median voter, government, Civic Trust, United States, voter dissatisfaction, voter preferences, Europe, decline, economic integration, welfare state, voters, voter satisfaction, OECD countries, nation-state, satisfaction, voter turnout, industrial democracies, political actors, public confidence, political systems, voter, polarization, European governments, advanced industrial, the government, system of government, integration, American voters, OECD
SEVEN The Economics of Civic Trust Alberto Alesina and Rornain Wacziarg MANY OBSERVERS have argued that the advanced industrial countries are undergoing a crisis of democracy. In spite of the expansion of democracy worldwide (Huntington 1991) and increased confidence in this system of government, some of the established democracies in Europe and North America are experiencing declines in both civic participation and confidence in the institutions of government. In the United States a secular trend of declining confidence can be found in poll data (Dalton 1999) as well as in shifts of voter turnout over time. In other countries trends in confidence as measured in polling data are both more varied and more difficult to establish because of a lack of reliable time-series data. The most convincing and general evidence for a decline in voter confidence comes from the United States.' Dalton (1999) finds a large and statistically significant
decline in popular confidence in the institutions of government in the United States between 1952 and 1994. Although there is some recent evidence that the decline may have stopped and perhaps even reversed itself since the early I 990s, other sources confirm this finding. The 1996 Survey of American Political Culture indicated that in 1966, 4lpercent of the population had "a great deal of confidence" in the presidency; this figure had fallen to 23 percent in 1976 and 13 percent in 1996. For Congress the figures are 42 percent in 1966, around 15 percent in the 1970s, and only 5 percent in 1996. Only 32 percent of Americans now have "a great deal" or "quite a lot" of confidence in the federal government
as a whole. Similarly, the Pew Research Center for thc People and the Press (1998) reports that "just 20 percent of Americans are satisfied with the state of the nation and only 34 percent basically trust the government." Surveying data from various sources, Dalton (1999) found that confi-- dence in government in many other advanced industrial democracies displays, if anything, a downward trend (see also Introduction, this volume). Outside of the United States, available poii data arc less directly concerned with confidence per se, but rather with questions such as "Do I Throughout this chapter we will use the terms `oti-r saci~fisction and p~iblicconfidenci interchangeably
150 Chapter 7 politicians care?" In all 14 OECD countries
surveyed except the Netherlands, the proportion of negative answers to this type of question has risen over time, indicating, in Dalton's words, that "there is clear evidence of a general erosion in support for politicians and political institutions in most advanced industrial democracies" (Dalton 1999, 63). The most striking examples of these trends are Germany and Britain, where declines in confidence are both large and significant statistically. Another clear example of decline in voter satisfaction is France, where the number of alternations of power between parties of the left and right has recently increased: after 20 years of Gaullist and conservative rule in the 1960s and 1970s, there have been five major shifts in political orientation since 1981. Although alternations of power are not in themselves undesirable in a democracy, this increase in instability demonstrates the impatience of voters, who are constantly sanctioning incumbents. This change has coincided with a decline in voter turnout and a growing perception that no clear policy alternatives are being offered. Several other countries in Europe are experiencing similar situations. In Italy the collapse of the so-called First Republic was accompanied by a decline in citizens' confidence in their government. It is probably too early, however, to establish any trends for the Second Republic. We can interpret these shifts as a sign that people in many OECD countries feel less and less confident in their governments. Although the data needed to establish this fact conclusively are hard to gather, a growing body of evidence supports this conclusion. The decline in confidence is probably larger for the United States than for other OECD countries, for which the evidence is more varied both across countries and over time. Both time-series and cross-countiy variations in the decline of confidence in government may allow the identification of some of its causes. And as suggested b Norris (1999), there are two different ways to assess this trend: either the nature of public policy
, public action, and government performance has changed in such a way that the declining confidence reveals a true "crisis of democracy," or voters are becoming more mature, and their loss of confidence in government is a healthy expression of a more critical electorate. THE MEANING OF VOTER DISSATISFACTION As noted above, voters in many advanced industrial democracies express increasing disconte ut about their governments. Before discussing possible economic explanations for this phenomenon, we address two questions: Can we formally define voter dissatisfaction and its change over time? When is the level of voter dissatisfaction "too low" or "too high"?
