شاخص های گمراه کننده؟ بحران ارز آرژانتینی

Despite the fact that Argentina has been suffering from recession for years, the timing and severity of the recent currency crisis surprised most observers. There is an extensive literature on early warning systems, which claim to be apt for a systematical prediction of currency crises. One of the most prominent “early warning system” approaches was first developed by Kaminsky, Lizondo, and Reinhart (KLR) [Leading Indicators of Currency Crises, International Monetary Fund Staff Papers No. 45, 1998, pp. 1–48]. This paper analyzes whether the KLR approach could have predicted the Argentinean currency crisis at an earlier point in time. Using a broad set of indicators, it is shown that the forecasting quality of this approach is poor in the case of Argentina.

In economics as in medicine, prevention combined with early detection of emerging problems comprises the better part of the cure. In the case of economic crises, the question is whether these abnormalities can be detected in time to allow preemptive political measures. One of the latest patients suffering from a far-reaching currency crisis is Argentina. The collapse of the Argentinean peso in 2002 was followed by a deep economic recession, which then began to spread within Latin America. The new Argentinean crisis revived old debates about potential causes, symptoms, and patterns of currency crises. In general, economic theory proposes at least three answers to the question of how currency crises emerge. According to “first generation models” (Krugman, 1979), currency crises come about due to a run on the international reserves, as speculators understand that fiscal and monetary policies are inconsistent with the chosen pegged exchange rate. In other words, these models explain speculative attacks against a currency as a consequence of unsustainable developments in the “fundamentals” of an economy, such as rapidly growing budget deficits, high inflation, large and growing current account deficits, etc. The “second generation models” of currency crises (Obstfeld, 1986 and Obstfeld, 1996) consider devaluation as an intentional decision of governments, which weigh costs and benefits of maintaining a fixed exchange rate system, with the costs of defending a pegged exchange rate depending on investors’ beliefs. Unlike the first generation models speculative attacks may occur and self-fulfilling crises arise even if fundamentals are not particularly unfavorable. These models suggest that among others, unemployment, output, and domestic interest rates, may contain information on the probability of the occurrence of a crisis. After the Asian crisis of 1997 a new generation of theoretical explanations evolved, linking currency crises and financial sector fragility (Aghion, Bacchetta, & Banerjee, 2001; Krugman, 1998a, Krugman, 1998b and Krugman, 1999). These models point out the role of financial intermediaries and asset prices with respect to the emergence of a currency crisis. In the case of Argentina there is no consensus about the reasons for the emergence of the currency crisis, apart from the well-known and long-lasting foreign debt situation. According to Calvo, Izquierdo and Talvi (2002), the capital flow retrenchment after the Russian crisis of 1998 created a major real exchange rate misalignment and fiscal difficulties in Argentina and thus broke the economy’s neck. De la Torre, Levy Yeyati, and Schmukler (2002) find that Argentina fell into a growth–debt trap after 1998: When economic activity did not increase and credit from abroad dried up, the crisis became unavoidable. Feldstein (2002) argues that the crisis was due to exchange rate overvaluation and an extremely high amount of foreign debt. There is a catch in all of these approaches, though: While numerous arguments for the potential emergence of a currency crisis in Argentina are found ex post, the concrete timing of the Argentinean crisis remains unexplained. It is still an open question whether the standard indicators gave hints for the evolution and timing of the currency crisis during the pre-crisis period or whether they failed to do so. To provide a substantial judgment concerning the usefulness of the prominent early warning system proposed by Kaminsky, Lizondo, and Reinhart (1998) for the anticipation of the Argentinean currency crisis, our paper analyzes the economic development in Argentina between 1992 and 2002, which to our knowledge, has not been done up to now. The “early warning system” itself is based on “leading indicators” which are expected to send “signals” prior to a crisis. We show that in the case of Argentina these indicators were mostly misleading. Signals, if any, were sent very late and they came too late for a crisis prevention, as effects of policy measures need time. The remainder of the paper is organized as follows. Section 2 presents some stylized facts concerning the economic development in Argentina. In Section 3, the “early warning system” approach developed by Kaminsky et al. (1998) is explained briefly. Section 4 is a presentation of the findings obtained by using this methodology to explain the Argentinean crisis. Finally, Section 5 concludes with a critical summary of the results.

After the emergence of the Argentinean currency crisis (2002) a magnitude of questions arose. When did the ambitious currency board start to become unreliable? What was the initial spark that set off the crisis? Was there nothing that could have been done to prevent this disaster? What message were standard economic indicators giving during the pre-crisis period? This paper analyzed the pre-crisis period in Argentina employing the “early warning system” developed by KLR (1998), Kaminsky (1998), and Kaminsky and Reinhart (1999). We found that this early warning system, which is based on a broad set of indicators, did not give enough indications for the emergence of a currency crisis. Neither the different indicators nor the composite indicators were able to predict a currency crisis in the given 24-month crisis window. Furthermore, some indicators even sent misleading information, especially during the immediate pre-crisis period. In addition, warning signals were sent very late, too late for any sufficient policy intervention. KLR (1998), Kaminsky (1998), and Kaminsky and Reinhart (1999) have shown that their indicators operated sufficiently in the case of many currency crises in many emerging economies. Why is Argentina different? To answer this question, further research is necessary. This should focus on at least two issues: First, expectations might have played a major role in the case of Argentina. Since the early warning system approach does not explicitly consider changes in expectations, the explanatory power of other methods might be more helpful. Second, other factors than macroeconomic ones, e.g., political turbulence and corruption, might be important to explain the crisis in Argentina. The standard indicators, however, seem to have failed to reliably predict the Argentinean crisis in any case.