Comparative Political Studies Volume 41 Number 4/5 April/May 2008 638-673 © 2008 Sage Publications 10.1177/0010414007313118 Beyond Clientelism http://cps.sagepub.com hosted at Incumbent State Capture http://online.sagepub.com and State Formation Anna Grzymala-Busse University of Michigan, Ann Arbor

In choosing strategies of state capture (the extraction of private benefits by incumbent officeholders from the state), rulers choose whether to share rents with popular constituencies and whether to tolerate competition. These choices are conditioned by existing organizational endowments, the costs of buying support, and the trade-off between the cost and probability of exit from office. In turn, both rent distribution and competition result in distinct configurations of state capture: clientelism, predation, fusion, exploitation, and the formation of specific state institutions and capacities.

Keywords: state capture strategies; rent distribution; competition

ncumbent elites capture private benefits from the public offices they hold. IBoth policy makers and scholars have focused on the corrosive effects of these practices (Ades & Di Tella, 1999; Keefer, 2002; Knack & Keefer, 1995; Persson, Tabellini, & Trebbi, 2001). Yet such extraction of state assets also forms state institutions and capacities. In choosing strategies of state capture, elites face two fundamental considerations. First, they must decide whether to share rents with potential constituencies in exchange for their support. Second, they must decide whether to allow competition, because contestation affects the levels of rent seeking. These two choices result in four distinct strategies of state capture, and the building of state institutions that further perpetuate the particular forms of capture. The most familiar examples of state capture are clientelism and preda- tion (Chandra, 2004; Ichino, 2006; Kitschelt, 2000; Kitschelt & Wilkinson, 2006; Magaloni, 2006; Piattoni, 2001; Robinson & Verdier, 2002; Scheiner, 2006; Stokes, 2005). Clientelism consists of the contingent and targeted

Author’s Note: I would like to thank Jim Caporaso, Cristina Corduneanu-Huci, Jonathan Hartlyn, Herbert Kitschelt, Lenka Siroka, Dan Slater, Erik Wibbels, and Steven Wilkinson, and the participants of the “Frontiers of Comparative Politics” Conference held at Duke University, April 2007, for their very helpful comments. 638 Grzymala-Busse / Beyond Clientelism 639 distribution of selective goods to supporters in exchange for their loyalty. These exchanges give clientelist organizations a strong advantage in politi- cal competition.1 In contrast, predatory rulers extract resources without sys- tematic targeting or delivering of goods to citizens, and opposition—much less democratic competition—is not tolerated. This dichotomy reflects a more basic proposition frequently found in the literature—that democracy itself is a redistributive regime, whereas authoritarian regimes tend to redis- tribute far less (Acemoglu & Robinson, 2006). Yet if clientelism and predation are distinguished by the degree of rent dis- tribution (sharing benefits with supporters) and their compatibility with com- petition, two other logical possibilities exist, as summarized in Table 1: redistribution without democratic competition and the capture of state assets under competitive conditions. As we will see, party-state fusion regimes heavily distribute rents in the absence of democratic competition, whereas exploitation abnegates redistribution despite a commitment to competition. In short, we cannot read political regimes off distributive practices. All these strategies involve the formation of distinct state institutions and capacities. State seizure does not simply corrode the state. Although extrac- tive rulers seek to maximize their discretion by weakening regulation and oversight, they also construct rules and durable practices of redistribution, budgeting, and authority. It is not simply the case that “clientelism thrives when government institutions are weak” (Manzetti & Wilson, 2007, p. 955), but rather, that specific institutions are built to serve the extractive goals of rulers, sometimes with unintended consequences (Tilly, 1992). For example, fusion strategies substitute party structures for the state’s, becoming the key agents of administration, distribution, and regulation and “hollowing out” of state institutions as the party takes over administrative roles. Clientelist rulers expand the structures of the welfare state. Predatory rulers deliberately weaken state institutions and in the process increase their own costs of exit. Exploitative political parties opportunistically reconstruct independent state institutions that offer access for the capture of state assets. This article first examines the different configurations of rent distribu- tion and competition in extractive regimes followed by an analysis of how and why rent distribution and competition arise. The article concludes by showing the impact on state capture and formation.

Characteristics of Extractive Regimes

I assume that in all cases of state capture (the elite extraction of state resources for private gain), rulers have direct access to state resources and the 640 Comparative Political Studies

Table 1 Party Strategies of Extracting Resources from the State

Compatible with Competition Incompatible with Competition

Rent distribution Clientelism Fusion of party and state No rent distribution Exploitation Predation

state is porous enough to allow such access (i.e., it is not the archetype of the Weberian legal–rational state). The rulers can be individuals, oligarchies, fac- tions, or parties (Levi, 1988). In many of the cases examined here, they are at least nominally political parties. Rulers are both opportunistic and risk averse, using all feasible and available means at their disposal to maintain power and lower their expected costs of exit. To maximize their returns, they can vary their targets, diversify their portfolio of extractive techniques, and try to alter their institutional environments (Levi, 1988; North & Weingast, 1989). The state is the set of formal institutions that administers citizen obliga- tions (e.g., taxes, military service), enforces legal sanctions, and regulates public provisions (e.g., infrastructure, rule of law, welfare, defense, etc.). It comprises both public finances and the channels of their distribution. These institutions provide the resources that generate popular support or compli- ance for rulers: housing, jobs, education, government services, and other provisions. Because the state also enforces property rights, the institutions of contract enforcement, property protection, oversight, and regulation can aid state capture directly and by punishing potential opponents. Elite state capture is the appropriation of state resources by political actors for their own ends: either private or political benefit. State capture was first conceptualized as “manipulating policy formation and even shaping the emerging rules of the game to their own, very substantial advantage” by economic agents (Hellman & Kauffman, 2001). In this analysis, the focus is on agents within the state. Extraction by the state, or the capture of resources held by society, is critical in building costly state institutions (Levi, 1988; North & Weingast, 1989; Tilly, 1992). It can result in a mutual contract between the rulers and society; taxes and military service are exchanged for security and the provision of public goods, such as education and infrastructure. Extraction from the state, in contrast, is the capture of resources that have already been accumulated by the state (Mann, 1988). Although some rulers will extract purely for themselves, others will obtain state resources for the sake of their political party or organization and may choose to share some of these gains with their supporters. State capture refers to this elite extraction from the state. Grzymala-Busse / Beyond Clientelism 641

The distribution of these gains and competition for office are two major considerations for potential rulers. The provision of public and private goods is a fundamental link to supporters. Similarly, competition for office and power is a fundamental relationship with other potential ruling elites. As a result, in the process of obtaining and maintaining power, rulers face two dilemmas. First, distribution of selective goods generates loyalty and compliance among supporters—but it cuts into the private gains the ruler can keep. Second, institutionalized competition decreases the costs of maintaining power and of exit from office—but it increases the probability of exit. Distribution and competition, in turn, rest on basic organizational investments and democratic commitments of rulers, without which neither distribution nor competition can succeed (Grzymala-Busse, 2007). The cases examined in this article vary on both dimensions. First, Table 2 shows the rankings for political rights and civil liberties. These are imper- fect and rough indicators, but they show that the potential for competition was far higher in the exploitative cases than in either fusion or predation, which scored very poorly on these dimensions. Clientelist cases include high scorers (Italy and Japan) and a relatively low scorer (Mexico). Second, the cases differ in the degree of rent sharing observed. Although comparable measures of such distribution are notoriously difficult to obtain, the scholarly consensus is that we see relatively little rent distribution in Nigeria, Indonesia, Philippines, or the postcommunist East-Central European cases. In contrast, extensive targeted and contingent rent distribution pre- vailed in Italy, Japan, Mexico, Russia, Singapore, and communist East-Central Europe, with a larger proportion of state employment and greater redistribu- tive spending and allocation of resources, such as housing or development programs (Chandra, 2004; Grzymala-Busse, 2007; Hicken, 2006; Magaloni, 2006; Scheiner, 2006; World Bank, 1997). At the same time, these differences are not a simple function of the levels of economic development: Mexico’s GDP per capita is lower than Italy’s or Japan’s, and yet all share clientelist features. The GDP of Singapore is far higher than that of either Russia or the communist East-Central Europe, and yet all are cases of party–state fusion. These patterns suggest that economic development alone is not responsible for generating the variation in state capture strategies. Different configurations of rent distribution and contestation produce distinct state capture strategies. Clientelism combines distribution and contestation, exchanging supporter loyalty for rent sharing, as targeted and contingent goods are delivered to select constituencies and individuals. The sharing of the spoils of office with constituents lowers the costs of staying in office and the threat of competition, even if it cuts into the profits 642 Comparative Political Studies

Table 2 Competition and Rent Distribution

Freedom House Scores

Political Civil Rent Country Rights Liberties Distribution

Italy, 1946-1994 1 2 High Japan, 1955-present 1 2 High Mexico, 1929-1999 4 4 High Russia under Putin, 2000-present 5 5 High Singapore, 1959-present 5 5 High Communist East-Central Europe, 1948-1989 6-7 6 High Nigeria under Abacha, 1993-1998 7 6 Low Indonesia under Suharto, 1967-1998 7 6 Medium, then low Philippines under Marcos, 1966-1986 5 5 Low Poland, 1989-present 1 2 Low Czech Republic, 1989-present 1 2 Low Slovakia, 1989-present 1 2 Low