The Economics of Civic Trust 151 Voter Satisfaction and the Median Voter Regarding the first question, one can think of three different ways in which citizens' satisfaction may decrease. First, voters may perceive politicians to be less competent and/or more corrupt than before. In other words, governments are now perceived as less able than in the past to deliver the same set of policies. Keeping voter preferences and the capacity of the political system to respond to those preferences fixed, we can equate this with a decline in the quality of policy. To illustrate the other two types of voter dissatisfaction, it is useftil to recall some simple notions of voting theory and rational choice. One of the most famous results in social choice theory is the "median voter theorem" of Black (`1956) and Downs (1957). According to this theorem, under certain specific conditions, ma~oritarianvoting will lead to the adoption of the policy preferred by the median voter. In the context of general elections the median voter theorem holds when two candidates (or parties) face each other and care only about winning; that is, they do not have policy preferences of their own. Under these conditions both parties will adopt the median voter policy platform preferred by the median voter, and full convergence of platforms and policies ensues. This result holds regardless of the distribution of voters' preferences. Figure 7.1 illustrates three different preference distributions, all of which share the same median. In all three cases, the equilibrium policy is the same. In case A the distribution is relatively tight around the median. In case B the distribution is much more spread out. Comparing B with A, it follows that on average, two randomly drawn individuals from distribution B will be farther from the median voter, thus farther from the chosen policy. As a result, voter dissatisfaction will he higher in case B. Analogous arguments apply to case C. Here the distribution is bimodal, but, once again, relative to case A, the median distance of a randomly drawn individual from the median voter will be higher. The point is that the second characterization of citizen satisfaction can be identified with median distance from the median voter (while keeping fixed the position of the median voter).2 In cases B and C the "median distance from the median" is higher than in case A. As a result, people's preferences are more spread out, and it is more costly for voters to agree on a common set of policies and pool their resources to achieve a given objective. 2 Alesina, Baqir, and Easterl~'(1997a) formally develop a voting model tn which the median distance from the median voter is a critical determinant of people's willingness to cooperate be pooling resources needed to produce useful public goons.
152 Chapter 7 A. Tight Distribution
B. Spread-Out Distribution
C. Bimodal Distribution
voter Figure 7.1. Distributions of Voter Preferences over a Single Policy Issue This analysis has two implications. In the context of cross-country comparisons, it suggests why the average level of confidence in government may be different in countries with different voter populations. In more homogeneous citizenries, which are closer to case A, the level of confidence is highest. in countries with preferences distributed as in cases
The Economics of Civic Trust 153 B or C, the level of confidence is lower. In this chapter, however, we are more concerned with a second, dynamic implication. 1f within a count~y, the distribution of voter preferences evolves over time from a case similar to A toward a situation more like B or C, then the average level of citizen confidence will decrease. There are two reasons why the political equilibrium might move from case A to case B or C. First, the distribution of voter preferences may have changed. Second, the issues at the forefront of the government's agenda may have changed, moving from those for which the distribution of voter preferences is like case A to other issues for which the distribution is like case B or C. Needless to say, these two sources of change are not mutually exclusive and may well operate simultaneously. In this chapter we argue that changes in the nature of government intervention in the economy may indeed have changed the public preferences from type A distributions to type B or C distributions. The third way in which dissatisfaction can increase has to do with how the political system translates the preferences of the median voter into actual policy goals. Consider Figure 7.2, which reproduces case A of Figure 7.1. For whatever reason the policy choice does not reflect the median voter's preference. Obviously, if the median policy were feasible, one of the two candidates would choose it, win the election, and this case would not emerge as an equilibrium. But if supra-national constraints, for example, rendered the median policy unfeasible, then policy P of Figure 7.2 may represent the only possible equilibrium. If this is the case, then in moving from case A in Figure 7.1 to Figure 7.2, the median distance from the policy choice increases, and the level of confidence in government policies correspondingly decreases. In this case> too, one can think of two nonexclusive reasons why the electorate may perceive the political outcome to be characterized by Figure 7.2. First, within the set of feasible policies, the policy choice is differ- median voter policy outcome Figure 7.2, Bias in the Transmission of Voter Preferences into Political Choice
154 Chapter 7 cut from the one preferred by the median voter. Second is a problem of perception and complexity of policy. In a world where governments do more and more and the various constraints on policy imply a growing complexity of government action, voters may find it more and more difficult to perceive adequately the relationship between government policy choices and the electorate's desired outcomes. In other words, voters face greater uncertainty about whether government policy really reflects their goals. Finally, a word of caution about the model. The median voter result is veiy stylized and does not always provide a realistic characterization of the complexities of multiparty democracies. The point here, of course, is not to develop a realistic model of multiparty democracies, or even of a twoparry political system. We are well aware that this task would require much more.3 The point of using the median voter model is simply to characterize different ways of explaining the decline in confidence. Indeed, the basic implication of our discussion carries over beyond the median voter model. Consider, for example, the case of two parties that have ideologies and care about their policies. For instance, in Figure 7.3 (which reproduces the preference distribution of case A in Figure 7.1), we have two parties, "Left" and "Right," with corresponding ideal policies. Under certain realistic conditions the two parties will adopt partially convergent policies; namely, they will converge toward each other, but not completely.4 If the distribution of voter preferences becomes more polarized, the average distance of a randomly drawn voter will be higherfor
Figure 7.3. Partially Convergent Party Platforms (In all figures, preferences refere to preferences regarding feasible policy outcomes.)
3 As a matter offact, much of the work of one of the authors of this paper is very critical of the median voter theorem as an unrealistic characterization of two-parry democracy. See Alesina and Rosenthal (1995) and Alesina, Roubini, and Cohen (1997). ~ See chapter 2 of Alesina and Rosenthal (1995) for a discussion of this type of model.
