Bayesian VAR Models for Forecasting Irish Inflation

In this paper we focus on the development of multiple time series models for forecasting Irish Inflation. The Bayesian approach to the estimation of vector autoregressive (VAR) models is employed. This allows the estimated models combine the evidence in the data with any prior information which may also be available. A large selection of inflation indicators are assessed as potential candidates for inclusion in a VAR. The results confirm the significant improvement in forecasting performance which can be obtained by the use of Bayesian techniques. In general, however, forecasts of inflation contain a high degree of uncertainty. The results are also consistent with previous research in the Central Bank of Ireland which stresses a strong role for the exchange rate and foreign prices as a determinant of Irish prices