© 2002 by The University of Chicago. All rights reserved.
0013-0079/2003/5101-0010$10.00

Building Political Consensus and Distributing Resources: A Trade-Off or a Compatible Choice?*

Yi Feng
Claremont Graduate University

Theodora-Ismene Gizelis
Academy of Athens and Chapman University

I.  Introduction

     In recent years, the rationale of most welfare programs has been the subject of intense political debate, not only in the United States but also in European countries (e.g., Germany, Sweden, and France) with well-established welfare systems. New global economic challenges and grim demographic trends present political leaders and workers with the difficult task of balancing government budgets while maintaining elaborate welfare programs. The prevailing opinion is that welfare has a negative impact on economic growth because of the disincentives it creates for the working population. Nevertheless, social scientists have often overlooked the political ramifications of welfare spending and its impacts on a country's income distribution. In this article, we argue that the level of welfare spending is fundamentally a political decision and, therefore, conditional on the sociopolitical and demographic structures prevailing within a country.

     Although extensive research shows effects of the size of income distribution on growth and vice versa, studies that examine how income distribution and welfare spending affect each other at different levels of economic development are almost nonexistent. Moreover, the existing literature does not seem to test adequately the hypothesis that demographic change affects the distribution of resources within each society.1

     Our point of departure is the contention that welfare transfers represent political choices. Specifically, they are the outcomes of a government's strategic choices to seek support by rewarding the interest groups critical for the government's survival. In this article we argue that governments choose to reallocate resources to increase their probability of remaining in office. Moreover, we explore the interrelationship between income inequality and welfare transfers by focusing on how governments choose to combine growth-enhancing policies with political strategies to maintain political stability. We conceptualize the emergence of welfare institutions as a necessary response to the underlying social and economic structures that delineate governmental choices.

     The article is divided into five sections. Section II reviews competing theories on the welfare state and income distribution by exploring the dynamics between income distribution and welfare transfers and incorporating the effects of demographic changes. Section III presents the article's theoretical argument. Section IV defines and describes the applied variables and the expected relations. Discussion and analysis of the results follow in Section VI. Finally, Section V suggests issues for future research.

II.  Conceptualizing Welfare State and Income Inequality

     The core of the Western welfare state has been political and social security and equality.2 In the modern welfare state, social security has been associated with the expansion of social rights and comprehensive social citizenship.3 Both elements of the political system are associated with social inclusion. The social welfare state distributes either direct funds or services in kind to secure the working class from contingencies such as involuntary unemployment, sickness, injuries, maternal or family leaves of absence, and old-age retirement. Despite the popular view that welfare payments mainly assist the poor, these payments are only a small portion of government transfers. Thus, in many cases, the term "social security spending" is preferred to "welfare state" so as to delineate the role of social security to maintain minimum income level.4

     One of the primary objectives of welfare programs is redistribution with a view to promoting equality and social justice. It is presumed that the state can intervene to redistribute resources to reduce inequality. The state is thus the only institutional mechanism in modern society capable of providing a form of collective intervention for demanding entitlements; it has the authority and the power to ensure that the redistribution is effected.

     The two other objectives of the welfare state—efficiency and consensus building—are separable from redistribution. Efficiency becomes pertinent because of market imperfections (e.g., adverse selection and externalities). Social unity or consensus building, through which the welfare state becomes a contributing factor to social integration, is one of the initial objectives of the welfare state.5

     There are four possible domains of redistributive policies. First, the state can choose to promote greater equality or follow a minimum standard to prevent segments of the population from falling below socially acceptable levels of living (e.g., housing, health care, and level of education). Another domain is the range of the beneficiaries and segments of society. The third domain differentiates the means to achieve redistribution (providing services such as medical care at schools) from the impact of policies on specific variables (e.g., health and food sufficiency). In the fourth domain, redistributive policies are divided between those that emphasize access and equality of opportunity (e.g., education) and those that focus on the amount of service on demand.6

     Many have regarded the emergence of the welfare state as the result of increasing industrialization and democratization and the consequent altered socioeconomic structures. It reflected the transition from autocracy to democracy, in which the protection of social rights was parallel to the provision of political and civil rights. Because of an increasing demand for socioeconomic security, the role of the state has been modified to secure a constant flow of social services rather than providing emergency relief. Governments employ three channels to pursue the desired goals of the welfare state: transfer payments, direct provision of services in kind, and tax credits for the working population.7 There are four primary societal developments accounting for the necessary conditions that give rise to the welfare state: demographic changes, the decline of rural populations (urbanization) because of industrialization, bureaucratization, and expansion of political and social rights.8

     The modernization approach is an extension of the industrialization thesis, combining economic conditions with political factors. The growth of the welfare state is associated with the evolutionary change of society to complex and developed forms. Modernization generates specialization and fragmentation in society. In highly differentiated societies, citizenship is the integrative mechanism of the society.9 The modernization approach considers the welfare state to be the outgrowth of two processes: the development of the industrial society and the emergence of mass democracy with universal political and social citizenship. Because of the complexity of the applied models in the modernization literature, empirical research, with few exceptions, is limited.10

     The next section develops a positive political theory of welfare transfers. Such a theory argues that welfare transfers are political choices that aim at increasing the probability of the incumbent remaining in power. We conceptualize this argument by examining the relationships between welfare transfers and different types of government change.

