engleski

Forecasting economic growth using financial variables - Comparison of linear regression and neural network models

According to both the theoretical and empirical literature, financial variables contain useful leading information regarding economic activity and thus can be used in forecasting GDP growth. However, empirical studies of the relationship between the financial development and the economic growth, as well as those of forecasting economic growth using financial variables are mainly based on linear econometric models. Since nonlinearities could exist in the relationship between the variables, in this paper we compare forecasting performance of the linear econometric models and the neural network model for panel data of European Union countries' economic growth. Our results show that at the 1-year forecasting horizon, according to three out of four valuation criteria, neural networks improve forecasting accuracy.

Forecasting; economic growth; financial variables; linear models; neural networks; panel data

nije evidentirano