.25 Low INC MIDINC .20 .06 .118 .113 -.39 .53 MIDINC '1o1D INC f-.31 -.u3 -.u8 .79 .38 -.85 -.78 TOP INC POP 50 .38 -.17 -.27 .15 .16 .26 .03 -.18 POP 50 %50 .06 -.22 -.62 .30 .55 .26 -.02 —.13 .33 $50 POP 75 .37 -.18 -.29 .17 .18 .27. .03 —.20 .999 .33 POP 75 9675 .02 -.26 -.6t+ .39 .65 .34 .006 -.22 .33 .95 .31. $75 254 Table 6—1 (continued) SOCSECEX SSEXGDP SSEXPOP LII LGI LOW INC MID INC TOP INC POP 50 %50 POP 75 %75 Social Security Expenditures, 1970 Social Security Expenditures as a 7a of GDP Social Security ExPenditures Per Capita Level of Income Inequality: low, medium, high* Level of Income Group Inequality: low, medium, high** Income Share of the Lowest 40% of Households Income Share of the Middle 40% of Households Income Share of the T0p 20% of'Households Millions of P0pulation with Per Cap Income below $50, U.S. %1P0pulation with Per Cap Income below $50 Millions of Population with Per'Cap Income below $75, U.S. % P0pulation with Per Cap Income below $75 * low = income share of lowest 40% households is 12%-+; medium.= income share of lowest 40% is between 12 and 12%; high = income share is less than 12% of national total income. ** low = GNP/cap income is up to $300, U.S.; medium = GNP/cap is $3oo-75o, U.S.; high = GNP/cap above $750, U.S. In sum, I will investigate the impact of the independent variables -- political. economic and social -- on SSEXPOP and on the income share of the lowest 40% of households (hereafter referred to as LOW INC). Furthermore, I will consider the empirical link between social security expenditures and benefits and the share of income received by the poor- est groUp in society. 255 The Impact of Politig§;_Structure Centralization on Socialggecurity Expenditures I begin the analysis by testing bivariate relationships and "partial theories." The first equation tests the role of constitutional status in determining SSEXPOP: Y = a + b1X1-+ e where X1 FU status, 1970 e stochastic disturbance. The literature reviewed above leads us to expect that federalism has a negative impact on social security expenditures. Results from our data reported in Table 6—2, however, challenge those arguments on empirical grounds. Our results show that federal/unitary status plays no role in determining SSEXPOP. Federalism has neither an expansionary effect (as Beer argues) nor a conservatizing impact on expenditures (as many other scholars argue). The E2 is negligible and the parameter estimate is much smaller than its standard error. Based on aggregate data from 52 nations for 1970, in other words, this aspect of structural centraliza- tion has no relationship to SSEXPOP. To test the relationship between fiscal centralization and SSEXPOP I employ the following equations: Y = a + 131x2 + 9 (linear form) Y = a + b1X3 + e (logarithmic form) Y = a-+ b1X2-+ b2X4-+ e (polynomial form) where x2 = FCI X = 1n FCI 3 256 Iaeuaété Regression of SSEXPOP on Federal/Unitary Status (N = 52) Independent Parameter Standard Standardized Variable Estimate E229! Estimate FU status -69.94 108.66 -.154 constant 191.16 96.25 R2 1'22 F .024 .000 .414 Recall that Wilensky found fiscal centralization (one aspect in- cluded in his composite index along with other characteristics of political structure centralization) to have a positive impact on social security spending. Similarly, others have hypothesized that as fiscal centralization increases so does social spending. Results from the present analysis clearly refute these arguments (see Table 6-3 and 6-4). FCI appears, in bivariate analyses, to have a strong negative impact on expenditures. As FCI increases, SSEXPOP decreases and then begins to level off at high levels of FCI. The logarithmic form of the equation, moreover, accounts for over 40% of the variance in SSEXPOP among nations. (The logarithmic form best fits the data we have, while the polynomial equation is not significant at the .05 level.) Beginning analyses, then, question earlier hypotheses. Next, I consider the impact of informal political authority con- centration -- the concentration of political party control in the legislature and the structure of interest articulation in society --on social security expenditures. those tested above. tralized the political control, 257 The equations are of the same form as Again, the literature suggests that the more cen- the greater the social expenditures. Table 6-:} Linear Regression of SSEXPOP on FCI (N = 52) Independent Parameter Standard Standardized Variable Estimate Error Estimate FCI -6.67* 1 .90 -.647 constant 616.05 141.20 R2 fig F .419 .384 12.25 * Starred estimates are at least twice their standard errors. Table 6-4 Logarithmic Regression of SSEXPOP on FCI (N = 52) Independent Parameter Variable Estimate ln FCI ~436-35* constant 1986.033 R2 .440 * Starred estimate is at least Standard Standardized Error Estimate 119.44 -.663 507. 52 fig F .407 13.34 twice its standard error. 258 Table 6-5 Polynomial Regression of SSEXPOP on FCI (N = 52) Independent Parameter Standard Standardized Variable Estimate Egggg Estimate FCI -19.05 14.70 —1.84 FCI, squared .090 .106 1 .21 constant 1010.04 485.05 R2 fiz F .444 .374 6.38 The data presented in Tables 6-6 to 6-9 indicate that the percen- tage of seats held by the majority party has no significant independent relationship to SSEXPOP. The parameter estimates are not significant and the Rz's are negligible. The level of interest articulation (indi- cating the extent to which power to voice interests is shared among groups in society), however, has a significant negative impact on expend- itures. That is, where interest articulation is negligible, indicating that most groups in society are prohibited from voicing demands, social security expenditures are lessened. Where there is free and open dis- cussion of needs and interests, on the other hand, SSEXPOP is higher. (IA accounts for almost 40% of the variance in SSEXPOP among nations in these bivariate analyses.) While federal/unitary status and SMP show no independent relation- ship to the level of social security expenditures, both fiscal centrali- zation and the degree of interest articulation negatively affect SSEXPOP. Earlier discussions in the literature about the impact of political 259 structure centralization on social welfare (or social policy generally) are refuted and qualified by this analysis. The aspects of centraliza- tion that do relate to SSEXPOP in our research, do so in an unexpected direction. Yet, these findings are tentative until multivariate analyses are performed. Table 6-6 Linear Regression of SSEXPOP on SMP (N = 43) Independent Parameter Standard. Standardized Variable Estimate ggzgg Estimate SMP -3.44 1.94 -.394 constant 344.41 124.76 R2 fiz F .155 .105 3.12 Table 6—2 Logarithmic Regression of SSEXPOP on SMP (N': 43) Independent Parameter Standard Standardized Variable Estimate Error Estimate ln SMP -199.89 113.55 -.392 constant 943.84 460.76 112 32 F .154 .104 3.10 260 Table 6—8 Polynomial Regression of SSEXPOP on SMP (N = 43) Independent Parameter Standard Standardized Variable Estimate Egzg; Estimate SMP -7.78 11.21 -.891 SMP, squared .0331 .084 .505 constant 471.07 346.58 R2 8'2 F .163 .058 1.56 Table 6-2 Regression of SSEXPOP-on Interest Articulation (N = 45) Independent Parameter Standard Standardized Variable Estimate E3323 Estimate IA -242.15* 67.39 -.657 constant 260.94 48.01 R2 '82 F .431 .398 12.91 * Starred estimate is at least twice its standard error. Finally, I consider the role of economic and social demographic factors in explaining SSEXPOP. Wilensky, Jackman, Cutright and others have shown that economic level has a significant impact on social welfare 261 deve10pment.4 Our datani e furth r support to this claim. Energy consumption per capita, in bivariate analyses, explains over 60% of the variance in SSEXPOP (the logarithmic form best fits the plotted data points as shown in Figure 6-1). Yet, as Wilensky notes, there is still some way to go in explaining social welfare program development. Multi- variate analyses will take us further toward the explanation. There is reason to believe that several social and demographic variables have an impact on social welfare expenditures. Several scholars have suggested, for instance, that population ppggp (indicated by population growth rate and age structure, for example) determine social policy expenditures to the extent that policy makers attempt to address needs by the policies they formulate. (Of course, population needs may outstrip the capacity to meet them.) Second, the literature on cultural pluralism suggests that extreme social heterogeneity leads to a lower commitment to social programs and expenditures due to communal conflicts and processes of ethnicization (of social policy out- puts). In separate bivariate analyses the following relationships are un- covered: while population growth rate and the percent of the population aged 14 or under have a ngggtive impact on SSEXPOP (Ra's are .406 and .618, respectively), the percent of the population aged 65 or older has a positive impact on SSEXPOP (R2 = .688). That is, social welfare policy makers seem to respond to the special needs of senior citizens. (But, a rapidly growing and largely youthful population makes it more difficult for policy makers to keep up with social needs and demands.) And presumably, pension and similar plans account for a large percentage of social security expenditures. Social Security Expenditures Per Cap, 1970 262 :.i;o Electric Power Consumption, 1970 (Kilowatt hours per capita, in thousands) Figure 6—1 (N = 50) 263 Finally, extreme social division and ethnic heterogeneity bears no significant relationship to SSEXPOP (R2 = .000 and the parameter estimate is less than its standard error). In contrast to the hypothesis, heterogeneity appears to play no role in determining the level of welfare spending. Table 6—10 Linear Regression of SSEXPOP on EPC (N = 50) Independent Parameter Standard Standardized Variable Estimate ggppp Estimate EPC .055* .0096 .811 constant 39.78 30.63 R2 fie F .658 .638 32.77 * Starred estimate is at least twice its standard error. L... _ Table 6-11 Logarithmic Regression of SSEXPOP on EPC (N = 50) Independent Parameter Standard Standardized Variable Estimate E339; Estimate 1n EPC 81.61* 15.24 .792 constant -369.29 98.08 R2 fig F .627 .605 28.65 * Starred estimate is at least twice its standard error. _._— __—-— ——_ 264 ¥ _ Table 6:12 Polynomial Regression of SSEXPOP on EPC (N = 50) Independent Parameter Standard Standardized Variable Estimate Eggpg Estimate EPC .108* .018 1.60 EPC, squared —.000005* .000002 —.871 constant .601 27.53 R2 R2 F -792 .766 30.41 * Starred estimates are at least twice their standard errors. Multiyarigte Analyses -jv;nteraction Relationshipg Following procedures adopted in the preceding chapters, here I shall test for the presence of interaction effects between economic level and the centralization measures and among the centralization measures themselves. Such special effects appeared in earlier analyses of policy. For example, we may argue that the impact of SMP varies by level of EPC and is evident only in joint occurrence with EPC. (Recall that SMP had no significance in bivariate analyses.) Similar joint effects may occur in cases of other types of centralization as well. Or, several of the centralization variables may interact to present special effects on SSEXPOP. For example, the impact of interest arti- culation (IA) on SSEXPOP may vary depending upon the level of fiscal centralization. Tables 6-13 and 6-14 report results of these multivariate regression 265 N.oom has. mo.mo Nr.om essensoo ooo.