Wealth Inequality in the United States since 1913 Emmanuel Saez (UC Berkeley) Gabriel Zucman (LSE) October 2014
Introduction US Income inequality has increased sharply since the 1970s Mixed existing evidence on wealth inequality changes Is inequality increase driven solely by labor income? We capitalize income tax return data to estimate new annual series of US wealth concentration since 1913 Key result: Wealth inequality has surged but phenomenon is concentrated mostly within the top.1% (=wealth above $20m)
U-Shaped Wealth Concentration 25% Top 0.1% wealth share in the United States, 1913-2012 20% % of total household wealth 15% 10% 5% 0% 1913 1918 1923 1928 1933 1938 1943 1948 1953 1958 1963 This figure depicts the share of total household wealth held by the 0.1% richest families, as estimated by capitalizing income tax returns. In 2012, the top 0.1% includes about 160,000 families with net wealth above $20.6 million. Source: Appendix Table B1. 1968 1973 1978 1983 1988 1993 1998 2003 2008 2013
Surge in top wealth shares concentrated in top 0.1% 14% Top wealth shares: decomposing the top 1% 12% % of total household wealth 10% 8% 6% 4% 2% Top 0.1%-0.01% Top 0.5%-0.1% Top 0.01% Top 1%-0.5% 0% 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015
Outline of the talk I.The capitalization method II. The distribution of wealth III. Robustness and comparison with existing estimates IV. Decomposing wealth accumulation: income and saving rates
I- The capitalization method
To obtain wealth, we divide capital income by the rate of return How the capitalization technique works: Start from each capital income component reported on individual tax returns Compute aggregate rate of return for each asset class (using Flow of Funds and aggregate tax data) Multiply each individual capital income component by 1/rate of return of corresponding asset class Simple idea, but lot of care needed in reconciling tax with Flow of Funds data Key assumption: uniform return within asset class Need detailed income components to obtain reliable results
Aggregate income and wealth Aggregate wealth W = Total assets minus liabilities of households at market value Excludes durables, unfunded DB pensions, non-profits Source: Flow of Funds since 1945, Goldsmith, Wolff (1989), Kopczuk and Saez (2004) before Aggregate income returns NIPA since 1929, Kuznets (1941) and King (1930) before 1929 Family unit Top 1% = Top 1% of all family units [as in Piketty and Saez] defs
A U-shaped wealth-income ratio The composition of household wealth in the U.S., 1913-2013 500% 400% % of national income 300% 200% 100% Equities Sole proprietorships & partnerships Pensions 0% 1913 1918 1923 1928 1933 1938 1943 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 1998 2003 2008 2013 Currency, deposits and bonds Housing (net of mortgages)
Distributional data: income tax returns Consistent, annual, high quality data since 1913: Composition tabulations by size of income 1913- IRS micro-files with oversampling of the top 1962- Various additional IRS published stats (estates, IRAs, trusts, foundations) Detailed income categories: Dividends, interest (+ tax exempt since 1987), rents, unincorporated business profits (S corporations, partnerships, sole prop.), royalties, realized capital gains, etc. A lot of income flows to individual income tax returns Mutual funds, S corporations, partnerships, holding companies, trusts, etc.
