Measuring Noise in the Permanent Income Hypothesis

Based on a number of deviation measures, Kim (1996) finds that postwar US consumption deviates from the Permanent Income Hypothesis (PIH) by only around 4 percent. In the present paper we investigate in more detail the extent to which the PIH provides a good approximation to US consumption data. We point out some unappealing features in the methods suggested by Kim, and we propose a method that does not have these drawbacks. In particular, we argue that due to the non-stationarity that characterizes consumption and income, deviation measure should be expressed in terms of saving rather than consumption. By applying our proposed method we find that in general US saving deviates fromPIH saving by substantially more than 4 percent. We also document that the behavior of US consumption in the 1990s has turned saving into a non-stationary process, which is strongly at odds with the PIH.