Heterogeneity in Economic Shocks and Household Spending March 2015 Sebastian Devlin-Foltz John Sabelhaus Federal Reserve Board The analysis and conclusions set forth are those of the author and do not indicate concurrence by other members of the research staff or the Board of Governors of the Federal Reserve System. Is Rising Inequality Affecting Economic Stability and Growth? • Inequality is rising, low-modest income families “kept up” their spending by borrowing excessively during boom years • House price boom reinforces increased spending and borrowing through wealth and/or collateral channel • Housing boom ends, consumers deleverage, spending falls • Incomes fall, especially for low-modest income (& young?) • Debt overhang (student loans?) slowing spending, recovery 2 Goals/Approach for this Paper • Does household survey data support the narrative? oMost favorable evidence is from geographically-constructed data oUse micro data to look directly at changes in income, wealth, spending, and debt accumulation across groups and time • Need comprehensive micro data that tracks macro trends o Survey of Consumer Finances (SCF) cross-sections, 1995-2013 • Use synthetic cohort analysis to deconstruct trends o Birth cohort groups gives “lifecycle” perspective o “Permanent” income groups within birth cohort groups speaks directly to inequality and related heterogeneity narratives 3 Survey of Consumer Finances (SCF) • Federal Reserve Board’s triennial household survey • Income and balance sheets, some spending information • Oversamples high wealth families • SCF has same basic structure since 1989 oPermanent (usual) income measure only available 1995+ • Does the SCF capture the trends in income, wealth, spending, and debt that we are trying to explain? 4 Aggregate Income 140 National Income and Product Accounts Index (2006=100) 120 100 80 Sources: National Income and Product Accounts, Statistics of Income, Survey of Consumer Finances. NIPA Income is the total of line 2.1.1 plus 7.2.21 minus 2.1.7, 2.1.19, 2.1.20 and 7.2.11. SOI Income is total AGI less deficit. SCF Actual Income is a household's total income in the year prior to the survey year. SCF usual income is a household reported income in a "usual" year. 60 40 1994 1997 2000 2003 2006 2009 2012 Aggregate Income 140 National Income and Product Accounts IRS Statistics of Income Index (2006=100) 120 100 80 Sources: National Income and Product Accounts, Statistics of Income, Survey of Consumer Finances. NIPA Income is the total of line 2.1.1 plus 7.2.21 minus 2.1.7, 2.1.19, 2.1.20 and 7.2.11. SOI Income is total AGI less deficit. SCF Actual Income is a household's total income in the year prior to the survey year. SCF usual income is a household reported income in a "usual" year. 60 40 1994 1997 2000 2003 2006 2009 2012 Aggregate Income 140 Survey of Consumer Finances Actual Income National Income and Product Accounts IRS Statistics of Income Index (2006=100) 120 100 80 Sources: National Income and Product Accounts, Statistics of Income, Survey of Consumer Finances. NIPA Income is the total of line 2.1.1 plus 7.2.21 minus 2.1.7, 2.1.19, 2.1.20 and 7.2.11. SOI Income is total AGI less deficit. SCF Actual Income is a household's total income in the year prior to the survey year. SCF usual income is a household reported income in a "usual" year. 60 40 1994 1997 2000 2003 2006 2009 2012 Aggregate Income 140 Index (2006=100) 120 Survey of Consumer Finances Actual Income Survey of Consumer Finances Usual Income National Income and Product Accounts IRS Statistics of Income 100 80 Sources: National Income and Product Accounts, Statistics of Income, Survey of Consumer Finances. NIPA Income is the total of line 2.1.1 plus 7.2.21 minus 2.1.7, 2.1.19, 2.1.20 and 7.2.11. SOI Income is total AGI less deficit. SCF Actual Income is a household's total income in the year prior to the survey year. SCF usual income is a household reported income in a "usual" year. 60 40 1994 1997 2000 2003 2006 2009 2012 Aggregate Net Worth 140 Financial Accounts of the United States Index (2007=100) 100 60 Sources: Financial Accounts of the United States, Survey of Consumer Finances. FAOTUS and SCF net worth measures adjusted for conceptual consistency as