مدلی از بحران ارز مسری و کاربرد آن در آرژانتین

This paper proposes a model of contagious currency crises where crises are transmitted across countries through a risk premium effect on government bonds. Three types of equilibria can occur in the model: a ‘no-collapse’ equilibrium in which a crisis is never transmitted from abroad, a ‘collapse’ equilibrium in which a crisis is inevitably contagious, and a ‘fundamentals’ equilibrium in which a crisis is contagious when domestic fundamentals are weak enough. The model is calibrated to the March 1995 Argentine crisis that followed on the heels of the December 1994 Mexican peso crisis. The results imply that the occurrence of the 1995 Argentine crisis reveals a combination of risk aversion of investors and weak credibility in the currency board arrangement. The Argentine crisis however could only be attributed to a ‘tequila effect’ alone (contagion from the Mexican crisis without any role for weak domestic fundamentals) if investors were excessively risk averse.

On 20 December 1994, Mexico was forced into a 15% devaluation of the Mexican peso followed by the adoption of a floating exchange rate, after a speculative attack reduced its stock of foreign exchange reserves to US$ 6 billion, down from US$29 billion in February of that year. Soon afterwards, financial problems occurred in Argentina, and, to a lesser extent, Brazil, as investors were reluctant to renew their loans to these two countries. Argentina lost 41% of its reserves and 18% of its bank deposits in the first quarter of 1995, while Brazil’s foreign exchange reserves declined by 20%. Similar events took place in the summer of 1997 after Thailand stopped defending the baht’s fixed US dollar value on 2 July: a wave of speculative attacks rocked South East Asian countries, leading to widespread depreciations that reduced the values of these countries’ currencies by 30–50% as of December 1997, in comparison with the beginning of the year1. The attacks in both episodes often seemed justified by fundamental problems in the countries concerned. For example, a large number of financial institutions in South East Asia were grappling with substantial amounts of unhedged foreign debt and nonperforming assets. But there were cases in which these currency attacks seemed unwarranted by the underlying fundamentals, as was the case in the run on the Hong Kong dollar in October 19972. In both instances of widespread financial turmoil, the temporal correlation of crises remains particularly striking. This phenomenon — that devaluation in a country can be the precursor and even one of the causes of financial troubles in other countries — has been dubbed ‘contagion.’ Prior to the 1990s, currency crises did not seem to spread across countries with the virulence and speed observed recently. It is therefore not surprising that in the 1980s the finance literature, while focusing on contagion in bank runs3, had little to say about contagious speculative attacks on currencies. In the 1990s, however, the literature took a step in the latter direction when the interest of researchers focused on the comovement of asset prices and capital flows across emerging financial markets4 This comovement was generally interpreted as supporting evidence for the presence of contagion in asset flows across countries. Calvo and Reinhart (1996), for example, argue that the significant correlation in bond and equity returns they found across emerging markets is indicative of such contagion. Valde´s (1996), after accounting for fundamental factors, is still able to identify excess comovement in country credit ratings and secondary market debt prices, hence ascertaining the contagion hypothesis. The focus on contagion in the context of currency crises alone occurred mainly in the past few years, inspired not only by the collapse of the ERM in 1992 but mostly by the Mexican peso crisis of December 1994 and by the Asian crises in 1997–98. There are, however, only a limited number of empirical studies that investigate the reasons why crises seem to be linked across countries5. Eichengreen et al. (1996) address this question using a sample of 20 industrialized countries over a 30- year period. They find a significant positive correlation in the incidence of crises across countries, after accounting for domestic economic and political factors. Looking at emerging markets, Sachs et al. (1996) examine the countries that suffered from financial problems in the wake of the Mexican crisis. They find that lending booms, real appreciation and low reserve levels were common priors across the countries that had severe crises at the time, hence explaining their extreme susceptibility to the ‘tequila effect’. Contagion in the aftermath of the Mexican crisis therefore seemed to have led to the worst turmoil in economies characterized by weak underlying fundamentals. Using a different approach, Schmukler and Frankel (1996) perform tests on country funds data and find significant adverse short-run spillover effects from the Mexican crisis to both Latin America and Asia. Overall, the empirical literature clearly cannot refute the hypothesis of contagion effects in specific episodes of currency crises. The theoretical literature, however, proposes few analyses of such effects. The existing theories generally link contagion either to (bilateral or third-party) trade channels, or to liquidity effects. Gerlach and Smets (1995), for example, model the role of trade channels as central to the transmission of crises. They show how a devaluation in a given country can lead to adverse monetary shocks and reserve losses in another country because of the inflationary effect of a rise in the latter’s import prices. (A rise in import prices leads to an increase in the general price level, which reduces money demand and leads to reserve losses in a fixed exchange rate system.) Alternatively, the devaluation can cause a crisis in the second country by leading to a current account deficit as the second country’s exports are priced out of the first market. The trade channel, however, cannot account for the observation of contagion effects across countries that do not have strong trade linkages with one another (or with the same third party). Valde´s (1996) provides an alternative framework, in which investors’ liquidity needs lead to contagion effects. According to his model, a crisis in a given country will drive investors to sell off another country’s assets in order to raise funds, either because these funds are needed today or because cash needs are expected to materialize in the future. His model, however, only explains the trigger effect of a crisis, namely the sell-off of assets in other countries. There is no analysis of how this initial effect leads to a full-fledged crisis6. Adopting a more general approach, Masson (1998) presents a balance of payments model in which a crisis could occur because of either adverse changes in domestic or external fundamentals or because of contagion effects. Masson’s model, however, captures the latter factor by a posi-tive shock to the exogenous probability of devaluation in a foreign country, and the transmission channel to the domestic economy operates through the bilateral real exchange rate. Therefore, as in Gerlach and Smets (1995), contagion in his model effectively operates through a fundamental economic link between the two countries. This paper takes seriously an alternative explanation for contagion and analyzes its implications on the incidence of currency crises. Contagion is introduced in the model through a risk premium channel: a currency crisis in a given country affects another country by causing a rise in the risk premium on bonds issued by the latter country. This transmission mechanism can be due to the herding behavior of investors, rather than to fundamental factors such as trade channels or liquidity needs7. Herding behavior is not necessarily an ad-hoc assumption: it can be justified by the globalization of capital markets in the 1990s, coupled with the presence of imperfect information. Indeed, as shown by Calvo and Mendoza (1997), by raising the number of investment opportunities and/or exerting pressures on managerial performances, globalization reduces the incentive to incur the cost of gathering country-specific information and induces conformity in investment decisions across fund managers. Both factors lead to herding behavior8.