Concentration of reported capital income has increased dramatically 45% 40% The top 0.1% taxable capital income share % of taxable capital income 35% 30% 25% 20% 15% 10% Including capital gains Excluding capital gains 5% 0% 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015
How we deal with non-taxable components Owner-occupied housing Home values set proportional to property tax paid Home mortgages set proportional to mortgage interest paid We assume (based on SCF) that itemizers have 75% of home wealth and 80% of home mortgages Pensions Pension wealth set proportional to pension distributions and wages above 50th percentile Consistent with SCF and with direct information on IRA wealth from IRS (IRAs 30% of pension wealth) Only matters for top 10% but irrelevant for top 1% and above, because pensions and housing very small there
How we deal with avoidance and evasion Tax avoidance: Systematic reconciliation exercise with national accounts to identify potential gaps in tax data kinc E.g., trust income imputations on the basis of distributions (Retained trust income 2% of household capital income) trusts Tax evasion: charitable Third-party reporting means all dividends and interest earned through domestic banks are reported Offshore wealth: If anything increases the trend in rising wealth top wealth shares by about 2 points offshore
Is the return constant within asset class? Two potential issues: Maybe the very rich have higher equity/bond returns (e.g., better at spotting good investment opportunities) level bias Maybe this differential has increased since the 1970s (e.g., due to financial globalization/innovation) trend bias Two checks show that return within asset class is flat and has remained flat
Check 1: No evidence that the wealthy have higher returns within asset class 10.0% 9.0% Returns by asset and wealth class, 2007 (matched tabulated estates and income tax data) 8.0% 7.0% Dividends + capital gains 6.0% 5.0% 4.0% Dividends yield 3.0% 2.0% 1.0% 0.0% Interest yield up to $3.5m $3.5m-$5m $5m-$10m $10m-$20m $20m+ Total net wealth at death
The very rich did collect a lot of dividends in the 1970s 8.0% Dividend yield by wealth class in 1976 (matched micro estate and income tax data) 7.0% 6.0% 5.0% 4.0% 3.0% 2.0% 1.0% P90-95 P95-99 P99-99.5 P99.5-99.9 P99.9-99.99 P99.99-100 Fractiles of the distribution of net wealth at death
Check 2: The capitalization method works for SCF and foundations Capitalization method can be checked with joint income and wealth micro-data: 1) SCF Data: provides individual micro-data for both wealth and (tax return) income component by component since 1989 2) Foundation Data: publicly available IRS micro-data with information on both market value wealth and income returns We apply same rates of returns & capitalization technique as for individual tax returns By capitalizing income we are able to reproduce the correct wealth distribution
Capitalization method works for the SCF % of household wealth excluding pensions and owneroccupied housing 100% 80% 60% 40% 20% 0% 1988 Top 10% 1990 Top household wealth shares: reported SCF wealth vs. capitalized SCF incomes Top 1% Top 0.1% 1992 1994 1996 Wealth 1998 Capitalized income 2000 The figure compares direct SCF wealth shares to wealth shares estimated by capitalizing SCF income. Wealth excludes pensions and owner-occupied net housing. Source: Appendix Table C1. 2002 2004 2006 2008 2010 2012
Capitalization works for foundations 80% 70% Top 1% Top foundations wealth shares: reported wealth vs. capitalized income Wealth % of foundation net wealth 60% 50% 40% 30% Top 0.1% Capitalized income 20% 1985 1987 1989 1991 1993 1995 The figure compares top foundation wealth shares obtained by using balance sheet wealth data as reported to the IRS and obained by capitalizing IRS-reported income. Source: Appendix Tables C11 and C13. 1997 1999 2001 2003 2005 2007 2009
