The Civil State: Trust, Polarization, and the Quality of State Government*
    
    Eric M. Uslaner Department of Government and Politics University of Maryland–College Park College Park, MD 20742 euslaner@gvpt.umd.edu
    
    For Jeffrey Cohen, ed., Public Opinion in State Politics (Stanford University Press, forthcoming)
    
    Uslaner, The Civil State (1)
    
    Does good government depend upon good citizens? Robert Putnam (1993) made this connection in Making Democracy Work and others (including myself) have echoed this claim. Yet, it has proved difficult to measure “good government” (much less to agree what it is) and the causal chain from a positive citizenry to governmental performance remains murky. Democracy depends upon a participatory citizenry, to be sure. Representation depends upon an alert citizenry. But what do citizens need to do to secure effective government? Putnam’s link between good citizens and good government is encapsulated in his concept of social capital, which encompasses social networks, formal organizations, and norms of trust. Putnam (1993, 115) argues that in Northern Italy: choral societies and soccer teams and bird-watching clubs and Rotary clubs. Most citizens in these regions read eagerly about community affairs in the daily press. They are engaged by public issues, but not by personalistic or patron-client politics. Inhabitants trust one another to act fairly and to obey the law. Leaders in those regions are relatively honest. But in Southern Italy: Engagement in social and cultural associations is meager.... .”Compromise” has only negative overtones. Laws (almost everyone agrees) are made to be broken, but fearing others’ lawlessness, people demand sterner discipline, nearly everyone feels powerless, exploited, and unhappy....it is hardly surprising that representative government here is less than in more civic communities.
    
    Uslaner, The Civil State (2) Putnam’s argument is that an engaged citizenry becomes more trusting and cooperative and more likely to demand that their representatives place great emphasis on the public good rather than clientelism or patronage. In one respect, this is not a new argument: Lane (1959, 164165) and Almond and Verba (1963, 3) suggested that “civic cooperation” was important for democratic government. Inglehart (1999) found a link between generalized trust and democratic government (as well as support for democracy). Knack and Keefer (1997), LaPorta et al. (1997, 1998), Uslaner (2002, ch. 8; 2004) find that countries with high levels of trust have lower levels of corruption, better functioning bureaucracies, more effective legal systems, lower rates of theft, and “better government” more generally. King, Zeckhauser, and Kim (2001) and Knack (2002) find that American states with higher levels of “social capital” have more effective governments. Effective government in the states is measured by a study by the Government Performance Project (GPP) of Governing magazine and the Maxwell School of Citizenship at Syracuse University. Program information, a survey, and “interviews [with] budget officers, auditors, public managers, auditors, academics, and legislative aides in every state” (Knack, 2002, 775). Each state was “graded” on its performance in financial management, capital management, human resources, “managing for results,” and information technology. I consider each of these measures in this essay, as well as three other measures: the number of Ford Foundation/Kennedy School of Government (Harvard University) awards for innovation a state has won; and the level of corruption in a state, as perceived by political reporters in state capitals (Boylan and Long, 2001). The GPP measures, the Ford Foundation/Kennedy school awards, and the level of corruption are all measures of the quality of a state government. King, Zeckhauser, and Kim (2001) and Knack (2002) have shown that states with higher
    
    Uslaner, The Civil State (3) levels of social capital have better functioning governments, using most of these same measures of state effectiveness. So what is new here? Two rationales for another look stand out. First, there is much conceptual confusion about social capital and why (or whether) it might effect government performance. I shall argue below that most of social capital is irrelevant to good government and the important aspect of social capital, generalized trust, has been poorly measured in other studies. Beyond social capital, I offer a vision of the civil state as a polity where people trust others who are unlike themselves, where there is minimal political divisiveness, where leaders work with each other toward finding some common ground, and where the political environment militates against confrontation. This civil state has better, more effective government because there is less back-biting and at least some of its foundations lead to greater honesty in government and more participation toward the common good by ordinary citizens. Better theory and better measurement lead to firmer support for the link between trust and performance. Second, trust is important because it leads to a more cooperative, less polarized society and polity: The decline in Congressional productivity over the past three to four decades has a direct linkage to falling levels of generalized trust; as we have less faith in each other, there are more filibusters and gridlock in the Congress and the ideological gap between the Congressional parties has grown (Uslaner, 1993, ch. 6; 2001; 2002, 214). A less polarized polity is more productive (see also Binder, 1999). Knack (2002, 778) briefly makes an argument about polarization, but his measures (divided government and racial heterogeneity) do not tap people’s preferences. I shall present models of good government in the civil state for the American states. The
    
    Uslaner, The Civil State (4) models emphasize political polarization, relying upon measures of partisan polarization from national surveys and from state legislative voting. More trusting and less polarized states should have more effective government. The evidence for trust is generally strong, but polarization effects are more sporadic (and, in one case, even positive). I also consider other political variables and find that high levels of party competition leading to divided government generally lead to less effective government, as do strong party organizations. Traditionally, we think of the most “innovative” (and perhaps most “effective”) state governments to be dominated by liberals, but instead I find that effectiveness in management is greatest in states where the dominant party is most conservative. “Social capital” matters mightily for good government, but not as Putnam’s argument would lead us to expect. It is trust, not civic engagement, that leads to good government. Trust is significant for most of the measures of government quality, but it has more powerful effects for corruption (see Uslaner, 2004). Corruption thrives where political competition is lacking–in one-party states and where there are strong party organizations–and where there some groups (mainly whites) are better off than others. You may need a vigilant public to guard against corruption, but as with government effectiveness more generally, civic engagement is not enough. No measure of civic engagement leads to better government performance (see below). A particularly demanding form of participation–the share of the public making political speeches (ranging from .05 percent in New Jersey to 9 percent in Oregon and South Dakota)–does combat corruption –but this is hardly the sort of political participation that we might expect of ordinary citizens.
    
