STUDY OF TIME SERIES FOR ECONOMIC FORECASTS USING MULTIFRACTAL ANALYSIS

Miftar RAMOSACO, Elmira KUSHTA

Abstract

In this study, we will analyze the dynamics of time series with the idea of predicting the trend of an economic variable covering intervals in different periods such as the interest rate and the exchange rate. The time series under study is the exchange rate Euro/Lek, Dollar/Lek, the basic interest rate for the local currency (Lek), and the birth rate in Albania obtained by INSTAT and the Bank of Albania with monthly frequencies. This study contributes to the field of time series forecasts in Albania and modeling using multiracial analysis. With the created models, complete modeling will be performed between the series, offers, and demands that are assumed to report changes in external influences and the effects of market forces. In addition to parallelism, non-linearity of the behavior of the data series, and the specifics of the system itself, we understood to conclude on some features of the time series. So, the daily series of the Lek/USD exchange rate is less stationary than the Lek/Euro. The evidence here supports the idea that the continuous injection of foreign exchange is among the dominant factors for the current level of the exchange rate level, but it is closer to the natural mean. Time series analysis was determined to be performed without considering the influence of external factors. This probably had its effects on the results of the models.

Pages: 380 - 391