Source: Freedom House, 1970-2006. Scale is 1-7, with 1 as high level of freedom/rights incumbents could otherwise gain from office in the short run. Examples of widely known clientelist regimes include Mexico, Italy, and Japan, as well as Austria and India. Their competitive configurations range from an authoritarian one-party hegemony (Institutional Revolutionary Party [PRI] in Mexico until the late 1990s) to a democratic two-party duopoly (Austria until the 1990s) and a volatile democracy dominated by one winning party (Italy until 1994).2 Clientelism targets spending and regulatory provisions to supporters to gen- erate sufficient support to stay in power without providing public goods to all. Through stable and repeated interactions with supporters, rulers credibly com- mit to providing selective goods contingent on support (Stokes, 2005). It is a familiar and widespread strategy: thus “virtually all electorally successful par- ties in Latin America, even the more ideological ones, have learned to cul- tivate clientelistic ties at the grassroots” (Coppedge, 2001, p. 176). And it provides a self-enforcing solution to the twin problems of ensuring adminis- trative loyalty and popular support (Crenson & Ginsberg, 2004). Distribution without contestation is exemplified by the fusion of party and state. To do this, rulers replicate and politicize state structures, subor- dinating them to incumbent representatives at every level. The ruling cohort distributes rents contingent on societal acquiescence and lowers its proba- bility of exit by resolutely eliminating the opposition. Fusion precludes Grzymala-Busse / Beyond Clientelism 643 competition through the repression of potential competitors, either de iure (the “leading role of the party”) or de facto (high entry thresholds, financ- ing and registration requirements).3 Examples of fusion include the com- munist states of the Soviet bloc and Singapore. As one analyst observed in Singapore, “PAP [People’s Action Party] has set up a . . . system of party cells at the grassroots level and combined it with a direct role in the admin- istration of the electoral wards and city districts. PAP and government, in this sense, are more or less one and the same” (Sachsenröder, 1998, p. 19). Elections are held, but “the ruling party obviously uses all available legal and organizational means to make it less easy for the few opposition politi- cians” (Sachsenröder, 1998, p. 11). As a result, the ruling PAP holds all but 4 to 6 seats out of 81 in the Parliament. Another example is post-2000 Russia, where the cohort around Vladimir Putin has effectively taken over state structures to produce “a merger of state budget and party organiza- tion” (Smyth, Wilkening, & Urasova, 2007, p. 9). In the absence of contestation, organizational investments are made in both the state and the party, but the goal is to eliminate the opposition as much as to ensure popular acquiescence. In both Singapore and Russia, for example, the rule of law was used selectively to eliminate rivals. Communist parties developed dense organizational networks that reach down to the level of local governments, workplaces, and residential units. These party organizations not only delivered (or withheld) welfare state services such as housing, education, vacations, and health care to individual workers (Fainsod, 1958; Hirszowicz, 1980; Kaplan, 1987, 1993), but they also directly moni- tored political opinion, named officials, controlled economic production, and subordinated government agencies. Institutional exploitation is one result of political competition without distribution of selective goods. It consists of deliberately building formal state institutions that allow direct extraction of benefits and perpetuate existing informal practices of rent seeking. Examples of exploitation include Slovakia until 1999, and the Czech Republic and Poland after 2004 (though even more severe cases are found in Latvia and Bulgaria). These rulers were committed to elections and to democracy: Having just emerged from decades of authoritarian rule, which had eliminated potential com- petitors, new postcommunist parties had little interest in pursuing strategies that could easily turn against them and give rise to another one-party hege- mony. Such commitments were further reinforced by the firm stance of the European Union (EU) that only democratic countries could enter the EU. At the same time, these rulers did not enter into distributive contracts with their constituents. Voters were relatively expensive to buy off, given 644 Comparative Political Studies

the relatively high economic development and income equality. Furthermore, parties had no capacity to develop the channels of monitoring and delivery. Members4 and organizations5 were both sparse, and the dominance of national media campaigns rendered party organizations superfluous. Finally, party instability made it difficult for voters and parties to engage in particu- laristic contracts: High volatility and fragmentation made party disappear- ance all too likely. Offering selective goods to supporters was both inefficient and implausible. In the absence of distribution, however, exploitative parties could not count on a loyal and dependent electorate, making exploitation very sensitive to changes in competition. As a result, the high levels of exploita- tion in Slovakia dropped significantly after 1998, when a reformist coalition swept the dominant party from office. Conversely, states with relatively low levels of exploitation became far more open to it once competition became more disorganized and muted in its criticism, as in Poland after 2004. With neither distribution nor contestation to constrain capture, extractive elites have the discretion to pursue capture directly, without making orga- nizational investments in welfare state delivery and monitoring channels or protecting contestation. Thus, in elite predation (kleptocracy or predatory rule; see Hutchcroft, 1998, for important distinctions among predatory regimes), rulers steal government funds and expropriate both state property and private assets (especially those of potential opponents). Distributive contracts between rulers and supporters do not arise, because rulers have not invested in requisite organizational linkages and voters are not as nec- essary to maintaining office as elite allies or armed forces. Classic exam- ples of countries that experienced predation include the Philippines, Nigeria, and Indonesia, where Ferdinand Marcos (1965-1986 in office), General Sani Abacha (1993-1998), and Mohamed Suharto (1967-1998) each stole billions USD while in office (Goldsmith, 2004; Transparency International, 2006). Competition is severely constrained by repression, electoral fraud, and anticompetitive institutions, and institutional invest- ments reflect the dominance of repression over redistribution. Such regimes “tried to eliminate, weaken, or take over any nongovernmental institution that might contest their legitimacy and authority” (Bratton and van de Walle, p. 72). Parties and factions that do not gain control of central exec- utive office find themselves starved of resources necessary for survival. As a result, rulers have greater opportunity to “ma[k]e little distinction between the public and private coffers, routinely and extensively dipping into the state treasury for their own political needs” (Bratton & van de Walle, 1997, p. 66). They have considerable discretion to shift strategies and rates of capture: For example, initial years of Suharto’s rule in Indonesia Grzymala-Busse / Beyond Clientelism 645

saw credible efforts to attract investment, as a way to thwart the communist threat and shore up his own rule. To that end, Suharto informally ensured the property rights of the Chinese minority entrepreneurs and maintained capital mobility throughout his early rule (see Lewis, 2007). Once the communist collapse made the domestic communist threat far less viable and the regime’s sphere of control was firmly established, however, Suharto sought chiefly to protect his financial interests. Special preferences, arbitrary decisions, and playing (and paying) off the army against Golkar, a corporatist civilian orga- nization, by Suharto all combined to accelerate the depredation of the economy, setting the stage for the country’s financial collapse in 1998. These examples serve as useful archetypes, but clear-cut examples of a single strategy are relatively rare. Rulers diversify and “catch-as-catch-can,” and these strategies are compatible with each other to an extent. Clientelism is especially adaptable: Thus, predation or exploitation can be augmented with clientelist distribution, as rulers buy off the support of potentially trou- blesome factions or individuals without systematically relying on such pro- visions to stay in office. It is also possible to use one strategy on the national level and another on the local, subnational authoritarian enclaves, or local exchanges that support national-level predation (Ichino, 2006). Third, these strategies can change over time, as the relative costs and benefits of distribu- tion and competition change. Finally, they can vary across state sectors: For example, the Singaporean PAP monopolizes the politics of redistribution but allows free markets to flourish. And lucrative state sectors (e.g., customs, infrastructure) are more tempting targets for capture.

The Origins of Rent Sharing and Contestation

How, then, do these configurations of rent sharing and contestation arise? What actor rationales and structural constraints underlie rent distrib- ution and competition? These choices are conditioned by existing organi- zational endowments, the costs of buying support, and the trade-off between costs and probability of exit from office.

Rent Sharing and Distribution One fundamental question for rulers is whether to share the rents obtained with supporters. The key distinction is whether to share with non- officeholders: The motivation to do so is to buy widespread popular support, whereas rent sharing with fellow officeholders buys the acquiescence of 646 Comparative Political Studies

direct and potentially powerful rivals. Rent sharing with constituents cuts into the share of gains the incumbents can keep, but the dependency on selective good provisioning reduces demand elasticity for rulers.6 It makes sense when (a) it is cost effective, (b) supporters are available and sensitive to the provision of such goods, and (c) the contracts between supporters and rulers are credible. The less these conditions hold, the more it makes sense for incumbents to extract for themselves, without sharing rents with non- officeholders. Poverty and income inequality increase the pay-offs of redistributive strategies (Stokes, 2005). Low levels of economic development and inequality mean that potential supporters can be bought off relatively cheaply (com- pared to buying off wealthier sectors of society). Poor voters are cheap and effective targets for clientelist rulers (Calvo & Murillo, 2004; Magaloni, 2006; Stokes, 2005).7 Moreover, the poor are less likely to challenge incumbent capture than the wealthy are. And under lower levels of economic develop- ment, the economy is not closely tied to the development of capitalism and less able to support public good provision or property right enforcement (Khan, 2005). Weak feedback from the capitalist sector gives considerable leeway to extract and redistribute, because there is less pressure for the provision of public goods. Potential supporters become more dependent on such rent distribution when they have few other options, thanks to low labor mobility, underpro- visioning of public goods such as education, and inability to enter into prof- itable economic activity. Moreover, “because of poor people’s limited physical mobility and clustered patterns of residence, politicians can also monitor the adherence of the poor to clientelist deals better than that of affluent individuals” (Kitschelt, 2000, p. 857). Where a sizable middle class exists, clientelist strategies are costlier, because middle-class voters are expensive to buy off and less likely to agree to bear the costs of rent distri- bution if it does not provide public goods (Kitschelt & Wilkinson, 2006; Robinson & Verdier, 2002). Rent distribution requires organizational investment. It relies on institu- tional channels of monitoring and benefit delivery (Grzymala-Busse, 2007). Potential supporters and rulers need to enter into credible exchange contracts, and these are monitored and delivered by state and party organizations. These organizational investments take a variety of forms: members, local activists and brokers, and affiliated organizations (see Levitsky 2003; Stokes, 2005). Members themselves are less important than the party’s ability to reach individuals through the organizational networks they command.8 Thus, the Japanese Liberal Democratic Party (LDP) developed a very dense network Grzymala-Busse / Beyond Clientelism 647 of affiliated local organizations (koenkai) in each electoral district that delivered services and mobilized LDP support (Richardson, 2001; Scheiner, 2006). And the Peronist vote in Argentina is higher and more stable in areas with dense party organizations and extensive public employment (Calvo & Gibson, 2001; Calvo & Murillo, 2004). Where rulers do not have such orga- nizational resources at their disposal, distribution of selective goods is not sustainable and rulers are more likely to turn to either predation or exploita- tion (Grzymala-Busse, 2007). Distributive networks are both costly to establish and to maintain, so rulers will not build them if they anticipate an unstable political environ- ment. External guarantees and domestic relationships can both serve as insurance: The Soviet support for the communist project in East-Central Europe, or existing patron–client ties in Latin America, made the develop- ment of the organizational networks and state institutions tenable. Thus, parties require stability to distribute rents and, subsequently, distribution breeds the dependence that may ensure stability. This cycle may be one rea- son why one-party authoritarian regimes are more durable than military or personalist ones (Geddes, 1999). Even under fully free competition and universal suffrage, the distribution of rents to supporters makes little sense where few party–constituency link- ages exist: When voters face a different menu of political parties at each elec- tion, under conditions of high electoral volatility, or when political parties are unable to identify loyal voters, rent distribution helps to ensure such loyalty but does little to identify its sources initially. High rates of party fragmenta- tion and voter volatility all make organizational investments needed for rent sharing costly and the pay-offs uncertain. Rulers then extract state assets and maintain competition, without redistributing to constituents: They can face electoral punishment if their rates of capture are too high (and become public), but they will not be punished by voters for not sharing rents, because there is no voter–ruler contract that exchanges support for selective goods. This scenario is most likely in new democracies, where parties arise in the modern era of mass media rather than mass mobilization and organizational investments, and where voter loyalties and party identities are fluid. At the same time, distribution acts as a buffer against competition: The dependence generated by selective good provision lowers the probability of exit. First, defection is costly for voters, placing would-be competitors at a disadvantage (unless they can either offer higher levels of goods or the elec- torate’s demand for such goods drops). As Medina and Stokes (2007) argue, clientelist incumbents can establish credible threats against clients who may defect to a challenger, thus reducing the potential opposition. As a result, even 648 Comparative Political Studies if voters discover corruption or elite rent-seeking, they may have little reason to punish the incumbents if the latter distribute rents (see Adserá, Boix, & Payne, 2003). Second, even if there is attrition among supporters, the exten- sive organization of supporters and activists acts as a buffer, so that a party may survive an electoral downturn more readily than a party with no infra- structure. Moreover, incumbents can pre-empt rivals by increasing their rates of distribution. If there is no competition, distributing selective goods makes monopoly rule less costly: Rulers can threaten to withdraw housing, educa- tion, and jobs from potential defectors and their families, a well-known penalty that dissuades potential opponents (see Smyth et al., 2007).

