Inequality Dynamics in France, 1900-2014: Evidence from Distributional National Accounts (DINA) Bertrand Garbinti 1, Jonathan Goupille-Lebret 2 and Thomas Piketty 2 1 Paris School of Economics, Crest, and Banque de France, 2 Paris School of Economics June, 21th 2016 1 / 54
Motivation 1 Previous work on long run income inequality trends in France (Piketty 2001, 2003, Landais 2007) focus on top income shares (top 10%, top 1%) and on fiscal income 2 Here we extend existing series up to 2013, and most importantly we combine fiscal data with national accounts and surveys in order to produce 1900-2013 series covering the entire income distribution (from bottom to top) and all forms of labor and capital incomes (taxable and tax exempt) 2 / 54
The DINA project Part of a broader multi-country project: World Wealth and Income database Distributional National Accounts (DINA) Provide long-term series on distribution of income and wealth Homogeneous across countries and over time Consistent with National Income and Wealth Accounts Covering all the distribution from bottom to top US: Piketty, Saez, Zucman (2016), Sweden: Lundberg and Waldenström (2016), UK: Alvaredo, Atkinson and Morelli (2016), Spain: Toledano (2016)... For France: two papers Last month: Wealth Today: Income 3 / 54
4 / 54
Literature Previous attemps to combine distributional tax data to national accounts: US: King (1927, 1930), Kutznets (1941, 1953), Piketty & Saez (2003) France: Piketty (2001,2003), Landais (2007), Landais, Piketty and Saez (2011) Some progress using tax data, but insufficient Tax data miss tax exempt income Silent on post-tax and transfer income Silent on distribution of the bottom 90 Other attempts to combine surveys with NA: OECD (Fesseau, Wolff and Mattonetti (2012), Fesseau and Mattonetti (2013) ); National level: France: Bellamy and ali (2009), US: Fixler et al. (2015),... Need for combining tax data, surveys and National Accounts in a more systematic way: Distributional National Accounts (DINA) 5 / 54
This paper We combine income tax data covering 1915-2013 period with national accounts, survey data and inheritance tax data 1800-2010 in order to produce: consistent unified income inequality series for France 1900-2013 (and 1800-2013 for wealth inequality series) detailed breakdown by age, gender, income and asset categories for 1970-2013 6 / 54
Main Findings 1 We confirm and update previous findings about long run inequality dynamics: huge fall 1914-1945, rise 1945-1968, decline 1968-1983, rise 1983-2013. Recent decades: moderate rise, except at the very top; reinforcing impact of missing capital income and changing family structure 2 Long run inequality fall: entirely due to fall in concentration of wealth and capital income. But rising inequality of saving rates and rates of return could lead to further increase in wealth concentration in coming decades. Steady-state simulations 3 Declining gender inequality... but not so much for high wages 4 France vs US: much bigger rise in inequality in the US; bottom 50% real income is now much smaller in the US than in France 7 / 54
Outline of the talk Data and methodology The long-run picture The role of capital income and wealth concentration Labor income: the limited decline of gender inequality France vs US and the bottom 50% Conclusion and perspectives 8 / 54
Outline Data and methodology The long-run picture The role of capital income and wealth concentration Labor income: the limited decline of gender inequality France vs US and the bottom 50% Conclusion and perspectives 9 / 54
Data and methodology (1/2) Income tax returns: 1915-2013 Income tax created in 1914, first applied in 1915 Finance Ministry estimate for 1900 and 1910 Annual exhaustive tabulations from 1915 to 2013 Microfiles from 1970 to 2012 ERFS surveys: 1970, 1975, 1979, 1984 (40,000 tax units) Annual from 1988 (400,000 tax units per year, all top incomes included) Exhaustive for recent years (2010-2012) National account series INSEE (French national statistical institute) annual series 1949-2015 Historical series since 1820 provided by Piketty-Zucman (2014) Household surveys: Wealth surveys ( Enquete Patrimoine ) 1986, 1992, 1998, 2004 and 2010 Housing surveys ( Enquete Logement ) 1970, 1973, 1978, 1984, 1988, 1992, 1996, 2001, 2006 and 2013 10 / 54