- amooo. *seoo.u cam * mam ooo.- moo. amo.- can no.a Nmo. more. one 00.m: was. 03.0mm 0m.mmm pompmnoo mos. no.m em.m ome * Hoe ca ooo.- oo.om *oe.ammu Hoe ca woe. moo. memo. 8m om.sm ash. ea.ra ea.e campuses ooo.- ego. oeo.- one a pa moo. em.mm mm.m: nausea ea so.e Neo. mono. one m mm amalgam gonna glam Emma's, ooueEmosocm Hangman someones esoosoeoofi mHocoz coapomnoch .monsmmoz coavmuHHmecoo one 0mm :0 momxmmm mo codmmoamom Ame u 20 0H 10 0.3.3. tr 266 mmém bl fimw. (12‘. Ill .moo. so osmoseasmsm ms assesses as .3890 09.0030 .305. 0003.... 9000..” 90 09.0 00.003300 dmhhmpm * III I sm.am oo.aofi osmemsoo mmof 000. 000... Euamosmem 88:30 Hopoamnmm eeoosoaoofi $856.88 310 canoe 267 330.- mm.:m mo.om- :0 $0. 08. £30. Em 00.00 :mm. 00.0m: *mm.mmmfi pampmcoo mo.fi mfi.mmfi 0m.mofi :0 * Hum ca 003.- 00.nofi *Hm.0mm- Hum ca 300.: 5:.wom mm.fimmu :0 0mm. moo. *mmzo. 000 mm.:« mom. :m.mmm mm.:mafi 0:000:00 Hm.m 30.0Nfi *mm.mmm «H * H00 :H mm.m- 0.03m *mm.mmdfiu «H Hm:.- No.05 *fim.m0m Hum 2H mum. poo. *0mo. 000 .M mm“ mmmammmm HonHm mmmmmmmm. mammmmmw 832%:30 28:30 0305000 235035 Am: u zv 0H0©oz coapomfi0ch .00Hdmm0z coapmwfiamupc0o no momxmmm Mo c0H000HM0m dale 0HQMB 268 00.00 000. 00.000 00.000 0:000:00 030. 30.0 00.0 00 * 0:0 003.- 0o.0 00.0- 020 003.- 00.000 0o.000- 00 000. 0oo. *00o. 000 00.00 000. o0.oo0 00.000 0:000:00 003. 0o.0 00.0 <0 * 0:0 000.- o0.000 00.300- 00 000.- 00.0 00.0- 020 030 . 0oo . 33o . 000 00.00 mom. Nn.0m 3m.0N0 0:000:00 000. 00.00 00.00 00 * 00 000.- 00.00 00.000- <0 .0 mm 00000000 00000 00000000 mammwmmm 00000000:000 0000:000 000000000 0:00:0000:0 0000:00:000 30-0 00000 269 .mHoHHo Hmongm H.303. meaty 0.0.8.0 0.0 90.0 0.0.0.0508 dmflfim * 00.3 Enl 000. 00.0- 030. 00.0 000. 0005000 000000000500 (I: o0.o00 30.0 00.000 00.00 moo. Henna 0000:000 03.0030- 0:000:00 *00.00- 0000 :0 0. 0000 00600 00.0 :0 035$: $3 .009 0.00 0005.000 mg 0000:0000 0800:3005 0000:0083 300 0000.0 270 analyses. Of the centralization interactions tested, two are signifi- cant. First. FCI (fiscal centralization index) and IA interact to affect SSEXPOP such that the impact of each is contingent on the level of the other; the impact of FCI depends on the level of IA and vice versa. Similarly, FCI and SMP (percent of legislative seats held by . the majority party) act jointly upon SSEXPOP. The impact of SMP varies depending on the degree of fiscal centralization in the nation. One other interaction term is very close to the .05 significance level and will be included in later analyses to test further its im- portance in explaining SSEXPOP. The SMP * EPC variable is significant at .062. That is, it appears to be the case that the impact of SMP on SSEXPOP varies depending on national economic level. This is an interesting finding in that SMP was found earlier to have no independent effect on the dependent variable. It may well be the case, then, that SMP works through EPC to determine SSEXPOP. The ultimate importance of the interaction effects will be tested in the next section where we specify a more complete explanation of social security expenditures. Multivagiate Analxses: Specifying a More Complete Model Up to this point, we have considered partial theories through bivariate analyses and have considered the possibility of joint effects of the main independent variables on SSEXPOP. I turn now to specifying a more complete picture of the explanation of social welfare spending. This is accomplished by entering all significant variables (from the bivariate and interaction regression models) into a single equation. Results are reported in Table 6-15. When SSEXPOP is regressed upon a range of political, economic and social variables, five variables 271 Table 645 Regression of SSEXPOP on All Surviving Independent Variables m=4m Independent Variable ln FCI SMP * EPC SMP * ln FCI Pop. Grow. Rt. IA Pct. Pop. 65 + Pct. Pop. 14 - EPC 1n FCI * IA constant Parameter Estimate -272.41* -.001* .278 30.74 ~905 - 67 26 . 54* 1.06 .089* 215.12 888.08 .868 E2. .833 standard Standardized Egggg Estimate 82-036 -.414 .0004 -.869 . 187 . 147 26.36 .155 501.95 -2.45 10.07 .544 4.90 .052 .0236 1.30 113.86 2.55 352.61 F 24.90 * Starred estimates are at least twice their standard errors. that were significant in earlier analyses become insignificant. While population growth rate and the percent of the population 14 and under are not significant in multivariate analyses, the percent of the popula- tion 65 and older remains a significant, positive factor in explaining SSEXPOP. Moreover, while three of the political variables drop out of significance (IA, SMP * 1n FCI, and 1n FCI * IA), two others remain in 272 the equation (ln FCI, SMP * EPC). Fiscal centralization has direct negative consequences for SSEXPOP, while the impact of SMP is contingent upon the economic level. Finally, EPC also has a direct positive im- pact on SSEXPOP. Our final equation is obtained from Table 6-16. SSEXPOP = 513.42 — 127.30 (ln FCI) - .0008 (SMP * EPC) + (245) (54.21) (.0003) 18.93 (% Pop. 65 +) + .0705 (EPC). (4.9) (.021) This equation explains over 82% of the variance among nations in levels of social welfare spending. It demonstrates that such expenditures are a function of political, social and economic factors. Economic level alone, in other words, will not accurately predict SSEXPOP. It is clear, however, that in the case of SMP, the political and economic variables each modify the impact of the other. As in other'policy areas, then, EPC proves to be an important contextual variable setting limits within which political factors must Operate. Tables 6—17 and 6-18 yield a better picture of the impact of the political structure centralization variables on SSEXPOP. First, it is clear that at each level of EPC, FCI has a negative impact on social welfare expenditures. As fiscal centralization increases, SSEXPOP decreases, no matter the level of economic deve10pment. It is obvious, at the same time, that at any level of FCI, as EPC increases, so does SSEXPOP. The impact of the percent of legislative seats held by the majority party on SSEXPOP is indirect in that its impact is dependent on the economic level of the country. SMP and EPC are mutually constrained. While at each level of SMP SSEXPOP increases as EPC increases, the 273 Table 6—16 Regression of SSEXPOP on Final Survivors (N = 43) Independent Parameter Standard Standardized Variable Estimate Egggg Estimate 1n FCI -127.30* 54.21 -.193 SMP * EPC -.OOO8* .00036 -.613 Pct. Pop. 65 + 18.93* 4.93 .388 EPC .0705* -.207 1.038 constant 513.42 245.10 R2 8'2 F .837 .821 50.27 * Starred estimates are at least twice their standard errors. Table 6-12 The Impact of FCI* on SSEXPOP At Various Levels of EPC: Expected Values of SSEXPOP** (holding other variables at their mean)@ (N = 43) E20. _F_‘_C_I_ 32 Third World 5': Total Sample E Developed Nations 329 1742 4557 54 134.05 165.12 227.02 72 95.22 126.29 188.19 90 64.03 95.10 157.00 274 Table 6-17 (continued) *- ** The values of FCI are the mean for the entire sample and one standard deviation above and below the mean. I use the log natural for each value since the logarithmic model best fits the data. In U.S. dollars, 1970. @ Cell entries are predicted values of social security expenditures per capita, 1970, at specified values of economic development and fiscal centralization, holding SMP and the percent of the population 65 or over constant at their means. Table 6-18 The Impact of SMP* on SSEXPOP At Various Levels of EPC: Expected Values of SSEXPOP** (holding the other variables at their mean)@ (N = 43) EPC gym; E Third World E Total Sample 2 Developed Nations 329 1742 4557 39.38 102.73 157.83 267.61 60.64 97.13 128.20 190.10 81.90 91.54 98.58 112.60 * The values of SMP are the mean for the entire sample and one standard deviation above and below the mean. ** In U.S. dollars, 1970. @ Cell entries are predicted values of social security expenditures per capita, 1970, at specified values of economic development and majority party dominance in the legislature, holding fiscal cen- tralization and the percent population aged 65 or older constant at their means. 275 increase in SSEXPOP is much less at high levels of SMP. SMP constrains the positive impact of EPC. Moreover, while at each level of EPC, SMP has a negative impact (as SMP increases, SSEXPOP decreases), that effect is much greater at high levels of economic deve10pment. These results give solid evidence that economic and political factors influence each other and they warn us against relying on either factor exclusively as a predictor of policy. Political Structupe Centralization and Income Distribution Several questions are addressed in this section. First, what role does centralization play in affecting the distribution of income (and therefore, the command over resources) in society? Do centralized or decentralized structures promote greater social equality? Here I am interested in the distribution and use of valued goods and services within society. To facilitate the investigation, then, I refer to the income share of the lowest 40% of households in each nation. Second, I consider the link between social welfare expenditures and the income shares of the poorest sector of society. As discussed earlier, a number of scholars argue that the result of social security programs is a redistribution of goods within society. Our data allow us to test this hypothesis. Toward An Explanation of the Income Share of the Poorest Because income and economic status usually determine an individual's abilities (opportunities) to satisfy needs and wants, the concepts of social equality and income distribution are tied together. Basic to this research is the question of the impact of political structure centralization on income distribution (and, particularly, the income 276 portion of the poorest 40% of households). Does the structure of political power, authority and decision making affect the real income status or relative income share of the poor? Tables 6—19 to 6—21 report the results of initial bivariate and multivariate regression analyses. Several relationships are suggested by these initial tests. In the bivariate tests, population growth rate (negative impact, R2 = .225) and the percent of the population 65 and older (positive impact, 8'2 = .149) are the only variables significant at the .05 level. None of the political variables appears to have independent effects on the income share of the lowest 40% of households. Nor, interestingly, does the level of economic development. There is no direct link, then, between centralized structure or decentralized structure and.greater income equality; nor is there a direct link between economic growth and income equality. (The "trickle-down of benefits" theory is again challenged by our data.) The pattern of causation of LOW INC appears to be more complex than is generally assumed. Although the effects of our main independent variables are not direct, we may consider whether the variables influence LOW INC jointly or through interaction relationships. In fact, three of the interaction variables are significant in multivariate analyses: SMP * IA, EPC * IA and SMP * EPC. This suggests that although they bear no direct or main relationships to LOW INC, SMP, IA, and EPC work through joint relation- ships to influence income distribution. The impact of EPC, for example, is a function of both the level of SMP and IA. The impact of SMP on LOW INC, moreover, varies depending upon economic level and interest articulation. 