Evaluating the Economic Growth Using Artificial Neural Networks and Panel Fixed Effects

This thesis uses a panel data to investigate the effects of eight macroeconomic variables on the evolution of growth rate of Gross Domestic Product per capita. The panel data consist of 23 years of observation for ten developed and ten developing countries. The years covered are from 1990 to 2012. The independent variables selected are: (i) initial GDP per capita (INIGDPPC) to account for the effect of convergence (ii) terms of trade (TOT), (iii) trade openness (OPEN), (iv) gross fixed capital formation (GFCF), (v) human capital (EDUC) measured as average years of schooling, (vi) inflation (INF), (vii) government size (GOVT) and (viii) population growth (POPUL). The thesis methodology is unique in combines cutting-edge data-driven models such as hybrid artificial neural network with genetic algorithm (ANN/GA) and fixed effect panel model. First, the impact of eight independent variables on growth is investigated and dominant variables are identified by using three data samples: developed countries only, developing countries only, and developed and developing countries together. Moreover the study uses three different data formatting for each sample: annual data, periodic data of 4 years overlapping and periodic data of 4 years non-overlapping. Second, two estimation methods are used to predict values of growth. This allows us to compare those forecasting methods with each other. The analysis indicates INIGDPPC, INF, GFCF, GOVT, EDUC, POPUL, TOT and OPEN variables have the statistically significant impact on growth in the panel regression. The INIGDPPC, POPUL, GOVT, and INF have negative and OPEN, EDUC and GFCF have positive statistically significant effects on the economic growth in developed and developing countries. Moreover, the results obtained from the study have shown that the power of the hybrid ANN/GA method (combined the artificial neural network method and genetic algorithm) is more than Panel fixed effect estimation method in predicting the economic growth.