The Triple Threat to Property Rights: Credible Commitment Dilemmas, Principal-Agent Problems, and Private Coercion

Transcription

1 The Triple Threat to Property Rights: Credible Commitment Dilemmas, Principal-Agent Problems, and Private Coercion Jordan Gans-Morse Northwestern University Currently under review Abstract Although prominent studies portray predatory rulers as the primary threat to property rights, lower-level state officials and unlawful private actors pose equally severe challenges. Secure property rights require mechanisms that (1) credibly constrain rulers, (2) resolve principal-agent problems between rulers and bureaucrats, and (3) reduce firms reliance on private coercion. Analyzing a formal model of this triple threat reveals three key findings. First, rulers credible commitment to respect property is insufficient to deter threats from lower-level officials and private actors. Second, lowerlevel officials predation and private actors coercion not only directly threaten property rights but also exacerbate rulers credible commitment dilemmas. Third, property security frequently depends more on state capacity than on constraining rulers. These findings suggest that a narrow focus on credible commitment underestimates threats to property rights, shedding light on why many countries struggle to secure property and build rule of law, even when rulers are institutionally constrained.

2 At least since the seminal work of North and Weingast (1989) on the Glorious Revolution in 17th century England, prominent studies of property rights have portrayed predatory rulers kings, dictators, and military juntas as the primary threat to secure property rights. Accordingly, these studies focus on credible commitment dilemmas: How can a ruler with a near monopoly on coercion credibly commit to respect property rights, given the ruler may face incentives to renege on agreements with investors or creditors once investments have been made or loans provided? A similar dilemma exists in many transactions between private actors, but in such cases institutions of third-party enforcement, such as courts and law enforcement agencies, frequently can deter both parties defection from the original bargain. When the state itself is involved, a third party with enough power to deter the state is unlikely to exist. North and Weingast (1989, 806) concluded that institutional arrangements to deter rulers must therefore be self-enforcing, meaning that major parties to the bargain must have an incentive to abide by the bargain after it is made. In 17th century England, they demonstrate how the creation of a powerful parliament and the rise of independent courts, backed by the credible threat of forcible dethronement, created self-enforcing checks and balances that limited the king s incentives to violate property rights. These institutional arrangements led to a dramatic improvement in the Crown s ability to raise funds. Following North and Weingast s contribution, the study of property rights became in many ways a quest to identify credible commitment mechanisms, both informal and formal, in a wide range of settings. 1 The importance of these findings is undeniable, but the focus on credible commitment has limited attention to two other equally severe threats to property rights: lower-level state officials and unlawful private actors. In contemporary societies, nearly all rulers rely on vast bureaucracies with extensive regulatory power over economic activities. As a consequence, the majority of firms, investors, and creditors face property 1 The credible commitment literature is discussed in greater detail in Section 1.1 below. 1

3 rights threats that emanate not from top-level rulers but from an army of lower-level predatory bureaucrats: regulators and inspectors who impose unjustified fines to extort bribes from entrepreneurs, licensing agencies that force firms to buy illegitimate licenses, and law enforcement officials who coercively expropriate private sector property rights. Meanwhile, whereas the credible commitment literature warns of the dangers of states strength, in today s developing countries it is often states weakness, as evidenced by the inability to prevent unlawful action by private actors, that results in threats to property rights. 2 Such private threats manifest in a number of forms, including mafia extortion rackets, private protection agencies, illegal corporate raiders, and unscrupulous firms seeking to gain a competitive advantage by employing illicit practices. When lower-level state officials and unlawful private actors are taken into account, it becomes clear that secure property rights require responses to a triple threat: (1) credible commitment mechanisms to constrain rulers, (2) mechanisms resolving principal-agent problems to monitor bureaucrats, and (3) mechanisms to reduce firms reliance on private coercion. This article develops a simple formal model to analyze how each of these threats individually undermines property rights, and how combinations of these threats create particularly novel challenges. The analyses produce three key findings. First, rulers credible commitment to respect property rights is insufficient to deter threats from lower-level officials and private actors. Second, lower-level officials predation and firms use of private coercion exacerbate the ruler s credible commitment dilemma. For example, predation by lower-level officials reduces the value of the assets on which law-abiding rulers collect tax revenue, making rulers commitments to refrain from expropriation less credible. Meanwhile, while firms incentives to rely on private coercion decline as the tax rate falls, lowering taxes 2 Consider the formulation of what is often referred to as the Weingast Dilemma : The fundamental political dilemma of an economic system is this: A government strong enough to protect property rights and enforce contracts is also strong enough to confiscate the wealth of its citizens (Weingast, 1995). But in much of the world, rulers are strong enough to confiscate wealth yet also too weak to prevent property rights violation by private actors and lower-level officials. 2