The Economics of Civic Trust 155 given parties' ideal policies. The implication here is that it may be unreasonable to assume that the two parties' policies will remain the same if the distribution of voters' preferences changes, because, at least up to a point, party policies must reflect the electorate's distribution of ideal policies. This two-party model with polarized policies can be used to make another point that the stricter interpretation of the median voter model does not encompass. Suppose that, for a given distribution of voter preferences, say the case of Figure 7.3, the policies advocated by parties become more polarized. A larger fraction of the voters, those in the middle of the political spectrum, will become increasingly dissatisfied with the two parties as they move away from the median. We argue below that this phenomenon may help explain the fall in voter satisfaction in the United States. In summary, we posit three types of voter dissatisfaction: first, dissatisfaction originating from the government's incompetence and the lack of honesty among elected officials (the policy quality hypothesis); second, dissatisfaction originating from greater polarization of preferences (the heterogeneity hypothesis); third, dissatisfaction originating from the deviation of policies from the electorate's median (the representation hypothesis). We discuss evidence for all three arguments. Before doing so, however, it is worth addressing our second question, namely, whether voters' satisfaction with their governments was "too high" in the past or is "too low" now The Optimal Level of Voter Satisfaction The question of an optimal level hinges on how one interprets voter satisfaction. If one views it as a measure of government perfbrmance, it can never be too high. This interpretation, however, is incomplete at best. Voter satisfaction reflects not only an evaluation of government performance, but also a comparison with voters' prior expectations, standards, and ability to monitor government. A change in voters' expectations about what governments can and should do would change voter satisfaction for a given level of government performance. Similarly, an increase in the ability to evaluate government activity (thanks to greater flows of information or enhanced human capital, for example) would lead to changes in voter satisfaction. Within this more articulated view of voter satisfaction, it is meaningful to ask whether the level of confidence in government at any given time is too high or too low. From an empirical point of view, this question is almost impossible to answer. From a theoretical point of view, the optimal level of confidence in government depends on one's conception of government. Neoclassical
156 Chapter 7 economists think of governments as social planners, namely, benevolent agents who maximize social welfare. An extreme Chicago version of this argument would be that inefficient governments will not survive and will be replaced by better ones. Thus, a high level of voter dissatisfaction should be a temporaiy phenomenon: in a steady-state equilibrium, the electorate should be happy with its social planners. At the opposite extreme, one can think of the public choice view of government as a Leviathan. In this case voters should deeply mistrust their governments. In fact, mistrust would a very good starting point for citizens wishing to defend themselves against the Leviathan. Trusting and satisfied voters would simply be gullible voters. The point here is not to offer a complete treatment of the role of government, but simply to argue that antithetical views of the economic role of the public sector
lead to diametrically opposed conclusions about the "correct" attitude of voters toward government. The simple fact that voters' confidence in government may be falling is thus insufficient for drawing a pessimistic conclusion. Perhaps electorates are achieving a more realistic view of their governments. Of course, this view still leaves us with the task of explaining why the level of confidence was too high in the past. GOVERNMENT COMPETENCE In this section we explore how government policy, especially macroeconomic policy, has failed since 1974 to perform as well as in the preceding three decades. We interpret this as an example of how our policy quality hypothesis can sometimes explain declines in civic confidence: for a given policy objective of the median voter, for a given distribution of voter preferences around those of the median voter, and for a given ability of the political process
to translate these preferences into policies, we argue that governments have been less able to deliver outcomes of the same quality as in the past. Macroeconomic Policy Outcomes A vast literature suggests that government popularity depends on basic macroeconomic variables, in particular income growth, unemployment, and inflation. The evidence is extremely solid and clear-cut for the United States, a bit murkier for other OECD countries.S For the United States the evidence suggests that the rate of GDP growth is the best 5 On the United States see Kramer (1971), Fair (1978), and Alcsina and Rosenthal (1995). On other OECI) countries see Lewis-Beck (1988), Powell and Whitten (1993), and Alesina, Perotd. and Tavares (1998).
The Economics of Civic Tmst 157 predictor among economic variables of the electorate's level of satisfaction. Inflation and unemployment are also important, but less so. For other OECD countries, the results vary depending on the institutional framework, the nature of government (e.g., a two-party or multiparty system), and the economic structure. While sweeping generalizations are difficult, broadly speaking, the evidence suggests that inflation, growth, and unemployment do affect electoral results, although in different ways and to a different extent in different countries. This is not the place for an in-depth analysis of the effects of economic conditions on voting behavior in a cross section of OECD countries; for this the reader is referred to Lewis-Beck (1988), Powell and Whitten (1993), Paldam (1991), and Alesina, Perotti, and Tavares (1998), among others. Here we are simply starting with the rather uncontroversial premise that voters in any country prefer higher to lower growth and lower to higher unemployment and inflation. Obviously there can be trade-offs among these variables, although only in the short run, and voters may disagree on the optimal trade-offs, hut the assumption is that a voter cannot be happier with, say, higher inflation and higher unemployment. In this chapter we are not particularly interested in cyclical fluctuations of popularity, but rather in secular trends
in confidence in government, where by "secular" we mean movements over the last three or Ibur decades. If voters react at the cyclical frequency to these three variables (unemployment, inflation, and growth), and if these same variables exhibit secular trends, then perhaps these long-term trends are correlated with secular trends in voter satisfaction. Table 7.1 summarizes the evidence on these three variables for the United States, Japan, the European Union
average, and the OECD average. All variables deteriorated in all samples: avcrage growth fell, average unemployment rose, and average inflation also rose from the pre-1974 to the post-1974 period. Admittedly, using the first oil shock as the watershed between periods conceals some important differences in the two decades following 1974. In fact, the 20 years following the first oil shock displayed a marked deterioration of macroeconomic performance before the mid-1980s and an improvement thereafter. Improvement over the last decade has not been even, however. Unemployment has worsened in Europe, for example, and the average rate of growth has been much lower in Europe than in the United States. Recent events in SouthEast Asia
as well as a major recession since 1992 have undermined the prospects fur a continuing miracle in Japan. The variability of macroeconomic outcomes has unquestionably risen since the early 1970s, and insofar as people care about the volatility of their environment (in addition to the absolute level of economic indicators), this also helps explain shifts in civic confidence.
