MEASURING WELFARE CHANGES IN BEHAVIOURAL MICROSIMULATION MODELLING: ACCOUNTING FOR THE RANDOM UTILITY COMPONENT

This paper presents a method of predicting individuals' welfare changes (compensating and equivalent variations) arising from a tax or social security policy change in the context of behavioural microsimulation modelling, where individuals can choose between a limited number of discrete hours of work. The method allows fully for the nonlinearity of the budget constraint facing each individual, the probabilistic nature of the labour supply model and the presence of unobserved heterogeneity in the estimation of preference functions. Yet it is relatively straightforward to implement. An advantage of welfare measures, compared with changes in net incomes, is that they take into account the value of leisure and home production. The method is applied to a hypothetical income tax policy change in Australia.