sign in sign up
<<
Home , Gabriel Zucman, Thomas Piketty

AEA Papers and Proceedings 2019, 109: 289–295 https://doi.org/10.1257/pandp.20191035

DISTRIBUTIONAL DIVERSITY IN THE ‡

Simplified Distributional National Accounts†

By , , and *

Piketty, Saez, and Zucman 2018 —hence- imputations are made at the individual level forth, PSZ—propose a method to( distribute) total based on a number of assumptions and combin- national income across individual adults in the ing information from income data and aux- . The method has recently been iliary datasets such as survey data and national applied to a number of countries as reviewed accounts data. Naturally, there are many in the 2018 Alvaredo assumptions involved and each assumption can et al. 2018 . The key advantage relative( to ear- be questioned. Because the number of assump- lier work using) fiscal income such as Piketty and tions made is very large, the methodology lacks Saez 2003 or survey data is that the national simplicity and hence the end results are not as income( concept) is comprehensive, homoge- transparent as the simpler earlier fiscal income neous over time, and comparable across coun- series by Piketty and Saez 2003 . tries. In particular, distributional national income As shown in Figure 1, (the Piketty) and Saez statistics can be used to study both growth and 2003 fiscal income series showed a huge inequality in a consistent framework that aggre- increase( ) in the top 1 percent fiscal income share gates cleanly to national income from national in recent decades. The top 1 percent fiscal income accounts. In contrast, fiscal income or survey share grew by about 10 points from 8.4 percent income aggregates display growth levels that in 1960 to 17.8 percent in 2016. The new PSZ are quite different from national income growth series based on national income also show a both in the short-term year-to-year fluctuations large increase in the share of national income and in the long-term growth rates averaged over going to the top 1 percent from about 10 percent decades see PSZ for a detailed discussion . in 1980 to about 20 percent today. This doubling The PSZ( methodology starts from individual) of the top 1 percent national income share in the tax return data providing information on fiscal PSZ series takes place both when the top 1 per- income at the micro-level and then imputes cent is defined based on individual adults with forms of income that are in national income equal splitting of income within married cou- but not in fiscal income such as fringe bene- ples the benchmark PSZ series and when the fits for employees, imputed rent of homeown- top 1( percent is defined based on) tax units as in ers, retained profits of corporations, etc. These the Piketty and Saez 2003 fiscal income series( . Recently, Auten and Splinter 2018 have pro)- ( ) ‡Discussants: Katharine G. Abraham, University of posed an alternative set of assumptions for Maryland; Claudia Sahm, Federal Reserve Board. distributing non-fiscal income and have found * Piketty: School of , 48 Boulevard a much more modest increase in top 1 percent Jourdan, 75014 Paris, email: [email protected] ; income shares. Auten and Splinter use the indi- Saez: University of California,( 530 Evans Hall, Berkeley,) vidual adult unit. All these series are depicted CA 94720 email: [email protected] ; Zucman: on Figure 1. University of( California, 530 Evans Hall, Berkeley,) CA 94720 email: [email protected] . We acknowledge To cast light on these discrepancies and help funding( from the Berkeley Center for) Equitable Growth, understand better the overall plausibility of the the Sandler Foundation, and the Institute for New Economic large set of assumptions in PSZ and Auten and Thinking. We thank Katherine Abraham and David Johnson Splinter 2018 , this paper develops a simplified for helpful comments and discussions. methodology( that) starts from the fiscal income † Go to https://doi.org/10.1257/pandp.20191035 to visit the article page for additional materials and author disclo- top income share series and makes very basic sure statement s . assumptions on how each income component ( ) 289 290 AEA PAPERS AND PROCEEDINGS MAY 2019

