Forecasting Fiscal Revenues in a Transition Country: The Case of Croatia

In this paper we asses the ability of alternative time series models to produce accurate fiscal revenue forecasts in a transition country and compare them to offi cial forecast. We take on a disaggregated approach and estimate separate models for seven revenue sources. Alternative time series models – trend model, random walk, ARIMA, regression and error correction models – are specified using quarterly data. One - and two - year ahead forecasts are calculated and compared against actual values and official forecasts. Results suggest that despite impediments, econometric methods produce forecasts that are in general more accurate than official forecasts prepared using expert judgment.