عوامل مؤثر بر تغییر رژیم نرخ ارز

Exchange rate policy in many emerging markets shifts between a stronger and weaker commitment to peg. This observation raises the following questions, which we address in our paper: Does intervention policy exhibit switching? And if so, what causes policy to shift? The theoretical literature increasingly points to the critical importance of financial fundamentals (Chinn, M.D., Kletzer, K.M., 2001. Imperfect information, domestic regulation and financial crises. In: Glick, R., Moreno, R., Spiegel, M. (Eds.), Financial Crises in Emerging Markets. Cambridge University Press, Cambridge, pp. 196–237; Dooley, 2000; Chang, R., Velasco, A., 2005. Monetary policy and the currency denomination of debt: a tale of two equilibria. Kennedy School of Government Working Paper Series, Harvard University, Cambridge, Massachusetts; among others). Direct empirical validation to date is limited, however, although relevant intuitions come from duration and indicators literature (Kaminsky, G.L., Reinhart, C.M., Vegh, C.A., 2004. When it rains, it pours: procyclical capital flows and macroeconomic policies. NBER Macroeconomics Annual, National Bureau of Economic Research, Inc.; Domac, I., Peria, M.S., 2003. Banking crises and exchange rate regimes: is there a link? Journal of International Economics 61, 41–72; Tudela, M., 2004. Explaining currency crises: a duration model approach. Journal of International Money and Finance 23, 799–816). We use a panel of quarterly data starting 1985 through 2004 for 14 countries, mainly from Latin America and East Asia, and adopt a novel two-step empirical strategy. First, we apply regime switching methods to two simple indices of central bank exchange rate policy. This generates likelihoods of high and low intervention. In the second step, we establish the variables that determine these probabilities. We find strong evidence that the economy's balance sheet and economic performance determine the likelihood of switching in the exchange rate regime. Specifically, an increase in reserve currency debt relative to assets raises potential capital losses from devaluation and reduces the probability of switching to a low intervention regime, as does a decline in growth. The purpose here is not to present an alternative regime classification (Levy-Yeyati, E., Sturzenegger, S., 2005. Classifying exchange rate regimes: deeds vs. words. European Economic Review 49, 1603–1635; Reinhart, C.M., Rogoff, K., 2004. The modern history of exchange rate arrangements: a reinterpretation. Quarterly Journal of Economics, Available at: http://www.puaf.umd.edu/faculty/papers/reinhart/papers.htm), but to establish the relative importance of different constraints on the central bank, which has policy implications. The empirical approach and specification is flexible to whether intervention is discrete or gradual and to whether or not the central bank sterilizes

Policy makers' in emerging markets have the usual macroeconomic objectives: high growth, low inflation, and a stable economic environment. In emerging markets, exchange rates have often been used to anchor inflation. Unfortunately, exchange rate based stabilizations have often ended in crises. The literature on exchange rate policy in small, open economies has developed largely in response to the currency and financial crises experienced by, among others, Mexico (1994–1995), several East Asian economies (1997–1998), Brazil (1999), and Argentina (2001). Quite a number of these countries moved from crawling pegs to fixed regimes. In other cases, currency boards terminated in large devaluations, which in turn gave way to managed floats or “hard pegs.” In each of these cases, crisis occurred because continued compliance to the peg carried a net cost. The latest generation of crisis models provides several explanations as to why the net cost of fixing the exchange rate might be high. For example, dwindling foreign reserves relative to the central bank's hard currency contingent liabilities may cause private agents to doubt the sustainability of the central bank's intervention policy, prompting a reallocation of portfolios away from domestic currency assets (Dooley, 2000, Chang and Velasco, 1998, Chang and Velasco, 1999 and Céspedes et al., 2004). This situation might easily arise if imperfect financial regulation distorts agent incentives by making it cheap to borrow (Chinn and Kenneth, 2001 and Hausmann and Velasco, 2004; among others). The literature increasingly focuses on the interaction between agents' portfolio choices and anticipated exchange rate policy given the regulatory environment (Chang and Velasco, 2005). Recent papers emphasize that external financing often dries up prior to crises. The “sudden stops” phenomenon (Calvo, 1998) highlights the key role of private agents' asset choices in influencing monetary and exchange rate policy. In the case of a “sudden stop” capital outflows or a sharp drop in capital inflows precede a currency crisis. This leads to a fall in the current account deficit, which must shrink as the external credit constraint tightens. In effect, the currency crisis coincides with a financial crisis. A disturbing implication is that crises associated with sudden stops are also associated with sharper recessions (Calvo and Reinhart, 1999). It is also widely acknowledged that exchange rate policy in emerging markets tended to switch between a stronger and weaker commitment to peg. These shifts have occurred even though volatile