To assess social welfare in the context of health insurance, most studies are based on an ex ante approach, which computes the level of well-being of each individual before the risk occurs. This approach has the pros to take into account risk preferences and thus the intrinsic value of insurance. However, most welfare economists do defend an ex post approach, that focuses on the distribution of welfare once all uncertainty is resolved, assuming that people may not be competent at thinking before a risk occurs. This approach gives also a greater weight to those who are unlucky, which is important if society cares about inequality. In this paper, we compare and discuss the empirical differences of using either the ex ante or the ex post normative approach to assess the effect of health insurance policies on social welfare. This paper focuses on a recent reform implemented in France that mandates all private sector employers to offer sponsored Complementary Health Insurance to all of their employees beginning on January 1st, 2016. The results show that the empirical measure of social welfare depends more on the level of aversion to inequalities of the social planner than on the adoption of the ex ante or ex post frameworks. Thus, taking or not into account individuals risk aversion does not seem to matter so much. Conversely, introducing aversion toward inequality modifies greatly the judgement the social planner may have on the reform.