The test of long-memory explanation of the deaton paradox by conducting unit root test of household income from PSID

Abstract

The Deaton Paradox is one of the most strikingconsumption behavior topics in recent years. One attractive explanation of the Deaton Paradox is the long-memory process hypothesis proposed by Diebold (1991). However, the existence of long-memory process in the aggregate level consumption data requires that the individual level data can be fitted by AR processes. As a result, I would like to investigate on the individual level income data of US families to check the availability of long-memory process explanation to the Deaton Paradox.