    Uslaner, The Civil State (5) Good Government and Good Citizens Good government means different things to different people and that is one reason it has proven so difficult to measure. We normally think of effective government as including strong service delivery, impartial bureaucrats and judges, and the honesty of government officials (Kaufmann, Kray, and Mastruzzi, 2003). These factors are terribly difficult to measure. There are multiple measures of government corruption, including the indictment and conviction rates of political leaders and newspaper reporters’ subjective estimates of corruption (Boylan and Long, 2001), which is closest to international estimates of corruption by Transparency International. However, measures of overall quality have been elusive. The GPP measures are an attempt to provide some data on governmental effectiveness. They are not measures of legislative productivity or policy gridlock in the states (which might be ideal). They are indicators of state capacity and bureaucratic functioning. The five measures and some of their components are (King, Zeckhauser, and Kim, 2001, 31-32; see also Knack, 2002, 775-776): * Financial management: accurate revenue estimates, state budget adopted on time, state has long-term budget planning; accurate financial reporting; management of goods and services; audited financial statements. * Human resource management: clear personnel policies, can hire new employees quickly; merit pay; can state discipline and terminate unproductive employees? * Information technology management: how widely available is technology and how quickly can it be ordered; IT training; use of web sites for citizen information. * Capital management: evaluation of planning process for construction; integration of
    
    Uslaner, The Civil State (6) capital and operating budget; planning for capital maintenance. * Managing for results: does state have a strategic plan and are citizens, businesses, unions, and other groups involved in planning? effectiveness of performance measures for policy makers. These are all measures of bureaucratic performance. Only Managing for Results, which includes citizen participation, includes any direct link between citizens and their government. The GPP project gave each state a letter grade--A, B, C, D, with pluses and minuses--on each category and King, Zeckhauser, and Kim (2001) translated these grades into numeric scores. The five indicators were all positively related to each other–with correlations ranging from .71 for financial and capital management to a low of .37 between capital and human resources management. Overall, the states with the highest grades were Missouri, Utah, Washington, and Virginia and those ranking at the bottom were Alabama, Hawaii, Connecticut, New York, and Rhode Island. The Ford Foundation/Kennedy School innovation awards (adminstered by the Kennedy School and the Council for Excellence in Government in Washington, D.C.) have been given annually since 1986 for public sector initiatives in areas including “customer focus” (meeting the demands of diverse constituencies); public-private competition; “performance benchmarking” (developing clear standards for evaluating governmental effectiveness); citizen participation; consumer choice, and privatization (Altshuler, 1997; Sparrow, 2000, 81). These awards measure not only performance, but also doing things differently than before–suggesting that the roots of innovation may be different from simple measures of performance such as the GPP indicators. These measures may not be ideal, but they are the best available on the quality of state
    
    Uslaner, The Civil State (7) government–with one exception, a measure of corruption by statehouse political reporters in 1999 reported by Boylan and Long (2001). Corruption is notoriously difficult to measure: If done well, it should be unobservable (unless, of course, the public doesn’t care about it). The most widely used measures in cross-national research are reputational measures from Transparency International.1 Reputational measures are always controversial, but the Transparency International measures are widely accepted for their face validity. The BoylanLong measures for 47 states have Rhode Island and Lousiana as the most corrupt and the Dakotas and Colorado as the most honest. An alternative measure of corruption is the share of public officials indicted or convicted (Meier and Holbrook, 1992).2 The most corrupt states in 1995 were Florida and Virginia and the least corrupt were New Hampshire and Vermont. The two measures are not identical–both Florida and Virginia rank 14th and 26th respectively on the Boylan-Long measure and the overall correlation between the two measures is just .259. The reporters’ measure seems to be the better one, since it has greater face validity. Prosecution indicators may reflect the personal priorities of prosecutors (Boylan and Long, 2001, 3-4)–and it may simply be more difficult to gain an indictment and conviction in a heavily corrupt state. Thus, I rely upon the reporters’ perception measure, which seems less troubled by endogeneity issues (such as whom to prosecute and whom to convict). I now turn to an examination of the role of trust in civic life, to other predictors of government performance, the results for the five GPP indicators and the Ford Foundation/ Kennedy school awards, and then a theoretical and empirical discussion of corruption. The Civil State and Good Government The civil state is a moderate state. Citizens in civil states trust people who are different
    