II- The US Wealth Distribution, 1913-2012
Wealth in 2012 is very concentrated Table 1: Thresholds and average wealth in top wealth groups, 2012 Wealth group Number of families Wealth threshold Average wealth Wealth share A. Top Wealth Groups Full Population 160,700,000 $343,000 100% Top 10% 16,070,000 $660,000 $2,560,000 77.2% Top 1% 1,607,000 $3,960,000 $13,840,000 41.8% Top 0.1% 160,700 $20,600,000 $72,800,000 22.0% Top.01% 16,070 $111,000,000 $371,000,000 11.2% B. Intermediate Wealth Groups Bottom 90% 144,600,000 $84,000 22.8% Top 10-1% 14,463,000 $660,000 $1,310,000 35.4% Top 1-0.1% 1,446,300 $3,960,000 $7,290,000 19.8% Top 0.1-0.01% 144,600 $20,600,000 $39,700,000 10.8% Top.01% 16,070 $111,000,000 $371,000,000 11.2%
Wealth inequality is making a comeback Main long-run trends in the distribution of wealth: Long run U-shaped evolution for the very rich (top 0.1%: >$20 million today) Long run L-shaped evolution for the rich (top 1% to 0.1%: between $4 million and 20 million today) Long-run -shaped for the middle-class (top 50% to 90%: less than $650K today) (Memo: Bottom 50% always owns 0 net wealth)
Wealth has always been concentrated 90% Top 10% wealth share in the United States, 1917-2012 85% % of total household wealth 80% 75% 70% 65% Capitalized income SCF 60% 1917 1922 1927 1932 1937 1942 1947 1952 1957 1962 The figure depicts the share of total household wealth owned by the top 10%, obained by capitalizing income tax returns versus in the Survey of Consumer Finances. The unit of analysis is the familly. Source: Appendix Tables B1 and C4. 1967 1972 1977 1982 1987 1992 1997 2002 2007 2012
Top 1% has gained more than top 10% 55% Top 10-1% and 1% wealth shares, 1913-2012 50% % of total household wealth 45% 40% 35% 30% 25% 20% 1913 1918 1923 1928 1933 1938 1943 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 1998 2003 2008 2013 Top 10% to 1% Top 1%
Top 1% surge is due to the top 0.1% 30% Top 1-0.1% and top 0.1% wealth shares, 1913-2012 25% % of total household wealth 20% 15% 10% 5% 0% 1913 1918 1923 1928 1933 1938 1943 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 1998 2003 2008 2013 Top 1% to 0.1% Top 0.1%
Top 0.01% share: 4 in last 35 years 12% Composition of the top 0.01% wealth share, 1913-2012 10% % of total household wealth 8% 6% 4% 2% Equities 0% 1913 1918 1923 1928 1933 1938 1943 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 1998 2003 2008 Fixed income claims Other
The rise and fall of middle-class wealth Composition of the bottom 90% wealth share 40% 35% 30% % of total household wealth 25% 20% 15% 10% Business assets Equities & fixed claims (net of non-mortgage debt) 5% 0% 1917 1922 1927 1932 1937 1942 1947 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 2002 2007 2012 Pensions Housing (net of mortgages)
Wealth is getting older, but at the very top remains younger than in the 60s- 70s 60% Share of wealth held by elderly households (65+) 50% Top 0.1% % of each group's total wealth 40% 30% 20% Total population Bottom 90% 10% 0% 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
Share of income and labor income of top wealth holders has grown a lot 8% 7% Share of income earned by top 0.1% wealth-holders % of total pre-tax income 6% 5% 4% 3% 2% 1% National income Labor income 0% 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 This figure shows the share of total pre-tax national income and pre-tax labor income earned by top 0.1% wealth-holders. Labor income includes employee compensation and the labor component of business income. Source: Appendix Tables B25 and B28.