Data and methodology (2/2) We start from income tax micro-files 1970-2012 Using income tax tabulations 1900-2013, we apply generalized Pareto interpolation techniques (Fournier 2016; = non-parametric generalization of techniques used in historical income distribution literature: Kuznets 1953, Piketty 2001, etc.) in order to generate series by income percentiles (from bottom to top) : results are quasi-identical to micro-files series over period 1970-2012 details We impute tax-exempt labor and capital incomes (including imputed rent, retained earnings, corporate tax, etc.) using national accounts and wealth and housing surveys Preferred series: distribution of pretax national income (before all taxes and transfers, except pensions and unemployment insurance), among adults (equal-split series: income of couples divided by two) We also compare with fiscal-income series and tax-units series, and fully individualized series Next step: after-tax after-transfers series (to be done) 11 / 54
Outline Data and methodology The long-run picture The role of capital income and wealth concentration Labor income: the limited decline of gender inequality France vs US and the bottom 50% Conclusion and perspectives 12 / 54
The long-run picture (1/2) Long-run fall in inequality: top 10% income share dropped from about 50% in 1910 to about 35% in 2010, to the benefit of bottom 50% share (15% 20%) and middle 40% share (35% 45%) Uneven and chaotic process: huge inequality fall 1914-1945 (capital shocks), rise 1945-1968 (reconstruction of wage hierarchy and capital share), decline 1968-1983 (compression of wage inequality, steep rise of minimum wage, declining capital shares), rise 1983-2007 (reverse evolution as 1968-1983) Rising top income inequality 1983-2013: moderate impact on top 10% income share, but very large impact on top 1% and top 0.1%. Exacerbates the perception of growth slowdown for the rest of the population (The 30 Glorious Years are not over for everyone). Drop in very top shares since 2008, but in 2013 they are still much higher than in 1980s. 13 / 54
The long run picture 40 000 35 000 30 000 25 000 20 000 15 000 The uneven rise of per adult national income in France, 1900-2014 ( 2014) Real growth rate 1900-2014 : +1.9%/year 1900-1945 : -0.1% 1945-1980 : +3.7% 1980-2014 : +0.9% Average national income per adult (2014) : 34 580 10 000 5 000 0 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 National income divided by adult population. National income = GDP - capital depreciation + net foreign income. 14 / 54
TABLE 1 Income thresholds and income shares in France, 2013 Income group Number of adults Income threshold Average income Income share Full Population 51 318 000 0 C 34 440 C 100,0% Bottom 50% 25 659 000 0 C 15 510 C 22,5% Middle 40% 20 527 200 27 420 C 38 920 C 45,2% Top 10% 5 131 800 60 970 C 111 230 C 32,3% incl. Top 1% 513 180 162 400 C 359 290 C 10,4% incl. Top 0.1% 51 318 536 410 C 1 308 290 C 3,8% incl. Top 0.01% 5 132 2 064 350 C 5 181 850 C 1,5% incl. Top 0.001% 513 9 562 310 C 18 990 120 C 0,6% This table reports statistics on the distribution of national income in France in 2013. The unit is the adult individual (20-year-old and over; income of married couples is splitted into two). Fractiles are defined relative to the total number of adult individuals in the population. 15 / 54
The long run picture 55% Top 10% income share, France 1900-2013: long-run fall in inequality 50% 45% Average national income per adult (2013) : 34 440 40% 111 230 35% 30% 25% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two). 16 / 54
The long run picture 55% 50% Income shares in France 1900-2013: rise of lower and middle classes Average national income per adult (2013) : 34 440 45% 40% 38 920 35% 30% 25% 20% 15% Top 10% Middle 40% Bottom 50% 111 230 15 510 10% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two). 17 / 54
The long run picture 24% Top 1% income share in France:long-run fall but upward trend since 1980s 22% 20% 18% Average national income per adult (2013) : 34 440 16% 14% 359 290 12% 10% 8% 6% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two). 18 / 54
The long run picture 220 210 200 190 180 170 160 150 140 130 120 110 100 Income shares 1983 = 100 Rising top income inequality in France, 1983-2013 Top 0.1% Top 1% Top 10% Bottom 90% 90 1983 1988 1993 1998 2003 2008 2013 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl.insurance) among adults. Equal-split-adults series (income of married couples divided by two). 19 / 54
The long run picture 5,4% 5,0% Top 0.1% income share in France 1970-2013 50x average 4,6% 4,2% 3,8% 3,4% 3,0% 23x average 1 308 290 38x average 2,6% 2,2% 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl.insurance) among adults. Equal-split-adults series (income of married couples divided by two). 20 / 54
