The proposed early warning system is designed to detect any early signs of currency crisis in the foreign exchange market and allow policy makers to devise necessary actions to abort it. The early warning system to avert currency crises consists of three parts: The first part is a diagnosis based on a composite vulnerability index constructed with 28 variables, which covers a wide range of economic activities. The second part of the prediction includes leading indices and associated probability measures over a specified time horizon. And finally, the last part is a rather sophisticated risk classification system that can monitor changes in the risk level associated with currency crises over time.
The system is designed to provide 6 to 12 months advance notice of any unusual sharp depreciation or pressure for depreciation in the foreign exchange market. The criteria for forecasting a currency crisis are based on exchange market pressures (Eichengreen, Rose and Wyplosz 1996), a weighted combination of changes in foreign reserves, interest rates, and the exchange rate. Initial candidate variables were selected based on previous panel studies by Kaminsky, Lizondo, and Reinhart (KLR 1997), Calvo and Mendoza (1995), and Frankel and Rose (1996). This monthly monitoring and signaling device is a reliable complement to Rose’s (1998) annual system that utilizes twelve Asian country’s panel data.
To construct a reliable EWS on a monthly basis, we need to have a reliable measure of vulnerability to currency crisis. Thus, the first step is to construct a composite index of vulnerability to currency crisis to evaluate the likelihood of crisis at time t. It allows us to compare the current aggregate level of risk or vulnerability to that which has led to currency crisis in the past. In addition, we need a leading index that can help predict future crises with relative accuracy. A signals approach (KLR 1997)
was applied to further reduce the number of variables that enter the leading index part of the early warning system. This approach sets the threshold level for each variable so as to minimize the noise/signal ratio. The resulting individual variables and composite index are useful in monitoring developments that are associated with the likelihood of a future currency crisis. Associated probability measures over a specified interval can also be calculated on the basis of a logit model. The results from the standard linear regressions were compared with those from the logit model to check the robustness of the probability measure. The composite index system and the logit analysis both show a lead time for primary indicators ranging from 6 to 20 months. Also as an extension, these variables, whether used in a composite index form or individually, can be fed into an ordered logit model to yield the probability of a specified type of risk or vulnerability over a pre-specified time interval. In sum, the combination of composite index and logit analysis allows us to monitor the level of diverse risk patterns contemporaneously with the associated probability of crisis over a desired forecast interval.
The early warning system for currency crises in Korea
[Seoul] : Korea Institute of Finance
|Series Title; No||금융조사보고서 / 2001-06|
|Subject Country||South Korea(Asia and Pacific)|
|Subject||Economy < Financial Policy|
|Holding||kif; KDI School|