2015 Series • No. 15–12
Research Department Working Papers
The Rise and Fall of Consumption in the '00s
The major portion of U.S. gross domestic product (GDP) is accounted for by consumer spending, which significantly affects the business cycle. Consumer demand has been extremely volatile since 2000, especially given the booms and busts in housing values and in subprime mortgage lending. While it is well-established that housing net worth, credit availability, and household debt levels help to explain changes in consumer spending, the roles played by other potential determinants of consumption are not well identified or understood. This paper uses county-level data and a multiple-regression framework to explore how fluctuations in consumption between 2000 and 2012 are correlated with these macroeconomic variables: income, unemployment, debt, income inequality, consumer expectations, housing wealth, credit access, cash-out refinancings, and foreclosures. Four subperiods are considered: the "dot-com" recession (2001-2003), the "subprime boom" (2004-2006), the Great Recession (2007-2009), and the "tepid recovery" (2010-2012).
- Unemployment has the highest explanatory power over the entire sample period, while income growth and debt overhang have consistent but lower explanatory power. Other individual variables exert a positive or negative influence on consumption growth only during some of the subperiods.
- Housing wealth had strong positive effects on consumption growth during the subprime boom, strong negative effects during the Great Recession, and insignificant positive effects during the dot-com boom and the tepid recovery.
- Changes in consumer expectations had a significant positive effect on consumption growth during the subprime boom: consumption was about 1.6 percent higher in regions with a one-standard deviation more optimistic expectations than average, though the estimated impact was negligible during other subperiods, even when expectations were quite negative during both recessions.
The role of income and housing wealth is much reduced in multiple regressions compared to univariate regressions of consumption using these variables. Studies that consider individual determinants of consumer spending may suffer from omitted variables bias, so in order to successfully model consumption growth over this entire twelve-year period, multiple determinants of consumer spending must be considered. The paper's results caution against making forecasts based on past statistical relationships when there has been a change in the economic environment. The results from this rich specification of consumption determinants provide important benchmarks for guiding fiscal and monetary policy and descriptive statistics to match when building structural economic models.
U.S. consumption has gone through steep ups and downs since 2000, but the causes of these fluctuations are still imperfectly identified. We quantify the relative statistical impact of income, unemployment, house prices, credit scores, debt, expectations, foreclosures, inequality, and refinancings on consumption growth for four subperiods: the "dot-com recession" (2001– 2003), the "subprime boom" (2004–2006), the Great Recession (2007–2009), and the "tepid recovery" (2010–2012). We document that the explanatory power of different factors varies by subperiods, implying that a successful modeling of this entire decade needs to allow for multiple determinants of consumption. Unemployment, income, and debt are important determinants of consumption during all four periods.