25% Second, this simplified methodology can also be used to assess the plausibility of the 20% PSZ assumptions. In particular, we will show that the simplified methodology can be used to 15% show that the alternative assumptions proposed by Auten and Splinter 2018 imply a drastic ( ) 10% equalization of income components not in fiscal income which does not seem realistic. Piketty-Saez fiscal unit ( ) 5% PSZ pretax national income tax unit ( ) PSZ pretax national income adult unit I. Simplified Distributional National Accounts ( ) Auten and Splinter 2018 adult unit 0% ( ) ( ) In what follows, we focus solely on pretax 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 national income defined as market income after the operation of public and private pension sys- Figure 1. Top 1 Percent Income Shares, 1960 –2016 tems i.e., of pension contributions either ( Notes: This figure displays i the top 1 percent fiscal income public through social security payroll or share from Piketty and Saez( )2003 using tax units and pretax private through defined benefit and defined con- fiscal income excluding capital( gains;) ii the top 1 percent ( ) tributions pension plans and including all pen- income share from PSZ using tax units and pretax national sion benefits public and private . Pretax national income; iii the top 1 percent income share from PSZ using ) the individual( ) adult unit with equal split of income within income is before all taxes except payroll taxes ( funding social security benefits( and before any married couples and pretax national income; iv the top 1 ) percent pretax income) share from Auten and Splinter( ) 2018 government transfers except public pensions . ( ) using the individual adult unit and pretax national income. As discussed extensively( in Alvaredo et al.) This paper shows how to reproduce approximately the PSZ 2016 , considering income after the operation and Auten and Splinter series starting from the Piketty and ( ) Saez fiscal income series and making simple assumptions on of pension systems allows to control for the how nontaxable income is distributed. effects of aging as retirees typically have very Sources: Piketty and Saez 2003 ; PSZ; Auten and Splinter little factor income( and whether a country orga- 2018 . ( ) ) ( ) nizes pensions privately or publicly. PSZ pretax income benchmark series also use this pretax national income definition. We consider tax units as opposed to individ- from national income that is not included in fis- ual adults as in the main PSZ( series because fis- cal income is distributed. This simplified meth- cal income series by Piketty and Saez) 2003 are odology has two main goals. based on tax units following the tax definition( ) . First and most important, it can be used to Moving from tax units( to individual units with) create distributional national income statistics income equally split within married couples the in countries where fiscal income inequality sta- benchmark series of PSZ is fairly easy to( do tistics are available but where there is limited but would require recomputing) fiscal income information to impute other income at the indi- series. As shown in Figure 1, using tax units ver- vidual level. Alvaredo et al. 2016, Section VII sus the individual adult with equal split within proposed a simplified methodology( for countries) married couples has only( a very minor effect with less data. The methodology proposed here on series as displayed) in Figure 1.2 can be seen as an even simpler method that can be applied to countries for which fiscal income top income share statistics exist1 and for which 2 national accounts and fiscal income aggregates Auten and Splinter 2018 estimates—as well as the Congressional Budget Office( ) estimates CBO 2018 —do are sufficiently detailed. not use consistent definitions when ranking( units to )define the top 1 percent and when defining income to compute top income shares. Incomes for ranking are normalized by size including children, but incomes for com- 1 Such series exist for a large number of countries and puting top shares are not. This inconsistency in definitions are available online in the World Inequality Database wid. mechanically biases downward the top 1 percent income world. See Atkinson, Piketty, and Saez 2011 for a review share as the incomes of the top 1 percent are actually not the of this literature. ( ) highest( top 1 percent incomes . This does not seem ­sensible ) VOL. 109 SIMPLIFIED DISTRIBUTIONAL NATIONAL ACCOUNTS 291

To simplify the exposition, here as in the Panel A. From taxable to total pretax national income rest of the computations presented below, 100% we exclude taxes on products and produc- 90% Tax-exempt capital income 80% tion primarily sales and taxes from Tax-exempt labor income ( ) 70% national income. This implies that we consider 60% factor-price national income instead of full 50% national income . This has no( consequence 40% Taxable income ) 30% on the distributional analysis as distributional 20% national income methodology distributes taxes national income 10% 0% on products and production proportionally to Percent of factor-price factor-price national income on a pretax basis. 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Conceptually, factor-price national income can Panel B. Separating taxable labor and taxable be seen as the income that would allow to buy all capital income 100% the production carried out by the factors labor 90% Other tax-exempt capital income ( Tax-exempt capital income in pension funds and capital owned by residents provided that 80% ) Taxable capital income this production can be bought at prices that do 70% Tax-exempt labor income 60% not include taxes on products and production. 50% The exclusion of taxes on products and produc- 40% Taxable labor income tion has only a very modest effect on the evolu- 30% 20% tion of the top 1 percent income share. national income 10%