4 makes commitment to property rights protection less appealing to rulers. Finally, addressing the triple threat to property rights requires not only constraints on rulers but also substantial state capacity. In other words, state weakness is just as likely as overweening state strength to undermine property rights. 3 These findings shed light on why so many countries even those with institutions that constrain rulers struggle to secure property and build the rule of law. They also present a conundrum for scholars and policymakers: State capacity may be essential for making property rights secure, but in many developing countries improving state capacity is politically or technically infeasible. The final section of this article therefore examines the important role that firms and citizens bottom-up strategies for protecting property can play in the development of secure property rights, particularly when the primary threats to property emanate from lower-level state officials or unlawful private actors. 1 The Triple Threat 1.1 Threats From Rulers and Credible Commitment Mechanisms Central to credible commitment dilemmas is a time-inconsistency problem. A ruler, recognizing that guarantees are necessary to induce investment or to obtain loans, may bargain with constituents or creditors to set a certain tax rate or pay a certain interest rate. But once the investment is made or the loan provided, the ruler may have incentives to unilaterally alter the terms of the agreement. Existing studies of property rights focus on two solutions to credible commitment problems: stationary banditry and limited government. 4 Olson (1993) famously suggested a distinction between rulers with long and short time horizons, or, in Olson s terms, between 3 Beyond debates over property rights, this article builds on and contributes to formal analyses of interactions among rulers, bureaucrats, and citizens (e.g., Egorov et al., 2009; Dixit, 2010) and models in which firms choose between predation and productive investment (e.g., Johnson et al., 1997; Roland and Verdier, 2003). 4 For an insightful discussion of these two mechanisms, see Haber et al. (2003, 3-6, 24-29). 3

5 stationary and roving bandits. In the case of stationary bandits, self-enforcing bargains between rulers and citizens can exist as long as rulers value tax income from well-protected producers over the long term more than the one-time immediate bounty from expropriating subjects assets. However, stationary bandit mechanisms are vulnerable to sudden shifts in a ruler s time horizons, as might occur with the advent of war (North and Weingast 1989, 807; Haber et al. 2003, 4). Considering the deficiencies of stationary banditry, most scholars recognize limited government as a preferable solution to credible commitment problems. As Haber et al. (2003, 4) explain, limited governments are those that respect individual rights as a matter of law, are bound by self-enforcing institutions to respect their own laws, and cannot arbitrarily alter the laws that constrain them. With limited government, attempts by the ruler to exceed authority require overcoming numerous hurdles and result in sanctions. In North and Weingast s (1989) initial study of credible commitment, competing branches of government the parliament and judiciary used the power of the purse and the legitimacy of the law to stifle attempts by the King to breach the rights of his constituents. Moreover, parliament, having successfully participated in the forcible removal of two kings in the preceding decades, could credibly back its legislative powers with the threat of coercion (pp ). In later work, Weingast (1995, 1997) examined the role of federalism and coordinated threats of rebellion in sustaining credible commitment, while other scholars identified mechanisms, such as coalitions of merchants in medieval Europe, that limit the authority of rulers by collectively punishing property rights violations (Root, 1989; Greif et al., 1994). 5 Much of the credible commitment literature emerged from analyses of the West s historical experience. But in contemporary developing countries, expropriation by dictators, central governments, and other rulers remains a major threat to property rights. Albertus 5 For discussion of other credible commitment mechanisms, such as hostage taking and access to specialized knowledge, see Diermeier et al. (1997, 24) and Haber et al. (2003, 24-26). 4

6 and Menaldo (2012, 974) find that in at least 11 Latin American autocracies, rulers conducted a large-scale expropriation of land, natural resource companies, or commercial banks at some point during the latter half of the 20th century. Meanwhile, in contemporary Russia tycoons often face threats to their property directly from President Vladimir Putin s inner circle, as was the case in the high-profile imprisonment of Mikhail Khodorkovsky and expropriation of his oil company Yukos (Sakwa, 2014). Moreover, attacks on property rights by high-level rulers are hardly limited to authoritarian regimes. Wellhausen (2014) documents the expropriation of dozens of foreign multinational companies in Ukraine, Moldova, and Romania throughout the 1990s and 2000s. Given such threats, studies of property rights in developing countries have largely continued to focus on credible commitment. Through analysis of the Porfirio Diaz regime and revolutionary Mexico, Haber et al. (2003) develop a theory of selective property rights enforcement based on vertical political integration, in which powerful private sector actors overcome informational asymmetries by serving in government agencies, helping to write policies and legislation, or acting as informal advisers to rulers. This proximity to power allows private sector actors to monitor rulers behavior and coordinate sanctions with other tycoons, labor organizers, or political actors should the ruler seek to infringe on property rights. Gehlbach and Keefer (2011) question how private sector investment occurs in modernday authoritarian regimes, in which limited government by definition does not exist. They show how single-party regimes, as in contemporary China, selectively create credible commitment via the institution of a ruling party, which helps ensure that a chosen coalition of investors will not face expropriation from the regime s leaders. Attention to the non-western world also led analysts to begin looking beyond top-level rulers in their analyses of property rights security, as in Firmin-Sellers s (1995) study of credible commitment mechanisms restraining lower-level tribal chiefs during the commercialization of land in colonial Ghana. However, as discussed in the next section, the threat 5