The long run picture 2,2% 2,0% Top 0.01% income share in France 1970-2013 213x average 1,8% 1,6% 1,4% 1,2% 5 181 850 150x average 1,0% 0,8% 80x average 0,6% 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl.insurance) among adults. Equal-split-adults series (income of married couples divided by two). 21 / 54
The long run picture 250 240 230 220 210 200 190 180 170 160 150 140 130 120 110 Income shares 1983 = 100 Top labor incomes vs top capital incomes in France, 1983-2013 Top 0,01% (capital income) Top 0,01% (labor income) Top 0.01% (total income) 100 1983 1988 1993 1998 2003 2008 2013 Distributions of total income, capital income and labor income among adults. Equal-split-adults series (income of married couples divided by two). 22 / 54
The long-run picture (2/2) New inequality series using national income show bigger rise in inequality than previous fiscal income series, because of rise of tax-exempt capital income (life insurance, retained earnings, capital gains, rent, etc.) Rising inequality would be even higher at the tax-unit level (rise of singles) than in our benchmark equal-split adult-level series (income of couples divided by two) 23 / 54
The long run picture 55% 50% 45% 40% 35% Income shares in France: national income vs fiscal income series Top 10% (national income) Top 10% (fiscal income) Top 1% (national income) Top 1% (fiscal income) 30% 25% 20% Rise of tax-exempt capital income 15% 10% 5% 0% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of pretax national income (incl. tax-exempt labor and capital income) vs pretax fiscal income (reported on income tax returns). Equal-split-adults series (income of married couples divided by two). 24 / 54
The long run picture 55% 50% 45% 40% Income shares: equal-split-adults vs tax-units series Top 10% (equal-split-adults) Top 10% (tax-units) Top 1% (equal-split-adults) Top 1% (tax-units) 35% 30% 25% 20% Rise of singles 15% 10% 5% 0% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two) vs tax-units series (singles and married couples). 25 / 54
Outline Data and methodology The long-run picture The role of capital income and wealth concentration Labor income: the limited decline of gender inequality France vs US and the bottom 50% Conclusion and perspectives 26 / 54
The role of capital income and wealth concentration 1 Long-run fall in income inequality is entirely due to the fall of top capital incomes: inequality of labor income did not change in the long run 2 Capital income has always been the main income source for very top incomes (including today); but because of the decline of capital concentration, the level of top capital incomes has dropped 3 Wealth inequality is still much higher than income inequality today: top 10% share around 60-70% for wealth and capital income, vs 35% for total income and 25% for labor income 27 / 54
55% Top 10% income share: total income vs labor income inequality 50% 45% Top 10% share (total income) Top 10% share (labor income) 40% 35% 30% 25% 20% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two). 28 / 54
24% Top 1% income share: the fall of top capital incomes 22% 20% 18% Top 1% share (total income) Top 1% share (labor income) 16% 14% 12% 10% 8% 6% 4% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two). 29 / 54
100% Income composition by income level, France 2012 90% 80% Capital income (Interest,dividends,rents,..) 70% 60% 50% 40% Labor income (Wages and pensions) Mixed income (Self-employed) 30% 20% 10% 9 800 26 000 34 500 50 900 102 600 193 600 1 134 000 0% P0-30 P30-40 P40-50 P50-60 P60-70 P70-80 P80-90 P90-95 P95-99 P99-99.5 P99.5- P99.9- Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. 99.9 100 Equal-split-adults series (income of married couples divided by two). 30 / 54
100% 95% 90% 85% 80% Top 10% share: total income vs labour vs capital income vs wealth Top 10% share (capital income) Top 10% share (wealth) Top 10% share (total income) Top 10% share (labor income) 75% 70% 65% 60% 55% 50% 45% 40% 35% 30% 25% 20% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of total income, labor income, capital income and net wealth among adults. Equal-split-adults series (income and wealth of married couples divided by two). 31 / 54
65% 60% 55% 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% Top 1% share: total income vs labour vs capital income vs wealth Top 1% share (capital income) Top 1% share (wealth) Top 1% share (total income) Top 1% share (labor income) 0% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of total income, labor income, capital income and net wealth among adults. Equal-split-adults series (income and wealth of married couples divided by two). 32 / 54
