ARCH-building models of time series prediction for investment
ARCH-building models of time series prediction for investment

Summary/Abstract: Nowadays, due to the increased internationalization of financial markets and the increasing openness of national economies, there is increasing cross-border movement of portfolio investments as a form of international economic relations. The objects of portfolio investment could be shares, bills, debt securities, derivatives etc. Direct and portfolio investments are different in nature and do not have the same influence on the economy of the resident. Since portfolio investments are providing the purchase of securities and other financial instruments and have a speculative nature due to the potential risk of rapid outflows, it can be concluded that they are largely volatile and, therefore, cannot be predicted by standard predictive models and ARIMA-models. The aim of our study is to attempt to apply econometric models of conditional and generalized heteroscedasticity (ARCH-model and GARCH-model) and to predict the volume of portfolio investments in Ukraine. The scope of portfolio investments in Ukraine is researched, and there is also built an econometric model of ARCH family. Evaluation of ARCH models options is made by maximum likelihood method. Checking the adequacy of models is done by selected the most optimal one and made for the forecasting period. Research was conducted using the package EViews 6.

• Details
• Contents