III.  The Welfare State, Income Inequality, and Political Stability

     A government engages in redistribution for two possible reasons. Either it seeks to improve its odds of remaining in power or it desires to facilitate the participation of various segments of the population in the larger economic and social life. For instance, unemployment benefits allow unemployed people to remain active economically, if on a limited basis. We believe that the second goal is rather unimportant to a government unless political benefits arise from such policy choices. We argue that welfare transfers may actually increase income inequality, as the governing elite uses redistribution as a political tool to remain in office.11

     Political violence is the intervening variable that explains why governments even care to redistribute resources. Though income inequality tends to reduce economic growth, the political interest of governments rests with the maintenance of their political power.12 Hence, resource redistribution can be essential in satisfying groups whose support is critical to the political survival of a government.13 State actors have their own interests in promoting welfare programs. The political elite seeks reelection, and welfare programs provide them with popular support.

     The recent work of Bruce Bueno de Mesquita, James D. Morrow, Randolph M. Siverson, and Alastair Smith shows that a government can redistribute two types of goods: public goods that are enjoyed by all members of the society and private goods that are distributed exclusively to the members of the pivotal group sustaining the government in power.14 Different types of political regimes have different structures in terms of the size and composition of a winning coalition necessary to sustain a government. For instance, authoritarian regimes require a small winning coalition to remain in power as compared with the coalition of democratic regimes. Hence, the former primarily favor the redistribution of private goods, while the latter favor a mixture of both private and public goods as essential to maintaining political power. The present article is an important extension of this finding. Our focus is that while the consensus-building function of the welfare state becomes extremely pertinent in the minds of the political elite, welfare programs do not necessarily target the most needy segments of the population but, rather, the ones critical for the regime's political survival. In this connection, we examine the joint relationships among welfare transfers, income distribution, and some very specific forms of government change. Furthermore, we argue that demographic changes leading to an increase in the dependent population may alter who absorbs resources within the society and, therefore, that the demographic profile of those supporting the regime modifies the choices of resource redistribution.

     In this article, we argue that governments redistribute resources with the hope that the probability of their political survival will increase. As a result, income inequality is unlikely to decrease; on the contrary, we believe that it may increase. There are two major reasons. First, the horizon of the political strategy to remain in office is short, while the reduction of income inequality takes a fairly long period of time. Second, the rationale of redistribution lies primarily in political rewards to the interest groups whose support is essential to a government holding on to its office. The beneficiaries tend to be those who have already been awarded by government redistribution and further transfers to them are needed for the government to remain in office, which increases wealth inequality. In the long run, growing unequal wealth distribution fertilizes the ground for the kind of profound political change that leads to the downfall of governments.

     For the purposes of this article, we focus on three types of government change to denote political stability or instability: minor regular government change, major regular government change, and irregular government change. Major regular government change occurs when a party of a different ideology gains office within the constitutional framework. The change in Britain's political scenery following Tony Blair's Labour Party winning the election is an example of a major regular government change. Irregular government change refers to unconstitutional government change. It indicates a fundamental change of the political system. The power transfer from President Salvador Allende to General Augusto Pinochet, or from the Shah Pahlavi to Ayatollah Khomeini, are classic examples of this kind of government change. Minor regular government change implies that the same ruling party remains in office without significant ramifications for the chosen policies. The People's Action Party in Singapore, the United Malays National Organization in Malaysia, and the Liberal Democracy Party in Japan provide relevant examples of minor regular government change. In light of our central theoretical argument, welfare transfers should improve the odds of the same party remaining in power, reduce the likelihood of fundamental political change, and decrease the probability of incumbents losing office in an electoral environment. These theoretical statements can be summarized in the following four hypotheses, which we test statistically in the following section:

     HYPOTHESIS 1.     Welfare transfers increase the levels of income inequality.

     HYPOTHESIS 2.     Welfare transfers decrease the probability of irregular government change.

     HYPOTHESIS 3.     Welfare transfers increase the probability of minor regular government change.

     HYPOTHESIS 4.     Welfare transfers decrease the probability of major regular government change.

IV.  Research Design and Data Specification

     To test these hypotheses, we define the following specification:















and



where P1 = probability of irregular government change, P2 = probability of major regular government change, P3 = probability of minor regular government change, G = welfare spending, D = income distribution, X = predetermined variables (which include older population, urbanization, government capacity, economic growth, initial levels of development, political rights, education attainment, and initial levels of political stability), and ε, μ, and vi = error terms.

     For the joint estimation of the five equations, including both developed and developing countries, a sample of 72 countries was selected.15 The estimation is through the three-stage least-squares method (3SLS), which generalizes the two-stage least-squares method (2SLS) to take into account the correlations between equations. It requires three steps: first, calculate the 2SLS estimates of the identified equations; second, use the 2SLS estimates to estimate the structural equation errors and then use these to estimate the contemporaneous variance-covariance matrix of the structural equation errors; and third, apply generalized least squares to the large equation representing all the identified equations of the system. The 3SLS estimator is consistent and, in general, is asymptotically more efficient than the 2SLS estimator. If the disturbances in the different equations are uncorrelated, so that the contemporaneous variance-covariance matrix of the the structural equations' disturbances is diagonal, 3SLS reduces to 2SLS.16 The range of time for the endogenous variables in our article is from 1970 to 1990, a period determined by the availability of the data.

Endogenous Variables

     Income distribution.     Various indices of income distribution have been used in the literature. They include the Gini coefficient, the top- or bottom-income quintile share, the ratio of the top- to bottom-income quintile shares, or the summation of the middle-income quintile shares. These indices are quite highly correlated. The result of using any one of them can usually be replicated consistently by using the others. This study uses the Gini coefficient because it captures the general level of income inequality. Klaus Deininger and Lyn Squire compiled the original data set for this variable.17 They address the problems that most of the previous data sets exhibit and expand the acceptable observations that are taken from primary surveys and official statistical documents. To compensate for the lack of observations within countries, they apply information on land holdings as a proxy for initial income distribution, thus increasing the sample of countries that can be studied.