    Uslaner, The Civil State (8) from each other. They seek to avoid confrontation and to seek common ground in decisionmaking. People in polarized societies don’t trust folks who disagree with them. Different views become a sign of heresy rather than simple disagreement. Fundamentalists, be they religious or political, are likely to trust only their own kind–and to favor confrontation over compromise. Good government requires a commitment to seeking common ground–and it will have difficulty flourishing in a confrontational political environment. So where the public or the elites (legislative leaders) are highly polarized, it will be more difficult to obtain effective government. In states where we find strong political party organizations, we may also find fewer incentives for the two parties to cooperate with each other. Especially when different parties control the legislative and executive branches–and the division between the parties is close (as we have seen in recent years in Congress), the opportunities for gridlock multiply and neither party has any incentive to cooperate with each other. The basis of good government, then, lays with more trusting, less polarized states, where party organizations are weak and the political opportunities for grandstanding and blocking legislation are minimal. Such foundations are partly societal (trust and polarization) and partly structural (party organizations and the partisan balance of power in a state). Together they amount to a simple lesson: It is difficult to have good government when there is polarization and where the political environment provides few incentives for cooperation. Perhaps the most important part of the civil state is the level of trust. In such a state, high levels of people say that most people can be trusted (as opposed to agreeing that “you can’t be too careful in dealing with people”). Elsewhere I show that this question, which has been asked in a large number of national surveys since 1960, reflects a variety of trust that is not based upon
    
    Uslaner, The Civil State (9) experience. Instead, it is based upon a world view of optimism and control: The world is a good place, it is going to get better, and I can help make it better. People who trust others are more tolerant of minorities, more willing to give of themselves through charity and volunteering time, and see interactions with people of different backgrounds as opportunities rather than risks. Generalized trust is faith in people who are different from yourself (in contrast to faith only in your kind, particularized trust, see Uslaner, 2002, chs. 2-3). Generalized trusters look for common ground with people of different backgrounds and views–so trust is the font of cooperation and compromise and is the opposite of polarization. Divided societies–through ideological polarization and especially from economic inequality (Uslaner, 2002, chs. 6, 8)–rank low on trust. Trusting societies have good government because their citizens put aside differences to work for common purposes–not because they are active in politics or even in civic affairs. When we join civic groups, and especially in our informal social life, we connect with people with similar interests, backgrounds, and worldviews. Our political participation usually polarizes us rather than builds trust with our opponents (Rosenblum, 1998, 48; Uslaner, 2002, chs. 2, 4). Political action may be important to make government more responsive to citizens, but there is no clear reason to believe that it leads to more effective government. And there is even less reason to believe that participating in choral societies or bowling leagues translates into better government. Membership in civic groups and trust are not the same thing–and there is scant evidence that one leads to the other. It is trust in other people that matters for good government, not membership in civic groups or other participation (with one exception)–and not trust in
    
    Uslaner, The Civil State (10) government. So it is essential that we separate out the component parts of “social capital.” King, Zeckhauser, and Kim (2001) use Putnam’s (2000) state-level measure of social capital, which is a mixture of estimates of “trust,” civic engagement, turnout, and informal social connections such as visiting friends and entertaining people at home. This is a veritable dog’s breakfast (a bit of this, a bit of that), making it difficult to isolate what aspect of social capital might lead to more effective government. As I argue elsewhere (Uslaner, 2002, chs. 2, 4), there is little reason to expect–and no evidence to support–the argument that most forms of civic engagement and especially informal social connections might lead to more trust or better government. Knack (2002, 780) does separate out the component parts of social capital and finds little evidence that volunteering, informal socializing, or attending club meetings shapes any of the GPP measures of government effectiveness and that trust has a weak effect only on human resources management. Trust is a significant predictor of a summary measure of government effectiveness derived from adding the five components, but not of most individual measures. So what are we to make of these conflicting findings? We need not worry quite so much. Neither Knack nor King, Zeckhauser, and Kim actually test the impact of generalized trust. Their indicator of trust comes from the DDB Needham Life Style surveys, which have large samples that can be aggregated to the state level from 1975 to 1998. However, these surveys are not random samples nor do they ask directly about trust. Instead, the DDB Needham surveys asked whether people agreed that “most people are honest”–and honesty (using the same question) is only modestly correlated (tau-c = .345) with trust in the 1972 American National Election Study. We cannot say much about trust
    
    Uslaner, The Civil State (11) because we have not actually tested it. I have calculated state-level measures of generalized trust using a variety of surveys conducted over time and here use the estimates for the decade of the 1990s.3 While there may well be concern about deriving state-level estimates fromn national surveys such as the General Social Survey and the American National Election Studies–which are not designed to be representative of state populations–Brace, Butler, Arcenaux, and Johnson (2002) have shown that such estimates are very reliable. In Table 1 below, I present the estimated shares of trusters in each state in the 1990s used in this analysis. _________________ Table 1 about here Beyond, or instead of, trust, good government is driven by state size, percent AfricanAmerican, and especially economic inequality (Knack, 2002, 779) and by legislative professionalism, the number of good government groups, the business environment, and neighboring states’s rankings (King, Zeckhauser, and Kim, 2001, 30). I considered each of these variables, but all fell to insignificance in models focusing on trust, polarization, legislative ideology, party organizational strength, and political competition.4 An effective government is not one that is bickering all the time. And a confrontational political environment should not be conducive to quality government. The measures of quality are all based upon objective assessments of performance and a conflictual environment makes objectivity difficult to get. So I suggest that states with polarized publics or polarized legislatures should be less likely to have quality governments. My measure of public opinion polarization is the absolute difference between the share of Democrats in a state who are liberal and the proportion of Republicans who are conservative. These measures come from exit polls
    
