ارزیابی تاثیر دوفاکتو PEG ها در بحران ارز

This paper empirically evaluates the treatment effect of de facto pegged regimes on the occurrence of currency crises. To estimate the treatment effect of pegged regimes properly, we must carefully control for the self-selection problem of regime adoption because a country's exchange rate regime choice is nonrandom. To address the self-selection problem, we thus employ a variety of matching methods. We find interesting and robust evidence that (1) pegged regimes significantly decrease the likelihood of currency crises compared with floating regimes, and (2) pegged regimes with capital account liberalization significantly lower the likelihood of currency crises compared with other regimes. From the standpoint of the macroeconomic policy trilemma, we can reasonably conclude that pegged regimes with capital account liberalization are substantially less prone to speculative attacks because they can enhance greater credibility in their currencies by maintaining strict discipline for monetary and macroeconomic policies.

n a world with increasingly integrated capital markets, which types of exchange rate regimes are more prone to crises? Do pegged regimes indeed have a higher risk of currency crises than floating regimes? Which exchange rate regimes can avoid currency crises? It is very important for economists and policymakers to obtain an answer to these questions. Accordingly, some empirical studies have investigated the links between exchange rate regimes and currency crises using various datasets and methods (e.g. Bubula and Ötker-Robe, 2003, Esaka, 2010, Ghosh et al., 2003, Haile and Pozo, 2006 and Husain et al., 2005). For example, Ghosh et al. (2003) estimate the occurrence of currency crises under alternative exchange rate regimes in IMF member countries from 1972 to 1999 using the data based on the IMF de jure classification, and they find that the probability of crises is the highest for floating regimes. Haile and Pozo (2006) investigate whether exchange rate regimes affect the incidence of currency crises in 18 developed countries from 1974 to 1998 by estimating probit models based on the data of the IMF classification and the de facto Levy-Yeyati and Sturzenegger (2005) classification. They find that while the probability of currency crises is significantly higher for pegged regimes than for other regimes when the IMF classification is used, there is no significant link between exchange rate regimes and currency crises when the Levy-Yeyati and Sturzenegger (2005) classification is used. Esaka (2010) examines whether de facto exchange rate regimes affect the occurrence of currency crises in 84 countries from 1980 to 2001 by estimating probit models using the data based on the de facto Reinhart and Rogoff (2004) classification. The study by Esaka (2010) finds that pegged regimes significantly decrease the likelihood of currency crises compared with floating regimes. Using the combined data of exchange rate regimes and the existence of capital controls, it also finds that pegged regimes with capital account liberalization have a significantly lower likelihood of currency crises compared with other regimes. Judging from the above, however, these previous studies provide a more mixed view of the impact of exchange rate regimes on the occurrence of currency crises. Therefore, it is very useful to investigate which types of exchange rate regimes are more susceptible to speculative attacks and currency crises and which exchange rate regimes can avoid currency crises. To properly estimate the effect of exchange rate regimes on the incidence of currency crises, we must carefully control for the self-selection problem of choosing the exchange rate regime to adopt. However, previous studies do not explicitly address the self-selection problem of regime adoption. In the estimation of these studies, self-selection bias can arise because a country's exchange rate regime choice is nonrandom (i.e. there are systematic differences between countries that do and do not adopt a certain specific regime). This suggests that previous studies may provide an inaccurate picture of the effect of exchange rate regimes on the occurrence of currency crises. Accordingly, this paper empirically evaluates the effect of de facto exchange rate regimes on the occurrence of currency crises by explicitly addressing the self-selection bias. To address the self-selection problem, we thus employ a variety of propensity score matching methods recently developed in the treatment effect literature. In addition, we apply the covariate matching method of Abadie and Imbens (2006) to test the robustness of the results from propensity score matching methods. The central concept of matching methods is to match each treated unit with control units (nontreated units) that have similar observed characteristics and then to compare outcomes between the treated and the control units. These matching techniques can avoid selection bias and provide unbiased estimates of treatment effects ( Dehejia and Wahba, 2002, Heckman et al., 1998, Imbens, 2004 and Imbens and Wooldridge, 2009). In the context of currency crises, it is often claimed that one of the major ingredients of the environment leading to currency crises is pegged regimes. Are pegged regimes really more vulnerable to currency crises? To answer to this question, we first estimate the average treatment effect of de facto pegged regimes on the likelihood of currency crises using matching methods. Moreover, it is often claimed that in particular, adopting the combined policy of pegged regimes and liberalization of capital flows is an underlying contributing factor behind the outbreak of currency crises experienced in recent years. Are pegged regimes with capital account liberalization really more prone to currency crises? To answer to this question, we then estimate the average treatment effect of pegged regimes with capital account liberalization on the likelihood of currency crises using matching methods. These analyses can provide useful policy implications of the link between exchange rate regimes and currency crises because the effect of exchange rate regimes on currency crises can be precisely estimated by using matching methods. From the results of estimating average treatment effects using matching methods, we find interesting policy implications for exchange rate regimes. First, pegged regimes significantly decrease the likelihood of currency crises compared with floating regimes. Second, pegged regimes with capital account liberalization significantly decrease the likelihood of currency crises compared with other regimes. These findings are robust to a wide variety of matching methods and samples. The paper is organized as follows. Section 2 presents our research hypotheses. Section 3 presents the data of currency crises, exchange rate regimes, and capital controls. Section 4 presents the methodology for propensity score matching. Section 5 estimates the average treatment effect of de facto pegs on the risk of currency crises using matching methods. Section 6 estimates the average treatment effect of pegs with liberalized capital flows on the incidence of currency crises using matching methods. Finally, Section 7 presents conclusions.

