Partial insurance, information and consumption dynamics

Abstract

Executive summaryThis paper uses panel data on household consumption and income to evaluate the degree of insurance to income shocks. Our aim is to describe the transmission of income inequality into consumption inequality. Our framework nests the special cases of self-insurance and the complete markets assumption. We assess the degree of insurance over and above self-insurance through savings by contrasting shifts in the cross-sectional distribution of income growth with shifts in the cross-sectional distribution of consumption growth, and analyzing the way these two measures of household welfare correlate over time. We combine panel data on income from the PSID with consumption data from repeated CEX cross-sections in a structural way, i.e. using conventional demand analysis rather than reduced form imputation procedures. Our results point to some partial insurance but reject the complete markets restriction. We find a greater degree of insurance for transitory shocks and differences in the degree of insurance over time and across education. We also document the importance of durables and of taxes and transfers as a means of insurance.

A Theory of Housing Collateral, Consumption Insurance and Risk Premia

“…Using US metropolitan area data, we find that the degree of insurance between regions decreases when the housing collateral ratio is low. This is consistent with evidence from Blundell, Pistaferri, and Preston (2002), who find evidence for time-variation the economy's risk sharing capacity. Regional income and consumption data provide direct support for the existence and importance of the collateral mechanism, whose asset pricing implications were the focus of this paper.…”

Errors in survey reports of consumption expenditures

“…7 The panel component of the IS survey is not exploited in this paper, that is quarterly observations for IS households are counted separately as if the four observations over the one year interview referred to different households. 8 It might be interesting considering how much robust our results are to variations in head's age definition (Deaton and Paxson, 2000). However, there is not particular reason to believe that any bias arising from such problem affects the two instruments in a different way and/or with a different sign.…”

“…However, income and assets data are known to be not as reliable as the expenditure data: the amount of incomplete income reporters is about 20 percent in the two surveys and missing values are currently not imputed. For this reason many applications in the literature have combined consumption information from the CEX to income information from complementary data sets (see, for example, Lusardi 1996 andBlundell et al 2002).…”

“…As stated above, within cohort inequality from the IS survey presents a mildly increasing pattern in the 1980s for those born between 1930 and 1949, but there is no evidence of any increase after 1990. This result can be directly compared to the numbers reported in Deaton and Paxson (1994) and Blundell et al (2002) where IS information for the 1980s is used. DS inequality is instead increasing over time uniformly for all cohorts.…”

“…As discussed in the last section, additional structure is required to identify the covariance term in (8). For example, Attanasio et al (2003) consider alternative identification strategies based on the specification of a demand system for C * 1 and C * 2 .…”