Advances in Accounting for Income Effects in the Derivation of Welfare Measures from Discrete Choice Experiments

Monetary valuation of both health and non health outcomes using stated preference data is gaining currency in health economics. Discrete choice experiments (DCEs) offer one form of stated preference data used for this purpose. In particular, the Hicksian compensating variation can be used to calculate willingness to pay (WTP) or willingness to accept (WTA) compensation arising from a policy change. This essentially involves a comparison of expected utility pre and post the policy change weighted by the marginal utility of income. To date in empirical applications in health, utility is usually assumed to be linear in income, or put another way, the marginal utility of income is assumed to be constant across respondents, producing constant WTP across individuals. That the marginal utility of income and WTP need not be constant across individuals was highlighted by the current authors in previous work which presented a method for calculating measures of welfare change from discrete choice random utility models allowing for non constant dependence of utility on income using evidence from a pilot study. The current paper builds on this earlier pilot work by presenting a more refined approach to esitmating WTP from choice experiments taking account of individual differences and demonstrates this in an empirical application with a larger sample size. Advantages of using this approach are outlined and further implications for the derivation of welfare measures from DCEs is discussed