    Uslaner, The Civil State (12) in the 1990s (see Erikson, Wright, and McIver, 1993) and were computed for me by Gerald C. Wright. Where most Democrats are liberals and most Republicans are conservative, there is less room for moderation or for compromise. The polarization measure is thus a variant on the wellknown measures of public opinion at the state level in Erikson et al.: Wright estimated the share of liberals, moderates, and conservatives for Democratic, Republican, and independent identifiers in each state. My measure of polarization is the absolute value of the share of Democratic identifiers calling themselves liberal minus the share of Republican identifiers calling themselves conservatives. My measure of elite polarization is more complex: It is based upon NOMINATE scores derived from roll call data from the American states for 1999-2000 also provided by Wright (see n. 4) . The measure I employ is the average for both legislative chambers of the absolute difference between the squared (quadratic) mean NOMINATE scores for Democrats and Republicans. I average these measures for each party for the two chambers (one in Nebraska). I then square the NOMINATE differences and then take the absolute difference between the parties. Why such a complex measure? Most spatial models of party competition employ squared distances: Quadratic measures make larger differences more “extreme” and smaller gaps less “extreme.” I believe that this formulation better captures the polarization in legislative politics than a simple difference of party positions. This is my measure of legislative polarization–and for both the mass and elite indicators, I expect negative relationships with government quality. I also employ the ideology of the majority party in the legislature from the same NOMINATE scores, with higher scores indicating greater liberalism. There are compelling
    
    Uslaner, The Civil State (13) arguments for hypotheses in either direction: Knack (2002, 774-775) argues that much of the existing literature on quality equates performance with innovation or liberalism. So we might expect that legislatures with more liberal majority parties would have higher levels of government quality. Alternatively, these specific measures of performance focus on management and in some cases a business-oriented approach to managing government (especially for human resource management, with its emphasis on merit pay and dismissing employees). This business orientation would support a negative relationship between legislative ideology (liberalism) and performance. Beyond these measures, I also expect that states with strong party organizations would be less likely to rank highly on these government performance measures. Strong party states, which have what Mayhew (1991) called “traditional party organizations,” place a heavy emphasis on patronage and on party control of the bureaucracy. “Scientific” management is an anathema to strong parties. I use Mayhew’s classification of party systems from weak to very powerful party organizations. Divided control of the legislative and executive branches of government may lead to less productive government. When one party controls the legislative branch and another the executive branch in the United States Congress, budget deficits spiral out of control (McCubbins, 1991). Overall, there seems to be little relationship between the enactment of major legislation and divided government (Mayhew, 1991; Binder, 1999), but there is a strong relationship between divided government and the ability of the executive to get what (s)he wants, especially in periods when the party system is highly polarized (Conley, 2003). Divided government may lead to more difficulties when the two major parties are relatively equal in strength. This makes
    
    Uslaner, The Civil State (14) it easier for the minority party to obstruct the agenda of the majority party. In contrast, single party government with an overwhelming majority for one party should have the least difficulty in enacting legislation. Preliminary estimations suggest that divided government per se does not shape the quality of government–however divided government interacted with the level of party competition in a state should be more likely to lead to less effective governments. Divided government might lead to greater compromise if each party is negotiating out of strength–as reflected in the dominance of one of the institutions. But where margins are tight, neither party wants to give any ground to the other. Each party might prefer standing on principle than letting the other party claim credit for government accomplishments. Parties will use these accomplishments in the next campaign and even a small gain in a tightly balanced environment might lead to a change in the balance in power. So divided government together with a competitive political environment should lead to less effective government. The five GPP measures are all indicators of a business-oriented approach to governing. This leads me to expect that polarization effects might not be as great as in the Ford Foundation/ Kennedy School Awards–which reward “innovation” rather than simply bureaucratic performance. Since innovation may reflect liberal rather than conservative ideologies (Knack, 2002, 774-775), the sign on legislative ideology may be positive rather than negative. Good Government: The Models Since these six models are not independent, I estimated the regressions using Zellner’s (1962) seemingly unrelated equation (SUR) technique.5 I report the SUR estimates in Table 2. ________________ Table 2 about here
    
    Uslaner, The Civil State (15) There is considerable, though not unconditional, support for the claim that civil states have better functioning governments in these regressions. Generalized trust is important for four of the six measures of state government performance. A low-trusting state that would otherwise receive a C on financial management would receive a B if it became the highest trusting state; for human resource management, the grade would increase to B+, for information technology to B+/A-, and for capital management to A-. Trust does not matter for managing for results or for receiving awards from the Ford Foundation and the Kennedy School of Government. Mass polarization does not seem to have deleterious effects on government performance. For human resources management, a more polarized environment seems to encourage higher rankings. The most polarized state would receive a ranking two grades higher than the state with the smallest partisan ideological division. The components of the human resources management index–including merit pay and terminating unproductive employees may make this dimension of management an ideological magnet. States where the majority party tilts to the right are also likely to rate highly on human resources management (see below). However, polarization does matter mightily for Ford Foundation/Kennedy School Awards. The most polarized state will receive five fewer awards than the state with the smallest partisan ideological division among the public.6 Where the party constituencies are further apart, state officials will perceive greater political pressures and will find it more difficult to innovate. While mass polarization does not seem to matter much, legislative polarization sends a very loud and clear symbol to the state bureaucracy: No factor matters as much as legislative polarization in shaping governmental effectiveness. An otherwise average state with the lowest level of elite division will receive a ranking two grades higher on financial management, capital
    
