وام های بین المللی آخرین علاج و مخاطرات اخلاقی: یک مدل مالی کاتالیزوری صندوق بین المللی پول

This paper analyzes the trade-off between official liquidity provision and debtor moral hazard in international financial crises. In the model, crises are caused by the interaction of bad fundamentals, self-fulfilling runs and policies by three classes of optimizing agents: international investors, the local government and an international official lender. Limited contingent liquidity support helps to prevent liquidity runs by raising the number of investors willing to lend to the country for any given fundamentals, i.e., it can have catalytic effects. The influence of the official lender is increasing in the size of its interventions and the precision of its information. Unlike the conventional view stressing debtor moral hazard, our model identifies circumstances in which official lending actually strengthens a government's incentive to implement desirable but costly policies.

In the last decade, many emerging market economies have experienced currency, debt, financial and banking crises: Mexico, Thailand, Indonesia, Korea, Russia, Brazil, Ecuador, Turkey, Argentina and Uruguay, to name the main ones. At the time of the crisis, each of these countries faced a massive reversal of capital flows and experienced a large drop in economic activity. In every case, large external financing gaps emerged because of strong capital outflows and the unwillingness of investors to rollover short-term claims on the country (including those on its government, its banks and its residents). A leading view is that these international crises are primarily driven by liquidity runs and panics, and could, therefore, be avoided via the provision of sufficient international liquidity to countries threatened by a crisis. According to this view, the global financial architecture should be reformed by creating an international lender of last resort (ILOLR). Not only would such an institution increase efficiency ex post by eliminating liquidation costs and default in the event of a run: by severing the link between illiquidity and insolvency, it would also prevent crises from occurring in the first place (see Sachs, 1995; Fischer, 1999). An opposing view doubts that international illiquidity is the main factor driving crises. When crises can also be attributed to fundamental shocks and policy mismanagement, liquidity support may turn into a subsidy to insolvent countries, thus generating debtor and creditors moral hazard (see the Meltzer Commission, 2000).1 Accordingly, IMF interventions should be limited in frequency and size so as to reduce moral hazard distortions, even if limited support would not prevent liquidity runs. The official IMF/G7 position is somewhere between the two extreme views described above: provided a crisis comes closer to being grounded in illiquidity than in insolvency, a partial bailout conditional on policy adjustment by the debtor country can restore investors’ confidence and therefore stop destructive runs—i.e., can have a ‘catalytic effect’. 2 If the ‘catalytic’ approach is successful, official resources do not need to be unlimited (i.e., so large as to fill in any potential financing gap), since some official liquidity provision and policy adjustment will convince private investors to rollover their positions (rather than run) while restoring market access by the debtor country. But can partial ‘catalytic’ bailouts ever be successful or, as argued by many, can only corner solutions of full bailouts or full bailins (i.e., debt suspension or standstill) be effective in preventing destructive runs? 3 This paper contributes to the current debate on these issues by providing a theoretical model of financial crises and the main policy trade-offs in the design of liquidity provision by an international financial institution. In our model, a crisis can be generated both by fundamental shocks and by self-fulfilling panics, while liquidity provision affects the optimal behavior of the government in the debtor country (possibly generating moral hazard distortions). Our study draws on the theoretical model by Corsetti et al. (2004) and the policy analysis by Corsetti et al. (2002), on the role of large speculative traders in a currency crisis. Consistent with these contributions, we model the official creditor (the IMF or ILOLR) as a large player in the world economy, with a well-defined objective function and financial resources. In our model, the strategies of the official creditor, international speculators and domestic governments are all endogenously determined in equilibrium. There are two major areas in which our model contributes to the debate on the reform of the international financial architecture: the effectiveness of catalytic finance and the trade-off between liquidity support and moral hazard distortions. As regards the first area, our analysis lends support to the hypothesis that catalytic liquidity provision by an official institution can work to prevent a destructive run—although in our model the success of partial bailouts is realistically limited to cases in which macroeconomic fundamentals are not too weak. In reality, the IMF does not have infinite resources and cannot close by itself the possibly very large external financing gaps generated by speculative runs, i.e., the IMF cannot rule out debt defaults due to illiquidity runs. According to our results, however, even when relatively small, contingent liquidity support lowers the likelihood of a crisis by enlarging the range of economic fundamentals at which international investors find it optimal to rollover their credit to the country. This ‘catalytic effect’ is stronger, the larger the size of IMF funds, and the more accurate the IMF's information. But our results, also, make clear that catalytic finance cannot and will not be effective when the fundamentals turn out to be very weak: as more and more agents receive bad signals about the state of the