     Welfare transfers.     Various indicators have been adopted to measure welfare programs. Some studies use government expenditures, exclusive of military and educational expenditures. Other studies focus on health care, social insurance transfers, and pensions. Still others look at education expenditures only.18 It should be noted that government expenditures include expenditures that do not belong to welfare transfers (i.e., health expenditures, housing, agriculture, transportation, and public investment). Although these government expenditures consist of the distribution of public goods, they do not target income contingencies.

     To capture the characteristics of welfare programs, as they are analyzed in this study, social security and welfare transfers stand as the most appropriate indicators. Social security and welfare transfers are short-term reallocations of resources that can be used conveniently for political purposes by helping incumbents remain in office. To measure the magnitude of the welfare state, this study focuses on the level of expenditures on welfare transfers and social security.19

     The data are from the Government Financial Statistics (GFS) by the International Monetary Fund (IMF). The IMF defines social security as "transfers in cash or in kind to protect the entire population or specific segments of it against certain social risk," which is "an event or circumstance that may adversely affect the welfare of the households concerned either by imposing additional demands on their resources or by reducing their incomes."20 Social security expenditures consist of pensions and retirement benefits. Welfare transfers are "transfers payable to household to meet the ... needs as social insurance benefits but which are not made under a social insurance scheme."21 Here, we use the ratio of welfare transfers and social security to GDP.22

     Political instability.     The three kinds of government change are all discrete phenomena. To account for the relationship between the all-or-nothing nature of the transfer of executive power and the relativity of political instability, we use the limited dependent variable estimation method. This method defines the probability of government change as a continuous variable, characterized as a function of economic and political conditions.

     The explanatory variables in the logit model fall into four broad classes. Economic variables are included to measure the recent economic performance of the government, political behavior variables account for significant political events that may lead to a government change, political institution variables reflect political structures and systems, and continent dummy variables control for regional factors. From the fitted values of the logit model using the pooled time-series and cross-country data, the probability of a government change for each country is estimated. There are three aggregate probabilities: irregular government change, major regular government change, and minor regular government change. The probability variables thus created have been used in recent publications to study economic performance and demographic change.23

Predetermined Variables

     The predetermined variables in our model include demographic, economic, and political variables. The demographic variables are the ratio of the population over age 65 to the total population and the ratio of urban population to the total population. The range of the data is from 1970 to 1990. All demographic data are gathered from the World Population Perspectives and The World Data of Social and Economic Information.24

     The economic variables are those that the literature indicates as determinants of income distribution. They are the real GDP per capita for 1970 (to avoid problems of endogeneity), the square of real GDP per capita, and the rate of economic growth. These variables are obtained or calculated from the data set compiled by Robert Summers and Alan Heston in 1995.25 The range of the data is from 1970 to 1990. We also include in the specification secondary school enrollment, which is an indicator for human capital accumulation. For developing countries, in particular, high levels of secondary school enrollment are critical for building the necessary human capital for economic growth.26 Enrollment data are applied as a proxy for the flow of investment in human capital. The data used in this study are from the R. Barro and J.-W. Lee data set, which is an adaptation of the Unesco statistics on educational attainment.27

     The political variables include the lack of political rights, the level of political extraction, and the initial levels of the probabilities of government change. The political rights variable is from the Freedom House, and it is ranked from one (highest degree of liberty) to seven (the lowest).28 The criteria for political rights include the meaningfulness of elections for the executive and legislative branches, the voting power of the electorate, significant opposition competition, decentralization of political power, the attempts of political agents to reach a consensus on major political issues, and minority self-determination.29 All these indicators measure the power that the electorate and the various interest groups have to express their political opinions and demands. The data range from 1970 to 1990.

     Political capacity is a composite measurement of the government's capacity to reach the population and execute its policy. It conceptualizes how capable the government is of extracting resources from the population such that the chosen policy goal of the government can be realized. A capable government is also able to marginalize or neutralize domestic opposition through its taxation policy.30 Here we adopt the measure used by Jacek Kugler and Marina Arbetman, who operationalize political capacity based on the difference between the economically expected performance of the government in collecting taxes and the actual tax collection resulting from political power.31 The ratio is estimated from 1960 to 1992. Finally, initial levels of political instability are also included to control for political violence, which existed at the beginning of the period of the study.

V.  Empirical Results and Analysis

The Transfer-Inequality–Irregular-Change Nexus

     The result in table 1 indicates that welfare transfers can be useful political and policy tools for a government working to prevent cataclysmic regime change, even at the cost of income inequality. In plain words, welfare transfers decrease the probability of unconstitutional government change. Therefore, a government can use welfare as a means of prolonging its tenure.

TABLE 1     WELFARE, INCOME INEQUALITY, AND GOVERNMENT CHANGE

     The positive effect of welfare transfers on income inequality might be the result of governments responding to specific groups that have more political clout than others. This result is consistent with our earlier positive, rather than normative, argument that the rationale of welfare transfers is political and not ideological. Transfers are handed out not because a society is unequal but because the government seeks to maximize social support so that it can remain in office. The result is also consistent with the selectorate theory by Bueno de Mesquita et al., that governing leaders transfer wealth to their winning coalition members in order to remain in power.32 This result may actually increase, rather than decrease, wealth disparities.33

     It should be noted, however, that the effect of income inequality on irregular government change is positive and significant at the 0.10 level. While the direct effect of transfers on irregular government change is negative—more transfers lead to a lower probability of being overthrown—its indirect effect on government change through income inequality is positive, which is to say, more transfers lead to higher inequality, resulting in a higher degree of political instability. The positive indirect effect of welfare transfers on irregular government change is, however, only marginally significant. The lack of major impact of income distribution on irregular government change in this nexus is consistent with the argument that the causality between income inequality and political instability is only tentative and dependent on other factors. Such factors may include demographic structures, political mobilization, cultural values, the balance between civil society and state power, and the stability of political institutions.34 Therefore, governments may actually use welfare transfers to prolong their tenure at the cost of income inequality. Although income inequality may also increase the probability that the government is overthrown, such probability is not salient compared with the benefits brought about by welfare transfers in terms of consolidating support for the government in the short run.