    Uslaner, The Civil State (16) management, and managing for results than a state with the highest level of conflict between legislative parties: An an otherwise C performance becomes an A grade when elite polarization is at a minimum. For human resources management, which seems to thrive on mass polarization, is more negatively effected by elite conflict: The least polarized state would receive an A, while the state with the most discord would get a low-end D. Perhaps most surprising is the powerful effect on information technology, where two states similar in all but elite conflict would receive an A and an F, respectively. Is there really so much conflict in computers? One might not think so, but perhaps parties might worry that the opposition would use technology to promote their own reelection: High-level technology will not only allow state officials to communicate with each other, but also with constituents. E-government may mean much more than renewing a fishing license online. It also allows legislators to reach out to their constituents and to promote themselves. Technology can thus become a key source of legislative conflict. Legislative polarization is not, surprisingly, connected to state innovation. The very strong impact of public discord may displace the impact of legislative conflict. It is not just polarization that matters a lot, but also the dominant ideology in a state’s legislature. On all five GPP measures, the more conservative a state is, the better functioning its government–which may not be so surprising, since the management criteria seem to be patterned on a business model. For innovation, however, a more liberal state majority party is important. The biggest impact for legislative ideology comes for information technology--an average grade change from C- to A as we move from the most liberal to the most conservative legislature. Here is further evidence for the polarization of technological innovation. Republican states seem to have made greater technological advances than northern states, so it is not surprising that IT
    
    Uslaner, The Civil State (17) management gets higher grades in states with conservative majority parties. Managing for results also is strongly affected by a conservative ideology–once again increasing from a C- to an A over the range of legislative ideology. Financial management is two grades higher where the majority party is the most conservative, while the human resources management grade is just over one rank higher in such states. Given the power of both mass and elite polarization for human resources management, there may be only a limited role for ideology itself on this dimension. The Ford Foundation/Kennedy School awards are higher when the governing state legislative party is more liberal. The most liberal state will receive an additional 3.5 accolades. There are also strong effects for three of the five GPP measures, but not for the Ford Foundation/Kennedy School innovation awards, for traditional party organizations. States with very powerful parties will have, on average, management scores two grades lower than those with weak parties for financial management, information technology management, and managing for results. When a state has divided government and close political competition, it will score two grades lower than a state with unified government on human resources management, information technology, and managing for results–and it will receive two fewer Ford Foundation/Kennedy School awards. The civil state does indeed seem to encourage more effective government. Across most of the measures, a trusting citizenry promotes better governance. So does a lower degree of polarization between the legislative parties–though partisan conflict among the public does not seem to matter as much–and, for human resource management, it even seems to promote higher rankings. Beyond greater trust and lower levels of elite polarization, strong parties and divided
    
    Uslaner, The Civil State (18) government with closely divided parties also lead to lower grades on effectiveness. The most effective government, then, seems to rest on a civil state that has high levels of generalized trust, lower amounts of partisan conflict among elites, and weaker party organizations. When one party has control of the legislative and executive branches, government is also more effective–especially in contrast to divided government with high levels of party competition. Better management depends upon a conservative majority party, but greater innovation reflects a liberal governing party. There is hardly uniform agreement on what constitutes good government, so it is not surprising that good “management” gets high scores in conservative states and “innovation” ranks more highly in liberal states. And since innovation may reflect policies more salient to the public, the greater link between Ford Foundation/Kennedy School awards to mass polarization makes sense. Since the bureaucracies respond directly to state legislatures, legislative polarization should matter more–and it does. The Civil State and Honest Government One criteria for good government that gains widespread support is honesty. Corrupt governments rob the public purse and take away funds that could be used for more productive purposes (Mauro, 1997, 7)–and corrupt officials look out more for themselves than for the public weal. Corruption leads to ineffective government (Mauro, 1997, 5; LaPorta et al., 1998, 32)–indeed, corruption is sometimes used as an indicator of ineffective government (Knack and Keefer, 1997). Across countries, corruption leads to higher tax evasion; lower growth; fewer expenditures for the public sector, education, or transfer payments from the rich to the poor; lower levels of government responsiveness, more bureaucratic red tape, and a less effective judiciary (Uslaner, 2004). Corruption is the scourge of good government.
    
    Uslaner, The Civil State (19) Corruption also depends upon a civil state. Countries with high levels of trust have low levels of corruption.7 So as with other measures of good government, I expect that states with high levels of trust will have less corruption. But beyond this measure of the civil state, there are fewer reasons to expect that ideological polarization should generate greater honesty. And it doesn’t. Instead, I argue that beyond a trusting public, we need a competitive polity with weaker party organization, vigilant elites, as well as greater equality among people. Corrupt regimes thrive on inequality. It transfers resources from the poor to the rich (Onishi and Banerjee, 2001)–and depends upon clientelistic relations between elites and masses. Where there is greater inequality, it is easier for leaders to exploit the public–and especially when there is great social and racial inequality, political leaders can pursue a divide and conquer strategy and spend more time enriching themselves than their constituents. The measure of racial inequality I use is the black/white poverty ratio in a state. Trust and inequality are intricately related: High levels of inequality lead to lower generalized trust.8 Both high levels of inequality and low generalized trust should contribute to a culture of corruption. A one party state, and especially a state with strong parties, will likely lead to greater corruption. In one party states, politicians don’t fear that someone is looking at their performance carefully–and they have little fear that they might be thrown out of office if they are found ethically wanting. A more vigilant citizenry can, in the words of former Virginia Lt. Governor Henry Howell, “keep the big boys honest.” I examined a variety of measures of political activity and include only one–the share of citizens in a state who make a political speech. This is a tiny fragment of the population, but it does point to an elite that is active in political affairs–that may have access to the media.9 Traditional party organizations are the
    
