Contributions to optimization modeling for pension fund ALM

PhD ceremony: Mr. M.H. Streutker, 14.30 uur, Aula Academiegebouw, Broerstraat 5, Groningen

Dissertation: Contributions to optimization modeling for pension fund ALM

Promotor(s): prof. M.H. van der Vlerk, prof. W.K. Klein Haneveld

Faculty: Economics and Business

Over the last decade, pension fund asset and liability management (ALM) has become much more complex due to, among others, new regulation and the need for an active indexation policy. Accordingly, models supporting the development of pension fund policy need to incorporate additional features. Traditionally, simulation models are used for this purpose. Policy is formulated in terms of decision rules which are then evaluated in a large number of scenarios representing possible future states of the world. However, the increased number as well as the complexity of the decisions to be made make it difficult to formulate and evaluate decision rules that capture all relevant policy options. Conceptually, the stochastic programming approach to pension fund ALM, as studied in this thesis, is able to capture all (relevant) policy options, as the aim of the stochastic programming approach is to compute the optimal policy for a certain goal given a set of scenarios and some constraints. In practice, however, such models are hard to formulate and solve.

To enhance the practical value of the stochastic programming approach to pension fund ALM a number of challenges need to be met. In this thesis we investigate some of them, like modeling issues as the implementation of regulatory rules and indexation, the economic basis of the models, and tractability limitations. Three of the four research questions posed concern these challenge areas, while the final question addresses the recently observed issue that current pension fund policies appear to be independent of the state of the fund.