Forecasting of foreign direct investments in Ukraine on the base of ARIMAmodels

Abstract

The article analyzes the problematic aspects of constructing the models of forecasting foreign direct investment in Ukraine. The expediency of using the autoregressive integrated moving average methodology for this purpose is justified. The transformation method of non-stationary time series of direct foreign investments into Ukraine to the stationary type is proposed. In order to be able to use the ARIMA-model, the output time series have been converted to static ones using the Statistica package by two-step transformation. Firstly, smoothing based on a simple moving average for the two points and then subtraction of the linear trend have been conducted. The ARIMA-model (2,0,0) has shown the lowest error values for MSE, AIC and BIC, indicating that it approximates the time series of FDI in Ukraine for the analyzed period in the best way and can be selected for short-term forecasting. The obtained ARIMAmodel (2,0,0) statistical parameters of foreign direct investment prediction in Ukraine prove the adequacy of ARIMAmodel and suitability for the formation of short-term forecasts. The forecast of foreign direct investment in Ukraine for 2018 based on the proposed ARIMA-model (2,0,0) in fact coincides with the forecasts of most investment companies and the National Bank of Ukraine. The article emphasizes that despite the series of measures aimed to deregulate the economy, which have been adopted by the government, investors are still afraid to invest in Ukraine due to the high level of corruption, slow implementation of reforms and unstable economy. It is stressed that without cardinal improvement of business climate and investment attractiveness, an increase in investment or rapid economic development in Ukraine should not been expected.