7 posed by lower-level government officials is better conceptualized as a principal-agent than a credible commitment problem, and until recently the literature on property rights has had little to say about this second component of the triple threat to property rights. 1.2 Threats from Predatory Bureaucrats and Principal-Agent Problems Although De Soto (1990) recognized bureaucratic obstacles as a threat to property rights more than two decades ago, scholars are just beginning to examine the role of principalagent problems and administrative capacity in relation to property rights (see Greif, 2008; Levi and Epperly, 2010; Markus, 2012, 2015). As McCubbins et al. (1989, ) explain, principal-agent problems result when one person (the principal) arranges for another (the agent) to take an action that is beneficial to the principal but costly to the agent, under circumstances when the principal cannot perfectly and costlessly enforce an ex ante promise by the agent to act in the best interests of the principal. Analysis of such dilemmas gained prominence among economists in the 1970s and 1980s, particularly with reference to the separation of management and ownership in contemporary firms (e.g., Jensen and Meckling, 1976). Scholars of American politics then adopted the principal-agent framework for analyzing congressional control of bureaucratic agencies (Weingast and Moran, 1983; Weingast, 1984; McCubbins et al., 1989). The principal-agent framework acquired traction early on and continues to wield influence in studies of bureaucratic corruption (Rose-Ackerman, 1978; Klitgaard, 1988; Shleifer and Vishny, 1993). However, scholars only recently have recognized the severity of the threat to property rights posed by lower-level state officials. Markus (2015) in particular has developed a theoretical framework that distinguishes between bureaucratic corruption and what he terms agent predation. Whereas bribe requests and other forms of corruption target income streams, often in a relatively predictable manner, agent predation targets ownership rights. This distinction between income and ownership rights is critical. As Markus (2015, 6

8 29) explains, While threats to income streams generally do not jeopardize the continued operation of a firm, even if they reduce its profits, threats to ownership rights fundamentally restrict the ways in which the owners can manage their firm, at the extreme jeopardizing the very existence of the enterprise (emphasis in original). For example, state officials including law enforcement agents, tax collectors, inspectors, and regulators may compel profitable enterprises to turn over an ownership stake at below-market prices or force firms out of business on behalf of private clients who pay for such services. In the context of post-communist Russia and Ukraine, Markus (2015) identifies numerous examples of agent predation, an observation confirmed by other scholars of the former Soviet Union. For example, despite the infamy of Russia s mafia protection rackets, firms in today s Russia find local government officials far more fearsome. In the words of a consultant to small businesses in Moscow: Who cares about criminals? Inspectors can close you in a matter of seconds. This is in itself a kind of mafia system (cited in Gans-Morse 2012, 284). Meanwhile, tax authorities and local government officials in post-soviet Georgia pressure entrepreneurs to give up real estate in favour of state or state-favoured private businessmen under the threat of... initiating criminal prosecution related with tax fraud, drug trade or other crimes (Kupatadze, 2010, 167), while in Kyrgyzstan formal and informal leadership of police units have used them in a struggle for control of economic activities (Kakachia and O Shea, 2012, 7). Nor is agent predation limited to the former Soviet Union. In the ostensibly more law-abiding context of Poland, Abrams (2014, 329) finds that local politicians and officials collude to make life impossible for firms that do not belong to privileged political networks. And in China, Lu (2000, 274) finds that despite local governments heralded involvement in promoting and adapting to the market economy, there is also extensive distortion and deviance epitomized by predatory behavior of local state agencies. In one striking example, Lu (2000, 286) draws attention to a mid-sized city in which firms on average pay 272 non-tax charges amounting to a sum equal to the city s total annual revenue. 7

9 With respect to resolving principal-agent problems, scholars have proposed numerous mechanisms, nearly all of which involve some combination of efficiency wages, improved monitoring of agents, and/or the instillment of professional norms and mores (Rose-Ackerman, 1986; Klitgaard, 1988; Besley and McLaren, 1993). Three points are of relevance concerning these approaches, all of which are elucidated in the models introduced below. First, none of these mechanisms involve credible commitment, attesting to the disconnect between the property rights literature s focus on credible commitment and one of the most prevalent types of property rights threats. Second, nearly all of these approaches require the existence of substantial state capacity and administrative resources; how states that lack such capacity should deal with lower-level threats to property rights has received much less attention. Third, to the best of my knowledge, no existing study has sought to integrate analysis of credible commitment and principal-agent problems. The formal analyses below examine how principal-agent dilemmas are not only a problem for the security of property rights in and of themselves, but how they additionally exacerbate credible commitment dilemmas, making it tougher to constrain top-level leaders. 1.3 Threats from Private Actors and the Challenge of Private Coercion Just as prominent property rights studies have focused disproportionately on threats from top-level rulers rather than lower-level officials, they have devoted minimal attention to threats from private actors, assuming the existence of a state strong enough to limit private predation. Almost by definition, modern states are assumed to have eradicated challengers to their monopoly on legitimate coercion and to have created the foundation for market economies in which firms compete on the basis of economic criteria (e.g., a product s quality or price) rather than on access to cheaper protection costs (Volkov, 2000, ). Yet the experience of lesser developed countries has more recently forced scholars to consider 8

