Precautionary saving in mean-variance models and different sources of risk

Abstract

We study the effects of first- and second-order risk increases on precautionary saving in a mean-variance model. In doing so, we reduce the gap between the theory of saving, which mainly stems from the expected utility model, and empirical estimations of the theory that are based on different measures of dispersion; these are atheoretical concepts that do not arise from optimal agent behavior. We then analyze what effects different risk sources have on saving and show that our results, derived in the mean-variance space, can easily be translated to conditions in the expected utility space. We argue that our contribution establishes a more solid ground for analyzing policies in highly risky environments, such as the COVID-19 pandemic.