Competition Competition or contestation consists of the capacity to contest the con- duct of government and to present distinct alternatives to the incumbent (Dahl, 1971). The more intense the competition, the more the opposition can be clearly distinguished from the incumbent, and its criticisms and pol- icy alternatives are credible. If voters object to rent seeking (e.g., they receive no benefits from it) and they are responsive (they are not fully dependent on the incumbent), then competition lowers incumbent rent seeking. It does so by increasing the threat of exit from office for the incum- bent (as a result of electoral backlash) and by increasing the rate of turnover (limiting any one incumbent’s opportunity to obtain rents). For incumbents, competition increases the likelihood of exit from office and lowers the portion of private gains rulers can keep. If institutionalized (mechanisms for succession are clear and losers are guaranteed safety and continued ability to compete), it also lowers the costs of exit, ensuring that losers survive and have the opportunity to re-enter office. Given the legit- imacy it can confer, it makes rule less expensive as well. However, where there are no clear mechanisms for succession nor guarantees (either domestic or international) for the losers of their safety and continued abil- ity to compete, competition can increase the rate of capture as a result. For example, in Nigeria, “incumbency has become everything . . . to lose office is to lose almost the only means of survival, as well as immunity from prosecution” (“Big men,” 2007). There is a vast literature on the factors and strategic interactions that underlie this trade-off between the probability and the cost of exit, which prompt some rulers to accept competition and others to reject it (e.g., Boix, 1999; North & Weingast, 1989; Przeworski, 1991). These include international conditionality (e.g., the Soviet Union precluded competition Grzymala-Busse / Beyond Clientelism 649 in its satellites, the EU insisted on it), the demands and mobilizational abil- ities of the potential electorate, the costs of repression, and whether all elite actors can accept competitive selection and the institutional guarantees for the losers. In postcommunist East-Central Europe, for example, the com- mitment to democratic competition was the result of shared fears of back- sliding into an authoritarian regime that would eliminate the inchoate democratic parties and elites (Grzymala-Busse, 2007). Fundamentally and simplistically, we see commitments to competition when the costs of repressive or noncompetitive rule become unsustainable and both elites and voters coalesce around the competitive equilibrium.

Impact on State Capture

Both rent sharing and competition result in distinct configurations of state capture: clientelism, predation, fusion, and exploitation, and in the formation of specific state institutions and capacities. Incumbents seeking to obtain private benefits from the state have a variety of extractive mechanisms at their disposal. Rewarding government contracts and tenders to allies (with the expectation of kickbacks and campaign funding), expropriation, asset nationalization, patronage, extrabudgetary funds and agencies, opaque accounting procedures, skimming or outright theft of foreign aid moneys and mineral resource profits are all ways of diverting state resources into the private coffers of ruling elites. To develop maximum control over state agen- cies, rulers will purge the civil service, weaken regulatory agencies, politi- cize fiscal bodies (such as central banks and budget offices), and take over the allocation of foreign aid, customs, and mineral wealth. Rent distribution and political competition constitute and constrain the choice of these strategies. Distribution of selective goods necessitates the building up of both state capacity and political party organizations to deliver and monitor selective good provision. Furthermore, because rent sharing relies on teams within political parties and the state, the personal- ization of office is reduced. Distribution does not affect the levels of extrac- tion from the state—but contestation can. Contestation further reduces the level of elite discretion and increases some state capacities. Distribution and contestation themselves interact; for example, distribution provides a buffer against competition. Rent distribution affects the mechanisms and agents of state capture. The mechanisms of rent distribution include (a) expanding partisan control over the welfare state, (b) expanding those provisions that can be selectively delivered, (c) monitoring the recipients’ support through both formal (state) 650 Comparative Political Studies

and informal (party) means, and (d) withdrawing benefits if necessary. These are needed both to ensure that rents can be systematically shared and that they buy supporter loyalty (and thus increase the expected utility of office). We observe this pattern in both clientelistic and fusion regimes. First, often under the guise of state reform, regulatory, budget manage- ment, and audit agencies are brought under partisan control, increasing access to state resources and their distribution. Incumbents pack the state administration with loyal supporters, who then further hire on the basis of loyalty. Control over housing, education, and other welfare provisions remains nominally at the ministry level, but local party representatives are put in charge of distributing these goods (see Chandra, 2004; Piattoni, 2001). In Singapore, the ruling PAP offers the electorate a set of constituency-specific programs, including welfare provisions, infrastructure, facility upgrades, and housing subsidies, “only offered if the PAP is voted in” (Seng, 1998, p. 389). These programs are under the jurisdiction of government ministries, but parliamentarians manage the housing estates in their constituency, and 86% of the population lives in public housing. Second, welfare state expansion in the service of elite state capture con- sists of increasing relative spending on those benefits and services that can be targeted and which are likely to have the greatest pay-off in terms of the number and loyalty of supporters (who are likely to be relatively poor, rural, and less educated). These include targeted economic development and rural credit programs (National Solidarity Program [PRONASOL] in Mexico; see Diaz-Cayeros, Estevez, & Magaloni, 2006), jobs within the public sec- tor (the Austrian proporz system), local infrastructure projects (the concrete clientelism of Japan; see Scheiner, 2006), and property rights (title and land registrations awarded at the discretion of local land-settlement officials; see Chandra, 2004). These sectors promote capture by targeting delivery and allow the incumbent to keep as large a portion of the rents as possible while maximizing the electoral pay-off. Finally, rent distribution further affects capture strategies through the development of (and payments to) an extensive organization within the state and teams outside of it. Both state and party institutions are used to monitor recipient support. Rural development officials, local registrars and electoral commissions, utility managers, and school teachers all obtain their jobs through the party. They are expected to be both loyal supporters and monitors of how widespread that support is among the populace. The number of state officials expands as the incumbents seek to develop their monitoring network, and the loyalty of these agents is ensured by their dependence on the party for state employment (Chandra, 2004). As a result, Grzymala-Busse / Beyond Clientelism 651 we would expect that the more extensive the rent distribution, the greater the state capacity to regulate and to monitor both citizens and their transac- tions. Incumbents also make their own extensive organizational investments in parties that act both as a mechanism for the delivery of goods from office to party to constituency and for the enforcement of the voters’ support (Chandra, 2004; Magaloni, 2006; Piattoni, 2001). Rent distribution perpetuates the rule of these party teams: Individual elites benefit, but the key aim is to ensure that the cohort can receive the resources and payments necessary to perpetuate its rule and the underlying system of rent distribution. Systematic distribution thus restricts the per- sonalization of office, or the process of increasing the decision-making weight of individuals relative to state institutions, legislatures, courts, or ministerial cabinets. Where distribution is sporadic, targeted, and driven by looming crises, few organizational investments are made: For example, the modal party organization in Nigeria under predatory rule was a one-person operation, with few regional–national linkages other than ad hoc coalitions with local leaders who deliver votes and intimidation in varying proportions (Ichino, 2006; Lindberg, 2004). As a result, if either competition or distribution is systematic and institu- tionalized, we are more likely to see capture on behalf of ruling organiza- tions rather than on behalf of individuals. A portion of the gains will be reinvested into the gain-seeking organization. This is because parties resolve collective action dilemmas inherent in policy making and elections—both distribution and electoral competition relies on organized parties (Aldrich, 1995; Kitschelt, 2000). Where neither distribution nor competition takes place, individual rulers gain private benefits without making organizational investments, as in predation. Contestation affects capture from the state in two main ways. First, it affects the choice of extractive mechanisms: Given their commitment to the rules of competition, incumbents are less likely to resort to tactics that would destabilize the competitive system: sudden nationalization of assets, outright expropriation of potential rivals, or the violent capture of assets. Without a commitment to competition and the relatively level playing field it requires, incumbents will discretionarily and arbitrarily use both repres- sive mechanisms and the rule of law to starve the opposition of resources and appropriating these for the incumbents. Second, intense competition increases the probability of exit from office. If rulers expect electoral pun- ishment when their rent seeking is revealed, they will directly curtail the rates of their extraction from the state and insulate state institutions from political influence, creating new, independent institutions of oversight and 652 Comparative Political Studies

regulation. Endangered incumbents fear that unless they do so, their rivals will be able to use state institutions to enrich and entrench themselves once in office. Thus, contestation can directly affect the rates of extraction in ways that distribution does not.9 Where competition is institutionalized and programmatic (or at least not based on delivering selective goods), and no rents are shared, incumbents are especially vulnerable, because they have no supporters that they can blackmail into loyalty. The costs of exit are relatively low, because losers are guaranteed safety, but there is no buffer of dependent supporters. Without the distribution of selective goods, voters have few reasons to tolerate exploitation but have the franchise necessary to punish the ruling parties (Grzymala-Busse, 2007; O’Dwyer, 2004). Competitors highlight incumbent shortcomings and are as credible in their appeals as the incumbents. As a result, rates of exploitation are highly vulnerable to changes in competition. And as Poland after 2004 shows, exploitation held in check by a robust opposition can quickly bloom once the opposition falters. Although these causal arrows can be logically reversed, it is unlikely that state institutions produce rent distribution and political contestation patterns. First, we see the same extractive strategies being adopted on the basis of very different institutional legacies: For example, fusion arose on the basis of postcolonial, postdemocratic, and postautocratic states, relying on the weakness of competitive pressures (a decimated civil society and absent alternative political elites) and demand for rent distribution. Second, there is the question of temporal precedence: As noted earlier, clientelism built on existing patron–client ties, exploitation arose simultaneously with the rebuilding of the postcommunist state, and so forth. State institutions subse- quently maintain elite state capture strategies: Decades of supporter depen- dence generated by clientelist practices reduces the plausibility of opposition competitors, for example. But state institutions themselves do not produce these extractive regimes.

