Prudence and Robustness as Explanations for Precautionary Savings : An Evaluation

This paper evaluates approximation methods to make manageable the numerical solution of overlapping generation models with aggregate risk. The paper starts with a model in which households maximize expected utility over their life cycle. Instantaneous utility is characterized by constant relative risk aversion. Prudence, a characteristic of the utility function, leads to precautionary saving. The first-order conditions include expectations. One source of uncertainty is not prohibitive for numerical integration of the expectation term. Because of its accuracy numerical integration results are used as a bench mark. Taylor series approximations can lead to the same results dependent on the linearization point. A linear quadratic approximation of the household model is evaluated subsequently. Alternatively, precautionary saving effects can be the result of robust decision making. This approach leads to linear policy functions and gives a rather good approximation of the bench mark model, although not as good as the Taylor series approximation