277 ms.a os.se pomemcoo on as.a mso. aom.n mam.a mc.mu aH mm .3 3.3 2338 can.. moo. soo.- penance .Hoe as ofi.a “Ho. 0mm. ems. nmo. Hos wo.mfi $3.3m vcmvmsoo as am.a Hmo. mmm.- mm.m mm.s- Hos ca mo.s nm.aa ecsemcoo as NN.N are. osm.- mam. ammo.u Hoa om.m mm.ma essencoo as mm.a sac. mom.u as.m ow.mu as m w. M oenfiemm Hosea 383nm cannons, eosaennocmpm onscreen soeosmnmm pcoocoaoecH moapeaso> ecoocoaoecH one use czH :oa coornom ncasmcoaemaom oesanmeam mate canoe 278 nmm.u sooooooo. sooooooo.- eohmsen .oam on No . a do. m8 . 88. So. cam was and 28.28 on mm.a mac. com. new. amp. cam ca me.e mm.ma essencoo or mm.m ammo. arm. sooo. oooo. cam mm.w m:.m pnevmcoo sa.au amoo. Noo.- eonnsem .azm on man. ooo. mo.a arm. mam. ash as.aa sa.sa ecsemcoo on moo. oHo. sao.u om.m mom.u arm as wH.m om.:H pampmcoo on are. ooo. Hso.- are. msao.u arm .m .m mm oemsaenm noses oposaemm mannahm> douddnmdnmpm saucepan Hovosmnem vsoddomousH Aeossaecoov mane capes 279 .wHOHHo .23:on H85. oofifi posed us who moposdpmo doflfim * ms.a Hm.ofi nememcoo on ne.s ass. mes. arm. anon. + no .com .eom Hm.m mm.wfi pcmencoo on mm.o mmm. mem.u arm. *Hs.mn .em .rono .aom mm.“ ms.mfi essencoo 0: mom . ooo . omfi . 1 fl . N mmm . 1 hpfiosomosopom am.a oa.ae essences m M Mm. 323mm Hoflnm opus. [Ems [m annuallaew ensign 3853 Horseshoe ecoccoaoecH 3350.83 Eur capes 280 on om.s omm. as.a as.mfi essencoo ans.u omoo. *ooo.u constanecnem Sarcasm soposfinm 3388ch 3n n 5 maoooz coayomnoch .mondmmoz coaamuaamnpeoo can use 0mm so 02H :09 mo cosmmoHMom omlw canoe 281 .mHoQHo vamucdpm Haonp cease pmmoa pm one moposdpmo doahmpm * mm 20 ad saw. om.m 00.N N0000. mmoa... ammo. mw.fiu mfioo. .mm mmmmmmmw Henna dosddhmdsdvm dnmcqopm sol: #00000. mmo.u *mmoo.u oemsaemm Hoposmumm vcmvmcoo 0mm * mzm mam 0mm canednm> psoosoaoocH Aconcaecoov omuo canoe 282 smn.u ms.o sa.m- as mam. mooo. sooo. cam mm.H mmo. mo.n wd.mfi vcmuweoo smo. ooa. oofi. 4H * Hos ooo.- am.s mn.su «H oom.u Nso. ooo.- Hoa Hum. mmooo. nmooo. oar am.m nma. am.r mm.m ecnemcoo mm.a- moo. emoo.- arm * Hoa mam. oma. soo. Hoe mo.a mos. Ham. arm mam. mooo. sooo. ems m mm Manly sonar. medians flog condensesoem 283m “speeches pcoocoaoecH Hmtm oHQdH maoooz soapomHovsH .mohdmmoz noapmuaamnvsoo do 02H :09 mo seammonmom 283 mn.m omm. no.m oo.oa essencoo Nw.a mmwo. roam. «H * mam mm.an no.2 *sm.meu acousomoocH Acocsaecoov Hmuo capes 285 Multivariate Analyses -- Spegifyingpa More Complete Model All of these results are tentative, Up to this point. That is, we have not regressed LOW INC on a more complete range of possible causal variables in an attempt to specify a more complete model. This is accomplished in this section. Table 6-22 reports results of regressing LOW INC on the variables that have survived earlier analyses. Both of the population characteris- tic variables drop out of the equation when considered along with the political and economic variables as determinants of LOW INC. The simplest explanation for this would be that those variables are highly related to economic level. Less developed countries are often character- ized by rapidly growing populations, while more highly developed nations usually have a greater’percent of senior citizens than do poorer nations. The interaction variables appear to be significant influencers of LOW INC. Although SMP * IA is just slightly beyond the .05 level of significance (.053) and the SMP * EPC variable is significant at .06, they are too close to our accepted significance level to be dropped from consideration. Moreover, we suspect that there are several joint effects working to influence LOW INC and that the inclusion of two superfluous variables (the population variables) is constraining true effects. Table 6-23, then, shows results of regressing LOW INC on the three interaction variables alone. In this case it is clear that EPC * IA and SMP * EPC are important determinants of LOW INC, while SMP * IA drops out of the equation. In other words, the impact of EPC on LOW INC is not direct, but is constrained by the levels of SMP.and of IA. Similarly, the impact of the political centralization variables depends 286 Table 6-22 Regression of LOW INC on All Surviving Variables (N = 34) Independent Parameter Standard Standardized Variable Estimate ggzgr Estimate Pop. Grow. Rt. -1.70 1.21 -.366 EPC * IA —.0088* .003 -.431 SMP * EPC .00001** .000005 .344 SMP * IA .O46*** .0228 .3999 Pct. Pop. 65-+ .0296 .374 .025 constant 15.31 4.86 R2 82 F .496 .409 5.17 * Starred estimates are at least twice their standard errors. ** Significant at .06. *** Significant at .053. on economic level. Table 6-24 gives our final equation: LOW INC = 13.15 - .0076 (EPC * IA) + .00001 (SMP * EPC). (.894) (.003) (.000004) This equation explains about 30% of the variance in the income shares of the lowest 40% of households across systems. That is, there is some distance to go in explaining income distribution. The political cen- tralization, social and economic variables considered here bring us only part of the distance toward a full understanding of income equality. What we do find is that EPC and IA, and EPC and SMP work jointly to 287 Table 6-23 Regression of LOW INC on Survivors of Table 6-22 (N = 34) Independent Parameter Standard Standardized Variable Estimate E332; Estimate EPC * IA -.0097* .00314 -.475 SMP * EPC .000015* .000005 .499 SMP * IA. .0349 .0202 .303 constant 11.91 1.11 R2 fiz F .404 .346 6.99 * Starred estimates are at least twice their standard errors. Table 6-24 Regression of LOW INC on EPC * IA and SMP * EPC (N = 34) Independent Parameter Standard Standardized variable Estimate Egggr Estimate EPC * IA -.0076* .003 -.3?6 SMP * EPC .00001* .000004 .367 constant 13.15 .894 R2 fiz F .346 .304 8.20 * Starred estimates are at least twice their standard errors. 288 Table 6—25 Regression of LOW INC on EPC * IA, SMP * EPC and SSEXPOP (N = 34) Independent Parameter Standard Standardized Variable Estimate ‘Errg; Estimate EPC * IA -.008* .003 -.407 SMP * EPC .00001* .000006 .474 SSEXPOP -.0036 .0054 -.155 constant 13.41 .977 R2 82 F ~356 .291 5.52 * Starred estimates are at least twice their standard errors. effect LOW INC in special ways. Tables 6-26 and 6-27 demonstrate several patterns of influence. When the right to voice interests is limited and informal political power is concentrated at the center (IA = 1) the income share of the poorest group is lessened. This is especially evident at middle and high levels of EPC. When IA = 1, as EPC_iggreases the income share of the poor greatly decreases. (The negative impact of IA is much greater in developed nations. Put another way, the limiting impact of centralized authority is much less evident at low levels of EPC.) When groups are free to articulate interests (IA = 0), on the other hand, not only is the income share of the poor larger, but the share increases with increases in economic development. Clearly, the political variable constrains the impact of the 289 Table 6—26 The Impact of IA at Various Levels of EPC: Expected Values of LOW INC* (holding SMP at its mean value) (N = 33) 3.139 IA E Third World E Total Sample E Developed Nations 329 1742 4557 IA = 1 10085 098 -18-72 IA = 0 13.35 14.21 15.91 * percent of total national income. Table 6:22 The Impact of SMP at Various Levels of EPC: Expected Values of LOW INC** (holding IA constant at its mean) (N = 33) Egg SEE? erhird World 'E Total Sample I; Developed Nations 329 1742 1+557 39.38 11.99 — 6.99 -2.95 60.64 12.06 7.37 -1.98 81.90 12.13 7.74 -1.01 * The values of SMP are the mean for the entire sample and one standard deviation above and below the mean. ** Percent of total national income. 290 economic variable, and vice versa. Table 6—27 demonstrates the impact of SMP on LOW INC, when all other variables are held at their means. At each level of SMP, §§_§§g increases, the income share of the lowest socio-economic group in the nation decreases. Regardless of the level of EPC, moreover, a§_§M§ increases so does LOW INC. In this case, centralized control leads to greater income shares for the poorest social group.5 The Role of SSEXPOP Earlier studies have suggested that social security programs and expenditures have a redistributive impact and that they change the distribution structure of valued goods and services within society. To test this notion, I regress LOW INC on EPC * IA, SMP * EPC egg SSEXPOP. Does SSEXPOP further'add to our explanation of the income status of the poor? Table 6-25 demonstrates that there is no evidence of an empirical link between SSEXPOP and LOW INC.6 Our data do not support Cutright's argument, for example, that with greater social welfare spending, the income distribution becomes more egalitarian. We have no evidence to SUpport the theory that social security expenditures affect the income status of the poor. (Recall that the income data reflect wage and non- wage income.) These findings, however, are limited in several ways and can only be considered preliminary. First, the sample for this aspect of the research is small -- only 33 countries are represented. This is due to the dearth of reliable income data available on a cross-system, comparable basis. Second, again due to data limitations, we were not able to con- sider a broad range of social security outcome variables. We could 291 better have considered the link between social security outputs and outcomes if we had data on actual social security program allocations, distribution of services, numbers of needy groups reached, et cetera. Summagy Results of this chapter further support earlier contentions that political structure centralization is a multidimensional concept and that different aspects of centralization -- say, fiscal centralization and centralization of legislative powers -- may have different effects on policy. By considering centralization as a single composite variable (including political and economic variables) we obfuscate important interactions and relationships among political, economic and policy variables. Moreover, such a conceptualization of regime centralization severely limits study of deve10pment strategies. We have uncovered evidence in the last three chapters of the futility of unilateral state- ments about the impact of centralization on policy (such as those state- ments made by early modernization theorists). The relationship between centralization and policy depends upon the type of centralization and the social policy area under consideration. In the case of SSEXPOP, FCI has a direct negative effect, but the negative impact of SMP varies somewhat, depending upon the economic level. SMP has a strong negative impact on the level of social security expenditures in industrially developed nation . On the other hand, in the case of LOW INC, FCI and FU have no impact and the effects of IA and SMP are juxtaposed. While SMP has a positive impact on LOW INC (and modifies the impact of EPC), IA has a negative impact on LOW INC that is more evident at high levels of EPC. In other words, where groups are not able to voice interests, the income share of the lowest 40% of 292 households lessens. The magnitude of the negative impact is greatest in deve10ped nations. Finally, analyses presented here yield no support for arguments that social security programs improve the general socio-economic status of the poorest groups. Nor, interestingly, does increased industrial and economic deve10pment translate directly or easily into greater in- come equality. Theories of Western modernization philosophers are again challenged. CHAPTER VI FOOTNOTES 1 Wilensky, Harold, The Welfare State and Equality, University of California Press, Berkeley, 1975, p. 52. 