10 environments in which an extensive state apparatus coexists with powerful private actors who successfully compete with the state to provide protection services. In some Russian cities in the 1990s, approximately 40 percent of small businesses reported recent encounters with criminal protection rackets, and around one-third of respondents indicated that such rackets helped to enforce contracts (Frye and Zhuravskaya, 2000, ). At the level of big business, Russia s emerging tycoons created their own private armies of security agents, bodyguards and commercial spies (Hoffman, 1997). This widespread reliance on private coercion not only created a number of physical threats for would-be entrepreneurs in 1993, 61 bombs exploded in Moscow as criminal groups battled for territorial control (Hoffman, 2002, 277) but also undermined the state s ability to develop effective institutions of law and order, as potential tax revenues instead flowed into the coffers of criminal protection rackets (Johnson et al., 1997; Roland and Verdier, 2003). The post-communist world has garnered significant attention for the extensiveness of private threats. But such threats are prevalent in a number of countries. In parts of Ghana, the private protection of property... has been contracted to people referred to by property owners as Land Guards for whom the use of force or threat of coercion is critical to their ability to enforce property rights (Joireman, 2011, ). In Indonesia private security companies not only guard land and buildings but also offer a wide range of services including the intimidation of a client s business rivals (Wilson, 2010, 255). Meanwhile, De Soto (2003, 155) finds that many firms in the Peruvian informal sector rely on the protection that local bullies or mafias are willing to sell them, a fate shared by informal sector entrepreneurs in cities ranging from Chicago (Venkatesh, 2006) to Taipei (Winn, 1994). And while mafia protection rackets role in protecting property and enforcing contracts in southern Italy has received extensive attention (Gambetta, 1996), less well-known is the similar role played by criminal rackets in Japan, where Milhaupt and West (2000, 66-67) find 9

11 that The influence of organized crime is readily apparent in bankruptcy and debt collection, property development, dispute settlement, shareholders rights, and finance. To the extent that scholars have addressed the question of how to mitigate private threats to property rights, most attention has been on developing state capacity, with an emphasis on improving law enforcement and courts (Johnson et al., 1997; Roland and Verdier, 2003). Even in the case of Japan, a country with relatively effective state institutions, Milhaupt and West (2000) suggest that a key component to combatting private coercion is to rewrite formal laws in order remove barriers to firms use of the legal system. As with principalagent problems, three points, all of which are central to the formal analyses below, are of relevance. First, credible commitment plays a minimal role in discussions of private coercion, again illustrating the limitations of analyses focused on constraining rulers. Second, many analysts concur that state capacity is essential not only for mitigating principal-agent problems but also for reducing firms reliance on private coercion. Third, existing studies have yet to consider the interaction of credible commitment dilemmas and private threats to property rights. The models introduced below show that firms reliance on private coercion, like principal-agent problems, exacerbate rulers credible commitment dilemma. In summary, property rights security requires addressing threats from three sources: toplevel rulers, lower-level state officials, and private actors. Examination of these three threats requires an understanding of credible commitment dilemmas, principal-agent problems, and the challenges resulting from competition to the state s monopoly on coercion. These problems individually undermine property rights security. When combined, the dilemmas of property rights security become even more severe, as elucidated in the following section. 10

12 2 A Model of the Triple Threat This section presents three versions of a simple formal model. 6 The analyses demonstrate (1) the insufficiency of credible commitment for containing threats to property from lowerlevel officials and private actors, (2) the extent to which threats from lower-level officials and private actors exacerbate the ruler s credible commitment dilemma, and (3) the importance of state capacity for the emergence of broad-based property security. The first model serves as a baseline in which principal-agent problems and threats from private actors are absent. The second incorporates the threat of predatory bureaucrats. The third then considers credible commitment in the presence of private threats but without principal-agent problems. For each model, I analyze the equilibrium outcomes for a credible commitment condition, representing the minimum level of constraints required to keep rulers from expropriating, and a profitability condition, representing the minimum return firms require in order to invest. In the latter two models I additionally analyze a state capacity condition, representing the minimum level of state effectiveness required to thwart threats from lower-level officials or private actors. 2.1 Baseline Credible Commitment (CC) Model Players and Timing: This section analyzes a baseline Credible Commitment (CC) model, consisting of a simple single-period perfect information game with two players: a firm (F) and a ruler (R). The ruler cares only about material enrichment and receives no utility from improving public welfare. In the spirit of North and Weingast (1989), the firm decides whether to make an investment of value V > 0. If the firm chooses not to invest, the game ends. If the firm invests, then the ruler decides whether or not to expropriate the investment. Expropriation can take a number of forms. Depending on the nature of the firm s investment, 6 For the sake of elucidating the key ideas, I present the following as three distinct models. However, I show in the Online Appendix that modeling the three games in a unified framework produces identical results. 11