Consequences for the State

If state formation reflects state capture strategies, we should see the same practices and mechanisms by which redistribution and contestation affect state capture mirrored in the creation of state institutions and capacities. This section argues that, accordingly, rent distribution bolsters certain aspects of the state’s infrastructural capacity (its ability to suffuse and implement its decisions within civil society) through the development of rent delivery Grzymala-Busse / Beyond Clientelism 653 channels and support monitoring. Contestation reduces the state’s despotic capacity, or the ability to act discretionarily, without consultation with society (Mann, 1988). Rent distribution has several effects on state capacities. First, partisan control weakens the institutions of audit or regulatory agencies, allowing incumbents to use them selectively. Second, the expansion of welfare state provisions, even if contingent and targeted, increases spending on health care, education, and other provisions. It further necessitates increasing state capacity to deliver goods and monitor compliance. Public administration expands both as a result of this increased demand and because patronage jobs are one of the goods distributed. Third, distribution and monitoring of constituency support requires that the state develop legibility, or the capac- ity to identify supporters and defectors and to process this information (Scott, 1998; Slater, 2003). In contrast to claims that redistribution does not require formal institutions (Gandhi & Przeworski, 2007), then, this analy- sis argues that rent distribution necessitates state institutions, not only party organizations.10 Fourth, institutions create opportunities for individual politicians to take credit for delivering rents to supporters. In turn, intense contestation builds up the quality of oversight and regu- latory institutions, where the electorate is not dependent on rent sharing and thus free to support the opposition. Incumbents lower their rate of rent- seeking and insulate state institutions from the political influence of their successors. Contestation without distribution also results in smaller state administrations relative to advanced democracies (given fewer incentives for patronage), with commensurately lower wage bills (given the lower rent-seeking). Finally, competition places additional pressures on rulers to increase legibility and infrastructural capacity, because elections require voter registration, ballot distribution, and the calculation of results. These configurations are summarized in Table 3. The impact of distribution and competition, and the mechanisms that translate state capture into state formation, are visible in four institutional domains of the state: formal state institutions (including property rights, regulation and oversight, the rule of law, and tax collection), state adminis- tration, development programs, and electoral institutions. These institutions fulfill fundamental state roles.11

State Institutions of Law and Oversight Formal institutions that protect property rights, ensure transparency in government spending, and regulate contracts between governments and 654 Comparative Political Studies

Table 3 Consequences of Extractive Regimes for State Institutions

Extractive Formal Public Developmental Electoral Regime Institutions Administration Projects Institutions

Clientelist Redistribution Large, Contingent, Credit taking: favored, weak commensurate targeted, personalized property rights, wage bill and deliver credit taking goods Fusion Relatively high Large, Contingent, Credit taking capacity, beholden commensurate more broadly but no to political party, wage bill targeted contestation credit taking Predatory Greatly Small, outsized Directly extractive, Noncredit taking weakened, wage bill or ruse for but also limited personalized embezzlement contestation Exploitative Relatively Large, No special Noncredit taking high capacity, relatively small emphasis discretionary wage bill without competition private actors can all constrain capture, and this is precisely why rulers are often loath to implement or enforce them (Dixit, 1996). Yet contestation and rent distribution can constrain these impulses. As Table 4 shows, strategies compatible with competition (clientelism and exploitation) go hand in hand with higher ratings for the rule of law and regulatory quality, whereas ratings for fusion and predation are both lower and display greater variance. Without rent distribution, contestation becomes the main impetus for building strong state institutions: Audit institutions and ombudsmen are powerful only where intense competition takes place (Poland, and also Hungary, Slovenia, and Estonia: see Grzymala-Busse, 2007). Without rent distribution or contesta- tion, predatory regimes have the weakest state institutions and greatest ruler discretion, as their poor rule of law and regulatory ratings suggest. Rent distribution simultaneously weakens individual property rights and expands the provision of contingent welfare to perpetuate supporter depen- dency on the ruler. The requirements of distribution necessitate the devel- opment of formal state institutions, but there are few incentives to make these institutions autonomous. Both fusion and clientelist states have con- siderable opportunity to develop strong state institutions of monitoring, tax collection, oversight, and so forth, but also have the discretion not to do so (given the popular dependence on state provisions). Thus, patronage is a Grzymala-Busse / Beyond Clientelism 655

Table 4 Indicators of State Governance and Capacities

World World Governance Governance Indicators Indicators Rule of Regulatory Tax Oversight and Regulation: Audit Law Rating Quality Revenues Country Courts and Ombudsmen (OECD: 91%) (OECD: 90%) as % GDP

Italy Audit court has limited investigative, 73 79 41 extensive punitive powers, no ombudsman Japan Audit court has limited investigative, 89 68 27 no punitive powers, no ombudsman Mexico Audit court has limited investigative, no 43 66 20 before 2000 punitive powers, no ombudsman Russia Audit court has investigative and 21 36 36 under Putin punitive powers, but can be immediately recalled by Duma or Federation council; Ombudsman established in 1997, no investigative powers, can only obtain information with consent Singapore Audit court has investigative but not 92 99 31 punitive powers, set up in 1970, no ombudsman. Communist Late communism; few ombudsmen, n/a n/a n/a East-Central parliamentary appointed, limited Europe powers information only 4 12 7 Nigeria under Audit court has no investigative or Abacha punitive powers; ombudsman established in 1975, investigative powers, but can be removed at any time Indonesia Audit court set up in 2001, limited 18 25 19 under Suharto investigative and no punitive powers; ombudsman established in 2000, but no investigative powers Philippines Audit court set up in 1973, limited 34 51 12 under Marcos investigative but punitive powers; ombudsman a presidential appointment since 1987, after Marcos; no direct enforcement Poland after 1989 Audit court established in 1994, 66 70 37 opposition-led, extensive investigative and some punitive powers; ombudsman established in 1987, increasingly extensive investigative powers Czech Republic Audit court established in 1992 but 71 84 38 after 1989 government controlled until 2000, no in-depth investigative powers; ombudsman established in 2000 (vs. limited investigative powers) Slovakia Audit court established in 1993, but 63 77 29 after 1989 controlled by government; no independent investigative powers; ombudsman established in 2002, apolitical