In this paper, we formally evaluate the treatment effect of de facto pegged regimes on the occurrence of currency crises in 84 countries from 1980 to 2001. Our important contribution is that we carefully address the self-selection problem of exchange rate regime choice that previous studies have not explicitly addressed. To address the self-selection problem, we employ a variety of matching methods. Using matching methods, we first estimate the average treatment effect of de facto pegged regimes on the incidence of currency crises. We then estimate the average treatment effect of pegged regimes with capital account liberalization on the likelihood of currency crises. The results of matching methods provide interesting policy implications for exchange rate regime choice. First, pegged regimes significantly decrease the likelihood of currency crises compared with floating regimes. Second, pegged regimes with capital account liberalization significantly decrease the likelihood of currency crises compared with other regimes. Therefore, our results support the findings of Esaka (2010). The results are in contrast to the conventional policy wisdom that pegged regimes under capital account liberalization are universally more prone to speculative attacks and currency crises. From the standpoint of the macroeconomic policy trilemma, it can be generally thought that pegged regimes with no capital controls have no monetary policy autonomy and have the strictest discipline for macroeconomic policy compared with other regimes. Therefore, we can reasonably conclude that pegged regimes with capital account liberalization are substantially less prone to speculative attacks compared with other regimes because they can enhance credibility in their currencies by abandoning monetary policy autonomy. Finally, our results suggest that policymakers need to achieve greater credibility in their currencies by maintaining strict discipline for monetary policy and having strict and clear rules for exchange rate policies if they truly want to avoid speculative attacks. Acknowledgments The author would like to thank the Editorial Board and the anonymous referees, Kazuo Ogawa, Kimiko Sugimoto, and seminar participants at the 2010 Japanese Economic Association Fall Meeting for useful comments and suggestions on an earlier draft. The author is also grateful to Carmen Reinhart for making the data from the study by Reinhart and Rogoff (2004) available on her website. Financial support from the Japan Society for the Promotion of Science (JSPS) through Grant-in-Aid for Young Scientists (B) and Grant-in-Aid for Scientific Research (B) is gratefully acknowledged. The author alone is responsible for any remaining errors.