The relevance of precautionary saving

The literature contains a number of applied works trying to assess the relevance of the precautionary motive for saving. Some papers, based on calibrated models and simulations, indicated potentially high levels of precautionary savings. Data based econometric models, on the other hand, have largely failed to confirm these findings. Moreover, most of the studies cited above were based on particular functional forms for utility and upon linearization of first order conditions of intertemporal optimization problems, which recently came under strong criticism in the literature (see, for example, the contributions by Ludvigson and Paxon (2001) and Carroll (2001)). The estimation contained in this work, using micro-data from a survey of Italian households, challenges that conclusion. In contrast to previous studies, we successfully estimate absolute and relative prudence using a relatively unrestricted form of utility, and relate these measures directly to precautionary saving by households. We find statistically significant evidence of precautionary saving, accounting for approximately 20% of total savings on average.