in Bricker et al (2015) 20 1995 1998 2001 2004 2007 2010 2013 Aggregate Net Worth 140 Financial Accounts of the United States Survey of Consumer Finances Index (2007=100) 100 60 Sources: Financial Accounts of the United States, Survey of Consumer Finances. FAOTUS and SCF net worth measures adjusted for conceptual consistency as in Bricker et al (2015) 20 1995 1998 2001 2004 2007 2010 2013 Aggregate New Car Spending 120 National Income and Product Accounts Index (2007=100) 100 80 60 Sources: National Income and Product Accounts, Survey of Consumer Finances. NIPA Table 2.4.5 Line 5 "New Motor Vehicles". SCF car spending totals new cars purchased with a loan in the survey year or year prior with a model year newer than the survey year minus 2, and new cars purchased with cash in the survey year or year prior with a model year older than the year prior to the survey year. 40 1995 1998 2001 2004 2007 2010 2013 Aggregate New Car Spending 120 National Income and Product Accounts Survey of Consumer Finances Index (2007=100) 100 80 60 Sources: National Income and Product Accounts, Survey of Consumer Finances. NIPA Table 2.4.5 Line 5 "New Motor Vehicles". SCF car spending totals new cars purchased with a loan in the survey year or year prior with a model year newer than the survey year minus 2, and new cars purchased with cash in the survey year or year prior with a model year older than the year prior to the survey year. 40 1995 1998 2001 2004 2007 2010 2013 Aggregate Recent Housing Purchases 120 National Association of Realtors / Department of Commerce Index (2007=100) 100 80 60 Sources: National Association of Realtors, Department of Commerce, Survey of Consumer Finances. National Associaiton of Realtors data is the sum of existing single family home sales in each quarter between survey year Q3 12 and survey year Q3. Department of Commerce Data is the sum of new homes sales in each quarter between survey year Q3 - 12 and survey year Q3. SCF measure is total weighted number of households that purchased their home within the last three years. 40 1995 1998 2001 2004 2007 2010 2013 Aggregate Recent Housing Purchases 120 National Association of Realtors / Department of Commerce Survey of Consumer Finances Index (2007=100) 100 80 60 Sources: National Association of Realtors, Department of Commerce, Survey of Consumer Finances. National Associaiton of Realtors data is the sum of existing single family home sales in each quarter between survey year Q3 12 and survey year Q3. Department of Commerce Data is the sum of new homes sales in each quarter between survey year Q3 - 12 and survey year Q3. SCF measure is total weighted number of households that purchased their home within the last three years. 40 1995 1998 2001 2004 2007 2010 2013 Aggregate Household Debt 120 Financial Accounts of the United States Index (2007=100) 90 60 Sources: Financial Accounts of the United States, Survey of Consumer Finances. FAOTUS and SCF debt measures adjusted for conceptual consistency as in Bricker et al (2015) 30 1995 1998 2001 2004 2007 2010 2013 Aggregate Household Debt 120 Financial Accounts of the United States Survey of Consumer Finances Index (2007=100) 90 60 Sources: Financial Accounts of the United States, Survey of Consumer Finances. FAOTUS and SCF debt measures adjusted for conceptual consistency as in Bricker et al (2015) 30 1995 1998 2001 2004 2007 2010 2013 And SCF tracks top 1% income share… 25% Administrative Data SCF Bulletin Income, Households Percent Share 20% 15% Sources: Survey of Consumer Finances (SCF) and Piketty and Saez (2003 + updates). SCF incomes are collected for the calendar year prior to each triennial survey. See Appendix B for details on Administrative, SCF Bulletin, and SCF Market income concepts. Income thresholds for identifying the top 1% of households and tax units are reported in Appendix C. 10% 1988 1991 1994 1997 2000 2003 2006 2009 2012 Especially after fixing income concept… 25% Administrative Data SCF Bulletin Income, Households SCF Market Income, Households Percent Share 20% 15% Sources: Survey of Consumer Finances (SCF) and Piketty and Saez (2003 + updates). SCF incomes are collected for the calendar year prior to each triennial survey. See Appendix B