Why is wealth inequality so large? Cumulative effects of unequal labor incomes, saving rates and rates of return Large multiplicative effects, especially with long horizon and inheritance Equation of wealth accumulation at time t + 1 for the wealth group p (for instance p = top 10% wealth group): W p t+1 = (1 + qp t )[W p t + s p t (Y p Lt + r p t W p t )] W p is the aggregate wealth for the wealth group p, Y p L labor income q p is the real rate of capital gain s p is the saving rate, r p is the after-tax rate of return (for group p) We infer group-level synthetic saving rates s p t from the observation of W p+1 t, W p t, Y p Lt, r p t, qp t 33 / 54
35% Synthetic saving rates by wealth group, France 1970-2012 30% 25% 20% 15% Top 10% Middle 40% Bottom 50 10% 5% 0% -5% 1970 1975 1980 1985 1990 1995 2000 2005 2010 34 / 54
Steady-state formulas for top wealth shares From the equation of wealth accumulation, with the same notations as above: W t+1 = (1 + q t )[W t + s t (Y Lt + r t W t )] and assuming q t has to be equal to 0 at steady state, we directly derive: sh p W = (1 + sp r p sr g s p r p )sp s shp Y L sh p W (resp. shp Y L ) is the share of wealth (resp. labor income) held by wealth group p (for instance p = top 10% wealth group) g is the growth rate, s the aggregate saving rate and r the aggregate after-tax rate of return If s p = s and r p = r (i.e. top wealth group has the same saving rate and rate of return as average), then sh p W =shp Y L : wealth inequality = labor income inequality but if s p > s and r p > r, then this can generate large multiplicative effects, and lead to very high steady-state wealth concentration 35 / 54
90% 80% 70% Simulating the top 10% wealth share, 1800-2100 Observed series 1800-2012 Steady-state trajectory with 1984-2012 saving rates: 27.5% for top 10%, 2% for bottom 90% 60% 50% 40% Steady-state trajectory with 1970-1984 saving rates: 24.5% for top 10%, 8.5% for bottom 90% 30% 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000 2020 2040 2060 2080 2100 36 / 54
Outline Data and methodology The long-run picture The role of capital income and wealth concentration Labor income: the limited decline of gender inequality France vs US and the bottom 50% Conclusion and perspectives 37 / 54
Labor income: the limited decline of gender inequality For subperiod 1970-2012, we have detailed breakdown by age and gender Age patterns did not change very much: age-labor income profile is always steeply rising (although less strongly than age-capital income and age-wealth profiles), and income inequality very high within each age group (like wealth) Main change over the period: large rise of female labor market participation, decline of gender inequality, but still very high at the top 38 / 54
180% Age-Income profile in France by age, 1970-2012 Average income by age (% average income 20+) 160% 140% 120% 100% 80% 60% 40% 20% 1970 1984 2000 2012 0% 20 25 30 35 40 45 50 55 60 65 70 75 80 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two). 39 / 54
180% Age-Labor income profile in France by age, 1970-2012 Average labor income by age (% average labor income 20+) 160% 140% 120% 100% 80% 60% 40% 20% 1970 1984 2000 2012 0% 20 25 30 35 40 45 50 55 60 65 70 75 80 Distribution of pretax labor income (incl. pensions, unempl. insurance and 70% of mixed income) among adults. Equal-split-adults series (income of married couples divided by two). 40 / 54
Average capital income by age (% average capital income 20+) 180% 160% 140% 120% 100% 80% 60% 40% 20% Age-Capital income profile in France by age, 1970-2012 0% 20 25 30 35 40 45 50 55 60 65 70 75 80 Distribution of pretax capital income (incl. 30% of mixed income) among adults. Equal-split-adults series (income of married couples divided by two). 1970 1984 2000 2012 41 / 54
65% 55% Income concentration by age group, France 1970-2012 Top 10% (all ages) Middle 40% (all ages) Bottom 50% (all ages) Top 10% (20-39-yr) Middle 40% (20-39-yr) Bottom 50% (20-39-yr) Top 10% (40-59-yr) Middle 40% (40-59-yr) Bottom 50% (40-59-yr) Top 10% (60-yr+) Middle 40% (60-yr+) Bottom 50% (60-yr+) 45% 35% 25% 15% 5% 1970 1975 1980 1985 1990 1995 2000 2005 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two). 42 / 54
1,7 Gender gap by age, France 2012 1,6 1,5 1,4 1,3 1,2 25 30 35 40 45 50 55 60 65 Ratio between average labor income of men and women by age (incl. non participants). Labor income includes wages, pensions, unemploy. insurance and 70% of mixed income. 43 / 54