2 It is appropriate to review these arguments as they are outlined in Chapter I. 3 As pointed out in Chapter II, the income data reflect wage and non-wage income where that information was available. That is, "income" from non-wage sources. such as property income or government programs is included. 4 See discussions in Chapter I on this point. 5 For a discussion of the impact of another set of political variables on relative income shares, see Robert W. Jackman, "Socialist Parties and Income Inequality in Western Industrial Societies," in Journalggf Politics (February, 1980), forthcoming. 6 It could be argued here that even though SSEXPOP does not effect the income share of the poorest 40% of the population, it may affect the income share of the top 20% (presumably the class paying for social welfare programs), or, the income share of the middle 40% (based on the argument that redistribution may be occurring even though it is not evident in the income share of the poorest 40% -- income may be moving from top to middle). To test these ideas, I performed analyses paralleling those performed on LOW INC on the income share of the top 20% and the income share of the middle 40%. In each case, SSEXPOP was not significantly related to the income variables. My data give no support to the idea that social welfare spending affects the overall distribution of income in society. SSEXPOP bears no relationship to social equality as we have defined it. 293 CHAPTER VI APPENDIX Summary Characteristics of the Social Welfare and Other Variables Variable LOW INC EPC FCI SMP FCI * SMP IA FCI * IA FU FCI * FU SMP * FU SMP * IA FU * IA FCI * EPC FU * EPC EPC * IA SMP * EPC Pop. Grow. Rt. Pct. P0p. 65 + SSEXPOP Mean 13.32 1741.73 71.96 60.64 4500.12 .5167 41.45 .7910 59-72 48.40 31.20 .4333 100489.8 1150.38 118.65 99414.30 2.10 6.14 135-83 294 mama Deviation 4.36 2733-27 18.01 21.26 22220.16 .504 41.70 .4096 34-13 32-07 37.82 .4997 154935-89 2365.48 214.35 147271 . 68 .940 3.81 185.74 Eases 44 63 67 56 56 60 60 67 67 56 49 6O 63 63 57 53 64 64 51 CHAPTER VII POLITICAL STRUCTURE CENTRALIZATION AND SOCIAL WELL-BEING: COMMENTS AND CONCLUSIONS In the preceding chapters I have investigated the impact Q5; political (and other) variables .3513. three separate social policy areas. Two final/tasks are to be accomplished in this chapter. First, I shall review the nature of the relationships between federalism, fiscal centralization and informal political power concentration and the social policy areas of health, education and social welfare. As part of this task, I also address the role of economic development and the impli- cations of the findings for'policy-making in various national settings. Second, because the results of this study very often differ from and sometimes contradict hypotheses and earlier study results presented in the literature, I shall, in the interest of theory-building, consider this study within the broad context of previous literature. How and why do our findings differ from the earlier research and writings (reviewed in the first two chapters)? How do our findings "fit" with what we "know" about the effects of politics on policy? Finally, what further investigations of the relationships we have studied might be undertaken? 295 296 Centralization and Sociaerolicy FederalAUnitary Status A large body of literature addresses the topics of intergovern- mental relations, federalism and unitary structure. An assumption of the literature is that formal-legal structure, the location of decision- making (center or locale) and the networks of authority and government relationships make a policy difference. Our own research, on the con- trary, leads us to question earlier claims. We have found, using aggregate data, that whether a country has a federal or unitary structure bears no significant relationship to the overall size of the public sector, to health policy outputs and outcomes, to education outcomes, and to social welfare outputs and outcomes. The one exception, of course, is the role of unitary status in affecting per capita education expenditures: unitary structures moderate the otherwise positive im- pact of economic level on education expenditures. Especially at high levels of economic deve10pment, commitment to education is lessened if a country has a unitary structure. This research, then, does not support the related notion of a trade-off between local control and social well-being; in the case of education expenditures, in fact, there is evidence that the opposite is true. Local control (federalism) leads to greaterreducation policy commitment. Nor have we uncovered evidence of a "federal solution" to the problem (referred to by some scholars) of the ethnicization of goods and providing social goods and services in extremely heterogeneous societies. On the average, heterogeneity has little or no impact on “~--—... I. v. r-aa-n ...-aw», \n . levels of policy outputs or outcomes and federalism displayed no special rmna.""" fin", , a-..“ moderating role on the impact of heterogeneity on social spending. 297 Mere interestingly, our results suggest that if extreme socio- cultural or ethnic fragmentation leads to communal conflict (a popular theory which we cannot address directly here), that conflict does not arise because of ggpppl limitations that are due to heterogeneity. In only one test did heterogeneity appear to affect social policy: extreme heterogeneity was found to have a negative impact on gross enrollment ratios for primary and first levels of education. (Of course, this does not preclude perceptions of discrimination or mismanagement.) Fiscal Centralization Our research indicates that the location of fiscal decision- making -- center or locale -- does make a difference in.§pp§ policy situations. As fiscal centralization and central control over taxation and expenditures increase, the size of the public sector, health ex- penditures, and social welfare expenditures decrease, other factors being equal. (In the case of health and social welfare expenditures, however, the negative effect levels off at high levels of fiscal cen- tralization.) In the case of secondary and higher education, the impact of fiscal centralization varies with economic level: at low levels of economic deve10pment, fiscal centralisation positively affects the percentage of the population enrolled in higher education; at middle and high levels of economic development, centralized fiscal decision-making leads to lower enrollments. Other policy situations are not significantly affected by the degree of fiscal centralization: fiscal centralization is not related to national life expectancy, nurse availability, nutrition levels, edu- cation expenditures, gross enrollment ratios or relative income shares of the poorest 40% of households. 298 Informal Political Authority Concentration Although the legislative strength of the majority political party (SMP) has a direct negative impact on life expectancy, in every other situation where SMP affects policy, it does so in interaction with economic development (EPC). Yet the direction of that impact varies by policy. In the case of health expenditures, majority party dominance in the legislatures of underdeveloped nations leads to greater per capita health spending, other factors being equal. At middle and high levels of national affluence, however, majority party dominance decreases commitment to health spending. In the case of secondary and higher education enrollments, on the other hand, majority party dominance in the legislature has a nggative impact in underdeveloped countries and a pgsitive impact in moderapaly. and highly developed nationa. The role of partisan strength (SMP) in social 1t_e_lfare Jolicy varies depending upon our focus: outputs or outcomes. Majority party control has a negative impact on social security expenditures and the strength of that impact increases with economic level. But, majority party dominance increases the relative income share of the poorest element of society. And, again, the strength of the effect increases with economic level, other factors being equal. Finally, partisan strength is not significantly related to the size of the public sector, nurse availability, nutrition levels, education expenditures nor gross enrollment ratios. The degree of interest articulation (IA) is related to three policy situations. First, negligible levels of interest articulation, where the power to voice interests is concentrated in the hands of a few, 299 lead to greater health spending. IA has a direct positive effect on per capita health expenditures. The impact of the degree of interest articulation on gross enrollment ratios, on the other hand, varies with the level of economic development. Although centralized control of interest representation has a negative impact on the percentage of school-aged children attending school in underdevelgped nations, in moderately and highly developed countries, it has the opposite effect. In these situations, negligible levels of interest representation or articulation lead to higher enrollments. Finally, the relative income share of the poorest 40% of households is increasing;y diminished as economic development increases gasp interest articulation is concentgated at the national center, controlling for effects of other causal factors. Income equality is best served by a free, unrestricted voicing of interests, needs and demands. The Role of Economic Level This thesis opened with reference to an ongoing debate among scholars over the role and importance of economic and industrial develop- ment in affecting social development, social well-being, and a redis- tribution of social goods in the underdevelOped nation. A growing con- sensus among students of the Third World is that economic and.appaa; development cannot be conceptually or theoretically separated; develop- ment strategies must simultaneously be quantitative (with concern for optimum industrial or technological development) and qualitative (with concern for population-wide social development, including moves to eradicate poverty and suffering). This research addresses questions growing out of these concerns. For example, what role does economic level play in affecting social 300 policy? More importantly, does ecgpgpigbgrpgth lead naturally to social deve10pment (as proponents of the "trickle-down theory" argue)? Our results indicate that although in a few policy situations such as nurse availability and nutrition levels, economic development has a positive (though curvilinear) impact, in most of the relationships that were significant: the impac’g ..af--39989194932Y919Pment,“1cmoderated: or even ne ég,by the political situation; rAn example of the latter phenomenon is in the area of health expenditures. Whereas at low levels of majority party dominance and, to a lesser extent, at middle levels of majority party dominance, economic deve10pment has a ppaative effgct on per