Staff Report, Federal Reserve Bank of New York 131
In a recent paper (“A Primer on the Economics and Time Series Econometrics of Wealth Effects,” 2001), Davis and Palumbo investigate the empirical relation between three cointegrated variables: aggregate consumption, asset wealth, and labor income. Although cointegration implies that an equilibrium relation ties these variables together in the long run, the authors focus on the following structural question about the short-run dynamics: “How quickly does consumption adjust to changes in income and wealth? Is the adjustment rapid, occurring within a quarter, or more sluggish, taking place over many quarters?” The authors claim that their findings answer this question, and imply that spending adjusts only gradually after gains or losses in income or wealth have been realized. We argue here, however, that a statistical methodology different from that used by Davis and Palumbo is required to address these questions, and that once it has been employed, the resulting empirical evidence weighs considerably against their interpretation of the data.