Scenarios of the Romanian GDP Evolution With Neural Models

This paper aims to explore the nonlinear relation between investments and GDP. The method of neural network is used to construct two nonlinear models of GDP in relation to domestic investments, foreign direct investments and real interest rate. The results show that the two neural models present good performance measures on the dataset. The improved forecast accuracy may be capturing more fundamental non-linearities between investment and financial variables and the real output for a longer horizon.