ECOPUB · Optimal pension design when individuals have different longevities

Periodic Report Summary - ECOPUB (Optimal pension design when individuals have different longevities)

Following population ageing and some recently engaged reforms of pension systems in several European countries, the question of how to link systems to longevity became a topic of major interest. Not only longevity was a crucial aspect to take into account in pension reforms but also differential mortality had to be considered. Indeed, life expectancy in all developed countries has been constantly rising after the Second World War; however not everyone has benefited from this increase in the same way. Apart from that, one of the missions of pension systems is to redistribute resources among agents. Yet, part of income redistribution is neutralised because, precisely, of differences in life expectancy. To be more explicit, individuals with lower wage get higher replacement rates but also have, on average, a lower life expectancy so that they do not benefit from this redistribution as much as they ought to. Thus, when talking about the link between life expectancy and pension design, several issues have to be considered. On the one hand, one has to consider that increased life expectancy makes more urgent the reform of our systems, since financial viability is at stake but, on the other hand, differential mortality has also to be taken into account in the design of social security schemes. This, of course, raises the question of the factors determining life expectancy. Is life expectancy the result of completely exogenous characteristics such as genes and, to some extent, such as luck? Or is life expectancy the result of some individual effort, either monetary, i.e. making health expenditure or undergoing expensive chirurgical operation, or non monetary, e.g. exercising, dieting, sleeping enough time etc.? Surely the answer lies in between. In determining the optimal link between life expectancy and pension design, these questions certainly need to be answered. For instance, there are characteristics for which the individual is not responsible, so that it is reasonable to think that the government would like to compensate unlucky individuals while for some other characteristics the government would not like to intervene. Our project wanted to focus on the issue of the optimal design of social security when individuals had different life expectancies and tried to answer the following questions: 1. how should differential mortality be included in pension systems? 2. how should contributions and benefits be linked to life duration? The project therefore lied at the frontier of public economics, health economics, behavioural economics and the economics of ageing. As such, our first objective was to study the normative problem of redistribution, i.e. how resources of the economy should be shared between agents who could influence their survival probability through private health spending, but who differed in their attitude towards the risks involved in the lotteries of life to be chosen. For that purpose, in the paper 'Optimal tax policy and expected longevity: A mean and variance utility approach', published in International Tax and Public Finance, 16, 514-537, 2009, Gregory Ponthiere of the Paris-Jourdan Sciences Économiques, Ecole Normale Superieure (PSE, ENS) and I developed a two-period model in which agents' preferences on lotteries of life could be represented by a mean and variance utility function allowing, unlike the expected utility form, some sensitivity to what Allais, in the Econometrica 21(4), 503-546, 1953, called the 'dispersion of psychological values'. We showed that in case agents ignored the impact of health spending on the return of their savings, the decentralisation of the first-best utilitarian optimum required intergroup lump sum transfers and group-specific positive taxes on health spending. Under asymmetric information, a differentiated taxation across agents was still required, nevertheless subsidising health spending might be optimal as a way to solve the incentive problem.

Data: CORDIS, © European Union