economy, massive withdrawals will cause a crisis regardless of whether the IMF intervenes. Our result runs counter to the hypothesis, first suggested by Krugman and then formalized by Zettelmeyer (2000) and Jeanne and Wyplosz (2001), that IMF bailouts can work only when there are enough resources to fill financing gaps of any possible size. These authors base their argument on models with multiple equilibria, in which partial bailouts cannot rule out the possibility of self-fulfilling runs. In such a framework, liquidity support is effective only insofar as it is large enough to eliminate all liquidation costs in the presence of a run.4 As regards the second area, contrary to the widespread view linking provision of liquidity to moral hazard distortions, our analysis identifies circumstances in which liquidity assistance actually increases incentives for a government to implement efficiency-enhancing but costly reforms. Specifically, the conventional view on debtor moral hazard is that, by insulating the macroeconomic outcome from ruinous speculative runs, liquidity assistance reduces the government's incentive to implement good policies. But this is not the only possible effect of an ILOLR. It is also plausible that some governments may be discouraged from implementing good but costly policies because speculative runs jeopardize the chances of their success. In this case, liquidity support that reduces liquidation costs in the event of a run can actually make socially desirable policies more attractive to the government. Our model shows that liquidity support can have either effect depending on circumstances. The following example conveys the two main points in this paper. In late 2002, as the Brazilian presidential elections were approaching and Luiz Inácio Lula da Silva—Lula—was expected to win, there was an incipient run on Brazil: foreign banks cut their exposure to Brazilian assets and there was a risk of a rollover crisis on government debt (which had very short maturity). Once elected, Lula had to choose between pursuing politically painful reforms and restructuring the country's debt, with the risk of triggering a financial meltdown similar to Argentina's. The IMF supported the former option with a US $30 billion loan conditional on the pursuit of sound fiscal policy and implementation of structural reforms. Arguably, the IMF loan had a ‘catalytic effect.’ Without it Brazil was very likely to experience a severe crisis, even if the government signalled its willingness to pursue sound policies: given the uncertainty surrounding the incoming government's goals and the size of the potential external financing gap, most investors were ready to ‘ rush to the exits’. Most importantly, without sizeable IMF support to fence off disruptive speculative runs and make the country less vulnerable to investors’ withdrawal, the incentive for Lula to pursue politically difficult policies would have been substantially weaker. By containing the risk of a run, instead, the expectations of a large IMF package raised the expected benefits from implementing sound policies: effectively, the IMF intervention induced good policies rather than triggering moral hazard.5 By reinforcing the willingness to pursue sound policies, contingent liquidity assistance was a key factor in avoiding an Argentine-style meltdown. Indeed, capital flight subsided and eventually reversed.6 Building on the main insights from the literature on global games,7 this paper contributes to our understanding of how and why catalytic finance could work in this and other crisis cases. Our analysis is related to a vast and fast-growing literature on the merits of alternative crisis resolution strategies and the arguments for and against an ILOLR. We contribute to this literature in a number of dimensions. First, we model the role of official financial institutions as large players whose behavior is endogenously derived in equilibrium. Relative to global games and the literature on the ILOLR building on them (see Morris and Shin, 2003b and the closed-economy models by Goldstein and Pauzner, 2005; Rochet and Vives, 2004), much of our new analytical insight stems exactly from this feature of our model. In specifying the preferences of its shareholders or principals, we model a ‘conservative’ IMF, in the sense that it seeks to lend to illiquid countries, but not to insolvent countries. As a result, in our equilibrium the IMF is more likely to provide liquidity support when the crisis is caused by a liquidity run, as opposed to crises that are closer to the case of insolvency. Second, in our framework, domestic expected GNP is a natural measure for national welfare—which may differ from the objective function of the domestic government because of the (political) costs of implementing reforms and adjustment policies. We can therefore analyze the impact on the welfare of domestic citizens and the government of alternative intervention strategies by the IMF. Third, we develop a model where a crisis may be anywhere in the spectrum going from pure illiquidity to insolvency. Most studies of ILOLR build on Diamond and Dybvig (1983)—D&D henceforth—and interpret crises as a switch across instantaneous (rational-expectations) equilibria, ignoring or downplaying macroeconomic shocks or any other risk of fundamental insolvency.8 Relative to this literature, we present a more realistic specification of an open economy where fundamentals, in addition to speculation, can cause debt crises. Fourth, in our global-game model the probability of a crisis and coordination among agents are endogenous, and the equilibrium is unique. We can therefore study the equilibrium implications of varying the size of the IMF support, the precision of its information and other parameters of the model without relying on arbitrary assumptions on the likelihood of a speculative attack. This is in sharp contrast with multiple-equilibrium models. While the model of catalytic finance shapes the traditional, official