158 Chapter 7
TABLE 7.1 Deterioration of Macroeconomic Performance
Growth of GDP OECD average EU1S average United States Japan Germany France
5.030% 4.941 3.842 9.170 4.234 5.268
2.481% 2.280 2.346 3.825 2.179 2.362
Unemployment Rate OECD average EU15 average United States Japan Germany France
2.354 2.338 4.899 1.251 0.563 2.014
6.139 7.148 6.920 2.270 4.599 7.852
Inflation Rate** OECD average EUI5 average United States Japan Germany France
4.903 4.533 3.158 5.904 3.329 4.663
9.845 8.589 5.606 4.299 3.459 7.164
Nores: OECD and EU1S averages are unweighted. *1974_91 for Germany. **~~Jjfl~~ as the growth rate of the consumer expenditure deflator. Source: OECD.
Government Size Although the macroeconomic performance of the last two decades has not been stellar relative to the previous period, government involvement in the economy has increased tremendously. Table 7.2 demonstrates that for the OECD sample, the share of government spending in GDP has basically doubled between 1960 and today, from 25 percent to almost 50 percent. Tax revenues lagged behind expenditures in the 1970s and early 198 Os, increasing more rapidly thereafter. As a result, budget deficits remained high in many countries between 1975 and 1985. The last decade or SO has witnessed several major attempts on the part of OECD governments with large public-sector deficits to balance their budgets. In many cases these adjustments were motivated by a desire to join the European
The Economics of Civic Trust 159
Revenue and Spending as a Percentage of GD?
Total Revenue OECD average EU15 average United States Japan Total Spending OECD average EUI5 average United States Japan Primary Spending* OECD average EU15 average United States Japan
1960 27.09% 28.16 26.36 18.77 25.73 26.86 25.70 16.86 24.03 25.16 24.40 16.33
1997 45.45% 46.55 31.98 33.05 46.66 49.90 31.98 34.72 41.47 44.07 30.13 30.96
Notes: OECD and EU1 5 averages are unweighted. *Exeludes interest payments on the public debt. Source: OECD.
Monetary Union. Even during periods of fiscal adjustments, however, expenditures kept increasing with very few exceptions, and taxes increased even more as a percentage of GD?, With lower macroeconomic performance and higher government involvement in the economy, the least one can say is that the productivity of government intervention in terms of its ability to deliver stable and favorable economic outcomes has fallen dramatically. Although it must be conceded that some of the increase in government size was used for redistributive or insurance purposes in the face of economic shocks like the oil crisis, it is hard to deny that it has played a part in reducing performance. It is not surprising that its social marginal product should shrink as government grows.6 Furthermore, as the public sector commands a larger and larger share of GD?, the electorate holds it more and more accountable for the state of the economy because its ~ On the optimai size of government from a growth perspective, see Barro (1990). This paper identifies an optimal size of government beyond which government consumption results in reduced growth. The optimal size of government is derived from a trade-off between the productive aspects of certain government expenditures and the distortionary eftиcts of the taxes needed to iinance them.