capita health expenditures (controlling for other variables), gpagg the majority party clearly controls the lagislature, health expenditures do not increase with economic growth. A similar situation exists with respect to gross enrollment ratios: the positive impact of economic deve10pment is moderated by the level of interest articulation. Where interest anacplatioggacmtmlflipgd; (IA = 1), and, controlling for the effects of other factors, economic development has a positive impact on gross enrollment ratios. Where the right to voice interests is un- restricted, on the other hand, increases in enrollment ratios do not follow increases in economic growth. Political variables moderate the impact of economic development on education and social welfare expenditures. While economic development has a positive impact on per capita education expenditures in both the federal and the unitary system, this impact is lessened by the unitary structure of intergovernmental relations. Likewise, while economic growth has a positive impact on per capita social security spending, this effect is more clearly seen where the degree of majority party 301 dominance in the legislature is low. At high levels of partisan control, the strength of the positive effect of economic development is diminished. Finally, for two policy outcome variables -- secondary and higher education enrollments and the relative income share of the poorest 40% of households -- the impact of economic development is dependent upon the nature of two separate political variables. The impact of economic growth on the percent of the population enrolled in secondary and higher education depends on the levels of majority party strength and fiscal centralization. First, when majority party dominance is low, and, holding other effects constant, economic growth has a negative effect on enrollments; when partisan strength is at mean or high levels, economic development has a positive impact on enrollments. Second, when fiscal centralization is low and, to a lesser extent, when it is at the mean level for the sample, economic development has a positive effect on enrollments. But, where fiscal decision-making is highly centralized, economic level has a negative effect on enrollments in higher education. The impact of economic development on the relative income share of the poorest 40% of households depends upon the degree of interest articulation and of majority party control of the legislature -- both indicators of informal power patterns. At each level of partisan aprength, economic gppwth has a negative impact on the income share of the poor (the effect of economic growth occurs in interaction with partisan strength). When interest articulation is centralized, more- over, economic development again has a negative effect. But, paaa interest representation is open and.unrestricted, economic growth has a positive effect on income equality, controlling for all other effects. 302 (In each case, we hold the other causal variables constant at their mean values.) In sum, the role of economic deve10pment in affecting social policy is not straightforward. Instead, in each of the three policy areas examined, the role of economic growth is moderated or otherwise affected by the structure of politics. Moreover, not only do the effects of the three dimensions of structural centralization on economic development change by policy area, they also change by the type of centraaization considered. We can speculate on the eauSes of the changing impact of economic growth. First, the impact of economic growth on policy and the degree to which political factors affect that impact may depend 99 thg level and.type of controversy surrounding the social issues. Our study results &&&&& W:r~'- -—-I’l"J u,"4 ”t I“), indicate that in two issue areas often surrounded by controversy -- education outcomes (as indicated by secondary and higher education en- rollment levels) and income distribution -- the impact of economic growth was moderated by two separate political variables. Second, the relation- ships may depend on what political factors come into play or what political forces are employed by interested groups to press for desired results. It is interesting to note, for example, that patterns of power in national law-making bodies (SMP) frequently moderate the role that economic growth plays in social policy. Political elites in the legislature use their'power and influence to control the way national resources will be used. We can argue, at this point, that economic development does not easily translate into social development, and, that the structure of politics plays a very important role in determining just how economic 303 strides will affect social well-being. Political strategies and plan- ning can determine how and whether economic development will benefit a. ._ f‘ ‘7'“ >..)~,,_’J, 'r‘lo L:_ . .3 ’ ' . ‘1 . < society. g: , p c .*7~__.__._ '.‘ ' 4 ~ . y.“ 5 w- Mr) “‘L to \e— Conclusions: Pppitics, Social Policy and SociplLWepgaeing A--\,_-.-,--__.,_z--.ef::_~e The central focus of this study is the investigation of the role r_»J' of political (and secondly, economic) variables in explaining social policy outputs and outcomes and, ultimately, social well-being or social equality. This research paralibls efdets being made by the World Bank, the United Nations, the Aspen Institute for Humanistic Studies, the Bariloche Foundation of Argentina and many other international, public and private organizations thatfiafgwfocusing on deyglgpgfigiafiifa' "fl-V’ tegies designed to meet basic human needs (housing, food, clothing, MJ—qw-M w- e x» health and education, for eagle)! Harland Cleveland writes: During three years of sudden conceptual change, from 1974 to 1977, national development strategies, international negotiations and global organizations have begun tgflhe deeply affétted by the simple notion that the purpose of economic deve10pment and international cooperation is to meet / e human needs of the neediest;l Combined with new hm.-- attitudes toward economic growth and environmental damage, the appearance of 'basic human needs' at center stage begins a new act in the continuing drama of world development.2 \ It has become.increasingly clear dyer the past two DevelOpment Decades, however, that meeting basic human needs, alleviating poverty and providing even a minimum level of social well-being will require both economic gro h and massive changes in the distribution of income J 304 and the opportunities, gogds and services that income provides.3 Sggial deyelopment/has not ftriékladfldown" from rising GNP's and national industrial growth. In this context, the question of the role of politics (and political structure specifically) in affecting social policy and distri- bution patterns is especially relevant. Several broad generalizations can be made, based on this research. First, unilateral statements about the impact of government centralization on policy are grossly midlead- ing. We have found that not only is governmental authority centraliza- tion a multidimensional concept, but the effects of the several aspects of centralization also differ from each other and differ by policy area (as is evident in the effects of partisan strength and interest articu- lation).1+ Second, in many policy situations economic and political variables interact to moderate each other's effects. The impact of the lo..- political variablesgonwpolicy outputs and outcomes ofteanepends upon Xi'an“), L the level of economic development; similarly, the impact of economic “...... -m Hv—v-u‘uqyo v - -—-.r development is/qualified by the palitical situation. The research results warn against prescribing generalized develop— ment strategies and point instead to the need for careful policy plan- ning. Development strategies that move toward social well-being and social equality must be chosen to fit the local (national) political and economic environment. Meet probably, ih9§9 strategies will involve some mixture of central and local decision-making and policy implemen- tatign. Designing optimum deve10pment strategies that focus on quantitative and qualitative aspects of growth, therefore, is a difficult task. This task is made more difficult, moreover, because of the gap that often 305 exists between expeggétures and policy ogtcomgs. Our research demon- M’em ' strates that appropgiations and commitments expressed by expenditure levels are not easily translated intgwpplicyvogtcomes. Apparently? there are breakdowns or othgrypppplems in the implementation process that prevent target groups and programs from being reafhed.‘ Red tape, bureaucratic delay and multiple layers of bureaucrats andipolitical officials may all be factors figuring in this gap. (Isolating causes of the gap between outputs and outcomes poses an interesting problem, but it is beyond the scope of our study.) The Research and Prgyagus Literature It became clear in Chapters IV, V and VI that our research results- diverge at several points from earlier study results. These differences in results are noted in the text. The task here is to speculate briefly on causes for varying findings. One obvious difference between this study and earlier ones is the variation in research designs and samples. First, this research goes somewhat beyond earlier studies of social policy in that three separate pglipypagaas are considered (where both outputs and outcomes are investigated) based on a relatively large, diversified sample of nations. In contrast, Wilensky studies social welfare policy in 22 rich nations, Cameron and Hofferbert study educa- tion finances in 8 Western European nations, Pommerehne studies fiscal centralization in 6 Anglo-European countries, and Oates considers fiscal functions in 58 countries. In addition, Peacock and Wiseman investigate the increased central government responsibility for social spending in Great Britain and Echols considers regional equalization programs in 5 capitalist and 5 communist nations. Results from these studies may be constrained by the nature of the sample used and the 306 Questions asked. The generalizations about fiscal centralization from Peacock and Wiseman's study are constrained by the fact that the relationships have not been tested in a variety of national situations. (The study of the United Kingdom is an historical one based from 1890 to 1955 and aimed at explaining why social spending has increasingly come under the domain of the central government.) Similarly, the applicability of Wilensky's results may be constrained by the sample he uses -- the 22 richest nations of the world. For example7/the impact of political structure centralization is an important question often raised by Third World policy-makers who face very different problems of national integration and sfdiai and.p6§iti;al development. We have found, in fact, that relationships among political and social policy variables often change depending upon thejaébnomic level. Further, while recognizing the different starting point and focus of Oates' study, we can argue that his results may be constrained by the fact that he fails to move beyond a consideration of fiscal cen- tralization's impact on the size of the public sector; results may have been different, for instance, if he had considered specific social policy programs. But, we must also point out that our measures of fiscal centralization differ to the extent