Note: OECD = Organisation for Economic Cooperation and Development. Source: Asher, 1989; World Bank, 2002. 656 Comparative Political Studies favored strategy of filling state positions, because it ensures loyal and dependent state officials. Similarly, rather than attempting to develop uni- versal welfare state systems, clientelist politicians will seek to keep services, infrastructure, and development benefits at their discretion. There are two reasons to do so: First, particularized benefits ensure continued sup- port, because they can be withdrawn. Second, politicians receive direct credit for patronage, rather than the party taking diffuse credit for universal policies (Chandra, 2004). Two results follow: (a) a set of state formal institutions that can collect taxes, regulate the economy, and enforce the rule of law but will not do so consistently or uniformly, and (b) a stronger state capacity to gather infor- mation and reach into remote areas. This legibility works both ways: patrons monitor support, and clients can assign credit. As a result, both rule of law and effective tax collection are feasible, but rulers are less likely to pursue these constraints on their extraction in less competitive and eco- nomically developed systems. As Table 4 shows, the result is that Japan and Italy have higher governance ratings than Mexico does. With fewer resources and a meager potential tax base, the PRI in Mexico concentrated on target- ing contingent goods rather than on developing these state capacities. In fusion regimes, state agencies are responsible to the ruling party alone and had few regulatory or oversight powers of their own.12 As a result, state institutions mirror party preferences and capacities. One extreme is Singapore, with a highly visible and enforced rule of law and regulatory system, designed expressly in mind with attracting investment and promoting eco- nomic growth. Another is Putin’s Russia, which strengthened legal and oversight institutions selectively. As a result, despite low overall regulatory ratings, powerful state institutions of oversight and regulation exist, such as the Tax Authority, which can conduct on-site audits of any company or indi- vidual, with no statute of limitations on the audits, duration limits, or right to appeal. The legibility capacities of such states grow in the service of the party’s control over state and society: registration requirements limit mobil- ity, job assignments are controlled by the party, and so on. The development and control of formal state institutions without con- testation also allows incumbents to legally neutralize rivals and rule by law, rather than follow the rule of law. For example, Putin and his cohort have limited competition through formal channels by raising the requirements for territorial presence, financial reserves, popular signatures, and party membership. As a result, the 60+ parties of the Yeltsin era have been reduced to fewer than 15. Putin has eliminated other potential competitors by using the rule of law, strictly but selectively applied, to question the Grzymala-Busse / Beyond Clientelism 657 property rights that were acquired in the mid-1990s in a much shakier legal framework (Allina-Pisano, 2005; Darden, 2001). This tactic eliminated potential challengers and helped to establish enormous new state-owned holding companies in critical sectors such as gas and mineral resources, automobile and aircraft manufacturing, shipbuilding, nuclear power, dia- monds, and titanium. Similarly, PAP actively limited legislative opposition and brought the legal system, state administration, and trade unions under its control, enabling it to use administrative law against its political oppo- nents (Seng, 1998). Predatory regimes combine the control of the state, made possible by the elimination of competition, with low pressures to build up state institutions of legibility or good provision delivery in the absence of rent distribution. Rulers are free to concentrate power in their own hands and rule by decree (Sidel, 1999). Such personalization still relies on building up specific state institutions, especially those that increase access to state resources and their diversion (such as unregulated privatization, extrabudgetary funds, and expropriation mechanisms). To maintain the incumbent in office, repressive mechanisms and state institutions are also strengthened (Slater, 2007). For example, in the Philippines, both the army and the Supreme Court were given additional power and resources under Marcos—but at the same time, they were made beholden to him. Although formal state institutions selectively enforce property rights in fusion and clientelist regimes, state institutions cannot do so in predatory regimes. Fearing expropriation, property owners purchase protection from rulers and ruling factions individually (Khan, 2005). Asset holders will seek to preserve asset mobility (through speculation and capital flight) and to collude with bureaucrats. Because insecure property rights and uncertainty surrounding contracting increase fears of expropriation and lower invest- ment rates, predatory rulers wishing to attract investment have to turn to membership in international organizations, convertible currency, or open financial markets: external commitments that trigger sanctions against expropriation (Lewis, 2007). Their domestic institutions are unable to offer such guarantees. Predatory rulers have few incentives to build infrastructural capacity. To ensure discretion, formal institutions are largely weak, with little or no inde- pendent oversight and regulation. With few resources and little efficiency (the result of administrative purges), state institutions have little capacity to enforce laws, regulate the economy, or collect taxes (Jackson & Rosberg, 1982). Oversight and regulatory agencies, the civil service, and the judi- ciary are left “weak and politicized, thereby compounding the uncertainties 658 Comparative Political Studies that flow from central policies” (Lewis, 2007, p. 20). Tax collection rates in predatory regimes are the lowest among the cases considered, even accounting for their lower income levels, as Table 4 summarizes. In exploitative regimes, competition underlies the building of formal state institutions. First, the same commitments to competition limit the strategies of state capture: Few cases of outright expropriation take place. Postcommunist incumbents, for example, benefited extensively from priva- tization through favorable awarding of bids and tenders but did not abnegate property rights once established (Ganev, 2007). Property rights are pro- tected, all the more so because privatization is a lucrative contributor of state assets. Second, because voters receive no selective goods and thus care more about programs and policy competence, competitors can profit electorally from incumbents’ malfeasance. The more intense the competition—the more voters are given clear, credible alternatives to the incumbents by an opposition that is highly critical of the incumbents—the more likely there will be a regulatory regime, with antinepotism and antipatronage laws, powerful regulatory agencies and courts, and oversight institutions such as securities and exchange commissions, anticorruption offices, and central accounting offices. Their terms do not coincide with the Parliament’s, and they are not controlled by the incumbents. Finally, competition places addi- tional demands on state legibility and other capacities, because representa- tion and contestation are ensured by formal state institutions. Incumbents committed to contestation need to ensure that state institutions can register voters, run local polling places, or gather ballots. Faced with a higher probability of exit, but knowing they were likely to compete for power again, incumbents built institutions that would restrain their successors. One result is higher rule of law ratings, as Table 4 shows. Another is that ruling parties rapidly established a slew of formal state insti- tutions, investing them with considerable powers and insulating them from party politics. Where a more robust competition existed in postcommunist democracies, 85% formal state institutions of oversight and regulation were implemented voluntarily, before the EU demanded adoption. Where com- petition was weaker, only 31% of such institutions were implemented vol- untarily. They were either weakened and politicized—or simply not built at all. Latvia, Slovakia, the Czech Republic, and Bulgaria all delayed impor- tant oversight institutions, such as audit courts, securities and exchange commissions, and ombudsmen. Unlike rent-distributing regimes that take credit for rent distribution, both predatory and exploitative regimes seek to remove swathes of public accounts out of public scrutiny to minimize the assignment of blame. To do Grzymala-Busse / Beyond Clientelism 659 so, they rely on a variety of quasistate organizations and funds to preserve discretion. Such off-budget items include various quangos, foundations, private armies and security forces, economic authorities, tax farming arrangements, and contracting firms. Predation is both aided by and results in budgetary opaqueness. Both because rulers preserve their discretion and because they do not take credit for rent distribution, the fiscal regimes are often kept deliberately obscure. Thus, the amount and allocations of Philippine’s national budget after 1973 were hidden from all but the high- est bureaucrats. Fiscal transfers were concealed so that “officers in particu- lar agencies knew when budgeted funds had not been received, but even they had no clear indication as to where the funds had gone” (Wurfel, 1988, p. 139; also see Sidel, 1999). Similarly, extrabudgetary funds underpinned predatory rule in Indonesia (Robison & Hadiz, 2005). Nigerian peacekeep- ing activities in Africa were conducted off budget and cost far more than UN estimates of these activities (Lewis, 2007). Exploitative incumbents either continued or established numerous funds and quasipublic agencies in the postcommunist democracies (Grzymala- Busse, 2007). Although some were inherited from the communist regime, exploitative incumbents expanded many of these to shift moneys out of the public budget, especially after the windfall associated with the privatization of state enterprises. Yet competition limited this expansion: Where compe- tition was weaker, the number of opaque extrabudgetary institutions and quasistate agencies increased by 100%—where competition was more robust, the increase was a relatively modest 28% (Grzymala-Busse, 2007).

State Employment Employment in the state sector is both a source of rents and a mechanism of rent distribution. The steady employment and wages of state jobs can be attractive rewards to supporters and to the rulers at whose discretion the employees serve, ensuring greater dependency on the ruler. As state agencies and positions multiply, they also offer the opportunity for state agents to mon- itor support for rulers, as noted above. For regimes that distribute rents, there- fore, expanding public employment provides both rents and the channels of state legibility. For regimes that do not distribute rents, public administration expansion is a ruse for either outright appropriation of funding or the expan- sion of state agencies to increase discretionary access to state resources. Actual hiring is incidental to these goals. Furthermore, by inflating the wage bill (through creative accounting, no-show jobs, and channeling funds into capacity development), rulers gain access to a major source of funds. Table 5 660 Comparative Political Studies

Table 5 State Administration and Wage Bills Across Extractive Regimes

Ratio of Spending % Population % Population in Wage to % Employed Employed by Employed in Bill as % of (Wage Bill/Total Central State General Government Government Extractive Regime Administration Government Expenditure Employment)

Clientelist 1.9 5.8 11.4 2.1 Fusion 1.1 9.9 19.4 2.9 Predatory .6 1.5 9.5 8.4 Exploitative 1.0 6.6 9.7 1.5 Organisation for 1.8 7.7 4.4 1.7 Economic Cooperation and Development average

Note: For the sake of comparability, these figures are reported in the percentage of the population; when computed as a percentage of the employed, these percentages are to be higher. For example, the average central state administration in exploitative regimes is over 5% of those employed (see Grzymala-Busse, 2007). shows that the ratio of spending to the size of state employment varies along with these patterns of state capture. Thus, clientelist and fusion regimes see a rise in overall public employ- ment, specifically in the welfare state sectors of health and education, which are largely controlled by the government (Calvo & Murillo, 2004; Gimpelson & Treisman, 2002). Exploitative regimes have smaller states and relatively small wage bills, further constrained by robust competition. Predatory regimes do not have large state administrations, because no rent distribution takes place and constant purges ensure both loyalty and little accumulation. However, the wage bill is disproportionately large, given ruler discretion. Both clientelist and fusion strategies seek to create dependence among supporters, and jobs in the state sector are contingent and targeted awards. It is not surprising that the percentage of the population employed by the general government (health, education, welfare) is high, at the levels of the Organisation for Economic Cooperation and Development average or higher (see Table 6). In Italy, for example, one study found that up to 500,000 employees were hired through patronage from 1946 to 1958. (Lewanski, 1997).13 Parties form and maintain loyal constituencies by handing out state jobs and then using these to create and monitor broader dependence on the ruler. Similarly, in fusion regimes, employment in the general public sector, especially in health care, education, and affiliated sectors, is relatively high. The wage bill as a percentage of government expenditure is also very high. Grzymala-Busse / Beyond Clientelism 661

Table 6 State Administration and Wage Bills Across Cases

Ratio of Employment % Population % Population in Wage to Spending Employed by Employed in Bill as % of (Wage Bill / Total Central State General Government Government Country Administration Government Expenditure Employment)

Italy 3.4 9.1 11.2 1.2 Japan 0.7 3.5 5.3 1.5 Mexico 1.7 4.8 17.8 3.7 Average 1.9 5.8 11.4 2.1 Russia 0.4 8.4 13.7 1.6 Singapore 0.8 6.1 25.0 4.1 Communist 2.0 15.3 na na East-Central Europe Average 1.1 9.9 19.4 2.9 Nigeria 0.1 0.1 1.3 13.0 Indonesia 0.7 2.3 13.3 5.8 Philippines 1.1 2.2 ~14.0 6.4 Average 0.6 1.5 9.5 8.4 Poland 0.6 4.1 7.6 1.8 Czech Republic 1.4 8.1 8.1 1.0 Slovakia 1 7.7 13.3 1.7 Average 1 6.6 9.7 1.5

Source: World Bank, 2000.

