Abhatoo - Centre Nationale de Documentation

In this article, the predictability performance of certain classical business cycle theories are tested against contemporary statistical methods by using Finnish macroeconomic data. Keynesian multiplier- accelerator model derivatives and neo-classical real business cycle models are compared to statistical stochastic time-series methods. Some philosophical considerations on the scientic principles and macroeconomic analysis are extended for applied econometric practice. VAR and SUTSE models are estimated and compared against classical theory implications. It is found that in this case, SUTSE model has a superior forecasting ability and that pure statistical algorithms are the most efficient alternatives for predicting Finnish business cycle data.