     In the literature on political instability, income inequality is considered one of the possible explanations for civil unrest. Some evidence shows that the poorest countries with the highest levels of inequality tend to be politically unstable as well. The reason is that poor countries are not able to generate enough wealth to satisfy the needs of all the society's members.

     Nevertheless, the causal link between income inequality and political instability is tentative. By itself, income inequality cannot fully explain incidents of political violence or the lack thereof. Arguably, income inequality is the necessary condition, but hardly the sufficient one, to cause political instability. This is critical to understanding governmental choices in redistributing resources. Since income inequality will not always lead to political instability, particularly in the short run, governments choose to provide private goods to the groups essential for their political survival. Government policy that leads to the relocation of resources to the groups that support them the most can result in an elevation of levels of income inequality. In the long run, this choice can be self-defeating if redistribution accentuates income inequality beyond what is socially acceptable.

The Transfer-Inequality–Minor-Regular-Change Nexus

     Similarly, the complexity of using welfare to prolong a government's tenure can also be seen through its effect on minor regular government change, namely, the same government remaining in office. First, transfers increase the probability of the same party retaining office, while such a probability decreases income inequality. Therefore, despite the direct positive effect of welfare transfers on income inequality, transfers also have an indirect effect on income inequality, which turns out to be negative. This indirect effect is through the channel of minor regular government change. On the one hand, welfare transfers consolidate the power base of the ruling party and increase the probability of the same party remaining in power. On the other hand, the scenario of the same party remaining in power decreases income inequality. When the ruling party is assured of its chance of reelection into office, it will endeavor to decrease income inequality, as it realizes that the increase of income inequality resulting from welfare transfers may eventually backfire on the ultimate goal of remaining in office. Examples of this scenario can be found in some Pacific Asian countries that have had the same party in office for a long time and have during that time practiced a strategy of "shared growth."35 Such countries include Singapore, Malaysia, and Taiwan. In Malaysia, for instance, welfare transfers are used along with differential health and educational expenditures, with the long-term goal of maintaining a stable political regime along with lower levels of income inequality.36

     In addition, the increase in the probability of the same party remaining in office directly increases the probability of unconstitutional change. The same party remaining in office could result from the lack of serious political competition (e.g., in Botswana) or from the lack of political mechanisms for replacing a governing party peacefully (e.g., in Zaire). If it were the latter, then the new governing party is forced to emerge in unconstitutional manners. Welfare transfers in this case may slow, but cannot stop, the ultimate violent government change.

The Transfer-Inequality–Major-Regular-Change Nexus

     The sign of the coefficient for welfare transfers on major regular government change is negative (as expected), but statistically insignificant. As a result, the lack of cross-effects between major regular government change and welfare transfers is interesting because it indicates that the choice of changing a political party in the constitutional framework is not affected by welfare transfers alone. Structural economic variables (such as the overall level of the economy, unemployment, and inflation) and political variables (such as the structure of institutions and electoral rules), as well as shifts in values may be more important than welfare transfers in altering the political party that rules.

     Major regular government change is a defining characteristic of democracies. "Democracy is a regime in which some governmental offices are filled by contested elections. ... It is a system in which incumbents lose elections and leave office when the rules so dictate."37 The lack of impact of welfare transfers or income inequality on major regular government change indicates that under democracy a government is effectively constrained by the legislature, opposition parties, and the constitution. These political mechanisms significantly reduce the tools that a governing party can use to attain its goal of remaining in office. Also, by definition, democracies have a large winning coalition.38 Whoever is in office has to satisfy a wide range of society, which limits the governing party's power to concentrate transfers to any single social group. Finally, income distribution is relatively equal under democracy compared with autocracy, which makes redistribution less salient in the policy of the former compared with that of the latter.39 All told, autocratic or semidemocratic governments have much better leverage than their democratic counterparts in using transfers to retain office.

Control Variables

     Distributing resources to older segments of the population can have very complex outcomes. An aging demographic structure tends to reduce income inequality (as the dependent population over age 65 coefficient shows in table 1). However, the support of an older dependent population increases income inequality as it absorbs large quantities of resources, pensions being the most characteristic example. These resources could crowd out investment essential for sustained economic growth and also education that not only enhances economic growth but also reduces long-term inequalities by increasing the opportunities of different groups to participate in the economic realm. This argument should be emphasized, as we also find a negative impact of education on income inequality in table 1. In addition to the direct and indirect relationships between the presence of a large older dependent population and income inequality, it should be noted also that such a population has strong implications for political change. Its existence is positively related to irregular government change and negatively related to the same party staying in power.40

     Among other control variables, urbanization is found to have a positive effect on welfare transfers, which reveals that the welfare system serves the urban areas more than the countryside. The economic side of this phenomenon may be that urban dwellers are more sensitive to changes in their finance (e.g., unemployment) than their rural counterparts. The political side of the argument is that urban dwellers tend to be better educated and better organized than their rural counterparts, thus making the former politically more valuable to the government than the latter. This article presents evidence that rural residents tend to fall outside welfare transfers in poor countries in particular and provides a political rationale that accounts for the disparities between the rural and urban areas.