III- Robustness and comparison with existing estimates
Findings are robust to different methodological choices Robustness checks: Different treatment of capital gains Capitalizing dividends only (Bill Gates world) Capitalizing dividends plus capital gains (Warren Buffet world) Capitalizing dividends plus capital gains for shares but not ranking (the best of both worlds) Allowing for bond yield rising with wealth Different imputations for pension wealth All show wealth inequalities rising fast at the very top, but not below the top 0.1%
Results robust to alternative treatment of pensions, capital gains, bond returns Figure B27: Top 0.1% wealth share, all methods 25% Top 0.1% Baseline Top 0.1% KG capitalized Top 0.1% KG not capitalized 20% Top 0.1% pensions proportional to pension distributions Top 0.1% higher bond return for the rich 15% 10% 5% 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
Link with previous studies using alternative data Forbes 400 rich list: large increase in wealth concentration Surveys: SCF shows increase in top 10% but less in top 1% SCF excludes Forbes 400 and under-estimates capital income concentration increases since 1989 Estate tax multiplier: No increase in top 1% wealth share since 1980s (Kopczuk-Saez 2004, SOI studies) Estate tax multiplier method fails to take into account widening mortality differential by wealth class Our capitalization analysis can help SCF weights and estate multiplier weights
Our estimate for top 0.01% is consistent with Forbes rankings Top 0.00025% share (% of total household wealth) 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% 1982 Forbes 400 (top.00025%) and top.01% Wealth Shares 1984 Top 0.00025%, Forbes magazine (left-hand scale) 1986 1988 1990 1992 1994 1996 1998 Top 0.01%, capitalized income (right-hand scale) The figure depicts the top.00025% wealth share as estimated from the Forbes 400 list on the left axis. For comparison, the figure reports our top 0.01% wealth share obtained by capitalizing income tax returns (on the right axis). Source: Appendix Table C3. 2000 2002 2004 2006 2008 2010 2012 14% 12% 10% 8% 6% 4% 2% 0% Top.01% share (% of total household wealth)
Estate tax returns fail to capture rising top wealth shares 25% Top 0.1% wealth share: comparison of estimates 20% Capitalized income % of total household wealth 15% 10% 5% SCF adjusted SCF baseline Estate multiplier 0% 1917 1922 1927 1932 1937 1942 1947 1952 1957 1962 The figure depicts the top 0.1% wealth share obained by capitalizing income, by using the Survey of Consumer Finances (SCF baseline and adjusted), and by using estate tax data (Kopczuk and Saez, 2004). Source: Appendix C4 and C4b. 1967 1972 1977 1982 1987 1992 1997 2002 2007 2012
SCF does not fully capture rising top capital income share Top 0.1% Capital Income Share in the SCF and Tax Data 45% Tax data 40% 35% % taxable capital income 30% 25% 20% 15% 10% 5% SCF 0% 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 The figure compares the top 0.1% capital income shares estimated with the SCF data vs. the income tax data. Capital income includes realized capital gains, dividends, interest, net rents, and business profits. Source: Appendix Table C2.
Estate multiplier issue: mortality gradient by wealth within top 10% Relative Mortality by Age and Wealth Group, Men, 1999-2008 Mortality (relative to full population) 100% 80% 60% 40% 20% 0% top 10% top 5% top 1% Kopczuk-Saez 30-49 50-64 65-79 80+ Age groups The figure depicts the relative mortality rate by age and wealth group for men in 1999-2008. E.g., male top 1% wealth holders aged 30-49 mortality rate is 40% of males aged 30-49 population wide. Kopczuk-Saez is based on the mortality of white college goers relative to population in the 1980s. The graph shows that mortality decreases with wealth (even within the top 10%) and that the wealth mortality advantage decreases with age. Source: Appendix Table C7.
Estate multiplier issue: mortality gradient by wealth widens over time Mortality (relative to full population) 100% 80% 60% 40% Evolu&on of Mortality Advantage, Men, Aged 65-79 top 10% top 5% top 1% Kopczuk-Saez 1979-1984 1984-1988 1989-1993 1994-1998 1999-2003 2004-2008 The figure depicts the relative mortality rate for men aged 65-79 by wealth group and period. E.g., male top 1% wealth holders aged 65-79 mortality rate is 90% of males aged 65-79 population wide in 1979-1984. Kopczuk-Saez is based on the mortality of white college goers relative to population in the 1980s. The graph shows that the wealth mortality advantage increases overtime and more so for the top 1% wealthiest. Source: Appendix Figure C7.
IV- Decomposing Wealth Accumulation: Saving Rates and Income Shares of Top Wealth Holders
Top 1% vs. bottom 90% wealth growth Real average wealth of bottom 90% and top 1% families Top 1% real average welath 14,000,000 12,000,000 10,000,000 8,000,000 6,000,000 4,000,000 2,000,000 Top 1% (left y-axis) Bottom 90% (right y-axis) 140,000 120,000 100,000 80,000 60,000 40,000 20,000 0 0 1946 1950 1954 1958 1962 1966 1970 1974 1978 1982 1986 1990 1994 1998 2002 2006 2010 Bottom 90% real average wealth Real values are obtained by using the GDP deflator, 2010 dollars. Source: Appendix Tables B3.
