بحران ارز و سیاست پولی در یک اقتصاد با محدودیت های اعتباری

This paper presents a simple model of currency crises which is driven by the interplay between the credit constraints of private domestic firms and the existence of nominal price rigidities. The possibility of multiple equilibria, including a ‘currency crisis’ equilibrium with low output and a depreciated domestic currency, results from the following mechanism: If nominal prices are ‘sticky’, a currency depreciation leads to an increase in the foreign currency debt repayment obligations of firms, and thus to a fall in their profits; this reduces firms’ borrowing capacity and therefore investment and output in a credit-constrained economy, which in turn reduces the demand for the domestic currency and leads to a depreciation. We examine the impact of various shocks, including productivity, fiscal, or expectational shocks. We then analyze the optimal monetary policy to prevent or solve currency crises. We also argue that currency crises can occur both under fixed and flexible exchange rate regimes as the primary source of crises is the deteriorating balance sheet of private firms.

Currency crises have been traditionally viewed as retribution for governments that have mismanaged the economy and/or lack credibility: Both the so-called first-generation models and the more recent second-generation models broadly answer to this description. However, the recent crises in East and Southeast Asia have led to a wide-spread questioning of this view.1 It is observed that most of the crisis economies enjoyed government surpluses and increasing foreign exchange reserves (unlike what the first-generation models would suggest) as well as low unemployment and booming exports (unlike in most of the second-generation models). Of course there are other forms of government failure. In the case of the East and Southeast Asian countries there is some evidence that the financial sector in these countries was not very well regulated. Without denying that this was an important element of the crisis, there is reason to doubt that it is the whole story: First because the lack of transparency in the financial sector of these countries was already well-known among market participants and second because these economies have now recovered and face interest rates not significantly higher than before the crisis, without any major overhaul of the financial sector. It is therefore not surprising that over the last two or three years, a third generation of models of financial crises has begun to emerge. These models have in common the idea that the crisis should be seen as a result of a shock that was amplified by what Bernanke et al. (1999) have called a financial accelerator mechanism. In some of these models (Aghion 1999a and Aghion 1999b) there is a real shock that gets amplified while in others (Krugman, 1999a; Chang and Velasco, 1999) there are multiple equilibria with the crisis brought on by a pure shift in expectations. The basic story is similar: A real currency depreciation can have a large effect on output if it affects the credit access of some subset of agents;2 moreover this effect on output may in turn affect the exchange rate, further amplifying the shock and causing it to persist. The present paper is a contribution to this line of research. It differs from the previous papers in that it is an explicitly dynamic monetary model with nominal rigidities playing an important role.3 This approach allows us to tell a very simple story of currency crises: If nominal prices are rigid in the short run, a currency depreciation leads to an increase in the foreign currency debt repayment obligations of the firms, and consequently a fall in profits.4 Since lower profits reduce net worth, it may result in less investment and lower output in the next period. This, in turn, brings a fall in the demand for money, and thus a currency depreciation. But arbitrage in the foreign exchange market then implies that the currency must depreciate in the current period as well. In other words, if people believe that the currency will depreciate, it may indeed depreciate. Multiple short-run equilibria in the market for foreign exchange are thus possible. A currency crisis occurs either when expectations change or when a real shock shifts the economy to the ‘bad’ equilibrium. This story of currency crises has the significant advantage that it is based on two well-known facts: First, the countries most likely to go into a crisis were those in which firms held a lot of foreign currency denominated debt. For example, Fig. 1 shows the ratio of claims to liabilities with respect to BIS banks; since these transactions are basically in foreign currency, this ratio is a measure of aggregate foreign currency exposure.5 It is striking that all the countries that had a ratio higher than 1.5 have experienced a serious crisis in the 1990s. The second fact is that there are substantial and persistent deviations from purchasing power parity following an exchange rate shock.6 By contrast Banerjee (1999) argues that the models such as Krugman (1999a), Chang and Velasco (1999) and our own previous work, require large changes in the relative price of tradeables and non-tradeables, as well as specific assumptions about the role of tradeable and non-tradeable goods in the economy.This credit-based approach to currency crises is consistent with numerous features observed in recent crises and left unexplained by the previous literature. For example, countries with less developed financial systems are more likely to experience an output decline during a crisis.7 Second, a currency crisis can also happen under a flexible exchange rate or without any significant decline in foreign exchange reserves. Third, crises may occur even in countries where governments face low unemployment and/or conduct sound fiscal policies and do not resort much on seigniorage. Obviously public policy variables such as fiscal deficits can play an important role in facilitating the occurrence of a currency crisis, as stressed by the existing literature on the subject. However, in contrast to first- and second-generation models, in the world described in this paper a deterioration of fiscal balances will lead to a crisis mainly through its impact on private firms’ balance sheets rather than through simple money demand adjustments as in the previous models.8 Moreover, the presence of public sector debt may exacerbate the problems of private sector debt, especially if a large fraction of public sector debt is in foreign currency. This result is in sharp contrast with the previous literature that finds that foreign currency (public) debt has a stabilizing role. Another advantage of our monetary model is that it lends itself very naturally to the analysis of monetary policy. There has been an important debate on the stance of monetary policy in the context of currency crises; in particular, the previous literature does not provide much guidance in the debate between those who emphasize past government failure and advocate monetary tightening,9 and those who blame shifts in expectations and bad luck (the multiple equilibrium view) and consequently support a more lenient approach to monetary policy.10 In our basic model in which the credit multiplier is either constant or dependant upon the real interest rate and price stickiness remains limited, a restrictive monetary policy is the optimal response to the risk of a currency crisis. However, this conclusion may cease to hold when credit supply is affected by the nominal interest rate and/or price stickiness is sufficiently persistent compared to the duration of debt contracts. The rest of the paper is organized as follows. Section 2 lays out the basic model. Some features of the model are taken as given, such as price stickiness, money demand, or the level of foreign currency debt. Microfoundations for these features are presented in Aghion et al. (2000b), where we also introduce commercial banks. Section 3 shows that this model naturally gives itself to graphical analysis. Using this graphical apparatus we examine the occurrence of currency crises and demonstrate the possibility of multiple equilibria. Section 4 introduces the public sector into the model, first by analyzing explicitly a fixed exchange rate system and second by introducing fiscal variables. In Section 5 we analyze the impact of monetary policy and we conclude in Section 6.