13 it could mean the outright confiscation of land, goods, or productive assets. It could also mean an arbitrary change in tax or regulatory policy. Or, in the context of a creditor offering the ruler a loan, it could mean the refusal to pay off the debt with interest. A simple way to model expropriation is that once the firm makes an investment, the ruler sets a tax rate t [0, 1]. The fact that the tax rate is set after the firm chooses whether or not to invest captures the ruler s commitment problem: The firm will not perceive any tax rate set before the investment as credible. With no constraints, the self-interested ruler naturally would be tempted to set t = 1 and fully expropriate the firm s investment. However, there exists a threshold tax rate, τ [0, 1], which is common knowledge. The ruler knows that if the tax rate is set above this threshold, firms and society will consider this a form of expropriation and the ruler will incur a cost C for violating property rights. The origins of this threshold are external to the model, but the threshold can be thought of as the outcome of a bargaining or coordination game between the ruler and the enforcer(s) of whatever credible commitment mechanism inflicts the sanction C. 7 The parameter C is of primary interest and can be interpreted in multiple ways. In the North and Weingast narrative, C is the cost the king would have to pay to overcome the constraints of an independent parliament and courts in order to expropriate property rights. In Weingast s (1997) analysis of collective action s role in restraining authority, C is the cost of the societal uprising the ruler will face if he expropriates too many assets. 8 Note 7 An alternative way to set up the model would be to endogenize the ruler s choice of the tax rate by replacing the ruler s strategy space s R = {expropriate, no expropriate} with a continuous strategy space s R = t [0, 1]. Given a threshold tax τ above which the ruler incurs sanctions C, the ruler s equilibrium payoffs will be the same as in the setup above, for the ruler optimally will choose to set t = 1 (corresponding to expropriating) or t = τ (corresponding to not expropriating). The reason is that for any t such that τ < t < 1, the ruler incurs the sanctions C but receives a payoff of tv C. This is strictly less than V C, so the ruler is better off raising the tax rate to t = 1. Meanwhile, for any t such that 0 t < τ the ruler is strictly better off choosing t = τ, which still does not incur sanctions C but brings in more tax revenue. 8 As discussed in Section 1.1, another source of credible commitment pertains to Olson s (1993) theory of rulers as stationary versus roving bandits. A stationary bandit cares about future tax revenue in a game of repeated play. The loss of this future revenue is the cost of expropriation, reducing the ruler s willingness to violate property rights. Diermeier et al. (1997) offer a simple formal model in the spirit of Olson s theory. Here, C can be considered a reduced form way of incorporating the loss of future income streams. 12

14 that C measures the difficulty of creating an effective credible commitment mechanism for a given society. When a low C is sufficient to incentivize the ruler to forgo expropriation, then creating secure property rights is easier. When a high C is required to prevent expropriation, the challenge of securing property rights becomes more difficult. Payoffs: As seen in Figure 1, if the firm chooses not to invest, both the firm and the ruler receive a payoff of 0. If the firm invests, payoffs are as follows: F invests, R does not expropriate: When the firm invests and the investment is not expropriated, the value of the assets increase to rv, where r > 1. The ruler receives a share of the resulting revenue proportional to the tax rate τ, while the firm receives the remaining revenue less the initial costs of investment. The ruler s payoff is therefore equal to tax revenues of τrv, and the firm s payoff is (1 τ)rv V. F invests, R expropriates: If the ruler chooses to expropriate, it is assumed that the firm realizes no returns on its investment (i.e., r = 1), representing the idea that private sector actors possess entrepreneurial skills that the ruler cannot replicate (Haber et al., 2003, 24-26). The ruler acquires the full value of the initial investment V but faces sanctions for violating property rights, giving the ruler a payoff of V C. The firm, meanwhile, loses its initial investment and receives a payoff of V. Analysis: In the baseline Credible Commitment game, it is straightforward to show that as long as the firm s profitability condition is satisfied, the ruler s credible commitment is necessary and sufficient for investment to occur. By backwards induction, first consider the ruler s credible commitment condition. The ruler prefers not to expropriate if and only if the sanctions for expropriating C are sufficiently large, taking into consideration the tax rate τ and return on investment r. More formally, the ruler will refrain from expropriating if and only if τrv V C, or equivalently if and only if: 9 9 I assume that τr < 1 so that the threshold sanctions for expropriating are strictly positive. 13

15 Figure 1: Baseline Credible Commitment (CC) Game F invest R no invest 0 0 expropriate no expropriate V V C (1 τ)rv V τrv C C cc = V (1 τr) (1) Meanwhile, the firm will invest if and only if sanctions C are sufficiently large to deter the ruler s expropriation and the return on investment r is sufficiently large such that the firm s post-tax payoff is greater than the firm s initial investment. Formally, the firm s profitability condition is (1 τ)rv V 0, or equivalently: r r cc = 1 (1 τ) (2) For any C C cc and r r cc, the unique Subgame Perfect Nash Equilibrium (SPNE) of the baseline Credible Commitment Game is for the firm to invest and the ruler to not expropriate. Note that when the credible commitment condition does not hold, both the firm and the ruler suffer. From the ruler s perspective, the optimal scenario is for the firm to invest and the ruler to expropriate. But the ruler is still better off under the scenario in which the firm invests and the ruler does not expropriate (yielding a payoff for the ruler of τrv > 0) than 14