The simplified methodology starts with the fis- Percent of factor-price 0% cal income top income share series developed by 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 20002004 2008 2012 2016 Piketty and Saez 2003 . As shown on Figure 1, ( ) Figure 2. From Taxable to Total Pretax National the top 1 percent fiscal income share excluding Income, 1960–2016 capital gains has increased from 8.4 (percent in ) 1960 to 17.8 percent in 2016. Everybody agrees Notes: Panel A decomposes factor-price national income defined as national income excluding taxes on produc- that the concentration of reported fiscal income ( has increased a lot since 1960. This is uncon- tion and production into taxable income reported on tax returns , tax-exempt) labor income not reported( on tax troversial because this fiscal income is directly returns), and tax-exempt capital income( not reported on tax observable in tax data. returns). Panel B further splits taxable income( into taxable Reported fiscal income excluding capital labor income) and taxable capital income and tax-exempt gains adds up to 64 percent( of factor-price capital income into tax-exempt capital income in pension national) income in 2016 down from 70 per- funds and other tax-exempt capital income. In both panels, realized capital gains are excluded from taxable income. cent in 1960. As shown in Figure 2 panel A, Source: PSZ, series updated to 2016. the majority of the pretax income not visible on individual tax returns is capital income corpo- rate retained earnings, corporate income( taxes, PSZ offers a sophisticated treatment of tax-exempt , imputed rents, property untaxed income that involves a detailed recon- taxes, income earned by pension ciliation with national accounts totals compo- funds, income paid to trusts, fiduciaries, etc. . nent by component. However it is possible to As shown in Figure 2 panel B, untaxed capital) reproduce the PSZ results quickly and to under- income accounts for the vast majority of total stand what their methodology amounts to doing capital income in the economy. As pension in a simple way. The 36 percent of pretax fac- funds grow over time and are more equally dis- tor-price national income not reported in( tax tributed than other forms of wealth, it is useful data in 2016) can be decomposed as follows: to split untaxed capital income into the untaxed capital income earned by pension funds and • 12.7 percent is untaxed labor and pension other untaxed capital income. income employer contributions to health insurance,( Social Security benefits, untaxed private pension benefits such as Roth IRAs, and after-tax Defined Contribution plans, to us. In our view, the top 1 percent should be the top 1 per- under-reported labor income most of which cent highest income earners. is from noncorporate business profits ; ) 292 AEA PAPERS AND PROCEEDINGS MAY 2019

• 10.6 percent is untaxed capital income 25% earned on pension plans including the frac- tion of the corporate income( tax, business 20% , and retained earnings attribut- able to pension plans ; and 15% • 13.0 percent is untaxed) capital income other 10% than earned on pension plans. Simplified PSZ tax units 5% ( ) PSZ tax units We use the following two assumptions in our ( ) simplified methodology: 0%