However, as the party takes over the state and its regulatory roles, one ironic result is that the actual state administration is tiny: well under the 4.5% of the developed world average. Because public administration employment is not chiefly used to enrich elites, the wage bill is also below average. In contrast, in exploitative and predatory regimes, in the absence of rent distribution, there is no systematic growth in employment in health care, educational systems, and other state services that have been the traditional target of patronage hiring. In postcommunist states, these actually shrank, thanks to employee departures and the increasing privatization of these sec- tors. The state in general lost its importance as a source of employment after 1989.14 And instead of mass patronage, both exploitative and preda- tory rulers engaged in prebendalism, rewarding elite allies with office (van de Walle, 2003). Rather than pursuing patronage, postcommunist exploitative incumbents established new state agencies and funds, increasing the staffing of central state agencies to expand their budgets and thus the funds under the control of the incumbent governing parties. In some cases, state administration 662 Comparative Political Studies employment more than quadrupled in the 15 years after the collapse of communism. Jobs, however, went to loyal elites, not supporters. The predatory regimes examined here have the largest wage bill but the smallest state employment. This paradox is the result of contradictory strategies: Predatory rulers try to build up the bureaucratic apparatus as an instrument of their will (Jackson & Rosberg, 1982) but at the same time demand absolute loyalty. Predatory rulers act to eliminate potential oppo- nents or resistance to their extractive projects partly by purging state admin- istrations. Technocrats and skilled officials are among the first to be fired, and the positions are either left vacant or replaced with loyalists, especially in sensitive (and lucrative) state sectors such as trade, customs, natural resource management, and so on. With no widespread distribution of jobs as rewards, the wage bill is diverted to pay for lucrative elite employment and discretionary account spending. The salaries of purged civil servants are often appropriated by the ruler and allied elites. For example, in the Philippines, state-owned enter- prises (“corporations”) were created by fiat and used as vehicles for sinecures. The biggest increases in bureaucratic hiring took place within the corporations: Their staffing grew from 41,000 in 1975 to over 106,000 in 1979, and “soaked up to 30% of public expenditures by the late Marcos years” (Hutchcroft, 1991, p. 446). These appointments were made possible by the appropriations policy: Bureau directors received appropriations from legislature in exchange for hiring. It is not surprising that the bureaucracy was described as subordinate to political interests (Hutchcroft, 1998; Kang, 2003; Wurfel, 1988). And once state revenues go into the pockets (or rather, the offshore bank accounts) of the rulers, state institutions find themselves with insufficient funding. Staff members go unpaid, weakening regulatory oversight and enforcement and creating incentives for state officials to com- pensate their meager salaries through selling services for additional fees (bribery) or skimming from government accounts. The high wage bill in predatory regimes is the result of resource diversion, and the small state employment is the result of low rent distribution and the maximization of ruler discretion through constant administrative purges.

Development Programs Rulers have distinct incentives to establish development programs. For nondistributing rulers, they offer enormous discretionary income. For rent- distributing rulers, they are a further means of targeting the most cost-effective constituencies. Development programs are an effective and attractive Grzymala-Busse / Beyond Clientelism 663 mechanism to deliver contingent and targeted goods, such as agricultural subsidies, loans, infrastructural spending and other policy benefits. As Chandra (2004) shows, development spending is used selectively to cement the loyalty of party supporters; development can become highly targeted. Rent sharing goes hand in hand with building up specific development programs. For example, agricultural spending typically targets the most vulnerable and numerous citizens. It is therefore most likely to pay off in the kind of loyalty and dependence that clientelist strategies seek, further affording considerable discretionary use. It is not surprising that in Mexico, Japan, and Italy, agricultural spending and development programs became a key delivery mechanism of selective goods (Magaloni, 2006; Scheiner, 2006). In another example, the Integrated Rural Development Program in India was a subsidized credit program that allowed indigent rural families to take up self-employment. The program is income based, but because income is difficult to verify, enormous discretion rests with the party-affiliated state agents who select the eligible (Chandra, 2004). We observe a similar pattern in fusion regimes (Allina-Pisano, 2007). However they are contingent or inefficient, such programs nonetheless deliver benefits to constituents. They help to perpetuate broad dependence on the ruling party and further require that the state develop its institutional reach down to the most rural or peripheral areas. In contrast, predatory regimes with large agricultural sectors use development programs largely to channel funding into the pockets and accounts of the rulers. There is little pay-off to either targeting or delivering these programs to their intended constituencies, narrow or broad. The history of such programs in the Philippines under Marcos, Nigeria under Abacha, or other cases such as Zimbabwe after 2000 is that of outright expropriation—in some cases, as with Marcos’s coconut and sugar levies, from the very agricultural produc- ers whom the development programs supposedly benefit. Where agricultural and infrastructure programs in rent-sharing regimes deliver benefits to local supporters, predatory versions of similar programs divert funds into the pockets of ruling elites at the national level, before dis- tribution can take place. For example, in the Philippines, Marcos and his cronies directly siphoned off resources, most notably foreign aid. The United States continually gave aid to assist in reforms of the state, but these were simply channeled into the private bank accounts of Marcos (Kang, 2003). The Marcos land reforms further expropriated opposition landown- ers in the name of development (Aquino family assets were the first to be dispossessed). Banks and other enterprises owned by regime opponents were transferred into cronies’ hands (Hutchcroft, 1991). 664 Comparative Political Studies

Predatory rulers can even reverse the flow of funds, expropriating agri- cultural actors, potential opponents, or lucrative sectors of the economy to benefit the incumbents. Thus, Marcos imposed a levy on coconut farmers in 1971. The initial levy levels quickly increased, and the purpose of over- seeing agency changed from price stabilization to “investment for coconut farmers.” The levies garnered close to $2.5 billion in less than a decade. In 1976, Marcos also established a similar Sugar Commission, which cost sugar producers an estimated $3 to $4 billion between 1974 and 1983. Unrestrained by either distribution or contestation, the scale of the depre- dation was vast: In 1977-1997, 43% of potential tax revenues went into pri- vate hands (Bhargava & Bolongaita, 2004, p. 88). Similarly, in Nigeria, working with a handful of retainers and cronies, General Sani Abacha seized funding from industries, construction, and other capital development projects (Lewis, 2007). Bankers were arrested, the assets of Yoruba regime opponents confiscated, and billions in oil revenues embezzled. In final years of the regime, nearly a quarter of the budget went into the pockets of Abacha and his cronies. Development programs became another tactic of expropriation that expanded the repressive capacity of the state without building its infrastructure.

Formal Electoral Rules Rent distribution creates incentives for credit claiming on behalf of the incumbents. It is clear to all actors that jobs, housing, higher education, and wel- fare state benefits are contingent on compliance with the regime. Institutions reflect this contract: the basic unit of communist parties, for example, was in each workplace, and food ration allotments, vacations, and other benefits were distributed through these party organizations. Similarly, the party units of the Singaporean PAP are organized in housing estates, making it clear that infra- structure improvements or housing transfers are contingent on party support. In competitive regimes, electoral institutions allow rulers to take credit or evade blame for their capture of state assets, as part of the institutional framework that buttresses extractive strategies. Electoral rules can explicitly aid state capture only where competition exists. At the same time, there is no simple relationship between institutions and clientelism (much as nation- alism, clientelism is a wanton mobilizational strategy that comfortably cohabits with diverse institutional and political bedfellows). Nonetheless, for rent-sharing strategies to establish constituent loyalty, politicians must be able to claim credit for the provision of goods in the eyes of the constituents—and everyone must be aware of the redistributive outcomes. Grzymala-Busse / Beyond Clientelism 665

Credit taking is so critical that transparency measures introduced to monitor earmarked pork allotments in the U.S. Congress actually increased the amount of earmarking, as congressmen competed not to fall behind in pro- viding for their districts and face electoral censure (see Andrews & Pear, 2007). Electoral institutions imposed on clientelist relationships by self- serving rulers thus tend to allow voters to identify patrons and patrons to monitor supporters. Many clientelist exchanges rely on a personal vote or the degree to which voters can reward or punish individual politicians (as opposed to entire parties). Moreover, clientelism favors small coalitions necessary to win office: The broader the swathe of voters whose support is necessary to win office, the lower the incentives (and the ability) to provide expensive club goods. In short, voters and politicians establish clear channels of responsi- bility, and narrow coalitions bring politicians into office. Therefore, we see some combination of personalized vote institutions (such as single non- transferable vote or open-list proportional representation; see Carey & Shugart, 1995). We observe these institutions in Japan (single nontransfer- able vote prior to 1994, a combination of single-member district [SMD] and open-list proportional representation [PR] after 1993), Italy (open-list PR until 1993, then a mixed SMD-PR system) and Mexico (60% of seats elected with SMD, but the rest in closed-list PR with large districts, along with a presidential system that enhanced the personal vote). The overrepre- sentation of areas dependent on clientelistic benefits can further benefit the ruling party (as it did the LDP in Japan; see Scheiner, 2006). Note that in all these cases, the electoral institutions were imposed on existing clientelist relationships: Patron–client relations dominated the pol- itics of Italy, Japan, and Mexico long before these countries adopted mod- ern electoral institutions. Clientelism appears to influence electoral system choice but not vice versa. And it is not the case that personalistic electoral institutions necessarily result in clientelism; open-list PR was also adopted in several countries with little clientelism: Chile, Denmark, Finland, Luxembourg Switzerland, and Uruguay, and in several postcommunist states, such as Slovakia, Estonia, Latvia, and Poland (until 1997). Electoral institutions do, however, foster the monitoring and credit assigning that is critical to clientelist relations. In contrast, in exploitative regimes, rulers have no incentive to form either close ties to the constituents or to engage in mutual monitoring of the exchange contract in the absence of rent sharing. Instead, their preferred insti- tutional solutions focus on programmatic competition, assignment of credit to parties rather than individual competitors, and avoidance of individual 666 Comparative Political Studies credit taking. Accordingly, in the new democratic states where exploitation is taking place, a variety of parliamentary electoral institutions were adopted: from closed list PR (adopted in Poland in 1997) to semiopen lists in Slovakia and to a mixed SMD-list PR system (Hungary). In all cases, parties have not developed strong ties to society, and the focus is on national campaigns and issues rather than on local mobilization and concerns. In contrast to clien- telism, electoral institutions were adopted before exploitation began in earnest, as part of a transitional package of economic and political reform.

