Financial Crisis Prediction: Specification of Pre-crisis Periods in Turkey, Argentina and Thailand[Predikce finanční krize: Specifikace předkrizového období v Turecku, Argentině a Thajsku]

In this article is studied hypothesis, that every period (time interval) before financial crisis is distinguished by co-movement of several variables. The study is based on monthly data (that means no quarterly data, like portfolio investment, were used). This hypothesis, tested with vector autoregression (VAR), was to define period of time, in which a specific country is under permanent risk of falling from the edge to financial crisis (only waiting for the trigger). Variables used in this study are foreign liabilities of domestic banks (showing foreign exposure of banking sector), gross international reserves (usually rises before crisis, because of central bank's defense against appreciation pressures caused by capital inflow), domestic credit (showing domestic indebtedness, in the study, this variable was lagged by 3 months) and profitability comparison with S&P 500 index in dollar recounted domestic main asset market indices. Studied countries were: Turkey, Argentina and Thailand. In every studied case, there was found several months-to-years lasting period, that confirmed accuracy of co-movement hypothesis at 1%level of significance and, also, asset market bubbles in those pre-crisis periods were found.