Subjective income risk and precautionary saving

Econometric studies have produced conflicting results on the relevance of precautionary saving. This ambiguity has been often ascribed to i) the difficulty of measuring key variables, like households' subjective risk in income and permanent income; ii) the occurrence of certain kinds of endogeneity bias associated to the unobservability of individual characteristics, like preferences and insurance possibilities. In the present work we investigate these estimation problems exploiting a particular wave of the Italian Survey of Household Income and Wealth which contains both type of information. Our results quantify the average precautionary saving as 4-5 percent of total net wealth. Moreover, excluding illiquid assets like the primary home and business equities, a figure of about 5 percent is statistically invariant to the alternative measures of wealth considered. This 'invariance' reveals a tendency of households to respond to a change in their perceived income risk by rescaling their (liquid) portfolio. We also discuss how the use of richer information on the household characteristics unveils how various kinds of 'bias' might affect estimates in opposite directions and with different intensities.