     China provides an interesting case in this light. D. Gale Johnson finds that wealth disparities between rural and urban areas have not narrowed, but perhaps widened, in China over the past 2 decades despite the rapid economic growth that the nation has achieved.41 In addition to all the sources of disparities he has identified—for example, restrictions on rural-to-urban migration, the poor quality and accessibility of rural education, and an allocation of investment and credit biased against the countryside—our findings here may help understand the political rationale of wealth disparities in China.

     Government capacity has a negative effect on irregular government change. This result confirms earlier findings that political violence is curbed under a capable government, although government capacity is found to have no effect on regular government change. Finally, the statistical results in table 1 show a strong pattern of the path dependence of political behavior. The three kinds of government change have a salient historical pattern. The initial levels of government change are all statistically significant. For instance, a coup-prone polity in 1970 tends to witness new coups. Similarly, a nation noted for its democratic procedures in changing its government in 1970 has largely maintained this tradition throughout the years. Those findings strongly support the coup-trap argument.42 They also provide some robust empirical evidence for the argument that political institutions tend to be permanent or semipermanent.43

     To conclude, the empirical results show that when welfare transfers, income inequality, and political instability are examined together in the context of demographic structures, governments choose to apply welfare transfers to reduce political instability and increase their survivability. Specifically, welfare transfers reduce political instability in two ways. First, by increasing the probability of a minor regular government change, welfare transfers enhance political changes that do not lead to major policy shifts. Second, through reducing unconstitutional political changes, welfare transfers decrease the likelihood of the events that remove a government from office by violence.

     Meanwhile, the redistribution of resources leads to an increase in income inequality as politically salient groups absorb a large part of the redistributed resources. We argue that welfare programs embody the incumbent's incentive to stay in office. They may also reflect the sources of fundamental structural changes that alter the nature of the political and the economic system.

VI.  Concluding Remarks

     This article explores the interactive relationships among welfare transfers, income distribution, and government change, controlling for the effects of demographic and economic variables. We argue that governments are involved in the redistribution of resources to enhance political stability and survival rather than enhancing economic performance or wealth equality. Hence, they tend to respond to groups that are essential for their political survival. Although the presented empirical results are not conclusive, they tend to support our first hypothesis—namely, that welfare transfers increase income inequality. They also provide important insights into the relationships between welfare transfers and three kinds of government change by testing the three other hypotheses. Welfare transfers reduce the likelihood of irregular government change, increase the chance of the same party remaining in office, and have no effect on the major regular government change characteristic of democracies. While the first two kinds of government change are typical of autocratic or semidemocratic societies, the third is characteristic of democratic nations. The implication is that the governments in the first two societies are more capable of using welfare transfers to prolong their tenure than their democratic counterparts. This finding is important in the sense that it provides a major reason why many autocratic governments are long-lived (e.g., in Zaire, Indonesia, and the Soviet Union) compared with democratic governments.

     In general, governments utilize welfare transfers to alter their political future. This goal can be successful in the long run only if it does not increase income inequality that can threaten the stability of the political regime. Hence, the policy choices that a government makes to maintain political power through redistribution of resources can be undermined by demographic structures and by the long-term effects of such redistribution of resources.

     The current study can be improved in various ways. First, additional case studies are required to improve the reliability of the statistical results. Second, additional demographic indicators should be included, such as fertility rates for the developing countries, so that we can further comprehend the impact of demographic variables on income distribution and welfare transfers. Third, other measures of welfare (e.g., health expenditures) can be included to enhance our understanding of the impact of state intervention on income distribution and the demographic structures within a country. Finally, while our article indicates strong and consistent evidence that welfare transfers increase political stability for political reasons, both welfare transfers and political stability could be the outcome of economic development.44 Though we have used control variables that are highly correlated with economic development, such as an aged population and urbanization, future work may retest the alternative hypothesis using additional economic control variables.

Appendix

Countries in the Sample

     Argentina
Australia
Austria
Bangladesh
Belgium
Bolivia
Botswana
Brazil
Burkina-Faso
Burundi
Cameroon
Canada
Chad
Chile
Colombia
Congo
Cyprus
Denmark
Dominican Republic
Ecuador
Ethiopia
Finland
France
Ghana
Greece
Guatemala
Honduras
Hungary
Iceland
India
Indonesia
Iran
Ireland
Israel
Italy
Jamaica
Japan
Jordan
Kenya
Korea Republic
Lesotho
Liberia
Madagascar
Malawi
Malaysia
Mali
Mauritania
Mauritius
Mexico
Morocco
Myanmar
Nepal
Netherlands
New Zealand
Nicaragua
Niger
Nigeria
Norway
Pakistan
Panama
Paraguay
Peru
Philippines
Portugal
Rwanda
Senegal
Sierra Leone
Singapore
Spain
Sri Lanka
Sudan
Sweden
Switzerland
Tanzania
Thailand
Togo
Tunisia
Turkey
Uganda
United Kingdom
Uruguay
United States
Venezuela
Zaire
Zambia
Zimbabwe