160 . Chapter 7 actions are rightly perceived to have a growing impact on economic outcomes. If, as we argued earlier, these outcomes have become less favorable as government has expanded, it is only natural that satisfiction with government should have declined. It is interesting to compare the United States and Europe with respect to government size and macroeconomic performance. While average economic performance since 1974 has deteriorated in both places, growth has been higher and unemployment lower in the United States than in the European Union in the last decade or so. How is this consistent with the fact that voter satisfaction seems to have decreased more in the United States than in Europe? One possible answer is that a more extensive welfare state in Europe may have dispersed the costs of unemployment to a broader base of taxpayers. In other words, high unemployment is more tolerable in the presence of Social Protection
, even when the fiscal costs of that protection may in fact contribute to keepingunemployment high. Although a generous welfare state may lessen the costs of unemployment, it also expands redistributive flows that generate political conflicts, which is the topic of the next section. In the United States the deterioration of the macroeconomic environment has led to increased income inequality rather than higher unemployment, and this may have helped reduce civic confidence by means of the polarization mechanism described above. A further indication that this may well be the case is that the prolonged expansion of the U.S. economy since 1992 has coincided with a stabilization--if not a slight reversal--of the downward trend in civic confidence. In summary, citizens of a typical OECD country have observed a massive increase in government economic intervention and a deterioration in macroeconomic conditions in the last few decades. Several observers, including the authors of this chapter, would argue that the massive increase in government intervention is at least partially responsible for some of our current macroeconomic ills. This is certainly the case in many European countries, where the burdens of taxation and extensive government regulation, particularly of labor market
s, are generally viewed as the main culprit for persistently high levels of unemployment. As we argue in the next section, however, while an increase in the size of government is observable in all OECD countries, the dimensions of the increase and the composition of government spending differ quite significantly in the United States and many European countries. Corruption The degree of honesty of politicians and bureaucrats is another measure of the quality of policy. Is there any evidence that corruption has in-
The Economics of Civic Trust 161 creased in OECD countries? Italy, Japan, and South Korea
are examples of countries in which, for a variety of reasons, voters became less tolerant of persistent corruption, and this became a crucial issue over the last two decades. It is hard to establish whether the actual level of corruption rose sharply in italy and Japan in recent years, however. On the one hand, it is extremely difficult to assemble time-series data on objective measures of political corruption in OECD countries. By its very nature, objective measurements of corruption are hard to obtain, so one has to rely on subjective measures, namely, surveys in which people are asked about their perceptions of corruption. Furthermore, most available data on corruption across countries rate OECD countries very close to one another, so there is very little variance from which to draw inferences. Finally, although these measures of corruption may be reasonably useful for the purpose of cross-national comparisons, they are not useful in establishing time trends within each country because they lack timeseries variation. On the other hand, even if objective measures of observed corruption were available, this would be of little help in assessing the true degree of underlying corruption. Perhaps the detection technology for such activities has improved over time. If that is the case, then a rising level of observed corruption would be perfectly compatible with a reduction in its true level because improved detection is likely to deter bureaucrats and politicians from breaking the law. Rational voters would not lose confidence in government as a result of this. But it is difficult to establish whether it is the true level of corruption that has increased in some OECI) countries or whether it is detection that has improved, so it is difficult to ascribe the decline in civic confidence to this phenomenon. One is left with the weaker conclusion that, fbr a given level of con-uption, citizens may have grown less tolerant of corrupt acts, but it would be difficult to provide anything but anecdotal evidence for this. POLARIZATION Redistribution and J'olarization Increasingly, governments in OECD countries are becoming redistributive machines; that is, they tax someone in order to redistribute to someone else. In fact, transfers account for virtually all of the increase in government size documented in Table 7.3. As a share of GDP, government consumption of goods and services is roughlv stable, and government investment is declining. Thus, while governments were mainly providers of public goods and infrastructure in the 1950s, today they are primarily involved in redistribution. The other component of government outlays
162 . Chapter 7 TABLE 7.3
Composition of Expenditures as a Percentage of GDP
Transfers OECD average EU15 average United States Japan
1960 7.95% 8.85 5.58 3.98
Notes: OECD and EUI 5 averages are unweighted. SDWvC: OECD.
1997 19.74% 21.34 13.71 15.08 11.95 12.48 9.73 7.24 2.77 2.49 2.80 5.72
that has increased is government wages. It is likely that a portion of public wages is also used as an instrument of redistribution toward politically powerful groups. In Italy public employment has been widely used to redistribute fiscal resources from the North to the South, for example.7 Similarly, Alesina, Baqir, and Easterly (1997b) document the use of public employment for redistributive purposes in U.S. cities. They find that public employment per capita is higher in cities with greater income inequality and ethnic fragmentation. An extended welfare state has two effects. First, as discussed above, it provides social insurance, removing some economic risks against which the market cannot adequately insure. Second, it also generates a complex web of redistributive flows. The more governments redistribute, the more they generate winners and losers: while everyone can take advantage of public goods and infrastructures, which are generally of a nonrival nature, it is necessarily the case that somebody gains and somebody loses in redistributive schemes. Given that redistribution is inherently conflicanal, it may polarize the interests of voters. In other words, the median distance from the median voter increases. Forced to support overex- 7 For an excellent discussion of the distortions and inefficiencies of public employment and the welfare system in Italy, see Rostagno and Utili (1998).