Wealth distribution Dynamics Individual i wealth accumulation can always be written: W i t+1 = (1 + q i t) (W i t + s i t Y i t ) where Wt i is wealth, Yt i is income, st i is net savings rate, 1 + qt i is pure price effect on assets in year t We define synthetic savings rate s p t for fractile p (e.g., top 1%): W p t+1 = (1 + qp t ) (W p t + s p t Y p t ) where 1 + q p t is price effect for fractile p based on W p t composition long-run steady state: sh p W = shp Y sp s where sh p W is fractile p share of wealth, shp Y is fractile p share of income, and s p /s is relative savings rate of fractile p
Saving rates typically rise with wealth % of each group's total primary income 55% 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% -5% -10% -15% 1917-19 Saving rates by wealth class (decennial averages) 1920-29 1930-39 1940-49 1950-59 1960-69 Top 10 to 1% 1970-79 1980-89 1990-99 Top 1% Bottom 90% 2000-09 2010-12
The bottom 90% massively dis-saved in the decade preceding the crisis 15% Saving rate of the bottom 90% % of bottom 90% primary income 10% 5% 0% 1975 1980 1985 1990 1995 2000 2005 2010-5% -10%
Bottom 90% wealth share decline due to (a) savings collapse, (b) income share fall Share of income and wealth of bottom 90% wealth holders 70% 60% Income share 50% 40% Simulated 1985-2012 wealth share (constant 3% saving rate and constant income share) 30% 20% 10% Simulated 1985-2012 wealth share (constant 3% saving rate) Observed wealth share 0% 1917 1922 1927 1932 1937 1942 1947 1952 1957 1962 Since the 1980s the share of total household wealth owned by families in the bottom 90% of the wealth distribution has fallen proportionally more than the share of total pre-tax national income earned by these families. Source: Appendix Tables B1, B25 and B33c. 1967 1972 1977 1982 1987 1992 1997 2002 2007 2012
Table 2: Rates of growth, saving and return by wealth group Real growth rate of wealth per family Real growth rate of income per family Private saving rate (personal + retained earnings) Real rate of capital gains Total pre-tax rate of return g wf g yf s = S/Y q r + q 1917-1929 All 1.8% 0.5% 10% 0.9% 9.0% Bottom 90% -0.4% 0.0% 1% 0.2% 7.9% Top 10% 2.3% 1.2% 23% 1.0% 9.2% Top 1% 3.6% 1.4% 28% 1.5% 10.5% 1929-1986 All 1.5% 2.0% 12% -0.6% 6.6% Bottom 90% 3.0% 2.3% 6% -0.2% 6.2% Top 10% 1.0% 1.4% 24% -0.9% 6.8% Top 1% 0.3% 0.5% 24% -1.1% 7.2% 1986-2012 All 1.9% 1.3% 9% 0.9% 7.5% Bottom 90% 0.1% 0.7% 0% 1.3% 7.5% Top 10% 2.7% 2.3% 22% 0.7% 7.5% Top 1% 3.9% 3.4% 36% 0.9% 7.9%
Effects of Savings and Income Inequality Bottom 90%: Since mid-1980s, plummeting savings rate s p for bottom 90% relative to aggregate s [due to surge in debt] Decline in bottom 90% wealth share, and expected to continue Top 1%: Since mid-1970s, surge in income share held by top wealth holders and solid savings rate s p (relative to aggregate s) Short-run: Large increase in top wealth shares, and expected to continue Long-run: Self-made wealth could become inherited wealth and lead to the patrimony society of Piketty (2014)
Policies to Reduce Wealth Inequality Top 1%: Progressive taxation (income, wealth, inheritance) is a proven historical tool to reduce income/wealth concentration Progressive income and wealth tax reduce income and savings incentives at the top Estate taxation can prevent self-made wealth from becoming inherited wealth Bottom 90%: Collapse in savings due to surge in debt [due to present bias for consumption? stagnating incomes? financial de-regulation?] Middle class income support + financial regulation Need to encourage savings / discourage debt [= nudged savings + borrow against yourself?]
