WEALTH EFFECTS REVISITED 1978-2009 By Karl E. Case, John M. Quigley and Robert J. Shiller February 2011 COWLES FOUNDATION DISCUSSION PAPER NO. 1784 COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY Box 208281 New Haven, Connecticut 06520-8281 http://cowles.econ.yale.edu/Wealth Effects Revisited 1978-2009 Karl E. Case Wellesley College Wellesley, MA [email protected] John M. Quigley University of California Berkeley, CA [email protected] Robert J. Shiller Yale University New Haven, CT [email protected] February 2011 We re-examine the link between changes in housing wealth, financial wealth, and consumer spending. We extend a panel of U.S. states observed quarterly during the seventeen-year period, 1982 through 1999, to the thirty-one year period, 1978 through 2009. Using techniques reported previously, we impute the aggregate value of owner-occupied housing, the value of financial assets, and measures of aggregate consumption for each of the geographic units over time. We estimate regression models in levels, first differences and in error-correction form, relating per capita consumption to per capita income and wealth. We find a statistically significant and rather large effect of housing wealth upon household consumption. This effect is consistently larger than the effect of stock market wealth upon consumption. This reinforces the conclusions reported in our previous analysis. In contrast to our previous analysis, however, we do find – based on data which include the recent volatility in asset markets – that the effects of declines in housing wealth in reducing consumption are at least as large as the effects of increases in housing wealth in increasing the course of household consumption. Keywords: consumption, nonfinancial wealth, housing market, real estate JEL Codes: E2, G1 We are grateful to Colleen Donovan and to Natasha Avendaño Garcia for research assistance.2 I. Introduction In the winter of 2000-2001, we made presentations at several professional meetings in which we sought to link household consumption expenditures to incomes and wealth, by relying on aggregate panel data on U.S. states and fourteen different countries. A formal paper was ultimately presented at the Summer Institute of the National Bureau of Economic Research (NBER) in July of 2001, and it was circulated as an NBER working paper (#8606) that fall. That research attempted to measure average consumption, income, housing wealth, and stock market wealth over time for U.S. states and foreign countries. The statistical relationship between consumption, income and wealth was estimated using standard multivariate techniques, and we interpreted the coefficients of the wealth variables as indicating the strength of the association between these two kinds of household wealth and household consumption. Our statistical results suggested that there were significant “wealth effects” upon consumption associated with both types of wealth, housing wealth and financial wealth, but that the stimulatory effects of housing wealth substantially exceeded the effects of financial wealth. This result persisted for a variety of specifications for both panels of aggregate data. These results received some notice in the popular media,1 in some part, presumably, reflecting concurrent trends in the macro economy. In due course, the paper, “Comparing Wealth Effects: The Stock Market versus the Housing Market,” was 1 This work was the subject of the “Economics Focus” column in the Economist (November 8, 2001) and formed the basis for a subsequent cover story (March 30, 2002).3 published, in Advances in Macroeconomics in 2005. Contemporaneously, the data were made available online, and they were used by John Muellbauer (2008) in his well-known paper presented at the Federal Reserve Conference at Jackson Hole, Wyoming in 2007. When our paper was originally presented, it relied upon the most recent data available. (The paper was first presented in January 2000, and it relied upon data through the second quarter of 1999). By the time the research was published, five years had elapsed, and by the time of the disastrous meltdown in mortgage markets, more than seven years had elapsed. The purpose of this paper is to update the empirical analysis using data through 2009, and thus to incorporate the past decade of unusual volatility in housing wealth, stock market wealth, and personal consumption. As before, we present a variety of econometric models linking consumption to income, housing wealth, and stock market wealth. As in our previous analysis, we make no effort to “deduce” a structural model reflecting these relationships, preferring again to observe the robustness of these relationships to plausible specifications of the association. In attempting to update our previous analysis, it was immediately apparent that comparable data from the panel of OECD countries previously analyzed could not be obtained. Hence, this analysis is confined to quarterly data on U.S. states, 1978:I-2009:II, extending our previous work which had analyzed these macro forces during 1982:I-1999:II. The principal results and interpretations in our previous work are largely unchanged, but the estimated magnitudes are larger and more important statistically and also economically. When the more recent volatile period is included in the analysis, we4 find that the relationship between housing market wealth and consumption is a good bit stronger, relative to the link between stock market wealth and consumption. This key finding is robust to a variety of reasonable specifications. One set of previous findings does not seem to hold up to replication – certainly not as strongly as during the earlier period. Previously, we noted an asymmetry in the association between housing market wealth and consumption. When housing market wealth was increasing, household consumption was increasing. But when housing market wealth declined, household consumption declined only marginally. For the most part, this asymmetry is absent from the longer panel which now includes substantially more variation in asset prices, notably periods of declining house prices and declining stock market indices. In Section II below we review the conceptual and measurement issues addressed in the original research paper, and we discuss our efforts to extend the time series for analysis. We also describe recent trends in