In order to do so, an algorithm (selected from the best of artificial neural network model, structural econometric model, dynamic general equilibrium model, nonlinear econometric model, golden triangle model) for cycle/growth sequencing and forecasting will be developed. Sequencing of the economic growth on an international scale (119 countries) will be performed using business cycle tracer methods and novel golden triangle model developed by Škare (2010). Using individual country data, we plan to isolate growth determinants using three dimensional golden triangle model and spectral analysis techniques. Output, inflation and employment form the three dimension of the cycle/growth triangle that will be used for sequencing growth. After growth sequencing is finished, a best forecasting algorithm is built. Pairing isolated growth bases help us to rebuild a growth. Using different algorithm models, reconstructed economic growth series is compared to the actual growth series (observed data). Algorithm resulting in the smallest forecast error (RMSE) is selected as optimal to explain economic growth and predict business cycles. Economic growth forecast and business cycles predictions will be compared to the actual data over the observation period. If data match poorly actual observed data on economic growth/cycles this signify economic growth follow random walk that cannot be predicted. Consequently, business cycles can not be avoided. This conclusion will have important theoretical and policy implications both for the science of economics and economic agents.

Project duration: 16 June 2014 – 15 June 2018

Call number: IP-11-2013

Principal investigator: Professor Marinko Škare