The Economics of Civic Trust . 163 tended welfare states, taxpayers become more vocal as the tax burden continues to increase. Recipients of net transfers, on the other hand, become accustomed to government generosity, and any attempt to reduce their benefits also invites vocal complaints. Thus, the more govermnents. become redistributive machines, the more they generate direct conflicts between winners and losers, and the more difficult it becomes to reverse the trend of increasing redistribution.8 If it is assumed that losers are more vocal than winners, this increased conflict will result in increased dissatisfaction with government. A particularly important example of this phenomenon is the case of pensions, one of the largest components of government transfers. As the population ages and pension systems start to face financial difficulties, governments try either to reduce the benefits of current or prospective retirees or to increase payroll taxes. The first option riles those who feel that the rules of the game are being changed just as they are about benefit from them. The second option generates dissatisfaction among the relatively young, who feel that they bear an ever-increasing tax burden to support preceding generations. Similar arguments apply to public employment. Public employees are very vocal in their outrage when faced with attempts to reduce their benefits. Taxpayers, on the other hand, are losing tolerance for increasingly large and inefficient bureaucracies. The Cases ofEurope and the United States The size of the welfare state varies from country to country. It is larger in Europe than in the United States, and larger within certain European countries than in others; it is much larger in Nordic countries than in Britain, for example. As Table 7.3 demonstrates, however, transfers as a share of GDP have more than doubled between 1960 and 1997 in both the European Union and in the United States, while they have nearly quadrupled in Japan. We argue that this increase helps explain the decline in Public confidence
, especially in the United States. In Countries with growing welfare states, such as many European nations, the growth of government is self-sustaining, leading to an excessive and ever-increasing tax burden.9 As the welfare state expands, a growing segment of the population receives more and more of its income from the government, that is, from other taxpayers. These recipients of public monies include pensioners, whose share in the total population is increasing for demographic reasons, beneficiaries of long-term unemployment See Olson (1982) for a similar point. ~ Meltzer and Richards (1981) were the first to formulate the idea of a political equi1ib~ rium that generates a self-sustaining increase in the welfare state. Sec also Alesina and Rodrik (1994L
164 Chapter 7 compensation, public employees, and recipients of income-support programs. In addition, a fraction of these groups, especially pensioners, are relatively wealthy and have the time to engage in political activities, so their weight in public policy-making may be greater than their weight in the electorate. Furthermore, it may be rational for each taxpayer or constituency individually to insist on preserving the transfers they receive even if it would be collectively optimal to overhaul the welf~restate. In other words, the economy suffers from a coordination problem. As a larger constituency comes to support the welfare state, it too becomes larger, inducing a vicious cycle. As this political equilibrium develops, taxpayers become increasingly unhappy. When governments attempt to reduce the weight of the welfare state due to mounting constraints on theirfinances, those people whose income depends on it are particularly vocal in their disapproval. In the United States the political equilibrium is quite different, and in some ways precisely the opposite. In this country the welfare state is narrower and public nonmilitary expenditures much lower than in the average European countiy A large majority of the American electorate views government as a burden and an impediment to market activities. But if the increase in the size of the weliIre state has something to do with increasing voter dissatisfaction, why has public confidence fallen more in the United States than in Europe when the American welfare state is smaller than its average European counterpart? Three details of the American case may help resolve this apparent con- tradiction and explain why a given increase in the size of the welfare state may have led to a larger decline in public confidence in the United States than in Europe. First, the welfare state and racial problems are deeply interconnected in the United States. Wilson (1996, 193, 202) eloquently puts it as follows: "Many White Americans have turned against a strategy that emphasizes programs that they perceive as benefiting only racial minorities. Public services are identified mainly with Blacks, private services mainly with Whites." Poterba (1997) convincingly demonstrates that the elderly object to increasing local taxes used to finance education, and this effect is much stronger when these funds are used to support public educational programs
in which minority children are perceived to be the main beneficiaries. Alesina, Baqir, and Easterly (1997a, 1997b) examine a large body of evidence on spending programs in U.S. localities and conclude that ethnic fragmentation is a major determinant of the allocation of public funds: public goods are less abundant in ethnically fragmented cities. in this context a smaller expansion of the welfare state than has occurred in Europe can result in a greater decline in voter satisfaction. A second peculiarity of the American case has to do with the degree of polarization of its two main parties, especially their most vocal activists,
The Economics of Civic Trust . 165 vis-г-vis the position of the median voter. This is particularly relevant to the size of the welfare state. Although a large majority of the American public probably favors limited welfare programs, the two parties offer polarized choices: While Democratic activists champion a large expansion of welfare, many Republican activists advocate drastic cuts. Similar examples of party polarization can be found on such issues as abortion and school prayer. Poole and Rosenthal (1984) have amply documented the increased polarization of American political activists. In fact, according to polling conducted by the Pew Research Center for the People and the Press (1998), excessive partisanship is one of the main reasons Americans give for distrusting the government.'0 This suggests that the excessive polarization of the two parties has much to do with voter dissatisfaction." American voters feel discontented because they face only extreme choices, the flip side of our earlier argument regarding the increased polarization of voter preferences themselves (which may apply more directly to Europe). Third, culture, traditions, and social norms also play a role. In the United States much more than in Europe, the idea of the "self-made man" is pervasive: Citizens are expected to be masters of their own fortunes with little help from others, especially the government. Thus, while European governments are often criticized for inadequately protecting some group of potential beneficiaries, America's federal government is often criticized on opposite grounds, for tlying to do too much. For a given increase in the size of the welfare state, the resulting decline in satisfaction due to heightened distributional conflict is thus likely to be greater in the United States than in Europe. This implies a paradox. In Europe, where governments occupy a larger share of the economy, they are often criticized for doing too little, particularly by those who are already well protected by government programs. In the United States, on the other hand, where the government is smaller, it is often criticized for doing too much. This observation points to the possible existence of two polarized equilibria. In the first, an increasing fraction of the population relies on government support and provides the backing for increasing the scope of the welfare state. In the second, a shrinking fraction of the population takes advantage of government services and provides shrinking support for taxation. tO More generally the survey reports that "When asked to say in their own words why they do riot like government, 40% of those with an unfavorable opinion of the government offer complaints about political leaders or the political system as the reason for their negative view." `` As argued by Fiorina (1992) and A]esina and Rosenthal (1995), divided government is the result of voters' attempt to implement moderate policies by splitting control of government institutiOfl5 between polarized pantc~..