Notes

     * This research was supported in part by a grant from the National Science Foundation (SBR-9730474). We thank D. Gale Johnson and two anonymous reviewers for their insightful comments and suggestions. We are responsible for any remaining errors.
     1. It has been pointed out that, although the expansion of the global markets and technical advances undermine welfare programs, the changing demographic structures are the most critical factor in determining the levels of welfare transfers. Paul Pierson, "Post-industrial Pressures on the Mature Welfare States," pp. 80–106; Torben Iversen, "The Dynamics of Welfare State Expansion: Trade Openness, De-Industrialization, and Partisan Politics," pp. 45–79, both in The New Politics of the State, ed. Paul Pierson (New York: Oxford University Press, 2000).
     2. For the context of the literature review and further citations, see Theodora-Ismene Gizelis, "Welfare Spending, Social Cohesion, and Relative Political Capacity: Exploring the Future of the Welfare State" (Ph.D. diss., Claremont Graduate University, 1999).
     3. The term "welfare state" was coined in Britain in 1941, while Great Britain alone was standing against Germany. The predecessor of the British welfare state was established in 1911 as the Liberal government's National Insurance Act, promoted by experts such as William Beveridge and prepared by Winston Churchill and Lloyd George. This act was targeting the "left out millions who were miserable" in the heyday of the British empire. In both cases of policy making, British administrators replicated the social policies established in imperial Germany. See Peter Flora and Arnold Heidenheimer, eds., The Development of Welfare States in Europe and America (New Brunswick, N.J.: Transaction, 1981), pp. 18–19. Most of the historians of the "welfare state" accredit imperial Germany as the forerunner of modern welfare programs. In 1881, Otto von Bismarck instituted the basis of the first social security legislation in an effort to shift the loyalties of the working class. Bismarck's system was a "carrot and stick" policy, targeting the social democratic movement. It dealt with prevailing social problems while simultaneously enacting disciplinary measures. The state had the monopoly on providing social insurance, which enhanced the paternalistic nature of the German state. Insurance reform was not an actual expansion of the state's functions but a qualitative transformation. See Hans-Peter Ullmann, "German Industry and Bismarck's Social Security System," in The Emergence of the Welfare State in Britain and Germany, ed. W. J. Mommsen and Wolfgang Mock (London: Croom Helm, 1981), pp. 107–33.
     4. See Fred C. Pampel and John B. Williamson, "Welfare Spending in Advanced Industrial Democracies, 1950–1980," American Journal of Sociology 93 (1988): 1424–56. On the conceptualization of the welfare state, see Flora and Heidenheimer, eds. For various policy applications of social citizenship, a concept initiated by Alfred Marshall in 1949, see Jytte Klausen, "Social Rights Advocacy and State Building: T. H. Marshall in the Hands of Social Reformers," World Politics 47 (1995): 244–67.
     5. There are concerns regarding the authority of the state to provide welfare programs. First of all, from a critical point of view, state distribution follows the predetermined distribution of resources. One view is that welfare programs create a condition of moral hazard since they provide incentives to people to adopt undesirable behavior. The counterargument remains that the state is the only institution that has the power to ensure that the system is not abused. Most of the arguments favoring state interference are based on moral and ethical considerations. The best-known argument is by John Rawls, A Theory of Justice (Cambridge, Mass.: Belknap, 1971). Thus, the trade-off is between a society's need to expand its production and the redistribution of the production goods in a socially appropriate way. See Peter Taylor-Gooby, "Welfare Futures," in Social Policy in Transition: Anglo-German Perspectives in the New European Community, ed. John Ferris and Robert Page (Aldershot: Ashgate, 1994), pp. 70–87, esp. 82–85.
     6. The distinction between equal access and the amount of a service can be summarized as follows: the former intends that everybody starts from the same point, and the latter aims at everybody ending at the same point. See Robert Goodin and Julian Le Grand, Not Only the Poor: The Middle Classes and the Welfare State (London: Allen & Unwin, 1987), pp. 5–8. This distinction is fundamental, because it denotes a different philosophy regarding inequality and types of policies that government should use to redistribute income. In the income distribution literature, one basic argument focuses on absolute differences between the well-off and the poor. The other position is that proportional differences should be relevant. See Goodin and Le Grand, p. 9.
     7. Flora and Heidenheimer, eds., pp. 22–25.
     8. See Christopher Pierson, Beyond the Welfare State? The New Political Economy of Welfare (Philadelphia: University of Pennsylvania Press, 1991), pp. 13–14. Some of the first specialists in the field who viewed the welfare state as a response to new social needs can be found in Harold L. Wilensky, The Welfare State and Equality: Structural and Ideological Roots of Public Expenditure (Berkeley: University of California Press, 1975); and Goran Therborn, "Welfare State and Capitalist Markets," Acta Sociologica 30, no. 3–4 (1987): 237–54. Manfred Schmidt tests the relationship between social policy development and industrialization empirically in a sample of 39 developed and underdeveloped countries. His small sample size, however, distorts the variation that the industrialization approach could explain across countries. See Manfred Schmidt, "Social Policy in Rich and Poor Countries: Socio-Economic Trend and Political-Institutional Determinants," European Journal of Political Research 17, no. 6 (1989): 641–59.
     9. Peter Flora and Jens Alber, "Modernization, Democratization, and the Development of Welfare States in Western Europe," in Flora and Heidenheimer, eds., pp. 39–40.
     10. See, e.g., ibid.