16 under the scenario in which the firm does not invest (yielding a payoff for the ruler of 0). Yet when an effective credible commitment mechanism is lacking, there is no way for the ruler to induce investment, illustrating the essence of the credible commitment dilemma. Given that C cc and r cc serve as a baseline for comparing the minimum sanctions on property rights violations needed to secure property rights and the minimum return rate at which investment is profitable, Equations 1 and 2 deserve additional attention. Note that the higher the value of the firm s investment V, the higher the sanctions required to prevent expropriation. As the asset s value rises, the more the ruler will be tempted to expropriate (which captures the entire increase in the asset s value) rather than settle for tax revenue (which captures only a share of the increase in the asset s value proportional to the tax rate). The more the ruler gains by expropriating, the more difficult it will be to secure property. Meanwhile, the higher the maximum tax threshold τ or the higher the return rate r, the greater the ruler s payoff to not expropriating. Correspondingly, the easier it will be to deter property rights violations. However, the higher τ, the larger the return r must be to induce the firm to invest. There is thus a tension with respect to τ: A higher τ makes it easier to secure property rights from the perspective of constraining the ruler, but it comes at the cost of making investment less likely from the perspective of ensuring profits for the firm. 2.2 Principal-Agent Model The model in this section, which I will refer to as the Principal-Agent (PA) game, now incorporates the risk of expropriation by lower-level state officials bureaucrats, inspectors, law enforcement agencies into the baseline Credible Commitment game in order to analyze how principal-agent dilemmas can complicate the credible commitment problem. The analysis reveals three key results. First, the ruler s credible commitment to respect property rights is insufficient to restrain lower-level officials from expropriating. Second, although investment is still possible even when lower-level officials expropriate, both the credible com- 15

17 mitment sanctions and the minimum return on investment needed to sustain such equilibria are higher than in the no-expropriation equilibrium in the baseline Credible Commitment game. In other words, the existence of a principal-agent problem exacerbates the ruler s credible commitment dilemma. And third, thwarting bureaucratic expropriation requires sufficient state capacity, rather than credible commitment. More specifically, the bureaucrat will continue to expropriate unless the ruler can offer a sufficiently high salary and/or effectively monitor and punish bureaucrats who threaten property rights. For elucidation of key arguments, the following setup does not allow for collusion between the ruler and bureaucrat, as in cases in which lower-level officials share corrupt revenues with their superiors, or in which rulers allow for a degree of lower-level corruption in exchange for political loyalty. However, in the Online Appendix I demonstrate that all findings discussed here are robust to extensions of the game that take collusion into account. Players and Timing: The PA model consists of a perfect information game among three actors: a firm (F ), a bureaucrat (B), and the ruler (R). The firm moves first and decides whether to invest. If the firm chooses not to invest, the game ends. If the firm invests, the bureaucrat then chooses whether to expropriate. In the final stage of the game, the ruler sets the tax rate, expropriating the firm s assets by setting t = 1 or forgoing expropriation and setting t at the threshold level τ above which sanctions are incurred, as in Figure 2. Payoffs: If the firm does not invest, then all actors receive a payoff of 0, as in the baseline Credible Commitment game. If the firm invests, then payoffs are as follows: F invests, B does not expropriate, R does not expropriate: When both the bureaucrat and ruler forgo expropriation, the payoffs for the firm and the ruler are nearly the same as in the Credible Commitment model, with the exception that the ruler must now pay the bureaucrat a salary S > 0. The investment is realized and the ruler receives a tax share 16

18 minus payment to the bureaucrat for a payoff of τrv S. The bureaucrat receives his salary S, and the firm receives after tax profits of (1 τ)rv V. F invests, B does not expropriate, R expropriates: If the ruler expropriates after the bureaucrat forgoes expropriation, he receives the full value of the asset. But returns on the investment are not realized, the ruler incurs a sanction for expropriation, and the ruler pays the bureaucrat s salary. The ruler s payoff is V C S, while the bureaucrat receives his salary S and the firm loses its initial investment and receives a payoff of V. F invests, B expropriates, R does not expropriate: When the bureaucrat expropriates, the ruler catches the corrupt bureaucrat and recovers the assets with probability q (0, 1). As punishment, the bureaucrat loses his salary S. If the ruler recovers the assets and decides not to expropriate, he returns them to the firm. The process of recovering and returning assets to the firm leads to a loss of resources, the extent of which depends on the effectiveness of the legal system, represented by the parameter α (0, 1). The ruler s expected payoff for not expropriating therefore is qατ rv (1 q)s. That is, with probability q the ruler catches the corrupt bureaucrat, pays no salary, acquires the assets, returns these to the firm and receives a tax share of the realized investment, minus the resources lost during the restitution process; with probability 1 q the bureaucrat successfully hides the theft of the assets and the ruler pays his salary. Correspondingly, the bureaucrat s expected payoff is (1 q)(v + S). With probability q the bureaucrat is caught and receives a payoff of 0; with probability 1 q the bureaucrat successfully hides the theft and keeps the assets and his salary. Finally, the firm s expected payoff after taxes now depends both on the probability the assets are recovered and the efficiency of the legal system: qα(1 τ)rv V. F invests, B expropriates, R expropriates: The ruler again catches the corrupt bureaucrat and recovers the assets with probability q, in which case the bureaucrat loses his salary S. If the ruler recovers the asset and decides to expropriate, then he faces a punishment of C. It follows that the ruler s expected payoff for expropriating is q(v C) (1 q)s. That 17

