Kerwin Charles and Melvin Stephens “The Level and Composition of Consumption over the Business Cycle” Comments by Christopher Jencks June 9, 2005 Main Findings • Mean expenditures hardly changed between 1988 and 2000 for any income group in the Consumer Expenditure Survey (CEX). • Expenditures in top three income quartiles were not affected by the business cycle. • In the bottom income quartile, a 2 point swing in unemployment only changed expenditures by about 6 percent. 2 Problem I: Trends • Hard to believe that living standards were no higher in 2000 than in 1988. • Comparisons between the CEX and the National Income and Product Accounts (NIPA) suggest that CEX underestimates spending in many categories. • This problem may have gotten worse between 1988 and 2000, although CEX and NIPA definitions differ (medical care). 3 Quarterly expenditures in 2000 dollars: NIPA versus CEX $18,000 $15,835 $16,000 NIPA personal consumption expenditure per household $14,000 $12,478 $12,000 $10,000 $8,300 CEX total expenditure per consumer unit 1988 1990 $8,200 $8,000 $6,000 $4,000 $2,000 $0 1992 1994 1996 1998 2000 4 Problem II: Recessions have changed • Historically, a recession meant both higher unemployment and lower consumption. • But this pattern changed in the 1990s. • The 1990-91 recession involved an unusually small drop in consumption. • The 2001-02 recession involved no drop at all in consumption (although rise slowed). 5 Mean Quarterly Personal Consumption Expenditure per Household: 1969 to 2004 in chained 2000 dollars $18,000 $16,000 $14,000 $12,000 $10,000 $8,000 $6,000 $4,000 $2,000 $0 1969 1974-75 1980-82 1991 2001 2004 6 What’s going on? In the 2001-02 GDP fell in only two quarters and consumption never fell. Employment leveled off because: More consumer goods were imported, Productivity rose Domestic investment fell. Unemployment rose because: Net immigration remained high New native job seekers exceeded retirees 7 Why doesn’t higher unemployment still lower consumption? Many trends predict that unemployment should lower consumption more than in past • Savings rate is down. • Unemployment insurance coverage is down. • Unsecured borrowing fell from 1990 to 1993 (Is this consistent with industry data?) 8 Why doesn’t higher unemployment still lower consumption? -- (cont.) Asset values matter more than saving rates for consumption smoothing Secured borrowing may have risen in 1991-92 and 2002-2003 Is the two earner family protective? How? Could inter-family transfers have risen? Did macroeconomic policy change in ways that protected more consumers? 9 Conclusions about paper • Need more checks on validity of CEX trend data for expenditures. Check NIPA. • Need data on whether CEX incomes fell in early 1990s and rose late 1990s. • Need to define “treatment.” • Estimate effect of “treatment” on cash and noncash income, total expenditure, and types of expenditure • Spell out how people protect themselves from downturns, and see why lowest quartile fares worst, if it does. Is it just low assets? 10