     11. In the study of government size, public choice models provide useful theoretical and empirical tools. Some of the public choice models target the role of the government in redistribution, while others focus on competition among interest groups and the role of bureaucracy and state actors. A plethora of public choice theorists have analyzed the redistributive functions of the government. The most interesting theoretical work comes from Allan H. Meltzer and Scott H. Richard, "Why Government Grows (and Grows) in a Democracy," Public Interests 52 (1978): 111–18, and "A Rational Theory of the Size of the Government," Journal of Political Economy 89 (1981): 403–18; Samuel Peltzman, "The Growth of Government," Journal of Law and Economics 23 (1980): 209–88; Peter H. Aranson and Peter C. Ordeshook, "Regulation, Redistribution, and Public Choice," Public Choice 37, no. 1 (1981): 69–100. An alternative approach points at interest groups as instrumental determinants to explain the size of the government. The dominant figure in this area is Gordon Tullock, "Some Problems of Majority Voting," Journal of Political Economy 67 (1959): 571–79, with his concept of logrolling under majority rule. Gary Becker develops an insightful analysis in linking government functions and interest groups. See G. S. Becker, "A Theory of Competition among Pressure Groups for Political Influence," Quarterly Journal of Economics 98 (1983): 371–400. One of the recent empirical works on the impact of interest groups and political fractionalization on government size is by Dennis C. Mueller and P. Murrell, "Interest Groups and the Size of Government," Public Choice 48, no. 2 (1986): 125–45. Other schools of thought emphasize the political or the societal forces in the distribution of welfare transfers. See, e.g., Walter Korpi, "Power, Politics, and State Autonomy in the Development of Social Citizenship: Social Rights during Sickness in Eighteen OECD Countries since 1930," American Sociological Review 54, no. 3 (1989): 309–28; Alexander Hicks and Joya Misra, "Political Resources and the Growth of Welfare in Affluent Capitalist Democracies, 1960–1982," American Journal of Sociology 99 (1993): 668–710; Pampel and Williamson.
     12. See, e.g., Torsten Persson and Guido Tabellini, "Is Inequality Harmful for Growth? Theory and Evidence," American Economic Review 84, no. 3 (1994): 600–621.
     13. Following Mancur Olson's arguments on collective action, some public choice theorists argue that bureaucrats respond only to well-organized groups when promoting and protecting their budgets. For Olson's arguments, see his seminal work, The Logic of Collective Action: Public Goods and the Theory of Groups (Cambridge, Mass.: Harvard University Press, 1965), and his macroeconomic application in The Rise and Decline of Nations: Economic Growth, Stagflation, and Social Rigidities (New Haven, Conn.: Yale University Press, 1982). Some of the most prominent scholars focusing on bureaucratization are W. A. Niskanen (Bureaucracy and Representative Government [New York: Aldine-Atherton, 1971], and Bureaucracy—Servant or Master? Lessons from America [London: Institute of Economic Affairs, 1973]) and Robert Bacon and Walter A. Eltis (Britain's Economic Problem: Too Few Producers [London: Macmillan, 1978]). See also Theda Skocpol, "Bringing the State Back In: Strategies of Analysis in Current Research," in Bringing the State Back In, ed. Peter B. Evans, Dietrich Rueschemeyer, and Theda Skocpol (Cambridge: Cambridge University Press, 1985), pp. 3–37; Theda Skocpol and E. Amenta, "States and Social Policies," Annual Review of Sociology 1, no. 1 (1986): 131–57; A. S. Orloff and T. Skocpol, "Why Not Equal Protection? Explaining the Politics of Public Social Spending in Britain, 1990–1911, and the United States, 1880–1920," American Sociological Review 49, no. 6 (1984): 726–50.
     14. See Bruce Bueno de Mesquita, James D. Morrow, Randolph M. Siverson, and Alastair Smith, "Political Institutions, Political Survival, and Policy Success," in Governing for Prosperity, ed. Bruce Bueno de Mesquita and Hilton L. Root (New Haven, Conn.: Yale University Press, 2000), pp. 59–84.
     15. All the oil producers are excluded for being outliers. Data availability has been a problem for Central and Eastern European countries. The political change variables were calculated using the Penn World Tables, which do not have the data for those countries. See Robert Summers and Alan Heston, The Penn World Trade Mark 5.6 (Cambridge, Mass.: National Bureau of Economic Research, 1995). The welfare data are likewise unavailable for those countries.
     16. William H. Greene, Econometric Analysis (Englewood, N.J.: Prentice-Hall, 1990), pp. 611–12. For statistical and software operational reference, see SAS Institute, "The SYSLIN Procedure," in SAS/ETS User's Guide (Cary, N.C.: SAS, 1993), pp. 815–68.
     17. Klaus Deininger and Lyn Squire, "A New Data Set Measuring Income Inequality," World Bank Economic Review 10 (1996): 565–91.
     18. Education expenditures also function as a form of investment in human capital. Therefore, education is not a means of income redistribution but rather a long-term policy to achieve economic growth through the improvement of human capital.
     19. Health expenditures tend to have longer and indirect effects on income distribution. Preventive health expenditures may have a strong progressive effect if they target young cohorts. On the other hand, an increasingly older population will absorb more pension-related and curative health resources.
     20. International Monetary Fund, GFS Manual (Washington, D.C.: International Monetary Fund, 2001), p. 71.
     21. Ibid., p. 72.
     22. The data are provided by Summers and Heston; and International Monetary Fund, International Financial Statistics (Washington, D.C.: International Monetary Fund, 2000). The paucity of welfare transfers and social security data prior to 1970 reduces the years of estimation.