policies may introduce costly uncertainty into the economic environment. Indeed, welfare gains from eliminating policy uncertainty could be as high as nine per cent of trend consumption (Mendoza, 2001). The model presented in Chang and Velasco (2005) provides a clue as to why there is policy variability in economies even when there are large potential gains from stabilization. In their model, domestic agents decide the currency composition of their borrowings while the central bank determines the exchange rate policy. These two decisions are made simultaneously: agents' portfolio choices affect the exchange rate and the central bank takes the currency denomination of debt as given when setting policy. Both fixed and floating exchange rate policies emerge as potential equilibrium solutions. In their model, “equilibrium” exchange rate policy and “fear of floating” are endogenous to the currency composition of the liabilities of the financial system and the private sector. In the absence of regulation designed to avoid “currency mismatches,” the existence of multiple equilibria leads to increased policy variability over time. The different strands of thought on failed exchange rate stabilizations and exchange rate regimes in emerging markets may be brought together as follows. Policy makers in credit constrained developing countries fix exchange rates with the intent to stabilize inflation, eliminate perceived exchange risk and stimulate capital inflows and foreign investment. The success or failure of such a high intervention regime depends on the extent to which stabilization is credible to foreign investors; how well growth picks up in response to capital inflows and stabilization; the effectiveness of debt management; and the duration of improved economic performance. If the economy is unable to absorb capital inflows effectively, over borrowing might occur during a “boom”, this is self-destructive in nature since it creates the necessary and sufficient conditions for a subsequent “bust.” The Asian currency crisis, where excessive capital inflows caused a bubble in property prices, is a well-studied case in point. If a bust occurs, intervention has eventually to be given up. Intervention policy thus becomes endogenous to performance. This endogeneity may help explain why, in many emerging markets, destabilizing policy switches have been a regular and debilitating occurrence. While this argument appears to contradict the “original sin” hypothesis, limited access to external finance may be an incentive to fix in spite of past failures (Eichengreen et al., 2002). Husain et al. (2005) present evidence that in developing countries with little access to international capital fixed exchange rates are durable and help achieve low inflation, while pegs appear to have the shortest life span in emerging market economies. Additionally, external factors determine between 30% and 60% of exchange rate and reserve changes in 10 Latin American countries (Calvo et al., 1993). The inverse relationship between US real interest rates and secondary market prices of emerging market paper for a sample of Latin American and East Asian countries (Calvo et al., 1996) confirms the role of external factors in exacerbating the sudden stop effect. At this point, we should clarify why the central bank needs to respond to reserve currency debt accumulation in the first place. As recognized in the literature lately, the accumulation of foreign debt by the private sector and/or the poorly regulated financial sector becomes the central bank's problem largely due to its role as lender of last resort.2 “Insurance” might be an additional mechanism through which the central bank becomes vulnerable to the economy-wide balance sheet. The central bank may provide explicit or implicit guarantees to foreign lenders thereby forging a direct relationship between the overall increase in privately owed foreign debt and the central bank's contingent liabilities (Dooley, 2000). This has had the additional effect of increasing the size of IMF bailouts aimed at protecting wealth losses (Bordo and Schwartz, 1999). The case for an international lender of last resort might be strong, but in the absence of conditionalities this may further compound moral hazard (Mishkin, 1999). How then would the central bank respond to growing net dollar liabilities? Foreign asset and liability mismatches can affect the central bank's exchange rate policy in offsetting ways. As foreign debt rises relative to reserves, the increased risk of default can cause capital inflows to slow or reverse. This raises the cost of a high intervention policy by putting pressure on central bank reserves and reducing resources available for intervention. A real shock that reduces growth rates could have the same effect: it is well documented that capital inflows are procyclical in developing countries (Kaminsky et al., 2004). Conversely, if the financial system has dollarized net liabilities, there is an incentive for the central bank to avoid capital losses that would result if the home currency were to depreciate. This balance sheet effect raises the benefits from staying in the high intervention regime and engenders “fear of floating” (Calvo and Reinhart, 2002 and Chang and Velasco, 2005). This effect is confirmed by Levy-Yeyati and Sturzenegger (2005), who find that the increase over 1974–2000 in the number of de jure “clean” floating regimes is matched by that in the number of de facto “dirty” floating regimes. Given