This paper investigates the use of dynamic factor model for forecasting headline and core inflation as well as food price index in Poland. Method applied in the study extend conventional approaches by using bayesian techniques to dynamic factors' estimation, way of handling "ragged edge" data structure and allowing for the model to change over time.

Forecasting results confirm that including current information extracted from data-rich environment improves inflation forecast precision and consequently DFMs perform better than the best autoregressive models. The analysis suggest also that applying dynamic model selection procedure can additionally reduce out-of-sample prediction errors.