     23. Presenting full information regarding the specification and results of the statistical model that led to the construction of the probabilities of government change is beyond the scope and space of our present article. For a detailed discussion and complete specification of the estimation of the probabilities of the three kinds of government change, see Yi Feng, "Democracy, Political Stability, and Economic Growth," British Journal of Political Science 27 (1997): 391–418. The results support many priors about using these variables to estimate probabilities of government change. Take irregular government change, for example. An increase in real GDP per capita decreases the likelihood of irregular government change, whereas general strikes and government crises increase the probability of such change. The variable "antigovernment demonstrations" has a negative effect on irregular government change and is statistically significant at the 10% error level. This may reflect the fact that antigovernment demonstrations are a legal means by which citizens in a democracy, where irregular government change is generally infrequent, can express their dissatisfaction and discontent with the government. One interesting finding regarding political structures and political instability is the role played by the legislature. Irregular government change is significantly lower where the legislature has a higher degree of responsibility and effectiveness, suggesting that a strong and democratically elected legislature inhibits irregular government change. The size of the cabinet also reduces the likelihood of an irregular government change. The number of officials at the ministerial level in an executive branch may indicate a mature bureaucratic system with well-defined divisions of labor, which should reduce those irregularities that breed political instability. Finally, countries with a premier as the executive and oil-producing countries tend to have a higher level of irregular government change than otherwise. The probability variables thus created have been used in numerous publications. For instance, see Baizhu Chen and Yi Feng, "Some Political Determinants of Economic Growth," European Journal of Political Economy 12 (1996): 609–27; Yi Feng, "Democracy, Political Stability, and Economic Growth"; and Yi Feng, Jacek Kugler, and Paul J. Zak, "The Politics of Fertility and Economic Development," International Studies Quarterly 44 (2000): 667–93.
     24. United Nations, World Population Perspectives (New York: United Nations, 1982), and The World Data of Social and Economic Information (New York: United Nations, 1994).
     25. See Summers and Heston (n. 15 above).
     26. The teacher-pupil ratio, which is another measurement, captures the quality of education. A lower ratio implies lower quality in education. Finally, literacy rates are another alternative, which provide information as to the stock of human capital rather than the flow of investment.
     27. Robert Barro and Jong-Wha Lee, "International Comparisons of Educational Attainment," Working Paper no. 5658 (National Bureau of Economic Research, Cambridge, Mass., 1993). The data are in 5-year intervals from 1960 to 1985. They were averaged from 1970 to 1985 in this study.
     28. Freedom House, Freedom in the World, 2001–2002 (New York: Freedom House, 2002).
     29. See Raymond Gastil, Freedom in the World (Westport, Conn.: Greenwood, 1982).
     30. See Michelle Benson and Jacek Kugler, "Power Parity, Democracy, and the Severity of Internal Violence," Journal of Conflict Resolution 42 (1998): 196–209.
     31. Jacek Kugler and Marina Arbetman, "Relative Political Capacity: Political Extraction and Political Reach," in Political Capacity and Economic Behavior, ed. Marina Arbetman and Jacek Kugler (Boulder, Colo.: Westview, 1998), pp. 11–46.
     32. Bruce Bueno de Mesquita and Hilton L. Root, "When Bad Economics Is Good Politics," in de Mesquita and Root, eds. (n. 14 above); Bruce Bueno de Mesquita, James D. Morrow, Randolph M. Siverson, and Alastair Smith, "International Conflict and Democratic Transitions," Journal of Conflict Resolution 43 (1999): 147–61.
     33. In our estimation, we tried continent dummy variables, but they were statistically insignificant. The political change variables represent the continent variation to a large degree, as we used the continent dummy variables to estimate the probabilities of government change. The political change variables already contain the continental information.
     34. Mark Irving Lichbach, "An Evaluation of `Does Economic Inequality Breed Political Conflict?'" World Politics 41, no. 4 (1989): 431–70; Will H. Moore, Ronny Lindstrφm, and Valerie O'Regan, "Land Reform, Political Violence and the Economic Inequality Political Conflict Nexus: A Longitudinal Analysis," International Interactions 21, no. 4 (1996): 353–63..
     35. For the concept of "shared growth," see Jose Edgardo Campos and Hilton L. Root, The Key to the Asian Miracle: Making Shared Growth Credible (Washington, D.C.: Brookings Institution, 1996).
     36. For a case study of Malaysia, see Yi Feng, Ismene Gizelis, and Jae Hoon Lee, "Managing Economic Development in the Presence of Ethnic Diversity: The Malaysian Experience," in Managing Economic Development in Asia, ed. Thomas K. Liou (London: Praeger, 2002), pp. 203–38.
     37. Adam Przeworski, Michael E. Alvarez, Josι Antonio Cheibub, and Fernando Limongi, Democracy and Development: Political Institutions and Well-Being in the World, 1950–1990 (New York: Cambridge University Press, 2000), pp. 35, 54.
     38. See de Mesquita et al. (n. 14 above).
     39. Yi Feng, Democracy, Governance and Economic Performance: Theory and Evidence (Cambridge, Mass.: MIT Press, 2003).
     40. For a discussion on how demographic changes may affect the structures of the welfare state in European countries, see Gosta Epsing-Andersen, "Interview on Postindustrialism and the Future of the Welfare State," Work, Employment, and Society 14 (2000): 757–69.
     41. D. Gale Johnson, "Agricultural Adjustment in China: Problems and Prospects," Population and Development Review 26, no. 2 (2000): 319–34. We also thank a reviewer for suggesting China as an example to illustrate our findings.
     42. See John B. Londregan and Keith T. Poole, "Poverty, the Coup Trap, and the Seizure of Executive Power," World Politics 32 (1990): 151–83.
     43. See Alex Cukierman, Sebastian Edwards, and Guido Tabellini, "Seigniorage and Political Instability," American Economic Review 82 (1992): 537–55; Alberto Alesina, Sule Ozler, Nouriel Roubini, and Philip Swagel, "Political Instability and Economic Growth," Journal of Economic Growth 1 (1996): 189–212.
     44. We owe this observation to a reviewer.