the evidence that floating regimes are better able to absorb terms of trade shocks and may be associated with higher secular growth (Edwards and Levy-Yeyati, 2005), this points to the growing importance of financial variables in determining the central bank's policy options. Once the costs of intervention outweigh the benefits, the central bank has no choice but to switch from high intervention to low intervention. In many of the countries in our sample, the shift to a floating regime occurs under crisis conditions or when capital flows are dwindling. However, many emerging markets that claim to be floating are in fact managing their exchange rates to a greater degree than the norm among the developed world. This distinction between de jure IMF classifications and de facto classifications is highlighted in papers that investigate fear of floating and the empirical validity of the bipolar hypothesis ( Reinhart and Rogoff, 2004 and Levy-Yeyati and Sturzenegger, 2005). These studies provide alternate de facto classifications of exchange rate regimes, which is not our aim here. We instead seek to determine the factors that increase the likelihood of being in a high or low state of intervention. Our application of a two state Markov methodology to construct the time series of probabilities of high and low intervention does not imply therefore that we are basing our study on the premise that hollowing out holds ( Eichengreen, 1994 and Fischer, 2001). To summarize, the growing consensus in the literature is that exchange rate policy emerges from the interaction between agents' portfolio choices, the central bank's role as lender of last resort, and weak financial regulation (Goldstein and Turner, 2004). This consensus highlights the need for an analysis of how central banks' exchange rate policies respond to changing financial and economic circumstances. Direct empirical validation of the theoretical models to date is limited however, although relevant intuitions come from duration and indicators literature and from theoretical models. For example, Domac and Peria (2003) show that adopting a fixed exchange rate diminishes the likelihood of banking crises among developing countries. Tudela (2004) estimates a duration model for OECD countries over 1970–1997 and finds that increases in import growth, claims on the domestic government and in foreign portfolio investment, and appreciated REER increase the probability of currency crises. We address the following questions in this paper: • Does intervention policy exhibit clearly defined cycles? • What drives these policy cycles? Our contribution in this paper is therefore the focus on how the central bank's exchange rate policy responds to financial and economic fundamentals. Even though exchange rate regimes are identified through a Markov switching model, each quarter there is a likelihood that the central bank will move towards greater or lower intervention, so we do capture a continuum of policy responses. The critical balance sheet variable in our paper is gross foreign currency liabilities relative to assets. The stock of assets is largely reserves. We believe this is an appropriate indicator of a currency mismatch in the balance sheet given that private sector foreign assets need not be appropriable by the central bank. The empirical strategy is as follows. We first obtain an explicit policy reaction function for the central bank based on the monetarist model (Girton and Roper, 1977, Weymark, 1995 and Kaminsky and Reinhart, 1999). In the monetarist model, money market disequilibrium causes exchange market pressure, which can therefore be relieved through direct intervention in currency markets. We measure intervention in two different ways, both of which we examine for regime switching behavior. This provides a robustness check. Policy choice, as reflected by the two measures of intervention, is modeled as an independent two state Markov process. The first state is stabilization or high exchange market intervention. The second state is letting it slide or low exchange market intervention. The policy response of the central bank for a given increase in exchange market pressure is larger in the “high” intervention state than in the “low” intervention state. This generates a continuous time series of the probabilities of switching from a high intervention regime to a low intervention regime. Next, we identify a set of variables that explains the regime switch. The explanatory variables capture potential “balance sheet effects” and include indicators of economic performance. The balance sheet indicators are the ratio of the stock of external liabilities to assets and the ratio of M0 to central bank's foreign reserves. The size of reserve currency debt relative to assets captures the extent to which the financial system is “leveraged” and is therefore a proxy both for risk of default and for the financial system's exchange risk exposure. The ratio of M0 to foreign reserves, in the absence of capital controls, indicates the size of the central bank's contingent liquid liabilities relative to assets, since all money in circulation is a liability of the central bank. Under a high intervention regime, the central bank must stand ready to swap domestic currency for reserve currency at the supported exchange rate. GDP growth, volatility in stock market returns, and inflation differentials measure economic performance. Our sample covers 15 countries, primarily from Latin America and East Asia,3 