Estimating household consumption insurance

Blundell, Pistaferri, and Preston (American Economic Review, 2008, 98(5), 1887–1921) report an estimate of household consumption insurance with respect to permanent income shocks 36%. In replicating findings for their model data, we find that this is distorted by a code error not robust weighting scheme generalized method moments (GMM) or consideration quasi maximum likelihood estimation (QMLE), which produces significantly higher at 55%. For sub‐groups age education, the differences between estimates across methods are even more pronounced, QMLE provides new insights into heterogeneity households compared original study. Monte Carlo experiments using non‐normal suggest accurate than GMM, including when correcting bias especially given smaller sample such as only available looking sub‐groups.