Mechanisms and Dynamics of Change Rulers have the greatest leeway to adopt particular capture strategies after crises of the state; democratization, regime collapse, or external shocks can alter the costs and benefits of distribution, competitive configu- rations, and institutional and organizational investments. For example, exploitative rulers had such leeway to form porous state institutions, because parliamentary parties initially assumed much of the legislative and administrative decision making after the collapse of communism in 1989 (Ágh, 1994; Nunberg, 1999). The sheer volume of laws to be drafted and the unresolved debates about powers and functions meant that political par- ties in Parliament had a unique opportunity to create state institutions and build in subsequent access to state resources. Similarly, Indonesia after the economic collapse and Suharto’s departure from power in 1998 can be described as having moved from a predatory to an exploitative regime, its politics centered on a fractious, rent-seeking democratic Parliament that tolerates competition but provides little distribution (Sherlock, 2002). Finally, Marcos’s targeting of oligarchic clientelist families that ruled the Philippines can be seen as an explicit attempt to shift to predation after an effective coup: The Philippines were traditionally a clientelist regime with enormously wealthy oligarchic families as the patrons. When Marcos assumed power in 1972, predation began in earnest, in favor of technocrats with no mass constituencies. Once the institutional upheaval subsides, initial experiments give way to decreasing oscillations between strategies. They generate their own con- stituencies and institutional support. Incumbents can incrementally change their strategies but face considerably more difficulties in changing the expec- tations of supporters and the institutional framework. Nonetheless, some equilibria may be undermined over the long run (e.g., economically compet- itive sectors may rebel against clientelism; Kitschelt, 2000). Predatory rulers can have long time horizons, but once they begin to fear the catastrophic cost Grzymala-Busse / Beyond Clientelism 667 of leaving office (itself the result of predatory rule), they will accelerate their rate of capture (Goldsmith, 2004). Other rulers may become the vic- tims of their own (relative) success: “growing equality and economic development encourage democratization because they reduce the value of the strategy of vote buying” (Magaloni, 2006, p. 30).15 In fusion regimes, once rulers establish the expectation of broad good delivery as the basis of compliance, an inability to fulfill this expectation endangers regime stability. Demographic change may also undercut capture strategies. Thus, Japan’s shrinking and aging rural population is lowering the electoral pay-offs of clientelist provision. Each of the strategies discussed above is sensitive to distinct incentives and constraints. Rent-sharing strategies respond to changes in the supply and demand of selective goods: If supporters gain alternative sources of these goods, if their economic standing improves, or if rulers become unable to supply the goods, support for incumbents can quickly wane. As a result, rent-sharing strategies are limited by the rulers’ ability to extract enough to distribute—but not so much that economic growth stalls. As communist rulers found, this can be a tall order. Changes in competitive configurations can alter both the type and level of capture. Thus, exploitation rates vary relatively directly with shifts in political competition, but predation rates vary inversely with the appearance of political challengers, as rulers attempt to extract all they can while they can. Competition can also ratchet up the rates of clientelist capture, as par- ties compete by bidding to provide more selective goods (Kitschelt & Wilkinson, 2006). However, this outbidding is limited by the credibility of the challengers, who do not have access to office and its benefits. Strategies that combine rent distribution with contestation are more resilient to changes in competition. Rent distribution provides a buffer of inelastic support and organizational resources, as noted above. As a result, democratization or changes in electoral competition tends to change rather than end clientelist relations (Khan, 2005). For example, in Japan, postwar land reforms meant that landlords could no longer control and deliver the vote, but politicians still pursued clientelist linkages. Politicians turned to koenkai and local politicians to deliver the vote and continued to deliver clientelist ben- efits even after subsequent electoral reforms in the 1990s (Scheiner, 2006).

Conclusion

This analysis advances three related claims. First, distribution and compe- tition, themselves rooted in earlier organizational investments and electoral 668 Comparative Political Studies

expectations, have a fundamental impact on extractive strategies. Rent distrib- ution promotes nonpersonalized elite extraction and the explicit building of state institutions of monitoring and delivery. Political Competition curbs the discretionary use of these infrastructural state capacities and can lower rent seeking when voters are both responsive and available. This analysis complements existing studies of the interaction between electoral institu- tions and clientelism, which argue, for example, that personalistic electoral institutions and intraparty competition favor clientelism. An open question is the subsequent impact of such institutions on other capture strategies, such as fusion, predation, or exploitation. Second, state capture and state formation go hand in hand. This is most explicit in exploitation but is a fundamental characteristic of other extractive strategies. Two mechanisms translate capture into state building: The expan- sion of welfare state institutions is used for targeting contingent benefits to supporters, and the weakening of oversight and regulatory agencies allows greater discretion. Where institutionalized competition for office is intense, it creates incentives for lowering the rates of capture (for fear of electoral pun- ishment) and for building stronger regulatory agencies (to prevent the next set of incumbents from entrenching themselves through extraction). What is less clear is the impact of state capture on economic development. Third, moving beyond clientelism and predation allows us to account for the diversity in extractive strategies and for the distinct configurations of state capacities that follow. The regimes that result from distribution and contestation can be further disaggregated: Vote buying, for example, com- bines rent distribution with far lower organizational investments than in clientelism (Hicken, 2006). Finally, the propositions developed here need to be tested on other cases of state capture (and lack thereof). In so doing, a fundamental point remains: Fitting a variety of strategies of state capture into the procrustean beds of clientelism or predation leads us to overlook both the variation in their redistributive and competitive aspects and their impact on state formation.

Notes

1. Clientelism has coexisted with various forms of political competition, from one-party hegemony (Mexico, see Magaloni, 2006), to one-party dominant systems (Japan) and to fractious multiparty democracies (Italy, dominated by the Christian Democrats until 1994). 2. Clientelism is not only compatible with contestation but can also buttress popular sup- port for democracy and its daily functioning. For example, the postwar Austrian system of pro- porz, which divided patronage into two main ruling parties, was part of new, postwar electoral commitments to avoid the antidemocratic fractioning and polarization of the Weimar era Grzymala-Busse / Beyond Clientelism 669

(Luther & Müller, 1992). Similarly, in Italy, patronage cemented the new democratic system: Popular support for democracy immediately after World War II was based on the “concrete benefits the [ruling Christian Democrats] provided to its social bases” (Tarrow, 1990, p. 312). 3. Some rulers using fusion strategies further bolster their power by nationalizing the econ- omy. The result is then both monopoly access to state resources and another lever of control over society, because most workers are now state employees. This was a central element of commu- nist state–party fusion, but is not a necessary component, as the case of Singapore illustrates. 4. Postcommunist party membership rates were below 3.0% of the population, less than half of the West European average of 8.2%, and only a tenth of the membership rates of coun- tries such as Austria or Sweden. 5. The Argentine Partido Justicialista’s dense clientelist organizational networks covered each square kilometer with an average of 1.8 base units, and its membership alone comprised 18% of the electorate (Levitsky, 2003). The most densely organized postcommunist party, the Czech Communist Party of Bohemia and Moravia (KSCM), averaged only .09 base units, and its membership never topped 3% of the electorate. 6. Elasticity is determined by the income available to spend, the available substitutes, and time. 7. Considerable debate exists on whether clientelist parties target swing voters (Dahlberg & Johanson, 2002; Dixit & Londregan, 1996; Schady, 2000; Stokes, 2005) or core supporters (Calvo & Murillo, 2004; Cox & McCubbins, 1986; Diaz-Cayeros, Estevez, & Magaloni, 2006; Magaloni, 2006). 8. Dense organization, however, does not necessarily indicate clientelism, as in Sweden, Finland, or Denmark, where they are the result of earlier labor or union mobilization and party ideology. 9. High rates of rent extraction can coexist with rent distribution, as in Chung-Hee Park’s South Korea from 1963 to 1979 (see Kang, 2003). 10. Gandhi and Przeworski (2007) argue that in contrast to distributive concerns, the resolution of internal regime conflict necessitates legislatures and other formal state institutions. Left unexplained are the differences in the institutional demands made by conflict resolution and distribution, or why only the former requires formal institutions. 11. I do not examine the maintenance of internal order and border security. 12. For example, the communist state had no civil service: Bureaucrats were subject to the general Labor Code rather than to a Civil Service code that could both ensure their political neutrality and offer state employees protection against political reprisal. Ministries hired their own employees, who were beholden to the party, not the state. 13. Carolyn Warner (2001) argues that 100,000 jobs were created. 14. This is not to say that parties did not attempt to target provision: Populist parties such as the Slovak Hnutie za demokratické Slovensko, the Polish Polskie Stronnictow Ludowe PSL, and the Hungarian Fidesz tried to reward loyal local governments with spending. But these efforts were too short lived and localized to account for the growth of national state administrations. 15. This argument further runs counter to the Acemoglu and Robinson (2006), who claim that equality breeds contentment with the regime and lowers pressures for democratization, as in Singapore.

References

Acemoglu, D., & Robinson, J. (2006). Economic origins of dictatorship and democracy. Cambridge, UK: Cambridge University Press. 670 Comparative Political Studies

Ades, A., & Di Tella, R. (1999). Rents, competition, and corruption. American Economic Review, 89(4), 982-993. Adsera, A., Boix, C., Payne, M. (2003). Are you being served? Political accountability and the quality of government. Journal of Law, Economics, and Organization, 19(2), 445-490. Ágh, A. (Ed.). (1994). The first steps. Budapest: Hungarian Centre of Democracy Studies. Aldrich, J. (1995). Why parties? Chicago: University of Chicago Press. Allina-Pisano, J. (2005). Informal politics and challenges to democracy: Administrative resource in Kuchma’s Ukraine. Unpublished manuscript, Harvard University, Cambridge, MA. Allina-Pisano, J. (2007). The post-Soviet Potemkin Village. Cambridge, UK: Cambridge University Press. Andrews, E., & Pear, R. (2007, August 5). With new rules, congress boasts of pet projects. New York Times, p. A1. Asher, M. (1989). Fiscal systems and practices in ASEAN. Singapore: Institute of Southeast Asian Studies. Bhargava, V., & Bolongaita, E. (2004). Challenging corruption in Asia: Case studies and a framework for action. Washington, DC: World Bank. Big men, big fraud, and big trouble. (2007, April 28). Economist. Retrieved January 8, 2008, from http://www.economist.com/world/displaystory.cfm?story_id=9070922 Boix, C. (1999). Setting the rules of the game: The choice of electoral systems in advanced democracies. American Political Science Review, 93(3), 609-624. Bratton, M., & van de Walle, N. (1997). Democratic experiments in Africa. Cambridge, UK: Cambridge University Press. Calvo, E., & Gibson, E. (2001). Federalism and low-maintenance constituencies: Territorial dimensions of economic reform in Argentina. Studies in Comparative International Development, 35(3), 32-55. Calvo, E., & Murillo, M. V. (2004). Who delivers? Partisan clients in the Argentine electoral market. American Journal of Political Science, 48, 742-757. Carey, J., & Shugart, M. (1995). Incentives to cultivate a personal vote: A rank ordering of electoral formulas. Electoral Studies, 14(4), 417-439. Chandra, K. (2004). Why ethnic parties succeed. Cambridge, UK: Cambridge University Press. Coppedge, M. (2001). Political Darwinism in Latin America’s Lost Decade. In Larry Diamond & Richard Gunther (Eds.), Political parties and democracy (pp. 173-205). Baltimore: Johns Hopkins University Press. Cox, G., & McCubbins, M. (1986). Electoral politics as a redistributive game. Journal of Politics, 48, 370-389. Crenson, M., & Ginsberg, B. (2004). Downsizing democracy. Baltimore: Johns Hopkins University Press. Dahl, R. (1971). Polyarchy. New Haven, CT: Yale University Press. Dahlberg, M., & Johanson, E. (2002). On the vote-purchasing behavior of incumbent govern- ments. American Political Science Review, 96(1), 27-40. Darden, K. (2001). Blackmail as a tool of state domination: Ukraine under Kuchma. East European Constitutional Review, 10(3), 67-71. Diaz-Cayeros, A., Estevez, F., & Magaloni, B. (2006). Vote-buying, poverty, and democracy: The politics of social programs in Mexico, 1989-2006. Unpublished manuscript, Stanford University, CA. Dixit, A. (1996). The making of economic policy: A transaction cost politics perspective. Cambridge MA: MIT Press. Grzymala-Busse / Beyond Clientelism 671

