Collective investments for heterogeneous individuals

Prior research shows significant variation in risk attitudes, partly reflecting actual differences, partly to be interpreted as uncertainty about true individual attitudes. Individuals also differ widely in their human and financial capital, and past employment histories, both affecting their risk capacity. Our aim is to study the implications of heterogeneity in risk preferences and risk capacity for pension investments. We study the possibilities of tailoring investment plans for pension savings to the individual characteristics of participants, either within a collective pension fund or in an individual arrangement.

For a pension fund with a collective investment portfolio we provide guidelines for aggregating individual characteristics into a collective profile. Not taking these characteristics into account leads to sub-optimal outcomes, since different participants have different optimal (lifecycle) plans with different portfolio allocations. Defining a representative participant with a representative plan depends on how the fund weights the costs of deviating from the collective portfolio for each individual. Part of the analysis therefore involves an assessment of the cost of a suboptimal investment portfolio.

For fully individual arrangements we analyse the trade-off between fully personalised portfolios and a limited number of portfolio strategies. Fully personalised plans may not be cost-effective, whereas a fully standardised product may induce welfare costs for various groups of individuals. In balancing the trade-off, we cluster households with similar characteristics into smartly designed life-cycle portfolios. Clusters and plan designs are simultaneously determined with a statistical learning algorithm, resulting in a small number of plans targeted at specific groups.

This project is a collaboration between Maastricht University and Tilburg University.

Read all project publications here.