that I have incorporated more fiscal information in my fiscal centralization index (see Chapter III). Finally, the Echols and Cameron and Hofferbert studies are based on subnational data. (Recall that both studies report that unitary structure has a positive impact on regional equalization of social pro- gram expenditures and suggest that federalism has a conservative impact 307 on social program implementation.) Within-nation patterns and varia- tions may thus become visible. Our study does not go beyond national- level and broad generalizations about aggregate patterns. Second, I have conceptualized political structure centralization as consisting of at least three separate dimensions and have tested for effects of each type of structural centralization on each of three social policy areas. This approach allows for tests of a number of relationships not considered earlier; For example, Wilensky employs a composite index to measure degree of centralization; he combined infor- mation on formal-legal status, centralization of fiscal decision making, and central appointments of local political officials into a single measure. Using this measure, it is impossible to discern what relation- ships the several aspects of centralization have to social welfare. Based upon his results, Wilensky argues that "centralization" leads to social welfare program development. Third, a number of hypotheses suggested by scholars (concerning the nature and impact of federalism, the special case of the hetero- geneous society, for example) are not supported in this research. Samuel Beer postulates that federalism leads to increased.public spend- ing due to the influence of interest groups located at the several levels of government (the "professional bureaucratic complex" and the "intergovernmental lobby"). When the hypothesis is tested empirically (as it is here), however, it is not supported. Federalism has no direct or main effect on any of the policy areas analyzed (although unitary status lessens the impact of economic development somewhat on per capita education expenditures). Other hypotheses concerning a "federal solution" to problems of providing services in the ethnically 308 heterogeneous society are, likewise, not supported. Again, these hypotheses had not been tested empirically in the earlier literature. (This lack of empirically based investigation is characteristic of the federalism literature; to that extent, political scientists have not gone far in showing what difference the structure of intergovernmental relations makes for policy.) How does this research, then, "fit" with earlier research into the relationships among politics, society, economic growth and social policy? And, what further investigations might be undertaken? J This study, along with others, refutes earlier claims that social a policy outputs are (exclusively) a function of economic or socio- economic growth, and that politics make little difference for policy. Our study suggests that determinants of policy may change by issue area and by level of economic development (Hofferbert, Hogan, Peters and others had suggested earlier that this might be the case). But more importantly, our research suggests that although economic growth is very important to social program deve10pment, the structure of politics moderates the impact of economic development. To state it more succinctly: political and economic factors interact to moderate each others' effects. Further, in an attempt to clarify the relationships between authority centralization and policy, we have suggested that several types or aspects of structural centralization exist, and, that the impact on policy changes depending upon what aspect of centralization is being considered. Thus, we have addressed several broad questions posed in earlier writings. Yet, this research seems to raise as many questions as it was 309 intended to address. We may suggest several topics for future research. First, are our results constrained by the lgzg; of study and the use of aggregate data? Might we have found significant relationships between the structure of intergovernmental relations (federal or unitary) and policy if we had been able to observe local, regional and within-nation data? Do aggregate data obscure important within-nation variations in these relationships? AlthOUgh this is an important question, social scientists have been limited in efforts to study it, due to the paucity of subnational data (especially for Third World countries). Second, do relationships similar to those that appear in the health, education and social welfare areas also exist in other policy areas, such as transportation, defense, commerce, et cetera? Do political, social and economic variables play different roles outside of the human services area? Does the impact and role of political structure centralization change significantly in other policy arenas? Third, what roles do other political variables play in these (and other) policy areas? For example, does political ideology make a difference for the impact that the majority party and legislative power patterns have on policy? Do these factors affect the type and degree of influence exerted to affect policy? Similarly, do the natures and types of interest groups active in society affect the role that interest representation will have on policy? Does the type of political party in control predict the level of interest articulation that will be tolerated? In short, a variety of political variables may significantly affect policy outputs and outcomes. Fourth, our investigation of the roles played by political and economic variables in affecting the distribution of income and the 310 income share of the poorest groups in society is limited by the availability and quality of the data. Our results in this area are constrained;to the extent that/they are based on information from only 44 nations and that there are discontinuities from nation to nation in how the data were gathered and recorded. Yet, the importance of under- standing distribution patterns and how patterns might be changed to promote social development leads us to continue our study. Obviously, there is much that remains to be done in this area of research. CHAPTER VII FOOTNOTES 1 The basic human needs approach considers education a basic human need (9 years of education are considered "basic") rather than a luxury. Education provides an important impetus toward the accumula- tion of human capital and potential. It presents an important, central stage in social development. At minimum, it opens new horizons to individuals. See MCHale and McHale, Basic Human Needs, Transaction Books, New Brunswick, N.J., 1977; and Hopkins and Scholnick in the I.L.O. publication, "Tripartite World Conference on Employment, Income Distribution and Social Progress and the International Division of Labour," Background Papers, Vol. I. 2 Harland Cleveland in McHale and MCHale, op. cit., p. 3. 3 See, for example, Hollis Chenery, et. al., Redistribution With Growth, Oxford University Press, London, 197#; and Hopkins and Scolnick, op. cit., p. 9. h The role of some political variables seems clear. While FU status, in general, has little or no effect on social policy, fiscal centralization usually impedes social policy development. The role of informal authority concentration, however, is not so clear. The effects of SMP and IA are not easily generalized. 311 BIBLIOGRAPHY 312 Aaron, Henry. "Social Security: International Comparisons," in O. Eckstein, editor, Studies in the Economics of Income Main- tenance. Brookings Institute, Washington, 1967: Adelman, Irma and Cynthia T. Morris. Economic Growth and Social Equity in Developing Countries. Stanford University Press, 1973. Ahmad, Mushtag. Government and Politics in Pakistan. Space Publishers, Karachi, Pakistan, 1970. Alford, Robert. Party and Society. Rand McNally, Chicago, 1963. Allardt, Erik. "Implications of Within Nation variations and Regional Imbalances for Cross-National Research," in Merritt and Rokkan, editors, Comparing Ngtions. Yale University Press, 1966. Almond, Gabriel and Bingham Powell. "A Developmental Approach to Political Systems: An Overview}'in Jason Finkle and Richard Gable, editors, Political Development and Social Change. John Wiley & Sons, New York, 1971. Anderson, C., F. vonder Mehden and C. Young. Issues 9f Political Development. Prentice-Hall, Inc., Englewood Cliffs, N. J., 1974. Anderson, Odin. "Are National Health Services Systems Converging? Predictions For The United States," The Annals, Vol. 43#. (November, 1977). Ashford, Douglas. "Territory vs. Function: Toward a Policy-Based Theory of Subnational Government." Paper delivered at the American Political Science Association meetings, 1976, Chicago, Illinois. Baster, Nancy and Wolf Scott. Levels 9: Living and Economic Growth. U. N. Research Institute for Social Development, Geneva, 1969. Becker, Gary and Barry Chiswick. "The Economics of Education: Education and the Distribution of Earnings," American Economic Reviewy Vol. 56 (May, 1966). Beer, Samuel. "A Political Scientist's View of Fiscal Federalism," in Wallace Oates, editor, The Political Economy gfiFiscal Federal- ism. D. C. Heath & Co., Lexington, Mass., 1977. Beer, Samuel. "Political Overload and Federalism," Policy, Vol. 10 (Fall, 1977). Beer, Samuel. The British Politic§l_§ystem. Random House, New York, 1973- Blau, Peter M. Exchange and POwer‘ig_Social_Life. John Wiley and Sons, New York, 196#. Blondel, Jean. Ag_lptroduction tg_Comparative Government. Wedenfeld and Nicolson, London, 1959. 313 Boaden, Noel. Urban Policy Making: Influences g9 Countpy Boroughs ip England.§gd_Wales. Cambridge University Press, Cambridge, 1971. Bowie, Robert and Carl Friedrich. Studies ig_Federalism. Little, Brown & Co., 1954. Break, George. Intergovernmental Fiscal Relations ig_the United States. Brookings Institute, 1967. Brunner, R. and G. Brewer. Organized Com lexit J. The Free Press, New York, 1971. Buchanan, James M. "Federalism and Fiscal Equity, " American Economic Review, Vol. 40 (September, 1950). Buchanan, J. and R. Wagner. "An Efficiency Basis for Federal Fiscal Equalization," in Margolis, editory The Analysis g§_Public Output. National Bureau of Economic Research, New York, Columbia University Press, 1970. Cameron, D. R. and R. I. Hofferbert. "The Impact of Federalin on Education Finance: A Comparative Analysis, " European Journal of Pplitical Research, Vol. 2(Sept. , 1974). Canadia Tax Foundation and.Institute of Intergovernmental Relations. Fiscal Need and Revenue Egualization Grants. Toronto, 1969. Census 9: Switzerland (by year). Berne (Federal Bureau of Statistics). Central Intelligence Agency. National Basic Intelligence Fagtbook. Office of Geographic and Cartographic Research, Washington, D.C., 1975- Chenery, Hollis, M. Ahluwalia, C. L. G. Bell, J. Duloy and R. Jolly. Redistribution With Growth. Oxford University Press, London, 197#. Cnudde, Charles and Donald McCrone. "Party Competition and Welfare Policies in the American States," American Political Science Review, Vol. 63 (September, 1969). Cockroft, James, Andre Gundera Frank and Dale Johnson. Dependence and Underdevelppmept. Doubleday, Garden City, N. Y., 1972. Cowart, Andrew T. "Anti-Poverty Expenditures in the American States: A Comparative Analysis," Midwest Journal of Political Science, Vol. 13 (May, 1969)- Cutright, Phillips. "Income Redistribution: A Cross-National Analysis," Social Forces, 46 (December, 1967). Cutright, Phillips.