Dixit, A., & Londregan, J. (1996). The determinants of success of special interests in redis- tributive politics. Journal of Politics, 58(4), 1132-1155. Fainsod, M. (1958). Smolensk under Soviet rule. Cambridge, MA: Harvard University Press. Freedom House. (1970-2006). Freedom in the world: Country rankings. Washington, DC: Author. Gandhi, J., & Przeworski, A. (2007). Authoritarian institutions and the survival of autocrats. Comparative Political Studies, 40(11), 1279-1301. Ganev, V. (2007). Preying on the state. Ithaca, NY: Cornell University Press. Geddes, B. (1999, September). Authoritarian breakdown: Empirical test of a game theoretic argu- ment. Paper presented at the American Political Science Association Meeting, Atlanta, GA. Gimpelson, V., & Treisman, D. (2002). Fiscal games and public employment. World Politics, 54(2), 145-183. Goldsmith, A. (2004). Predatory versus developmental rule in Africa. Democratization, 11(3), 88-110. Grzymala-Busse, A. (2007). Rebuilding Leviathan. Cambridge, UK: Cambridge University Press. Hellman, J., & Kauffman, D. (2001). Confronting the challenge of state capture in transition economies. Finance and Development, 38(3), 1-9. Hicken, A. (2006). Party fabrication: Constitutional reform and the rise of Thai Rak Thai. Journal of East Asian Studies, 6(3), 381-408. Hirszowicz, M. (1980). The bureaucratic leviathan. Oxford, UK: Martin Robertson. Hutchcroft, P. (1991). Oligarch and cronies in the Philippine state. World Politics, 43(3), 414-450. Hutchcroft, P. (1998). Booty capitalism. Ithaca, NY: Cornell University Press. Ichino, N. (2006). Thugs and voters: Political tournaments in Nigeria. Doctoral dissertation, Department of Political Science, Stanford University, CA. Jackson, R., & Rosberg, C. (1982). Personal rule in Black Africa. Berkeley: University of California Press. Kang, D. (2003). Crony capitalism. Cambridge, UK: Cambridge University Press. Kaplan, K. (1987). The short march: The Communist takeover of power in Czechoslovakia, 1945–1948. New York: St. Martin’s. Kaplan, K. (1993). Aparát ÚV KSC v letech 1948-1968 [Apparat of the Central Committee of the Czechoslovak Communist Party, 1948-1968]. Prague, Czech Republic: Sesity Ústavu pro soudobé Dejiny AV CR. Keefer, P. (2002). Clientelism, credibility, and democracy. Washington, DC: World Bank Development Research Group. Khan, M. (2005). Markets, states, and democracy: Patron-client networks and the case for democracy in developing countries. Democratization, 12(5), 704-724. Kitschelt, H. (2000). Linkages between citizens and politicians in democratic polities. Comparative Political Studies, 33(6/7), 845-879. Kitschelt, H., & Wilkinson, S. (Eds.). (2006). Patrons, clients, and linkages. Cambridge, UK: Cambridge University Press. Knack, Stephen, & Keefer, Philip. (1995). Institutions and economic performance: Cross- country tests using alternative institutional measures. Economics and Politics, 7, 207-227. Levi, M. (1988). Of rule and revenue. Berkeley: University of California Press. Levitsky, S. (2003). Transforming labor-based parties in Latin America. Cambridge, UK: Cambridge University Press. Lewanski, R. (1997, April). Italian civil service: A pre-modern bureaucracy in transition? Paper presented at the Conference on “Civil Service Systems in Comparative Perspective,” University of Indiana, Bloomington. 672 Comparative Political Studies

Lewis, P. (2007). Growing apart. Ann Arbor: University of Michigan Press. Lindberg, S. (2004). The power of elections: Democratic participation, competition, and legit- imacy in Africa. Lund, Sweden: Lund University. Luther, K., & Müller, W. (1992). Politics in Austria: Still a case of consociationalism? London: Frank Cass. Magaloni, B. (2006). Voting for autocracy. Cambridge, UK: Cambridge University Press. Mann, M. (1988). States, war, and capitalism. Oxford, UK: Basil Blackwell. Manzetti, L., & Wilson, C. (2007). Why do corrupt governments maintain public support? Comparative Political Studies, 40(8), 949-970. Medina, L., & Stokes, S. (2007). Monopoly and monitoring: An approach to political clien- telism. In H. Kitschelt & S. Wilkinson (Eds.), Patrons, clients, and policies (pp. 68-83). Cambridge, UK: Cambridge University Press. North, D., & Weingast, B. (1989). Constitutions and commitment: The evolution of institu- tional governing public choice in seventeenth-century England. The Journal of Economic History, 49(4), 803-832. Nunberg, B. (Ed.). (1999). The state after Communism. Washington, DC: World Bank. O’Dwyer, C. (2004). Runaway state building: How political parties shape states in postcom- munist Eastern Europe. World Politics, 56(4), 520-553. Persson, T, Tabellini, G., & Trebbi, F. (2001). Electoral rules and corruption. In S. Piattoni (Ed.), Clientelism, interests, and democratic representation. Cambridge, UK: Cambridge University Press. Przeworski, A. (1991). Democracy and the market. Cambridge, UK: Cambridge University Press. Richardson, B. (2001). Japan’s ‘1995 System’ and beyond. In L. Diamond & R. Gunther (Eds.), Political parties and democracy (pp. 143-172). Baltimore: Johns Hopkins University Press. Robinson, J., & Verdier, T. (2002). The political economy of clientelism (Discussion Paper Series No. 3205). London: Centre for Economic Policy Research. Robison, R., & Hadiz, V. (2005). Reorganising power in Indonesia. London: Routledge Curzon. Sachsenröder, W. (1998). Party politics and democratic development in East and Southeast Asia: A comparative view. In W. Sachsenröder & U. Frings (Eds.), Political party systems and democratic development in East and Southeast Asia (pp. 1-35). Aldershot, UK: Ashgate. Schady, N. R. (2000). The political economy of expenditures by the Peruvian Social Fund (FONCODES) 1991-1995. American Political Science Review, 94(2), 289-304. Scheiner, E. (2006). Democracy without competition in Japan. Cambridge, UK: Cambridge University Press. Scott, J. (1998). Seeing like a state. New Haven, CT: Yale University Press. Seng, O. (1998). Singapore. In W. Sachsenröder & U. Frings (Eds.), Political party systems and democratic development in East and Southeast Asia (pp. 342-402). Aldershot, UK: Ashgate. Sherlock, S. (2002). Combating corruption in Indonesia? Bulletin of Indonesian Economic Studies, 38(3), 367-383. Sidel, J. (1999). Capital, coercion, and crime. Stanford, CA: Stanford University Press. Slater, D. (2003). Iron cage in an iron fist: Authoritarian institutions and the personalization of power in Malaysia. Comparative Politics, 36(1), 81-101. Slater, D. (2007, October). Can a Leviathan be a democrat? Competitive elections, robust mass politics, and state infrastructural power. Paper Presented at the 2007 Annual Meeting of the American Political Science Association, Chicago. Smyth, R., Wilkening, B., & Urasova, A. (2007). Engineering victory: Institutional reform, informal institutions, and the formation of a hegemonic party regime in the Russian Federation. Post-Soviet Affairs, 23(2), 1-21. Grzymala-Busse / Beyond Clientelism 673

Stokes, S. (2005). Perverse accountability: A formal model of machine politics with evidence from Argentina. American Political Science Review, 99(3), 215-225. Tarrow, S. (1990). Maintaining hegemony in Italy: The softer they rise, the slower they fall! In T. J. Pempel (Ed.), Uncommon democracies: The one-party dominant regimes (pp. 306-332). Ithaca, NY: Cornell University Press. Tilly, C. (1992). Coercion, capital and European states, AD 990-1992. Cambridge, MA: Blackwell. Transparency International. (2006). Global corruption report. London: Author. van de Walle, N. (2003). Presidentialism and clientelism in Africa’s emerging party systems. Journal of Modern African Studies, 41(2), 297-321. Warner, C. (2001). Mass parties and clientelism: France and Italy. In S. Piattoni (Ed.), Clientelism, interests, and democratic representation (pp. 122-151). Cambridge, UK: Cambridge University Press. World Bank. (2000). Cross-national data set on government employment and wages. Washington, DC: Author. World Bank. (2002). World governance indicators. Washington, DC: Author. Wurfel, D. (1988). Filipino politics: Development and decay. Ithaca, NY: Cornell University Press.

Anna Grzymala-Busse is an associate professor of political science at the University of Michigan. Her second book, Rebuilding Leviathan: Party Competition and State Politicization in Post-Communist Democracies, was published by Cambridge University Press in 2007.