A Test of the Habit Formation Hypothesis Using Household Data

Numerous studies have confirmed the importance of habit formation, as represented by a lagged dependent variable, in demand analysis. Although all work to date has been based on aggregate time series data, this study uses household level BLS Interview Panel data to test the habit hypothesis. An interrelated demand system for seventeen goods is estimated from cross-section data and compared with a similar system based on time series data. The results show that the habit component is significantly different between the two data sets and much smaller in the cross-section data. Habit effects, while not as large in cross-section data as time series, are still highly significant. Several explanations are offered concerning why the two sets of estimates differ. Copyright 1991 by MIT Press.