166 Chapter 7 SOVERI~IGNTY We argued above that growing voter dissatisfaction could also originate from increased discrepancies between the preferences of the median voter and the policies actually implemented, keeping fixed the distribution of voter preferences and the quality of policy.'2 This idea provides a good framework with which to understand the effects of international economic integration on voters' confidence in government. Globalization challenges democratic institutions in two distinct ways. First, the international environment reduces governments' latitude to implement preferred policies. Second, certain policy prerogatives tend to be transferred to supra-national institutions without a concurrent transfer of sovereignty, reducing democratic control over policy outcomes. This point is especially valid in the case of Europe. Rising International Conctraints on Policy Even if policy-makers and the median voter seek to maximize the same objective function, a tightening of the constraints on policy will lower voter satisfaction. Since World War II economic integration has risen sharply. As measured by the ratio of imports plus exports to GDP averaged over a sample of 61 countries, the volume of trade has increased from 43.2 percent in 1950 to 60.6 percent in 1992. But economic integration is not limited to trade: financial markets have become more tightly linked as the share of domestic assets owned by foreign entities has increased sharply in most industrial countries. These trends have been supplemented by a move toward "deep integration" in most developed nations, the breaking up of the process of production into different stages, each potentially occurring in a different country. Falling transportation costs
permit a slicing up of the value-added chain: multinational firms can "produce a good in a number of stages in a number of locations, adding a little bit of value at each stage" (Krugman 1995). This allows producers to locate each step of the production process in the most advantageous location. The flip side of this development is to undermine governments' ability actively to monitor and direct industrial activity. And whatever the objective desirability of active industrial policies (it may not be great), this limitation on governments' potential influence is sufficient to drive a 12 A Rasmussen Research poll published in April of 1998 found that only 9 percent of American adults believe that U.S. politicians represent the American people. The polling agency also reports that "34% ofArnericans believe that the government reflects the will of the people. Half (47%) say it does not, while 20% are not sure."
The Economics of Civic Trust 167 wedge between the preferences of the median voter and feasible policy outcomes. The impact of globalization on public policy is not limited to industrial policy, of course. International constraints also affect macroeconomic policy, for example. It has often been pointed out that capital Market Integration
across the OECD has reduced the effectiveness of monetary policy; countries face the threat of capital flight and devaluation if they do not maintain a strict anti-inflationary stance. This was especially true in Europe during the 1980s, when many countries such as France would immediately have been sanctioned by global financial markets had they not closely followed the same monetary policy as the German central bank. Thus, in this dimension, too, globalization has reduced the ability of governments to conduct autonomous policies as prescribed by the domestic system of political decision-making. And while these constraints may have led to more virtuous policies in the long run, they have also put a growing range of potentially important economic decisions under constraints that lie outside of the control of nation-states. Indeed, there is mounting evidence that the average European voter is and has for some time been less enthusiastic about European integration and monetary union than his or her leaders. Germany provides a good example of this point, with Chancellor Helmut Kohl actively supporting monetary union despite the objections of a majority of voters. The Declining Authority of Nation-States Beyond the constraints that globalization imposes on domestic policies, international economic integration has also required some explicit transfers of political prerogatives to supra-national institutions, often without a concurrent transfer of sovereignty. In other words, as powers are transfened further away from citizens, the objective functions of the median voter and relevant policy-makers are becoming more and more distinct. The prime example of this shift is the European Union, where policies decided in common now range from agricultural policy to monetary matters and competition regulation.'3 Because the European Parliament has a very limited ability to control the actions of the European executive branch, these transfers have occurred without a corresponding transfer of popular sovereignty. Executive decisions made at the Council of Ministers
result from negotiations among representatives of member nations, each presumably representing the preferences of a different median voter. It is very unlikely that these decisions could efh~ct.ivelyreflect the choices ° Sec Alesina and Wacziarg (1998) for a recent discussion of European political integration
168 Chapter 7 favored by the median voters of each country, and it is unlikely that the European median voter (defined as the median voter obtained by pooling the electorates of Member States
) would see his or her preferences implemented either. Although they are still the basic units in which democratic sovereignty operates, the authority of nation-states is thus threatened from above by the rise of supra-national institutions. But they are also threatened from below by the rise of separatist and regionalist movements in Europe.'4 Alesina, Spolaore, and Wacziarg (1997) argue that globalization in the form of a liberalization of the trade regime provides heterogeneous groups with access to large markets without their having to share institutions with other groups. Indeed, if the optimal size of countries is determined by .a trade-off between the gains from scale and the costs of ethnic, cultural, and linguistic heterogeneity, trade liberalization should result in a multiplication of countries because the benefits of country size, in terms of returns to scale, are reduced when access to foreign markets is assured. In other words, economic integration leads to political disintegration. In fact, there is considerable empirical evidence
to support this claim. In the interwar years, a period of increasing protectionism, practically no countries were created. But in the aftermath of World War II, when global trade started to rise sharply as a result of liberalization and falling transportation costs, the number of countries exploded from 74 in 1945 to 192 in 1997. The problem with these tendencies `toward greater autonomy is that they take time to result in actual secessions. In fact, most episodes of country creation have occurred in lumps as the result of a major political event like the end of a war. The end of World War II resulted in a first wave of decolonization, and the end of the Cold War brought an explosion in the number of countries in Eastern Europe
and the former Soviet Union
. The process of country formation is far from smooth, largely because of centers' resistance to letting go of their regions. When the fundamental conditions determining the optimal number
of countries change, as happens when the world becomes increasingly integrated economically, yet there is resistance to secessions or mergers, there will be heterogeneous groups that would like to secede but cannot. In such a situation these groups are likely to voice growing dissatisfaction with the nation-state from which they would like to secede. Even if these demands for political autonomy do not always take the form of explicitly separatist movements, as is increasingly the case in Europe, they may take the form of more subdued demands for greater de- 14 In the United States this takes the fbrrn of greater demands for decentralization by individual states.