    Uslaner, The Civil State (20) homes of old-line political machines, sources of patronage and dishonesty. ________________ Table 3 about here I report the regression for perceptions of corruption in the states in Table 3 below. The corruption measure I use is the one that seems the most reliable: the reporters’ perceptions of the level of corruption (Boylan and Long, 2001). Each of the independent variables strongly shapes corruption levels in the 30 states for which there are complete data on all of the variables. The truncated sample does not seem unrepresentative of the 44 states in the earlier analyses, at least based upon the one variable common to all estimations, trust. The mean share of trusting respondents in the 30 states in this model is .359 compared to .383 for the 44 states in the earlier analyses.10 Additionally, the reporters’ corruption mean for the full 47 states for which data are available has a mean of 3.484 and a standard deviation of 1.038, while the estimated sample has a mean of 3.578 and a standard deviation of 1.114 (with higher scores indicating greater corruption on a scale from 1.5 to 5.5). The omitted states, then, are slightly more trusting and a bit less corrupt–but the differences do not seem great. High levels of trust and greater black/white inequality lead to lower levels of corruption–while we find more corruption in one party states and in states with very powerful political parties. The index of corruption ranges from 1.5 (least) to 5.5 (most). Moving from the least to the most trusting state, corruption increases by 1.17–a change equivalent to the distance between Pennsylvania and Rhode Island. As the black-white poverty ratio increases, corruption rises by 1.25, the gap between Utah and Rhode Island. The most dominant party system has a corruption ranking .862 higher than the least and strong party states are .98 “more corrupt”–an
    
    Uslaner, The Civil State (21) increase equivalent from moving from Ohio to Rhode Island. As the share of citizens reporting making a political speech rises from less than one percent to slightly more than 9 percent, perceived corruption falls by 1.3 percent, the difference between Maryland and Rhode Island.11 Reprise Honest government, like good government, rests upon a civic population. But polarization, either in the legislature or in the executive, doesn’t matter so much for honesty as it does for performance and innovation. Both good government and honest government do rest upon social capital–but not social capital as Putnam traditionally perceives it. Membership in civic organizations has no significant effects on good government or on honest government. Political participation, outside of the handful of people making political speeches in the corruption model, has no impact on any of the GPP measures, the Ford Foundation/Kennedy School awards, or corruption. There is also little evidence that institutional structures matter. Structural variables such as gubernatorial power or legislative professionalism (King, Zeckhauser, and Kim, 2001) fade to insignificance once trust and polarization are taken into account. Legislative professionalism does not even predict corruption (or the lack of it) when added to the model in Table 3. We might expect that Interest group strength in the states could lead to less effective government–since interest groups fight for their own interests against the common weal–or “good government” groups such as Common Cause might promote better government (King, Zeckhauser, and Kim, 2001). In my models above, the good and the bad effects of interest groups wash out in the face of trust, polarization, and competition. Corruption is rather sticky. Across 52 countries, the correlation of a corruption
    
    Uslaner, The Civil State (22) perceptions measure in 1980-83 by Business International (Monitola and Jackman, 2002) and the Transparency International indicator for 2003 is .867. There are no early measures for corruption in the American states, but it makes sense to expect an inverse relationship between present-day corruption and support for political reform in the past. A good measure of support for reform is the state-level vote for Robert LaFollette in 1924. In Figure 1 below, the LaFollette vote in 1924 does clearly track reporters’ perceptions of corruption in the late 1990s ( r2 = .284). The LaFollette vote is also highly correlated with trust in the 1990s ( r = .624) and moderately correlated with traditional party organizations ( r = -.263)–so the historical measure drops out of a multivariate estimation. But there is clear support for the link of corruption to a civic culture (and a civil state), reaching back to the 1920s. What you get out of government is what you put into it. You need a civic citizenry to get a civil state. Possibly, the argument might work the other way around–good government produces good citizens (Brehm and Rahn, 1997). But this seems less likely. Government management is out of sight for most people, so it seems unlikely that citizens learn to trust one another, and especially people who are different from themselves, by judging financial auditing or capital budgets. More generally, people use rather different criteria for judging government performance and deciding whether to trust others (Uslaner, 2002, ch. 5). Government performance and honesty are important and its link to generalized trust, inequality, polarization, and political competition suggest that reform is more than an exercise in political engineering. Some states are simply better candidates for good government than others. Structural tinkering, such as a more professionalized legislature, may be in George Bernard Shaw’s message about second marriages, may represent “the triumph of hope over experience.”
    
    Uslaner, The Civil State (23) TABLE 1 Trust Estimates by State for the 1990s
    Alabam a Alaska Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Illinois Indiana Iowa Kansas Kentucky Louisiana Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana New Ham pshire New Jersey New York North Carolina North Dakota Ohio Oklahom a Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Verm ont Virginia W ashington W est Virginia W isconsin W yom ing .162 .304 .413 .105 .384 .413 .440 .250 .351 .447 .400 .398 .412 .488 .296 .293 .447 .448 .448 .513 .188 .380 .500 .630 .305 .362 .223 .594 .333 .290 .453 .436 .316 .259 .516 .266 .279 .560 .558 .382 .413 .263 .521 .417
    
    Uslaner, The Civil State (24) TABLE 2 Seemingly Unrelated Regressions for Government Quality Measures
    
