Trends, Random Walks, and Tests of the Permanent Income Hypothesis

Recent studies find that consumption is excessively sensitive to income. These studies assume that income is stationary around a deterministic trend. The data, however, do not reject the hypothesis that disposable income is a random walk with drift. If income is indeed a random walk, then the standard testing procedure is greatly biased toward finding excess sensitivity. Moreover, if income is borderline stationary, this procedure is also seriously biased.

JEL Classification: 131, 211, 921

Keywords: Non-stationary time series, Detrending, Permanent income hypothesis, Small sample bias