for details on Administrative, SCF Bulletin, and SCF Market income concepts. Income thresholds for identifying the top 1% of households and tax units are reported in Appendix C. 10% 1988 1991 1994 1997 2000 2003 2006 2009 2012 And correcting for unit of observation… 25% Administrative Data SCF Bulletin Income, Households SCF Market Income, Households SCF Market Income, Tax Units Percent Share 20% 15% Sources: Survey of Consumer Finances (SCF) and Piketty and Saez (2003 + updates). SCF incomes are collected for the calendar year prior to each triennial survey. See Appendix B for details on Administrative, SCF Bulletin, and SCF Market income concepts. Income thresholds for identifying the top 1% of households and tax units are reported in Appendix C. 10% 1988 1991 1994 1997 2000 2003 2006 2009 2012 Synthetic Cohort Analysis • Given micro data that tracks aggregates, what can we say about shocks and responses across types of families? o Narrative focuses on old v. young and rich v. poor • Six ten-year birth cohorts per chart, lifecycle view • Within birth cohorts, put families into three “usual” income groups, separate charts for bottom 50, next 45, top 5 • Show graphically, then use synthetic cohort data to answer direct questions pertaining to narrative, for example oWho borrowed so much between 2001 and 2007? oWho stopped buying cars between 2007 and 2010? Actual and “Usual” Income in the SCF • Synthetic cohort analysis by income requires a plausible sorting of families into groups, actual income too volatile • SCF actual income is sum of respondent-reported earnings from wages, businesses, investments, pensions, transfers • After reporting actual, respondents asked to estimate “usual” income in a “normal” year, if different from actual • Self-reported version of “permanent” income • Most report actual=normal, sizeable minority in each year report deviations, varies predictably with business cycle Statistical Properties of Self-Reported Transitory Shocks • Traditionally, model log household income using observables (xit) and permanent (µit) and transitory (εit) shock components: ln(yit) = βxit + µit + εit with µit= µi,t-1 + ηit evolving over time. • Assume transitory shocks zero mean, variance σε2, and permanent shocks orthogonal with zero mean, variance σµ2 • Ackerman/Sabelhaus (2012) show estimated transitory variances, percent negative and positive, and skewness of gap between actual and usual all match panel data Income and Wealth Change Heterogeneity • Three charts per variable, one for each usual income group • Each birth cohort represented by up to seven points corresponding to the seven (midpoint) ages at which that cohort was observed between 1995 and 2013 • Focus on three aspects of lifecycle trajectories oGrowth during the boom period (through 2007) oChange in trajectory for that cohort in the bust (after 2007) oRelationship of that cohort’s values to predecessor cohort as observed 10 years earlier in a different SCF 23 $45,000 Average Income, Bottom 50% by Usual Income 1951-1960 $40,000 1961-1970 1941-1950 $35,000 $30,000 1971-1980 1931-1940 $25,000 $20,000 1981-1990 $15,000 $10,000 $5,000 Source: Survey of Consumer Finances, 1995 to 2013 $0 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 $45,000 Average Income, Bottom 50% by Usual Income 1951-1960 $40,000 1961-1970 1941-1950 $35,000 $30,000 1971-1980 1931-1940 $25,000 $20,000 Indicates how far below previous cohort at the same age ten years earlier 1981-1990 $15,000 $10,000 $5,000 Source: Survey of Consumer Finances, 1995 to 2013 $0 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Average Income, Next 45 % by Usual Income $160,000 $140,000 1951-1960 1941-1950 1961-1970 $120,000 1931-1940 $100,000 1971-1980 $80,000 $60,000 1981-1990 $40,000 $20,000 Source: Survey of Consumer Finances, 1995 to 2013 $0 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Average Income, Top 5% by Usual Income $1,200,000 1941-1950 $1,000,000 1951-1960 $800,000 1931-1940 1961-1970 $600,000 $400,000 1971-1980 $200,000 Source: Survey of Consumer Finances, 1995 to 2013 1981-1990 $0 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Average Net Worth, Bottom 50% by Usual Income $300,000 $250,000 $200,000 $150,000 $100,000 $50,000 Source: Survey of Consumer Finances, 1995 to 2013 $0 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Average Net Worth, Next 45% by Usual Income $1,200,000 $1,000,000 $800,000 $600,000 $400,000 $200,000 Source: Survey of Consumer Finances, 1995 to 2013 $0 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Average Net Worth, Top 5% by Usual Income $12,000,000 $10,000,000 $8,000,000 $6,000,000 $4,000,000 $2,000,000 Source: Survey of Consumer Finances, 1995 to 2013 $0 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Spending and Borrowing Heterogeneity • Again, three charts per variable, one for each usual income group; variables of interest for narrative are • New car spending (average unconditional $s) • Buying new primary residence (percent) • Debt to income ratio (average debt/average income) • Focus again on shape of lifecycle trajectories, and behavior relative to previous cohort ten years earlier • Also consider whether spending or borrowing behavior consistent with or divorced from income and wealth trends 31 Average New Car Spending, Bottom 50% by Usual Income $3,500 $3,000 $2,500 $2,000 $1,500 $1,000 $500 Source: Survey of Consumer Finances, 1995 to 2013 $0 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Average New Car Spending, Next 45 % by Usual Income $10,000 $9,000 $8,000 $7,000 $6,000 $5,000 $4,000 $3,000 $2,000 $1,000 Source: Survey of Consumer Finances, 1995 to 2013 $0 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Average New Car Spending, Top 5% by Usual Income $30,000 $25,000 $20,000 $15,000 $10,000 $5,000 Source: Survey of Consumer Finances, 1995 to 2013 $0 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Fraction of Families Buying Primary Residence Past 3 Years, Bottom 50% 60% Source: Survey of Consumer Finances, 1995 to 2013 50% 40% 30% 20% 10% 0% 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Fraction of Families Buying Primary Residence Past 3 Years, Next 45% 60% 50% 40% 30% 20% 10% Source: Survey of Consumer Finances, 1995 to 2013 0% 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Fraction of Families Buying Primary Residence Past 3 Years, Top 5% 60% 50% 40% 30% 20% 10% Source: Survey of Consumer Finances, 1995 to 2013 0% 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Debt to Income Ratio, Bottom 50% by Usual Income 250% 200% 150% 100% 50% Source: Survey of Consumer Finances, 1995 to 2013 0% 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 250% Debt to Income Ratio, Next 45% by Usual Income 200% 150% 100% 50% Source: Survey of Consumer Finances, 1995 to 2013 0% 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Debt to Income Ratio, Top 5% by Usual Income 250% 200% 150% 100% 50% Source: Survey of Consumer Finances, 1995 to 2013 0% 20 25 30 35 40 45 50 Age 55 60 65 70 75 80 Decomposing Aggregate Trends • Synthetic cohort estimates ultimately useful for calibrating dynamic models that seek to explain changes in behavior • Differences in behavior across groups become moments to match in calibration, for example oDid new car and primary residence buying decline because certain types of families were suddenly constrained? oOr, was it more widespread, maybe expectations driven? • In the meantime, use simple examples of how synthetic cohort data supports or refutes various aspects of narrative 41 Who Borrowed So Much Between 2001 and 2007? Share of Debt, 2001 Birth Cohort Usual Income 1981-90 1971-80 1961-70 1951-60 1941-50 1931-40 Bottom 50 0% 2% 6% 8% 4% 2% Next 45 0% 6% 16% 20% 12% 5% Top 5 0% 2% 4% 6% 5% 2% All 21% 60% 19% 42 Who Borrowed So Much Between 2001 and 2007? Share of Debt, 2001 Birth Cohort Usual Income 1981-90 1971-80 1961-70 1951-60 1941-50 1931-40 Bottom 50 0% 2% 6% 8% 4% 2% Next 45 0% 6% 16% 20% 12% 5% Top 5 0% 2% 4% 6% 5% 2% All 21% 60% 19% Share of Debt Growth, 2001-2007 Birth Cohort Usual Income 1981-90 1971-80 1961-70 1951-60 1941-50 1931-40 0% Bottom 50 1% 4% 6% 3% 2% 13% 11% -1% Next 45 4% 25% 18% 1% 0% Top 5 1% 3% 5% 6% All 16% 68% 15% 43 Who Borrowed So Much Between 2001 and 2007? Share of Debt, 2001 Birth Cohort Usual Income 1981-90 1971-80 1961-70 1951-60 1941-50 1931-40 Bottom 50 0% 2% 6% 8% 4% 2% Next 45 0% 6% 16% 20% 12% 5% Top 5 0% 2% 4% 6% 5% 2% All 