19 Figure 2: Principal-Agent (PA) Game F no invest invest B expropriate no expropriate R R expropriate no expropriate expropriate no expropriate V (1 q)(v + S) q(v C) (1 q)s qα(1 τ)rv V (1 q)(v + S) qατrv (1 q)s V S V C S (1 τ)rv V S τrv S is, with probability q the ruler receives the full value of the asset minus the costs imposed by the credible commitment mechanism; with probability 1 q the bureaucrat successfully hides the theft of the assets and the ruler pays his salary. The bureaucrat s expected payoff is again (1 q)(v + S). Whether or not the ruler catches the thieving bureaucrat, the firm loses its investment and receives a payoff of V. Analysis: Whenever the ruler expropriates, the firm loses its investment and receives a payoff of V. The firm is strictly better off not investing and receiving a payoff of 0. Therefore, without the ruler s credible commitment, no equilibrium exists in which the firm invests. However, while the ruler s credible commitment remains necessary for investment, credible commitment alone will not prevent bureaucratic expropriation. Proposition PA-1. Even when C C pa, Sub-Game Perfect Nash Equilibria exist in which the bureaucrat expropriates (i.e., credible commitment is insufficient to deter bureaucratic expropriation). Consider first the conditions for the ruler to not expropriate in the history of the game 18

20 in which the bureaucrat expropriates. Whether or not the ruler expropriates, he will pay the bureaucrat s salary S with probability 1 q. The decision to expropriate thus rests solely on whether expected tax revenues when not expropriating are greater than expected revenues from expropriating and incurring sanctions, that is, on whether qατ rv q(v C), which implies the following credible commitment condition: C C pa = V (1 ατr) (3) For any C C pa, the ruler will respect property rights. 10 But from the bureaucrat s perspective, the fact that the ruler credibly commits is inconsequential: When expropriating, the bureaucrat s payoff does not depend on whether or not the ruler chooses to expropriate, but only on whether or not the ruler manages to punish the corrupt act. The bureaucrat will thus forgo expropriation if and only if the state capacity condition S (1 q)(v + S) is satisfied, or, equivalently, if and only if: 11 q q pa = V V + S (4) Thus, for any C C pa and q < q pa, the ruler s credible commitment is satisfied, but the ruler cannot credibly guarantee that the bureaucrat will refrain from expropriating. The firm s decision to invest or not depends on whether its profitability condition (discussed below) is satisfied. When satisfied, and when C C pa and q < q pa, straightforward backward induction shows that the strategy profile in which the firm invests, the bureaucrat expropriates, 10 Note that the ruler s credible commitment condition in the history of the game in which the bureaucrat does not expropriate will be satisfied whenever the credible commitment condition (Equation 3) in the history of the game with bureaucratic expropriation is satisfied. 11 State capacity here has two dimensions: resources to pay salaries (S) and technology to monitor agents (q). The state capacity condition could be written in terms of S or q. I choose to write the condition in terms of q for the sake of comparability with the conditions in the Private Coercion game below. 19

21 and the ruler does not expropriate in any history of the game is an SPNE in which credible commitment exists but bureaucratic expropriation persists. 12 Although it is still possible to support investment in an equilibrium in which the bureaucrat (but not the ruler) expropriates, the threat of bureaucratic expropriation raises the level of sanctions needed for the ruler to credibly commit and increases the firm s minimum rate of return needed to invest. Proposition PA-2. Investment with bureaucratic expropriation requires C C pa > C cc (i.e., bureaucratic expropriation exacerbates credible commitment problems: relative to the baseline Credible Commitment game, investment requires stronger constraints on rulers). When lower-level officials expropriate, the ruler receives a tax share only after the assets have been expropriated and returned to the firm in a costly restitution process. Given that expropriation is therefore relatively more appealing to the ruler than in the absence of principal-agent problems, preventing expropriation requires harsher sanctions. More formally, to see that sanctions needed to maintain the ruler s credible commitment are higher in equilibria with bureaucratic expropriation than in the no-expropriation equilibrium from the baseline Credible Commitment game, compare C pa = V (1 ατr) in Equation 3 and C cc = V (1 τr) in Equation 1. Meanwhile, the firm will invest, knowing that the bureaucrat will expropriate the assets but that the ruler will catch the bureaucrat with probability q and return some of the assets, as long as expected after tax profits outweigh the initial costs of investment, such that q(1 τ)αrv V 0. The profitability condition is therefore: r r pa = 1 αq(1 τ) (5) For any C C pa, q < q pa, and r r pa, the unique SPNE in the Principal-Agent game is for the firm to invest, the bureaucrat to expropriate, and the ruler to not expropriate 12 When the profitability condition is not satisfied, the strategy profile in which the firm does not invest, the bureaucrat expropriates, and the ruler does not expropriate in any history of the game is an SPNE. 20