This paper presented a framework illustrating the transmission of currency crises across countries. Contagion here is the consequence of rational herding behavior of investors who diversify risk by holding assets of several countries and form ex-ante expectations about the comovement of exchange rates across these countries, possibly because of lack of accurate information. A crisis in one country leads to an increase in the likelihood of a crisis in another country when investors expect a positive covariance of the two countries’ exchange rates. This phenomenon of contagion translates into an increase in the risk premium required to hold domestic bonds. The resulting pressures of a crisis are transmitted to the money market through the domestic interest rate. For a benchmark of parameter values, calibrated to reproduce the Argentine economy, the model yields multiple equilibria. The different equilibria that the economy could settle at are associated with different levels of uncertainty prevailing in the economy, as captured by investors’ expectations regarding the comovement of the two countries’ exchange rates. In particular, the ‘nocollapse’ equilibrium corresponds to a relatively low degree of uncertainty. If, therefore, one believes that the currency board in Argentina was credible enough to reduce exchange rate uncertainty to near-zero levels then the model implies that contagion was not the main cause for the Argentine financial turmoil of 1995. However, if one is to extrapolate from historical values of the exchange rate variance in Argentina and its covariance with the Mexican rate over the non-fixed periods in 1970–1995, then the model provides a rationale for the argument that contagion contributed to the speculative attack, but only if investors are believed to be sufficiently risk averse. It is important to note that the economic fundamentals accounted for in this paper, namely the stock of public debt, do not seem to play a critical role in determining whether or not the Argentine economy was driven to a crisis equilibrium. This is not surprising because that aspect of the economy was not problematic at the time. Argentina’s financial sector fundamentals, however, had deteriorated in 1995. Hence, a useful extension to this paper would be to model the banking sector explicitly in order to account for a more comprehensive spectrum of fundamentals.