The Option Value Model in the Retirement Literature: the Trade-off between Computational Complexity and Predictive Validity

This study gives an overview of retirement modelling, starting from the single-period consumption/leisure model up to the recent life-cycle multiple-decisions and joint retirement models, paying particular attention to the role played by the option value model in the economic literature on retirement. The option value model was initially interpreted as a sub-optimal solution of the dynamic programming rule. But its simpler implementation and similar theoretical background soon attracted economists’ attention to the trade-off between computational complexity and predictive validity in retirement modelling. Supporters of the option value model underlined how “complex specifications may presume computational facility that is beyond the grasp of most real people ...” (Lumsdaine, Stock, and Wise, 1992). Moreover, they provided evidence that the option value model was at least as good as the dynamic programming in terms of predictive validity. A comparison between option value and more recent models, which explain much more than the first dynamic programming applications, depends on the specific context of the analysis. Even if, in many cases, it is likely that more complex models better approximate reality, there can be situations in which rules governing actual behaviour are simpler and more suitably modelled within the option value framework. At the same time, the option value itself can to some extent be modified and adapted to given circumstances.