The data available to economists is rarely the outcome of natural or quasi experiments. Inaddition, it is common for distinct individuals to exhibit similar responses in a given environment whileobservationally identical individuals will respond differently to similar incentives. In such situations the useof an economic model fitted using maximum likelihood methods provide a general approach to thedescription of the observed data whatever its nature. The predictions obtained from a fitted model providecrucial information about the distributional consequences of economic policies.