view of liquidity provision by the IMF, until very recently there was no theoretical analysis of it. Haldane et al. (2002) present a model that allows for fundamentals-driven runs, and assess the arguments in favor of debt standstills, relative to official finance, as crisis resolution mechanisms. These authors discuss the implications of moral hazard but do not develop a model of the trade-off between these objectives and the optimal intervention policy. Gale and Vives (2002) study the role of dollarization in overcoming moral hazard distortions deriving from domestic (but not international) bailout mechanisms (such as central bank injection of liquidity in a banking system subject to a run). Allen and Gale, 2000 and Allen and Gale, 2001 introduce moral hazard distortions in a model of fundamental bank runs, but do not consider analytically the role of an ILOLR. Rochet and Vives (2004) study domestic lending of last resort as a solution to bank runs in a global-game model.9 They find that liquidity and solvency regulation can solve the creditor coordination problem that leads to runs but that its cost is too high in terms of foregone returns. Thus, emergency liquidity support is optimal in addition to such regulation. However, they do not model the lender of last resort as a player and do not analyze the trade-off between bailouts and moral hazard that is central to our study. In independent research, Morris and Shin (2003b) develop a model of catalytic finance and moral hazard, reaching conclusions that are close to ours. However, they do not model the IMF as a large strategic player and do not employ a two-period bank run framework.10 The structure of the paper is as follows. Section 2 introduces the model. Sections 3–5 present our main results regarding the effect of IMF lending on the likelihood and severity of crises and the trade-off between IMF assistance and moral hazard distortions. Section 6 concludes.

In the last decade, the IMF systematically adopted a catalytic approach of large financial loans to support the main emerging market economies that experienced a crisis (see Cottarelli and Giannini, 2002). Arguably, ‘catalytic interventions’ succeeded in some cases (Mexico in 1995, Brazil in 1999 and again in 2002); they clearly failed in other cases (Russia in 1998, Argentina in 2001, Indonesia in 1998); there are a few examples that could be dubbed as ‘partial success,’ such as the IMF intervention in Korea during the 1997–1998 crisis. A number of contributions have attempted to perform systemic empirical studies of why and to what extent catalytic finance may work. While there is a clear need for further research, these studies point out formidable methodological difficulties in this area—whereas the actions by investors, governments and the IMF are all endogenous and interdependent. Because of these difficulties, the available empirical evidence is still not fully conclusive about catalytic finance even if some evidence is consistent with the implications of our model. Two arguments have been commonly presented against the provision of liquidity as a way to resolve crises. First, it has been argued that ‘limited’ liquidity support cannot work. Unless the IMF package can match a financing gap of any size—in most cases exceeding the amount of resources realistically available to an international institution operating as an ILOLR—self-fulfilling speculative runs leading to bad equilibria with default and high economic costs cannot be ruled out (see e.g. Zettelmeyer, 2000; Jeanne and Wyplosz, 2001). The second argument is that IMF liquidity support necessarily induces debtor moral hazard, i.e., expectations of a bailout always exacerbate welfare-reducing policy distortions. Based on these arguments, some have argued that ‘standstills’ (the international equivalent of bank holidays in a domestic bank run model) could be a superior approach to stem liquidity crises, and could possibly lessen moral hazard distortions due to bailout expectations. The model in this paper contributes to the debate providing reasons to refute the two arguments against international liquidity provision. First, it shows that ‘corner solutions’ in the form of exceptionally large and potentially unlimited liquidity provision are not necessary to reduce the incidence of liquidity runs. The presence of limited contingent liquidity support can be effective in inducing a fraction of private investors to decide to rollover their exposure to the country. Thus, partial support that does not fill ex ante the whole possible financing gap for a country can have an impact on individual portfolio decisions and therefore on the likelihood and the possible incidence of a crisis. Second, the model suggests that the standard argument that liquidity support always induces moral hazard distortions is similarly incorrect. In our results, the availability of contingent liquidity funds may tilt the incentives of a government towards implementing desirable but politically difficult policies and reforms—whereas the same government would have found them too costly and risky to implement if the outcome of its efforts were highly exposed to disruptive speculative runs. Thus, liquidity support may encourage good policy behavior—rather than discouraging it. Are ‘standstills’ superior to ‘catalytic finance’ as an international policy strategy to contain destructive liquidity runs? Standstill solutions to liquidity runs have been studied in closed economy models (see e.g. Goldstein and Pauzner, 2005), as well as in open economy models (see e.g. Gai et al., 2004; Shin, 2001; Martin and Penalver, 2003; Gai and Shin, 2004), but without providing a systematic comparison of costs and benefits of alternative solutions (i.e., standstills versus catalytic finance). An important direction for future research consists of addressing this issue within a rigorous analytical framework. Our model—we believe—provides a simple yet rich setup to undertake such a study.