Forecasting Economic Crises Using Fractal Characteristics

The paper studies the problem of modelling and forecasting economic crises. The fractal dimension local values ​​time series serves as an indicator of the economic system stability. The paper analyses the fractal characteristics of exchange rates time series of 20 national currencies in different time periods. The relationship between the dynamic stability violation and the deviation of the fractal dimension currency fluctuations from the optimal value has been proven. The boundaries of the normal state of economic systems was also obtained. These stability boundaries are built on the basis of a linearly weighted moving average and fractal dimension time series. The crisis predicting was reduced to determining the intersection points of the fractal dimension local values curve with the curve of the normal system state boundaries. The informative ranges of time scales reflecting short-term and long-term tendencies in violation of information efficiency of foreign exchange markets have been determined. The macroeconomic systems were classified by the efficiency level of foreign exchange markets.