4,0 Gender gap by age, France 1970-2012 3,5 3,0 1970 1984 2000 2012 2,5 2,0 1,5 1,0 25 30 35 40 45 50 55 60 65 Ratio between average labor income of men and women by age (incl. non participants). Labor income includes wages, pensions, unemploy. insurance and 70% of mixed income. 44 / 54
100% Labor market participation by gender, France 1970-2012 90% 80% men 70% Women 60% 50% 40% 1970 1975 1980 1985 1990 1995 2000 2005 2010 Fraction of men and women 25-to-65-year-old with positive labor income. Labor income includes wages, pensions, unemploy. insurance and 70% of mixed income. 45 / 54
50% 45% 40% 35% 30% Share of women in fractiles of top labor incomes in France, 1970-2012 Top 50% Top 10% 25% 20% 15% 10% 5% Top 1% Top 0.1% Share of women in top 1%: 10% in 1994, 16% in 2012, 50% by 2102? Top 0.1%: 50% by 2144? 0% 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 46 / 54
45% Income shares in France: equal-split income vs individual income 40% 35% 30% 25% Top 10% (equal-split income) Top 10% (individual income) Bottom 50% (equal-split income) Bottom 50% (individual income) 20% 15% Rise of female participation 10% 1970 1980 1990 2000 2010 Distribution of pretax national income: equal-split income series (income of married couples divided by two) vs individual income series (capital income of married income divided by two, but labor income allocated to each individual). 47 / 54
Outline Data and methodology The long-run picture The role of capital income and wealth concentration Labor income: the limited decline of gender inequality France vs US and the bottom 50% Conclusion and perspectives 48 / 54
France vs US and the bottom 50% 1 Top income shares increased much more in the US than in France since the 1980s 2 Complex combination of factors: education system (more unequal in the US?), labor market rules (fall in US minimum wage), changing governance and incentives for top executive pay-setting (huge fall in US top income tax rates). Not analyzed here (see Piketty 2014) 3 Distribution matters: per adult national income is 25% smaller in France (more hours of work in the US, similar productivity), but bottom 50% average income is 30% higher in France 4 This would probably be reinforced if we look at after-tax after-transfer inequality (to be done). But it is interesting to see that this is already the case for pre-tax pre-transfers inequality. More generally, long-term changes in inequality reflect large changes in both pretax inequality (itself influenced by policies and institutions) and after-tax inequality 49 / 54
55% Top 10 % income shares: France vs US, 1910-2013 243 650 (PPP) 50% USA 45% France 40% 35% 30% 111 230 25% 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two). 50 / 54
25% Top 1 % income shares: France vs US, 1910-2013 1 010 000 (PPP) 20% USA France 15% 10% 359 290 5% 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two). 51 / 54
55% 50% 45% 40% 35% 30% 25% Top 10 % and bottom 50% income shares: France vs US, 1910-2013 Top 10% USA 243 650 (PPP) Top 10% France Bottom 50% France Bottom 50% USA 111 230 15 510 20% 15% 12 140 10% 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Distribution of pretax national income (before all taxes and transfers, except pensions and unempl. insurance) among adults. Equal-split-adults series (income of married couples divided by two). 52 / 54
Outline Data and methodology The long-run picture The role of capital income and wealth concentration Labor income: the limited decline of gender inequality France vs US and the bottom 50% Conclusion and perspectives 53 / 54
Conclusion and perspectives 1 Main contribution: by combining fiscal data, national accounts and survey data, we have constructed "Distributional national accounts" (DINA) for France, i.e. unified series for the distribution of total income, labor income, capital income and wealth over the 1900-2013 period 2 We observe large changes in inequality, both over time and across countries, largely due to different institutions and public policies. Inequality is political, not natural 3 This work is due to be extended to lots of countries: World wealth and income database 54 / 54
Appendix Appendix 55 / 54
Appendix 55% Figure C3. Income shares: tabulations vs micro-files (equal-split individuals, fiscal income excl. cap.gains) 50% 45% 40% 35% 30% 25% 20% 15% 10% Top 10% (micro-files) Top 10% (tabulations) Middle 40% (micro-files) Middle 40% (tabulations) 5% Bottom 50% (micro-files) Bottom 50% (tabulations) 0% 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 back 56 / 54