21% 60% 19% Share of Debt Growth, 2001-2007 Birth Cohort Usual Income 1981-90 1971-80 1961-70 1951-60 1941-50 1931-40 0% Bottom 50 1% 4% 6% 3% 2% 13% 11% -1% Next 45 4% 25% 18% 1% 0% Top 5 1% 3% 5% 6% All 16% 68% 15% 44 Who Stopped Buying Cars Between 2007 and 2010? Share of New Car Spending, 2007 Birth Cohort Usual Income 1981-90 1971-80 1961-70 1951-60 1941-50 1931-40 Bottom 50 0% 1% 2% 4% 4% 2% Next 45 1% 2% 7% 12% 14% 7% Top 5 0% 2% 6% 15% 14% 9% All 11% 43% 46% 45 Who Stopped Buying Cars Between 2007 and 2010? Share of New Car Spending, 2007 Birth Cohort Usual Income 1981-90 1971-80 1961-70 1951-60 1941-50 1931-40 Bottom 50 0% 1% 2% 4% 4% 2% Next 45 1% 2% 7% 12% 14% 7% Top 5 0% 2% 6% 15% 14% 9% All 11% 43% 46% Share of New Car Spending Decrease, 2007-2010 Birth Cohort Usual Income 1981-90 1971-80 1961-70 1951-60 1941-50 1931-40 Bottom 50 0% 1% 3% 10% 6% 1% Next 45 0% 6% 10% -9% 19% 21% Top 5 -1% 1% -22% 19% 14% 20% All 21% 47% 32% 46 Conclusions • Shocks to income and wealth between 2007 and 2010 were widespread across birth cohorts and income groups • Decreased spending and changes in the other outcome variables also very widespread o Support for heterogeneity in circumstances and behavior driving aggregate trends weaker than previously thought o Common behavioral changes suggest factors like expectations about house prices, income growth probably more important • In any event, narratives based solely on balance sheets, income fundamentals, or credit constraints, for only a subset of families seem incomplete 47 Thanks! john.sabelhaus@frb.gov Additional Slides Median Income with Counterfactual Lifecycle Growth $90,000 $80,000 1941-1950 Cohort $70,000 $60,000 1951-1960 Cohort $50,000 1961-1970 Cohort $40,000 1931-1940 Cohort 1971-1980 Cohort $30,000 1981-1990 Cohort $20,000 $10,000 Note: Counterfactual growth rate estimated using observed growth for earlier cohorts at same ages, 1995 through 2007. Source: Survey of Consumer Finances, 1995 to 2013 $0 20 25 30 35 40 45 50 55 60 65 70 75 80 Is SCF Missing Some of the 2013 Wealth Gains? • Financial Accounts (FA) aggregates show stronger gains in household sector net worth o FA (‘10 Q1 to ‘13 Q1) +21% aggregate nominal o SCF shows about +12% for 2010-2013 • SCF respondents may not have fully built in recovery in housing and stock prices o National house price indexes disagree on timing o Some correction for prior survey over-valuations? • Also, continued strong growth since survey conducted 2014-04-01 2014-01-01 2013-10-01 $339K (2013 $) 2013-07-01 2010 SCF Field Period 2013-04-01 2013-01-01 2012-10-01 2012-07-01 2012-04-01 2012-01-01 2011-10-01 150 2011-07-01 The small nominal change in SCF average house values between 2010 and 2013 is in line with FHFA, and below the increase reported by Case-Shiller 2011-04-01 225 2011-01-01 250 2010-10-01 FHFA Case-Shiller CoreLogic 2010-07-01 2010-04-01 2010-01-01 2009-10-01 2009-07-01 2009-04-01 2009-01-01 2008-10-01 275 2008-07-01 2008-04-01 2008-01-01 2007-10-01 2007-07-01 2007-04-01 2007-01-01 2006-10-01 2006-07-01 2006-04-01 2006-01-01 2005-10-01 2005-07-01 2005-04-01 2005-01-01 2004-10-01 2004-07-01 175 2004-04-01 2004-01-01 Index 1989 Q1 = 100 FHFA, Case-Shiller, and CoreLogic Nominal House Price Indexes and SCF House Values 300 2013 SCF Field Period SCF Average House= $316K (nominal) 200 SCF Average House= $319K (nominal) 2013 SCF Field Period SCF Average Equity Holdings= $213k (nominal) $228K (2013 $) SCF Average Equity Holdings= 2014-07-01 15000 2010 SCF Field Period 2014-04-01 2014-01-01 2013-10-01 2013-07-01 2013-04-01 2013-01-01 2012-10-01 2012-07-01 2012-04-01 2012-01-01 2011-10-01 2011-07-01 2011-04-01 2011-01-01 The Wilshire index increased 36% between 2010 Q1 to 2013 Q1, while SCF aggregate nominal holdings rose 29% between the 2010 and 2013 surveys. 2010-10-01 10000 2010-07-01 20000 2010-04-01 2010-01-01 2009-10-01 2009-07-01 2009-04-01 2009-01-01 2008-10-01 2008-07-01 2008-04-01 2008-01-01 2007-10-01 2007-07-01 2007-04-01 2007-01-01 2006-10-01 2006-07-01 2006-04-01 2006-01-01 2005-10-01 2005-07-01 2005-04-01 5000 2005-01-01 2004-10-01 2004-07-01 Aggregate Stock Market Valuation Wilshire 5000 Stock Value Index and SCF Mean Equity Holdings 25000 (nominal) $270K 0