over the period encompassing the first quarter of 1985 through the third quarter of 2004. The sample provides a control group of mature floaters and a set of diverse emerging markets that nonetheless satisfy the common criteria of the Morgan Stanley Capital International (MSCI) classification. We did not include a larger sample of developing countries in order to restrict our analysis to countries that have global market access. We find that balance sheet effects are highly significant in explaining the switch from high to low exchange market intervention, even though the central bank faces offsetting incentives in this context. The results indicate that fear of floating does prevail, and high external liabilities relative to assets significantly increase the likelihood of remaining in the high intervention state. This is true for both measures of intervention. In the next section, we motivate two measures of intervention and sketch the Markov switching model that we fit on these two measures. In Section 3, we discuss the results of the model outlined in Section 2 and present the probabilities that the central bank will switch away from a high intervention policy. In Section 4, we provide the underlying theoretical motivation for our empirical strategy, an overview of the data, the results of the regression analysis, and the estimates of the thresholds over which the regressors signal a policy switch. We conclude in Section 5 that financial and economic fundamentals are the critical constraints on the central bank's choice of exchange rate regime.

This paper is a step forward in the empirical validation of the theoretical literature that models the interaction between the economy's balance sheet and the sustainability of the exchange rate regime. Our results are consistent with the insights provided by the duration and indicators literature. We adopt a novel empirical strategy in order to shed light on the key constraints on the central bank's exchange rate policy. First, we model two standard measures as independent Markov processes. This enables us to identify well-defined switching behavior between states of high and low intervention. Next, we condition the likelihood of the regime switch on a wide range of macroeconomic while controlling for country fixed effects. The empirical approach and specification is flexible to whether intervention is discrete or gradual and to whether or not the central bank sterilizes. We find strong support for the thesis that financial responsibility determines the sustainable level of intervention across our sample of 15 countries. This was found to be true not only around points of crisis. Boom-bust cycles less dramatic than those observed during crises are observed on a regular basis for most of the countries in our sample, as evident from the graphs displayed in Section 2 and in Appendix 2. Our primary focus is to predict the likelihood of a change in the central bank's policy choice conditional on rising external debt and contingent liabilities. We find that higher reserve currency debt constrains the central bank's intervention policy in at least two different, opposite ways. For one, given limited reserves and unreliable capital inflows, the burden of intervention is more difficult to sustain as the central bank's uncovered contingent liabilities rise. This raises the costs of a peg and generates an incentive to cease intervention. At the same time, devaluation implies capital losses on external debt and generates “fear of floating” (Calvo and Reinhart, 2002). This raises the benefits of a peg and provides an incentive to cling to intervention. Our econometric results indicate that the fear of floating effect dominates: the likelihood of moving from high to low intervention is significant and decreasing in the ratio of external liabilities to assets. However, as the threshold analysis in Section 3 indicates, there is a large increase in the ratio of external liabilities to assets in the pre-floating period. High reserve currency debt does therefore reduce the central bank's willingness and ability to intervene in currency markets. This constraint on the policy maker is tighter when capital inflows are dwindling, or in a sudden stop situation, or when US interest rates are increasing. Interestingly enough, for most variables we found no significant difference in performance across fixed and floating regimes that are sustained (turning points are one-year windows at both ends of the regime); though as expected, inflation and stock market volatility are significantly higher under floating relative to fixed. The higher macroeconomic volatility under floating creates the desire to stabilize, which is typically the rationale for adopting the high intervention regime. Our results are robust across the two policy response indices and changes in specification. We identify the following country-specific constraints, which appear to drive policy regimes in a systematic way: High inflation and the desire to gain access to international capital markets create the incentive to fix. Worsening debt, increasing contingent liabilities, rising stock market volatility, and declining rates of growth determine the likelihood of switching away from the fixed regime. In validating the constraints on central bankers and decision-makers in emerging markets, we find strong support for policy recommendations that focus on improved debt management and development of domestic capital markets (Goldstein and Turner, 2004).