    Independent Variable Generalized Trust Mass Polarization Legislative Polarization Legislative Ideology Traditional Party Organization Divided Government* Competition Constant R2 Root Mean Square Error
    
    Financial Management
    
    Human Resources Management 5.486*** (2.46) 5.908+ (2.527) -5.851*** (1.987) -.819** (.427) -.146 (.180) -4.974*** (1.918) 6.712**** (1.302) .436 1.451
    
    Information Technology Management 6.543*** (2.82) 1.907 (2.629) -7.500**** (-3.63) -1.455**** (.444) -.500*** (.187) -5.204*** (1.995) 8.220**** (1.354) .556 1.509
    
    Capital Management
    
    Managing for Results
    
    Ford Foundation/ Kennedy Awards -1.068 (3.912) -16.843**** (4.432) -2.741 (3.484) 2.124 (.748) .242 (.316) -5.274* (3.363) 7.672*** (2.283) .378 2.544
    
    4.298** (1.67) 3.573 (2.912) -3.422* (2.291) -1.157** (.493) -.452** (.208) -1.019 (2.212) 8.858**** (1.501) .284 1.674
    
    6.986** (2.28) 1.541 (.44) -3.584* (2.734) -.963* (.587) -.309 (.248) -3.294 (2.639) 7.623*** (1.791) .258 1.996
    
    .879 (3.036) -3.988 (3.440) -3.624* (2.704) -1.532*** (.581) -.480** (.245) -4.336** (2.610) 10.785**** (6.09) .307 1.975
    
    Entries are regression coefficients, standard errors under coefficients in parentheses. + Significant in opposite direction from hypothesized. All tests one-tailed except for constants. N = 44
    
    **** p < .0001 *** p < .01 ** p < .05 * p < .10
    
    Uslaner, The Civil State (25) TABLE 3 Model of Corruption Perceptions in the American States
    
    Independent Variable Generalized trust Dominant party share Traditional party organization Black/white poverty ratio M ake political speech Constant
    
    Coefficient -2.578*** 3.850** .232***
    
    Standard Error 1.080 1.751 .079
    
    t Ratio -2.39 2.20 2.98
    
    .435***
    
    1.634
    
    2.66
    
    -15.063*** 2.900***
    
    6.195 .728
    
    -2.43 3.98
    
    R2 = .733 Adjusted R2 = .678
    
    RMSE = .589 N = 30
    
    *** p < .01 ** p < .05
    
    Uslaner, The Civil State (26) FIGURE 1 The Historical Roots of Corruption in the American States
    
    Uslaner, The Civil State (27) REFERENCES Almond, Gabriel and Sidney Verba. 1963. The Civic Culture. Princeton: Princeton University Press. Altschuler, Alan A. 1997. “Public Innovation and Political Incentives.” Unpublished paper, John F. Kennedy School of Government, Harvard University, at http://www.innovations.harvard.edu/research/papers/altshuler.pdf, (March 4, 2004). Binder, Sarah A. 1999. “The Dynamics of American Gridlock, 1947-96.” American Political Science Review 93 (September): 519-34. Boylan, Richard T. And Cheryl X. Long. 2001. “A Survey of State House Reporters’ Perception of Public Corruption.” Unpublished manuscript, Department of Economics, Washington University in St. Louis. Brehm, John and Wendy Rahn. 1997. "Individual Level Evidence for the Causes and Consequences of Social Capital." American Journal of Political Science 41 (July): 8881023. Brace, Paul, Kellie Sims-Butler, Kevin Arcenaux, and Martin Johnson. 2002. “Public Opinion in the American States,: New Perspectives Using National Survey Data.” American Journal of Political Science 46 (January): 173-89. Conley, Richard S. 2003. The Presidency, Congress, and Divided Government. College Station, TX: Texas A&M Press. Erikson, Robert S., Gerald C. Wright, and John P. McIver. 1993. Statehouse Democracy. New York: Cambridge University Press.
    
    Uslaner, The Civil State (28) Inglehart, Ronald. 1999. “Trust, Well-Being and Democracy.” In Democracy and Trust. ed. Mark Warren. Cambridge: Cambridge University Press. Kaufmann, Daniel, Aart Kray, and Massimo Mastruzzi. 2003. “Governance Matters: Governance Indicators for 1996-2002.” Washington, DC: The World Bank, at http://www.worldbank.org/wbi/governance/pdf/govmatters3.pdf. (February 24, 2004). King, David C., Richard J. Zeckhauser, and Mark T. Kim. 2001. “Explaining the Management Performance of U.S. States.” Unpublished manuscript, John F. Kennedy School of Government, Harvard University. Knack, Stephen. 2002. “Social Capital and the Quality of Government: Evidence from the States.,” American Journal of Political Science 46 (October):772-87. Knack, Stephen and Philip Keefer. 1997. “Does Social Capital Have An Economic Payoff? A Cross-Country Investigation.” Quarterly Journal of Economics 112 (November): 125188. LaPorta, Rafael, Florencio Lopez-Silanes, Andrei Schleifer, and Robert W. Vishney. 1997. “Trust in Large Organizations,” American Economic Review Papers and Proceedings 87 (May): 333-38. ___________________1999. “The Quality of Government.” Journal of Law, Economics, and Organization 15 (March): 222-79. Lane, Robert E. 1959. Political Life. New York: Free Press. Mayhew, David R. 1991. Placing Parties in American Politics. Princeton: Princeton University Press.
    
