Leading indicators of country risk and currency crisis: The recent Asian experience

Leading indicators have been a successful forecasting tool adopted by the National Bureau of Economic Research (NBER) since the work of Burns and Mitchell (1946). In recent years, new econometric models such as dynamic factor models have been used to explore more formally potential dynamic differences across cycle phases in several variables. This dissertation applies the idea of leading indicators and uses such model to construct an early warning system to recent Asian currency crises. In particular, one dynamic factor with Markov regime switching model is used to construct leading composite indicators of country risk and currency crises for Thailand, Indonesia and Korea. The dynamic factor, which follows an AR(1) process and is unobserved, represents common distress signals underlying financial and banking system. Moreover, this factor switches between two unobserved regimes ruled by a Markov process, representing the states of the financial and banking system. That is, periods of high currency risk (or pressure) and periods low currency risk. Technically, we are able to obtain recursive statistical inferences on the unobserved factor and probability of the unobserved Markov state by casting the model into State Space Representation and applying the synthesis of Kalman filter and Hamilton nonlinear filter, which is proposed by Kim (1994).;The major results are as follows. Firstly, although a 'country risk' of currency crisis is not directly observable, but prior currency pressures can be detected in several sectors of the economy. In particular, financial and banking variables (yt) reflecting financial and banking distress are highly sensitive (lambda terms) to changes in the economic environment such as a change in investors' risk perception and/or business confidence (vt term). Secondly, country risk of currency crisis index is highly volatile prior to crises ( s2n,crisis ). Thirdly, country risk and currency crisis index (DeltaFt ) gives earlier signals of increases in country risk and subsequent currency crises in terms of probabilistic inferences (Pr[St=crisis|Information at t] for Thailand and Indonesia. Therefore we term them leading composite indicators of country risk and currency crisis. Fourthly, while crises that result from weak fundamentals make a country vulnerable to adverse shocks may be predictable, crises that arise from pure contagion effects are less likely to be foreseen by economic models. Currency crisis in Korea is likely to be the second case.;We understand that further refinement of leading indicators would require a large number of observations on the 'rare' events. Mere use of monthly data is not enough.