In this paper we have developed a simple framework to study currency crises and assess the effects of monetary policy. This ‘third generation’ model is particularly well suited to analyze the case of economies like in Asia, where the source of currency crises lied primarily in the deteriorating balance sheets of private domestic firms and commercial banks rather than in uncontrolled budget deficit policies by local governments (e.g., see Mishkin, 1999). Five main conclusions emerged from our analysis. First, an economy with a large proportion of foreign currency debt is more likely to face currency crises associated with large recessions and currency devaluations. Second, a currency crisis may occur both under a fixed or a flexible exchange rate regime as the primary source of such a crisis is the deteriorating balance sheet of private firms. Third, public sector imbalances can have destabilizing effects on the domestic currency through the crowding-out effects of public debt (especially public foreign currency debt) on the balance sheet and credit access of private firms. Fourth, unless credit supply does not strongly react to changes in the nominal interest rate, it is always desirable to increase the nominal interest rate if the primary objective is to avoid a currency crisis; this in turn vindicates the IMF approach. This result, however, may cease to apply if credit supply reacts too strongly to changes in the nominal interest rate, for example in the presence of signaling effects or as a result of persistent price rigidity. Fifth, a tight monetary policy will always produce a debt-burden effect on medium-term economic activity. A natural next step if this framework is to be used for policy purposes, is to empirically assess the relative importance of the various effects pointed out in the paper. In particular, besides the determination of actual foreign currency debt ratios, we need to get a better sense of how credit supply reacts in practice to changes in the real or the nominal interest rate; we also need to assess the elasticities of money demand with respect to income and to the nominal interest rate. For example, our analysis indicates that monetary tightening should be used to avoid a currency crisis if the credit multiplier does not react too strongly to changes in the nominal interest rate that leave the real interest rate unchanged. We thus need to understand the actual behavior of this multiplier before drawing definite policy conclusions. Our priority at this stage – but again this requires further empirical investigation – is that the credit multiplier should not dramatically respond to changes in the nominal interest rate alone, at least insofar as those changes are not too dramatic and/or interest rate increases are accompanied by complementary policies aimed at maintaining the credit multiplier, in particular adequate government support to banks and bank restructuring. Finally, we need to evaluate the relative speeds of price versus interest rate adjustments as our analysis suggests that the optimal design of monetary policy, is potentially sensitive to the degree of price stickiness, or more precisely to the duration of the deviation from PPP following the initial shock.