Unemployment rate forecasts: evidence from the Baltic States

ENThis paper compares forecasts of unemployment rates in the Baltic states using time-varying parameter models. We are particularly interested in dynamic asymmetries in the unemployment rates and the forecasting performance of different models at different times. We consider each predictor in the time-varying model as either stochastic or deterministic and use a rolling shortterm forecast experiment as an out-of-sample test of forecast accuracy. The time span of our data from 2001 to 2014 includes the global financial crisis. We find that the forecasting ability of the models depends on both the forecasting horizon and the moment in time when the forecasts are done. The empirical evidence derived from our investigation suggests that no single model is the best one, but models that include a cyclical component tend to perform better than others. Our findings show that the preferred models differ in the time of increase or decrease in unemployment rates.