Title: Some new evidence on the response of households to predictable changes in income
Abstract: Several tests of the Permanent Income Hypothesis have examined whether predictable changes in income affect current consumption growth. The predictable income change is constructed using past variables as instruments in a two-stage regression. However, finding instruments that are truly exogenous but nevertheless predict income growth is very difficult. Moreover, identification requires a large number of time periods. This paper, using the CEX, shows that previous studies have given insufficient attention to properly testing the instruments. Using a longer number of time periods, and a valid set of instruments, the paper finds that households are indeed excessively sensitive to predictable changes in income. But if households are partitioned by their level of education, the study finds highly educated households behave according to the predictions of the Permanent Income Hypothesis.