22 in any history of the game. 13 But in this equilibrium, investment requires not only more stringent credible commitment sanctions, but also a higher return rate than in the noexpropriation equilibrium from the baseline Credible Commitment game (compare r pa = 1 in Equation 5 and αq(1 τ) rcc = 1 (1 τ) in Equation 2). Because the firm knows that its assets will be expropriated with positive probability, investment projects that were feasible in the no-expropriation equilibrium are no longer profitable. The no-expropriation equilibrium from the baseline Credible Commitment game can be recovered in the Principal-Agent game. But in addition to credible commitment, this requires state capacity. Proposition PA-3. Given C C cc, q q pa, and r r cc, there exists a Sub-Game Perfect Nash Equilibrium in which neither the ruler nor bureaucrat expropriates and the firm invests (i.e., with sufficient state capacity, bureaucratic expropriation can be deterred and the lower sanctions from the baseline Credible Commitment game are sufficient for investment). When the state capacity condition in Equation 4 holds, the bureaucrat s salary and/or the risk of punishment will be sufficiently high to deter bureaucratic expropriation. With no risk of bureaucratic expropriation, it is straightforward to see by backward induction that the decisions faced by the ruler and firm are identical to those faced in the baseline Credible Commitment game. In the history of the game in which the bureaucrat chooses not to expropriate, the ruler also will forgo expropriation as long as τrv S V C S, or equivalently, as long as C C cc = V (1 τr), the credible commitment threshold from the baseline Credible Commitment game in Equation 1. Similarly, the firm will invest as long as (1 τ)rv V 0, or equivalently, as long as r r cc = 1, the minimum return on (1 τ) investment threshold from the baseline Credible Commitment game in Equation 2. There 13 Note again that the ruler s credible commitment condition in the history of the game in which the bureaucrat does not expropriate will be satisfied whenever the credible commitment condition (Equation 3) in the history of the game with bureaucratic expropriation is satisfied. 21

23 will be an SPNE in which the firm invests, the bureaucrat does not expropriate, and the ruler does not expropriate on the equilibrium path. 14 In summary, when principal-agent problems are considered in tandem with the credible commitment dilemma, three key findings emerge: (1) Credible commitment mechanisms are insufficient to deter bureaucratic expropriation. (2) Investment remains possible in the presence of bureaucratic expropriation, but only if sanctions for rulers expropriation are particularly stringent and returns on investment are particularly lucrative. In other words, the ruler s inability to restrain the bureaucrat s expropriation exacerbates the credible commitment dilemma. And (3) reducing threats to property rights from lower-level officials requires state capacity, not merely constraints on rulers. 2.3 The Private Coercion (PC) Game This section presents a third model, the Private Coercion (PC) game, that incorporates the threat of private actors into the baseline Credible Commitment game. As discussed in Section 1.3, in many countries private security agencies and criminal protection rackets constitute not only a direct threat to firms property rights but also are widely used by entrepreneurs and investors to protect their assets. This widespread reliance on private force constitutes a threat to the state s monopoly on coercion with important ramifications for the security of property. More specifically, the analyses that follow reveal three key results that parallel the findings from the Principal-Agent game. First, just as credible commitment is insufficient to mitigate bureaucratic expropriation, it is also insufficient to deter private threats to property rights. Second, when private coercion plays a significant role in an economy, the ruler may find it more difficult to credibly commit, particularly when state capacity is low. The reason is that when state institutions are relatively ineffective, firms will abandon private coercion in favor of state protection only if state institutions are cheap (i.e., if the tax rate is low). 14 Off the equilibrium path, in the history of the game in which the bureaucrat expropriates, the ruler will expropriate if C pa > C C cc and not expropriate if C C pa > C cc. 22

24 But the lower the tax rate, the less a law-abiding ruler benefits from forgoing expropriation, increasing the ruler s incentives to resort to expropriation. Third, as with efforts to reduce bureaucratic expropriation, mitigating firms reliance on private coercion requires a sufficient level of state capacity. Players and Timing: The PC model is a two-player perfect information game consisting of a firm (F) and a ruler (R). The firm moves first, but unlike the baseline Credible Commitment game, the firm now has three choices. It may choose not to invest, to invest but hide its output from the state and rely on private coercion for protection, or to invest and rely on state protection of its property rights. If the firm chooses not to invest or to invest and rely on private protection, the game ends. If the firm invests and relies on state protection, then the ruler faces the choice of whether or not to expropriate. Payoffs: If the firm chooses not to invest, both the firm and the ruler receive a payoff of 0. Otherwise, the payoffs are as follows: F invests/uses private protection: If the firm chooses to invest and utilize private protection, its payoffs depend on the effectiveness of private coercion relative to law, represented by the parameter β > 0. When β < 1, coercion is less effective than law; when β > 1, coercion is more effective. Additionally, with probability q the state detects this unofficial economic activity and confiscates the firm s assets. Thus, as can be seen in Figure 3, the firm s expected payoff to investing and relying on private coercion is (1 q)rβv V. The ruler s expected payoff when the firm invests and uses private coercion is qv. Here, no return on the firm s investment is realized (i.e., r = 1), reflecting the fact that the ruler confiscates illegal firms assets whenever law enforcement identifies illegal economic activity, not necessarily after firms investments come to fruition. Confiscation of illicitly produced assets is not considered expropriation, so the ruler does not incur sanctions. F invests/seeks state protection, R does not expropriate: If the firm invests and seeks state protection and the ruler does not expropriate, then the firm s investment comes to fruition. 23