    Uslaner, The Civil State (29) Mauro, Paolo 1997. “Why Worry About Corruption?” Washington: International Monetary Fund. McCubbins, Mathew D. 1991. "Government on Law-Away: Federal Spending and Deficits Under Divided Party Control." In The Politics of Divided Government, eds., Gary W. Cox and Samuel Kernell. Boulder, CO: Westview. Meier, Kenneth J. and Thomas M. Holbrook. 1992. “‘I Seen My Opportunities and I Took’Em:’ Political Corruption in the United States.” Journal of Politics 54 (February): 135-55. Montinola, Gabriella R. and Robert W. Jackman. 2002. “Sources of Corruption: A CrossCountry Study.” British Journal of Political Science 32 (January): 117-70. Onishi, Norimitsu with Neela Banerjee 2001, Chad’s Wait for Its Oil Riches May Be Long, New York Times (May 16), http://www.nytimes.com/2001/05/16/world/16CHAD.html, (May 17, 2001). Putnam, Robert D. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton: Princeton University Press. ________________. 2000. Bowling Alone. New York: Simon and Schuster. Rosenblum, Nancy L. 1998. Membership and Morals. Princeton: Princeton University Press. Sparrow, Malcolm K. 2000. The Regulatory Craft. Washington, D.C.: Brookings. Uslaner, Eric M. 2002. The Moral Foundations of Trust. New York: Cambridge University Press. ____________. 2004. “Trust and Corruption.” In Corruption and the New Institutional Economics, eds. Johann Graf Lambsdorf, Markus Taube, and Matthias Schramm. London: Routledge.
    
    Uslaner, The Civil State (30) Uslaner, Eric M. and M. Mitchell Brown. 2005. “Trust, Inequality, and Civic Engagement.” American Politics Research 33. Zellner, Arnold. 1962. “Estimators of Seemingly Unrelated Regressions: Some Exact Finite Sample Results,” Journal of the American Statistical Association 58 (June): 977-92.
    
    Uslaner, The Civil State (31) NOTES * This research was supported by a grant from the Russell Sage Foundation and the Carnegie Corporation under the Social Dimensions of Inequality Project. Some of the data come from the Inter-University Consortium for Political and Social Research, which is not responsible for any of our interpretations. I am also grateful to the General Research Board, University of Maryland—College Park, for support on related projects and to M. Mitchell Brown for research assistance, to David C. King for both providing data and important bibliographic help, and to an anonymous reviewer for Stanford University Press–and the editor, Jeffrey Cohen. 1. 2. 3. For more on these measures, see http://www.transparency.org. These data have been updated by Richard F. Winters, who kindly provided them to me. See Uslaner and Brown (2005) for a discussion of the data and a more comprehensive list of the pre-1990 surveys we used for generating trust for other decades. Fengshi Wu helped with the data aggregation at the beginning of the project and M. Mitchell Brown did most of the work.. We used the following surveys for generating the trust estimates: General Social Survey or GSS (1990, 1991, 1993, 1994, 1996, and 1998), American National Election Study (1992, 1996, and 1998), the Washington Post trust in government survey (1995), the Pew Civic Engagement survey (1997), the New York Times Millennium survey (1999). We are grateful to Robert Putnam and John Robinson for making the state codes for the GSS available to us and to Richard Morin of the Washington Post for the 1995 survey and to Michael Kagay of the New York Times for his paper’s Millennium survey.
    
    Uslaner, The Civil State (32) 4. The GPP and Ford Foundation/Kennedy School measures as well as measures of party competition were kindly provided by David C. King. The divided government variables come from the 1994, 1996, and 1998 editions of the Book of the States (computed by myself). The data on public opinion and legislative polarization were graciously provided by Gerald C. Wright and the data on racial inequality in poverty (below) came from Rodney Hero. Arthur Jones, Jr. of the United States Bureau of the Census for provided the state level Gini indices (also see below). 5. Since the models are identical, the coefficients will be identical to those of ordinary least squares (OLS). However, the standard errors will be incorrectly estimated using OLS. The correlations among the residuals range from moderate (.273) to high (.651), except for the Ford Foundation/Kennedy School awards measure (ranging from .058 to .293). The Breusch-Pagan test of independence has Chi square= 91.433, p < .00001, so OLS would be less efficient. 6. 7. The overall range of the Ford Foundation/Kennedy School awards is from 0 to 15. And countries with low levels of corruption have high levels of trust. Uslaner (2004) sorts out the causal order by looking at changes in trust and corruption over time and finds that trust comes first in the causal order. 8. 9. Both over time in the United States and cross-nationally. See Uslaner (2002, chs. 6, 8). The measure comes from the Roper Social and Political Trends Archive; the Roper survey asked national samples of 2000 ten times a year for 21 years (1974-94) about levels of political activity. Hence there are sufficient cases to aggregate activities for the states by decade. Here I use the estimates for making political speeches in the 1990s.
    
    Uslaner, The Civil State (33) 10. The standard deviations, respectively, are .110 and .117. The 14 states excluded from this model are: Alaska, Arizona, Kansas, Kentucky, Massachusetts, Montana, New Hampshire, New Jersey, North Dakota, Oklahoma, Rhode Island, Vermont, Wisconsin, and Wyoming. 11. As a long-time resident of Maryland, this seems less consequential than the data suggest.

Add Comment

To add a comment, you need to login or signup (you can signup without leaving this page)