‘ "Inequality: A Cross-National Analysis," American Political Science Review, Vol. y(August,1967). 314 Cutright, Phillips. "Political Structure, Economic Development and National Social Security Programs," American Journal Lf Sociology, Vol. 70 (March, 1965). Daalder, Hans. "The Consociational Democracy Theme," World Politics, Vol. 26: a (July, 197i»). Dahl, Robert. Polyarohy. New Haven, Yale University Press, 1971. Davie, Bruce and Joseph White. "Equalization Alternatives in Grant-in- Aid Programs: Allotment of Formulas and Measures of Fiscal Capacity," National Tax Journal, Vol. 20 (June, 1967). Davis, Rufus. "The Federal Principle Reconsidered," in Aaron WildaVSKy, editor, American Federalism in Perspective. Little, Brown & Co., Boston, 1967. Dawson, Richard and James Robinson. "Inter-party Competition, Economic Variables, and Welfare Policies in the American States," Journal Lf Politics, Vol. 2 (May, 1963). Despres, Leo A. Cultural Pluralism and Nationalist Politics in British Guiana. Rand-r McNally, Chicago, 1967. Duchacek, Ivo. Comparative Federalism: The Territorial Dimension of POIitics. IHolt, Rinehart and.Winston, Inc., New York, 1970. Duesenberry, James. Income Saving and the Theory Lf Consumer Behavior. Oxford University Press, New York, 1967. Dunn, James A. Jr. "Consociational Democracy and Language Conflict: A Comparison of the Belgian and Swiss E eriences," Comparative Political Studies, Vol. 5 (April, 1972. Dye, Thomas. Politics, Economics and the Public: Policy Outcomes in the American States. Rand McNally, Chicago, 1966. Easton, David. "An Approach to the Analysis of Political Systems," World Politics, Vol. 9 (April, 1957). Echols, John M. "Politics, Budgets and Regional Equality in Communist and Capitalist Systems," Comparative Political Studies, Vol. 8 (October, 1975). Echols, John M. Politics, Policy and Equality Under Communism and Democracy. Unpublished disseration, University of Michigan, 1976. Eckstein, Harry. "Authority Patterns: A Structural Basis for Political Inquiry," American Political Science Review, Vol. #7 (December, 1973 Economic Council of Canada. Fifth Annual Review, Queen's Printer, Ottawa , September , 1968 . 315 Elazar, Daniel J. "Federalism," in International_Encyclopedia 9§_the Social Sciences, Vol. 5, New York, McMillan, 1968. El Fathaly, Omar, Monte Palmer and Richard Chakarian. Political Development éfli Bureaucracy ig_Libya. D. C. Heath & Co., Lexington, Mass., 1977. Emerson, Rupert. From Empire to Nation. Harvard University Press, Cambridge, Mass., 1950. Epstein, Leon. "Comparative Study of Canadian Parties," American Political Science Review, Vol. 58 (March, i96u). Epstein, Leon. Political Parties ig_Western Democracies. Praeger Publishers, New York, 1937. Europa Yearbook, 1971: A_World Survey, The. Europa Publications Limited, London, 1971. Feldman, Elliot. "Comparative Public Policy: Field or Method?" Comparative Politics, Vol. 10 (January, 1978). Feldman, Herbert. Revolution in Pakistan. Oxford University Press, London, 1967. Ferry, W. H. "The Case for a New Federalism," Saturday Review, June 15, 1968. Fishman, Joshua. Bilin Education. Newbury House Publications, Inc., Rowley, Mass., 197 . Fishman, Joshua, C. Ferguson and J. Das Gupta, editors. Langu_age Problems g£_Developing Nations. John Wiley and Sons, New York, 1968. Fox, Thomas and S. M. Miller. "Occupational Stratification and Mobility," Studies ig_Comparative International Development, Vol. 1, 1965. Franck, Thomas, editor. Why Federations Fail. New York: New York University Press, 1968. Frey, Frederick. "Political Development, Power and Communications in Turkey," in Lucian Pye, editor, Communications and Political Development, Princeton University Press, N. J., 1963. Frey, Rene L. "The Interregional Income Gap as a Problem of Swiss Federalism," in Wallace Oates, The Political EconomyfgfiiFiscal Federalism, Lexington Books, D. C. Heath & Co., Mass., 1977. Fry, Brian and Richard Winters. "The Politics of Redistribution," American Political Science Review, Vol. 64 (June, 1970), p. 508 ff. 316 Cans, Herbert. Mere Eguali y. Pantheon Books, New York, 1973. Geertz, Clifford, editor, 01d Societies and New States. The Free Press, New York, 1963. Gilb, Corinne L. Hidden Hierarchies. Harper & Row, New York, 1966. Glendening, Parris and Mavis Mann Reeves. 'Eragmatic Federalism. Palisades Publishers, Pacific Palisades, Ca. Glenn, Norval. "Social Security and Income Redistribution," Social Forces, 46 (June, 1968). Gordon, Margaret S. The Economics of Welfare Policies. Columbia University Press, New York, 1933. Gramlich, Edward. "Intergovernmental Grants: A Review of the Empirical Literature," in Wallace Oates, editor, The Political Economy of Fiscal Federalism, Lexington Books, D. C. Heath Co., Mass., 1977- Greene, K. V. "Some Institutional Considerations in Federal-State Fiscal Relation," Public Choice, Vol. 9, 1970. Grodzins, Morton. The Amerigag_§ystgm. Chicago, Rand MdNally, 1966. Gross, Neal. "Theoretical and Policy Implications of Case Study Findings About Federal Efforts to Improve Public Schools," The Annals g§_the Amerigag Academy g£_Political and Social Science, Vol. #34 (November, 1977? Grubb, W. Norton and Stephan Michelson. States agd_Schools. Lexington Books, D. C. Heath & Co., Toronton, 1974. Gurr, Ted Robert. Cross National Studies 9§_Civil Violence. The American University Center for Research in Social Systems, Washington, D. C., 1969. Curr, Ted Robert. "Persistence and Change in Political Systems, 1800-1971," American POlitical Science Review, Vol. 68 (December, 197a)- Gurr, Ted Robert. Politimetrics. Prentice-Hall, Inc., Englewood Cliffs, N. J., 1972. Hage, Jerald and J. R. Hollingsworth. "The First Steps Toward the Integration of Social Theory and Social Policy," The Annals, Vol. 43L; (November, 1977). Harrison, Russell. Equality ig_Public School Finance. Lexington, Mass., D. C. Heath & Co., 1973. Haug, Marie. "Social and Cultural Pluralism as a Concept in Social System Analysis," Ih§_American Journal g§_Sociolog1 (November, 1967 . 317 Heisler, Martin. "Ethnic Conflict in the World Today," The Annals, V01. 433 (September, 1977). Heisler, Martin O. and B. Guy Peters. "Toward a Multidimensional Frame- work fOr the Analysis of Social Policy," The Annals, Vol. 434 (November, 1977). Heller, Walter W. New Dimensions of Political Economy. Cambridge, Mass., Harvard University Press, 1965. Hermann, Charles, M. East, M. Hermann, B. Salmore and S. Salmore. Creon: A_Foreigo Events Data Set. Sage Publications, Beverly Hills, 1973. Hibbs, Douglas. "Political Parties and Macroeconomic Policy," American Political Science Review, Vol. 71 (December, 1977). Hicks, U. K., F. G. Carnell, J. R. Hicks, W. T. Newlyn, and A. H. Birch. Federalism and Economic Growth in Underdeveloped Countries. Oxford University Press, 1961. Hill, Richard C. "Separate and Unequal: Governmental Inequality," American Political Science Review, Vol. 68 (December, 1974). Hinrichs, ,Harley. ,A_General Theory of_Tax Structuro_Change Duriog Economic Development. The Law School of Harvard University, Cambridge , 1986.— Hofferbert, Richard and Ira Sharkansky. Chapter in Allan Kornberg, editor, Legislatures io_Comparative_Pgr§pective. McKay Co., New York, 1973. Hofferbert, Richard and Ira Sharkansky, editors. State and Urban Phlitics. Little, Brown & Co., Boston, 1971. Hofferbert, Richard. "The Relation Between Public Policy and Some Structural and Environmental Variables in the American States," American Political Science Review (March, 1966). Hogan, J. B. "Social Structure and Social Policy," Comparative Politics, vol. 4 (July, 1972). Holmes, Jean and Campbell Sharman. The Australian Federal System. Allen and Unwin, Ltd., Sydney, 1977. Holzner, Burkhart and L. Salmon-Cox. "Conceptions of Research and Development fbr Education in the United States," Too Annals, Vol. 434 (November, 1977). HOpkins, M. J. D. and A. Scolnik. "Basic Needs Growth and Redistribu- tion: A Quantitative Approach," in International Labour Office publication, World Conference og_Employment, Income Distribution ood_Social Progress and the International_Division of_Labour. Background papers, Vol. I, p. 9. 318 Horowitz, Donald L. "Cultural Movements and Ethnic Change," 1192 Annals, Vol. 433 (September, 1977). Hudson, Michael. Conditions of Political. Violence and Instabilipy: _A_ Preliminary Test of Three Hypotheses. Sage Publications, Inc. , Beverly Hills, 1970. Huntington, Samuel. Political Order in W Societies. Yale University Press, New Haven, 1968. Inman, Robert, Martin McGuire, Wallace Oates, Jeffrey Pressman and Robert Reischauer. Financing The New Federalism. John Hopkins University Press, 1975. International Bank for Reconstruction and Development, World Tables, 19m; 1926- International Labour Office. "Costs of Non-Statutory Social Security Schemes," International Labour Review, Vol. 78 (October, 1958)- International Labour Office. Labour Costs _i_I_1_ European Industry. Geneva, 1959. International Labour Office. Social Security - A Workers' Education Nahual. Geneva, 1958- International Labour Office. 11113 Cost or; Social Security, won-go, 1262-Zl0 Geneva. International Labour Office. Yearbook of labour Statistics, 1971. Geneva, 1973. International Studies of Values in PolitiCS. Values god the Active Community. The Free Press, 1971. Jackman, Robert W. "On the Relation of Economic Development to Demo- cratic Performance," American Journal o_f_ Political Soience, Vol. 17 (August. 1973)- Jackman, Robert W. "Political Democracy and Social eqality: A Com- parative Perspective," American Sociological Review, Vol. 39 (February, 1974). Jackman, Robert. "Political Parties, Voting and National Integration," Congrative Politics, Vol. 4 (July, 1972). J ackman, Robert W. Politics and Social Eguality. John Wiley and Sons, New York, 1975. J ackman, Robert W. "Socialist Parties and Income Inequality in Western Industrial Societies," in Joumal of Politics (February, 1980). Forthcoming. 319 Jackman, Robert W. "The Impact of Outliers on Income Inequality," in American Sociological_Review, Vol. 45 (February, 1980). Forth- coming. Jackman, Robert W. "The Predictability of Coups D' Etat: A Model With African Data," American Political Science Review, Vol. 72 (December, 1978 - Jacob, Herbert and Michael Lipsky. "Outputs, Structure and Power: An Assessment of Changes in the Study of State and Local Politics," in Sharkansky and Hofferbert, editors, State and Urban Politics, Little, Brown & Co., Boston, 1971. Jain, S. Size Distribution of Income: A_Compilation of Data. World Bank, Washington, D. C., 1974. Janda, Kenneth. "Internation Comparative Political Parties Project," Northwestern University, ENanston, Illinois, ICPP reports #7. Jencks, Christopher. Inequality. Basic Book, Inc., New York, 1973. Jennings, Edward Jr. "Party Constituency and Welfare Policy Change in American States," Paper delivered at the Midwest Political Science Convention; Pick Congress Hotel: Chicago, 1978. Joseph, Lawrence B. "Normative Assumptions in Educational Policy Research: The Case of Jencks' loequality," The Annals, Vol. 434 (November, 1977). Katz, Ellis and Benjamin Schuster, editors. Dialogge Qo_§opparative Federalism. Center for the Study of Federalism, Temple University, 1978- Kaufman, Herbert. Politics and Policies i2 State and Local Governments. Prentice-Hall, 1933. Key, V. 0. American State Politics. Alfred A. Knopf, New York, 1967. Koropeckyj, I. S. "Equalization of Regional Development in Socialist Countries: An Empirical Study," Economic Development and Cultural Change, Vol. 21 (October, 1972). Kothari, Rahni. Politics ig_India. Little, Brown & Co., Boston, 1970. Lampman, R. J. "The Effectiveness of Some Institutions in Changing the Distribution of Income," American Economic Review, Vol. XLVII (May. 1957). p. 519-528. Lane, David. The End of Ideology: Stratification Under Socialism. Penguin Books, London, 1971. Legum, Colin and John Drysdale. Africa Contemporarprecord. Africa Research Limited, Exeter, United Kingdom, 1970. 