Social Expenditure Projections: a Stochastic Approach

This paper develops a stochastic approach to the projection of ratios of social expenditure to GDP in order to examine their sampling properties using simulation methods. The approach involves the translation of informed judgements about the range of variation of variables into distributional parameters. It is suggested that the method can be used in other contexts where projections need to be based on variables whose distributional characteristics cannot be obtained from a time series of data, either because there are few data or because there are good reasons for believing that past variations give little insight into potential future variations. The results suggest that while the standard approach generates unbiased values, these may be associated with quite large standard errors. In making projections, it is therefore recommended that serious consideration should be given to the likely variability in the component distributions. In particular, the results are sensitive to assumptions regarding the growth rate of productivity.