Conclusion
A first step toward DINA We are constructing new, consistent series on the distribution of wealth W and income Y = Y K + Y L fully consistent with flow of funds and national accounts Next step: construct a microfile with individual-level income (pre-tax and post-tax) and wealth consistent with macro flow of funds and national income accounts = distributional national accounts (DINA), reconciling macro growth and inequality studies
Need for better wealth and savings data Using additional data would enable us to refine our estimates: E.g., matched property and individual income tax data Modest additional administrative data collection effort could have high value: 401(k) taccounts balance reporting (and not only IRAs) Mortgage balances on forms 1098 Market value of portfolio securities on forms 1099 Purchases and sales of securities (to measure saving and consumption) Necessary to obtain fully accurate distributional national accounts
Supplementary Slides
Wealth categories definition back Equities: corporate equities, including S corporation equities, and money market fund shares (treated as dividend-paying for income tax purposes) Fixed claims: currency, deposits, bonds, and other interest-paying assets, net of non-mortgage debts Business assets: sole proprietorships, farms (land and equipment), partnerships, intellectual property products Housing: owner- and tenant-occupied housing, net of mortgage debt Pensions: funded pension entitlements, life insurance reserves, IRAs. Excludes social security and unfunded defined benefit pensions
Rates of returns on wealth around 7% No long-run price effects 14% Figure A8: Yield and total return on U.S. private wealth (decennial averages) 12% Pure yield 10% 8% 6% 4% 2% Total return = pure yield + asset price effect 0% 1913-19 1920-29 1930-39 1940-49 1950-59 1960-69 1970-79 1980-89 1990-99 2000-09 2010-13 back
What tax data miss 0.35 0.3 From reported to total capital income, 1920-2010 Retained earnings % of factor-price national income 0.25 0.2 0.15 0.1 Non-filers & unreported sole prop. profits Corporate income tax Imputed rents Income paid to pensions & insurance 0.05 Didivends, interest, rents & profits reported on tax returns 0 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 back
Most trusts generate income taxable at the individual level 12% Wealth held in estates & trusts 10% % net household wealth 8% 6% 4% 2% Total estate & trust wealth Estate & trust wealth that does not generate distributable income 0% 1952 1962 1972 1982 1992 2002 2012 back
Charitable giving follows top incomes Surge in top incomes is real Mean charitable giving of top 1% incomes / mean income back 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1962 1966 Charitable Giving of Top 1% Incomes, 1962-2012 Mean charitable giving of top 1% divided by mean income [lea y- axis] Top 1% Income Share [right y- axis] 1970 1974 1978 1982 1986 Source: The figure depicts average charitable giving of top 1% inomes (normalized by average income per family) on the left y-axis. For comparison, the figure reports the top 1% income share (on the right y-axis). 1990 1994 1998 2002 2006 2010 25% 20% 15% 10% 5% 0% Top 1% income share
Off-Shore Tax evasion, if anything, has probably increased since the 1970s 10% U.S. equities held by tax haven firms and individuals % of U.S. equity market capitalization 8% 6% 4% 2% 0% 1940 1950 1960 1970 1980 1990 2000 2010 In 2012, 9% of the U.S. listed equity market capitalization was held by tax haven investors (hedge funds in the Caymans, banks in Switzerland, individuals in Monaco, etc.). Source: Zucman (2014) using US Treasury International Capital data. back
Total returns of foundations grow with wealth but realized returns do not Figure C4: Return on foundation wealth, 1990-2010 average Returns including realized & unrealized gains 7% 6% 5% Realized return Unrealized capital gains 4% 3% 2% 1% 0% 1m-10m 10m-100m 100m-500m 500m-5bn 5bn+ back