The Economics of Civic Trust . 169 centralization and local empowerment, as in the United States. The bottom line is that, in many cases, the nation-state does not seem to be the optimal unit for decision-making anymore. For a wide range of economic issues, the optimal unit for decision-making lies above the nation-state, at the supra-national level, and for a growing range of political or local issues the optimal unit lies below the nation-state, at the regional level. Until both sovereignt and political prerogatives are transferred to these new units of political and economic decision-making, voter satisfaction and confidence in the traditional institutions of the nation-state are likely to decline further. Finally, discrepancies between voter preferences and governmental policy can arise from the increased complexity of public policy A global economy
, supranational institutions, national entities, and local governments interact in ways that give rise to a complex division of political prerogatives, conflicts among levels of government, and difficulties in assessing who should be held accountable for which policy outcomes. One can argue that the multiplication of levels of government, the rapidly expanding amount of information that voters need to process in order to understand the actions of government, and the increased expertise needed to exercise governmental power have all led to more complex political systems. One response to this trend is that governments have had to become more technocratic, which is another way of saying that the distance between citizens and policy has increased. This certainly explains much of the decline in confidence in a country like France, where complaints about the excessive technocracy of ruling elites is a recurring theme at election time. This is another way in which the objective functions of policy-makers and voters can come to differ. CoNcLUSION This chapter has suggested three possible interpretations for the decline in public confidence in the advanced industrial democracies. First, the quality of government policy may have declined. Second, the distribution of voter preferences may have become more polarized. Finally, the ability or willingness of political actors to deliver the policies favored by the median voter may have been reduced. For each of these explanations we have provided examples iliustrating how recent economic shifts could help explain the drop in public confidence. Comparing the economic performance of OECI) countries before and after the 1974 oil shock, we showed that growing voter dissatisfaction rna\' be due in part to the weaker performance of the last 25 years in terms of average inflation, Economic Growth
, and unemployment. Although the decline in perforinance is not necessarily attributable to inappropriate government activ-
170 Chapter 7 ities, we argued that the observed correlation between government approval ratings and measures of economic performance indicates that governments are generally held accountable by voters for the state of the economy. We then argued that the rise of the welfare state has polarized voter preferences. In Europe each constituency is happy to receive government handouts and reduce other constituencies' share of the public-sector pie. This struggle among pressure groups has resulted in an upward spiral of government transfers, leading to increased discontent with the privileges of others and a diminished willingness to surrender one's own. In the United States the game has been played out differently: the unwillingness of voters to be taxed in order to finance transfers to certain constituencies has limited the government's ability to expand the welfare state, and this tendency is increased by the influence of racial conflict on welfare politics. This explains why a given rise in the size of the welfare state has resulted in a greater fall in public confidence in the United States. It is the increased polarization of America's main parties over the proper extent of government involvement rather than the polarization of voter preferences themselves that has irked relatively moderate voters. Finally, we argued that globalization tends to drive a wedge between the preferences of the hypothetical median voter and the actual policies being implemented. First, globalization sets limits on the set of feasible policies. Second, economic integration has resulted in transfers of powers to supra-national institutions as well as demands for local autonomy. This has not, however, led to corresponding transfers of sovereignty. Because it is increasingly viewed as "too big to run everyday life and too small to manage international affairs" (Newhouse 1997), the nation-state may be the real target of voter dissatisfaction. The organization of public life among local, national, and supra-national government
s has been undergoing profound transformations, especially in Europe. In this context de-- dilning voter confidence may result from natural worries about a rapidly changing polity. ACKNOWLEDGMENTS This chapter was prepared for the Conference on Public Trust and Governance in the Trilateral Democracies, at the Rockefeller Study and conference center
, Bellaglo, Italy, June 29--July 3, 1998. We thank our discussant Peter Katzenstein as well as Susan Pharr, Robert Putnam, and many other conference participants for their useful comments, but all errors are ours alone.