320 Lenski, Gerhard. Power and Privile e: A Theory o§_Social Stratification. MCGrawbHill Co., New York, 196%. Levy, Frank 3., Arnold Meltsner and Aaron Wildavsky. Urban Outcomes. University of California Press, Berkeley, 1974. Levy, Marion Jr. Mbdernization and the Structure of Societies. Princeton, New Jersey, Princeton University Press, 1966. Two volumes. Lijphart, Arend. "Linguistic Fragmentation and Other Dimensions of Cleavage: A Comparison of Belgium, Canada and Switzerland," IXth World Conference paper, Montreal, August, 1973. Linz, Juan and Amando de Miguel. "Within-Nation Differences and Com- parisons: The Eight Spains," in Merritt and Rokkan, editors, Comparing Nations, Yale University Press, 1966. Lipset, Seymour Martin. Political Man. Anchor Books, Doubleday and Co., New York, 1960. McConnell, Grant. Private Power'ooo American Democra_y. Vintage Books, New York, 1953. McHale, John and Magda Cordell McHale. Basic Human Needs: ,A_Framework for.Action. Transaction Books, New Brunswick, New Jersey, 1978. McWhinney, Edward. Comparative Federalism. University of Toronto Press, 1965. Margolis, Julius, editor. The Analysis of Public Output. National Bureau of Economic Research, New York, Columbia University Press, 1970. Maxwell, James and R. Aaronson. Financing State and Local Governments. 1977. Mayer, Lawrence. oomparative Political_lnquiry. Dorsey Press, Home- wood, Illinois, 1972. May, R. J. Federalism and Fiscal Adjustment. Oxford, The Clarendon Press, 1969. Mead, Lawrence. "Health Policy: The Need for Governance," The Annals, Vol. 434 (November, 1977). Merritt, Richard and Stein Rokkan, editors. Comparigg Nations. Yale University Press, 1966. Millikan, M. and D. Blackmer, editors. The Emerging Nations. Little, Brown & Co., Boston, 1961. Milner, Murray. The Illusion of_ ualit : The Effect of Education op: portunity, Inequality, and Social Conflict. Jossey-Bass, Inc., San Francisco, 1972. 321 Merrill, R. C. The Spacio;,0rgonization of_Society. Second Edition. Duxbury Press, 1974. Morrison, D. and H. M. Stevenson. "Integration and Instability: Patterns of African DevelOpment," Americao Political Science Review, Vol. 66 (September, 1972). Moynihan, Daniel P. "The Relationship of Federal to Local Authorities," Daedalus, V01. 96 (Summer. 1967). Mulder, Mauk. "Power Equalization Through Participation?" Administrative Science Quarterl , Vol. 16 (March, 1971). Munger, Frank J. and Richard Fenno Jr. National PolitiCS and Federal AlQ.EQ Education. Syracuse University Press, 196?. Musgiave, Richard A. Fiscal Systems. New Haven, Yale University Press, 1969- Musgrave, Richard A. The Theory of Public Finance. New York: McGraw- Hfll 9 19590 Neumann, Franz. "Federalism and Freedom: A Critique," in Arthur Macmahon, editor, Federalism mature and Emergent, Doubleday and Co., Garden City, N. Y., 1955. Oates, Wallace, editor. The Political Economy of Fiscal Federalism. Lexington Books, D. C. Heath & Co., Lexington, Mass., 1977. Oates, Wallace. Fiscal Federalism. Harcourt, Brace, Jovanovich, Inc., New York, 1972. Okun, Arthur. Eguality ago_Efficiency: The Big Trade-Off. Brookings Institute, Washington, 1975. Organization of Economic Cooperation and.Development. Expenditures and Trends i§_OECD Countries. 1960-i980. OECD publiCation, Paris. Peacock, Alan T. and Jack Wiseman. The Growth of Public Expenditures ig_the United Kingdom. Princeton University Press, Princeton, 1961. Paukert, Felix. "Social Security and Income Redistribution: Comparative Experience," in E. Kassalow, editor, Too Role of Social Security io_Economic Development, U. S. Social Security Administration, Government Printing Office, Washington, 1968. Peters, B. Guy. "Economic and Political Effects on the Development of Social Expenditures in France, Sweden, and the United Kingdom," Journal oi; Political Science, Vol. 16 (May, 1972). Peters, B. Guy, J. C. Doughtie and M. K. MCCulloch. "Types of Demo- cratic Systems and Types of Public Pelicy," Comparative Politics, vol. 9 (April, 1977)- 322 Pommerdlne, Werner. "Quantitative Aspects of Federalism: A Study of Six Countries," in Oates, editor, The Political Economy o£_Fiscal Federalism, D. C. Heath's Co., Lexington, Mass., 1977. Pryor, Frederic. Public Expenditures in Communist and Capitalist Nations. Allen and Unwin, Ltd., London, 1968. Rabushka, A. and Kenneth Shepsle. Politics go Plural Societies: :A Theory‘of_Democratio_lnstabili_y3 Charles E. Merrill PubliShers, Ohio , 1972. Rae, Douglas and Michael Taylor. The Aoalysis of Pelitical Cleavages. New Haven, Conn., Yale University Press, 1970. Rawkins, Phillip M. "Outsiders as Insiders: The Implications of Minority Nationalism in Scotland and Wales," Comparative Politics, Vol. 10 Reagan, Michael. The New Federalism. Oxford University Press, New York, 1972. Rhodes, Robert, editor. Imperialism and Underdevelopment. MOnthly Review Press, New York, 1970. Riker, William. Federalism: Origin, Operation, Significance. Little, Brown & Co., Boston, 1964. Rose, Richard, editor. Electoral Behavior. New York: Free Press, 1974. Rothenberg, Jerome. "Local Decentralization and the Theory of Optimal Government," in Margolis, editor; The Analysis of Public Oupput, Columbia University Press, 1970. Rubin, Leslie and Brian Weinstein. Introduction.to_African.PolitiCs. Praeger PubliShers, New York, 1974. Sacks, Seymour’and Robert Harris. "The Determinants of State and Local Government Expenditures and Intergovernmental Flows of Funds," National Tax Journal, Vol. 17 (March, 1961:). Sawyer, Malcolm. "Income Distribution in OECD Countries," OECD Occasional Studies, July, 1976. Schnore, Leo and Henry Fagin, editors. Urban Research and.Polioy Planning. VOlume I, Sage Publications, Beverly Hills, 1976. SharkanSky, Ira. "Economic and Political Correlates of State GOVern- ment Expenditures: General Tendencies and Deviant Cases," Mid- west Journal of Political Science, Vol. 11 (May, 1967). Sharkansky, Ira. "Government Expenditures and Public Services in the American States," American Political Science Review, Vol. 61 (December, 1967). 323 Sharkansky, Ira. ‘Spogdipguip_the Americao States. Rand McNally, Chicago , 1968 . Sheridan, L. A. and Harry Groves. The Constitution of Mala sia. Oceana Publishers, Inc., Dobbs Ferry, New York, 19 7. Shibutana, Tamotsu and Kian Kwan. Ethnic Stratificotion. MacMillan Company, London, 1965. Silver, Brian D. "Levels of Sociocultural DevelOpment Among Soviet Nationalities: A Partial Test of the Equalization Hypothesis," American Political Science Review, Vol. 68 (December, 1974). Silverstein, Josef. Burma: Military Rule and the Politics o§_Stagpa- tion. Cornell University Press, Ithaca, 1977. Singer, H. W. "Social Forces in DevelOpment: An Overview With Special Emphasis on Social Security," in Everett Kassalow, editor; Tho Role of Social Securipy.ip Economic Development, U. S. Dept. HEW, Social Security Administration, U. S. Government Printing Office, Washington, 1968. Smiley, Donald. Canada.ip_guestion: Federali lo the Seventies. McGraWHHill, Ryerson, Toronto, 1976: Stokes, Donald. "Parties and the Nationalization of Electoral Forces," in W. Chambers and W. Burnham, editors, The Amorican Party Systems: Stages of Politiogl Development, Oxford University Press, New York, 1967. Strouse, James and Philippe Jones. "Federal Aid: The Forgotten variable in State Policy Research," Journal‘oijolitics, Vol. 36 (February, 1974). Taylor, Charles and Michael Hudson. World Handbook of_Political and Social Indicators. Second Edition. New Haven, Yale University Press, 1973. Teune, Henry. "Macro Theoretical Approaches to Public Policy Analysis: The Fiscal Crises of.American Cities," The Annals (November, 1977). Textor, Robert and Arthur Banks. A_Cro§§ Policy Surv y. Cambridge, Mass., 1963. Thomas, Clive. ‘Qopendence and Transformation. Monthly Review Press, New York, 1974. Tinbergen, Jan. "Tendencies and Determinants of Income Distribution in Western Countries," in Bhagwati and Echaus, editors, Develop- ment pg; Flaming, London, Allen and Unwin, Ltd. , 1972. Tingsten, Herbert. The Problem of Democracy. The Bedminster Press, Totowa, N. J., 1965. 324 Titmus, Richard. Commitment _to Welfare. Pantheon Books, New York, 1968. Truman, David. "Federalism and the Party System," in Arthur Macmahon, editor, Federalism Mature and Emergent, Doubleday, New York, 1955. Uphoff, Norman T. and Milton J. Esman. Local Organization for Rural DevelOpment: Analysis of Asian Eigerience. Rural DevelOpment Committee, Center for International Studies, Cornell University, Ithaca, New York, 1974. United Nations, Report op the World Situation, U. N. Publication, no. 61, IV, Ll” 19610 United Nations, Statistical Ykearbook, 1973. New York, 1974. United Nations, UNESCO. Statistical Yearbook, 12:2. Louvain, Belgium, 1974. U. S. Census Bureau. Governmental Finances _io 192-1973. Government Printing Office. U. S. Census Bureau. Historical Statistics of tho U. _S_.: Colonial Times _to 1957. Government Printing Office, 1966. Vieirra, Paulo Reis. Toward _a Theo_r_'y of Decentralization: A Compara- tive View op Forty-Five Countries. Unpublished dissertation, University of California, 1967. Wagner, Richard E. The Fiscal Organization of American Federalism. Farkham Publishers, Chicago, 1971. Walker, Jack. "The Diffusion of Innovations Among the American States," in Hofferbert and Sharkansky, editors, Statg and Urban Politics, Little, Brown & Co., 1971. Watts, Ronald. Multiculturil Societies and Federalism. Studies of the Royal Commission on Bilingualism and Biculturalism, no. 8, Ottawa: Queen's Printer, 1970. Well, Gordon. The Benelpx Nations. Holt, Rinehart and Winston, Inc. , New York, 1970. Weinberg, Carl. Education and Social Problems. The Free Press, New York, 1971 . Weisband, Burton. External Benefitg of Public Education. Princeton University Press, 1964. Wilensky, Harold and C. N. Lebeaux. Industrial Society and Social Welfare. Glencoe, 111.: Free Press, 1965. Wilensky, Harold. "The New Corporatism, Centralization and the Welfare State," Sage Publications, Beverly Hills, 1976. 325 WilenSky, Harold. The Welfare State god Equality. University of California Press, Berkeley, 1975. Wiles, P. G. D. and S. Markowski. "Income Distribution Under Communism and Capitalism," Sovioi,§§udies, Vol. 22 (January, 1971). Williamson, J. G. "Regional Inequality and the Process of National DevelOpment: A Description of the Patterns," in L. Needleman, editor, Regional Analysis, Penguin Books, 1968. World Factbook,‘The. Publishing Sciences GrOUp, Inc., Acton, Mass, 1974. World Survey of Education, Volume V. Eoucational Policy, Legislation and Administrapion (Paris, UNESCO, 1971). Yearbook 2f. Swiss Statistics. Berne (Federal Bureau of Statistics). Yearly. Young, Crawford. The Politics of Qulturai Pluralism. University of Wisconsin Press, Madison, 1976.