Methods of Revenue Forecasting: An International Comparison

Little is known in Germany about the methods of revenue forecasting that are employed in other countries. Hence, this report compares 14 revenue forecasts across 12 industrialized countries with respect to institutional and methodological aspects and their impacts on the quality of forecasts. The international comparison reveals a range of differences between the approaches that are used in the analyzed countries. A first point is the institutional arrangement and thereby the independence of forecasts. Some countries increase independence by consulting external experts, either when the revenue forecast itself or the underlying macroeconomic forecast is prepared. Other countries delegate the revenue forecast to research institutes. Regarding the organizational framework there are fewer differences. Only the time span between forecast and the forecasted period differs considerably. Concerning the methodology there are more or less similar approaches in use. Indirect methods are principally employed, where the forecast is a function of macroeconomic variables, such as GDP or consumption. Among these methods, which are predominantly used for taxes being strongly dependent on economic circumstances (as income and corporation taxes), methods using elasticities turn out to be the most popular ones. In contrast, direct methods, that use data of the respective tax from previous years to predict the expected revenue, are used particularly for taxes that are less dependent on the economic situation (as excise taxes). Changes in tax law are usually taken into account. Most countries underestimate revenues, but just in one case at a significant level. The quality of forecasts measured by means of the size of the prediction error differs notably at first sight. A more precise analysis, however, shows that more than two thirds of the differences in accuracy regarding the whole revenues can be explained by differences in the tax structure, the time span, and the degree of independence. A more differentiated tax structure has a considerable positive effect on the accuracy. Also a forecast taking place closer to the forecasted period is accompanied with smaller estimation errors. Concerning the independence another positive effect on the accuracy can be observed.