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 i Untaxed labor and pension income and ( ) untaxed capital income earned on pen- Figure 3. Top 1 Percent Pretax National Income Share: sion plans is distributed like taxable PSZ versus Simplified Computations labor and pension income. Notes: This figure displays the top 1 percent pretax national income share from PSZ in solid line and using our proposed ii Other untaxed capital income is distrib- ( ) simplified computation in dashed line. The simplified com- uted like taxable capital income. putation uses realistic assumptions and reproduces closely the PSZ series both in levels and trends. With these two assumptions and using the com- Sources: PSZ and authors’ computations. position of top fiscal incomes broken down in a labor and pension income,( and b capital income( ) provided by Piketty and Saez( ) 2003 compositional) series, we can compute (pretax) the PSZ methodology delivers results that are national income top income shares as follows about the same as the results one would obtain see the supplemental materials for complete by decomposing national income into taxable computations( . income and three categories of untaxed income, The share ) of untaxed labor and pension and making simple assumptions about how these income and untaxed capital income earned on three categories are distributed. pension plans accruing to top 1 percent earners The PSZ series and the simplified series track is assumed to be the same as the share of labor each other closely with an almost perfect match and pension income in fiscal income accruing to in 1960 and 2016. The main difference is in the top 1 percent earners. It grows from 6 percent late 1970s and early 1980s when very little tax- in 1960 when labor income had low concentra- able capital income was reported due to large tion to 15( percent in 2016 when labor income business losses due in large part (to the devel- is much) more concentrated (. The share of other opment of tax shelters . Such business losses untaxed capital income in national) income accru- were corrected for in PSZ) by ignoring business ing to top 1 percent earners is assumed to be the losses but not with this simplified( method. As a same as the share of taxable capital income in result,) the simplified series undershoot slightly fiscal income accruing to top 1 percent earners. the PSZ series in that period. Such taxable capital income has always been More generally, the simplified assumptions highly concentrated with the share accruing to are clearly too coarse. In particular, the aggre- the top 1 percent growing from 40 percent in gate of untaxed labor and pension income is a 1960 to 48 percent in 2016. These two calcula- mixed bag of heterogeneous income categories tions use the two assumptions stated above and that are distributed very differently e.g., Social assume that any effects due to re-ranking when Security benefits are equally distributed,( while moving from fiscal income to national income( under reported labor income, most of which is in are negligible. ) businesses, is unequally distributed. Similarly, Figure 3 displays the resulting simplified top untaxed capital income includes elements) that 1 percent pretax national share. It shows that this are very concentrated such as retained earnings simplified top income share tracks very closely of corporations not owned( by pension plans and the corresponding sophisticated PSZ income elements that are much less concentrated )such share in both levels and trends. Put another way, as imputed rent of homeowners . Therefore,( ) VOL. 109 SIMPLIFIED DISTRIBUTIONAL NATIONAL ACCOUNTS 293 the sophisticated PSZ approach is required to results? One can recover the Auten and Splinter deliver more accurate results. But at least our 2018 top 1 percent pretax income shares by two basic assumptions are a reasonable way changing( ) our two assumptions as follows. to distribute the aggregate amount of untaxed Under our assumption 1, nontaxable labor income, and for all its complexity the PSZ meth- and pension income and capital income earned odology amounts to making roughly these sim- on pension plans is distributed like taxable ple assumptions. labor income. The share going to the top 1 per- cent grows from 6 percent in 1960 when labor ( II. Comparison with Auten and Splinter (2018) income had low concentration to 15 percent in 2016 when labor income is much) more concen- Auten and Splinter 2018 also propose to trated(. Instead, to replicate Auten and Splinter, distribute national income( by) income groups we assume) that the concentration of nontaxable but make different assumptions than PSZ along labor and pension income and capital income many dimensions. They start from the fis- earned on pension plans remains frozen at its cal income series of Piketty and Saez 2003 , 1960 level, i.e., the top 1 percent get only 6 per- make a number of definitional changes,( and) cent of such income throughout the full period add various income components not included 1960–2016. This is a very low level that essen- in fiscal income. In the end, they find a top 1 tially states that the rich have been largely left percent income share of 14.2 percent in 2015. out of the explosion of pension funds, fringe Using the national income at factor prices total benefits, and the surge of business income that of $14.98Tr in 2015, this means that their top is under-reported on tax returns.3 1 percent earns $2.13Tr in 2015. In the Piketty Under our assumption 2, the share of other and Saez 2003 fiscal series including realized untaxed capital income in national income capital gains,( the) top 1 percent income share in accruing to top 1 percent earners is assumed 2015 is 21.6 percent of a $10.26Tr fiscal income to be the same as the share of taxable capital total, which means that the top 1 percent earns income in fiscal income accruing to top 1 percent $2.22Tr in fiscal income in 2015. Therefore, earners. It is very concentrated with the share Piketty and Saez 2003 find more income going accruing to the top 1 percent growing from 40 to the top 1 percent( by) simply looking at their percent in 1960 to 48 percent in 2016. Instead, reported fiscal income including realized cap- to replicate Auten and Splinter, we assume that ital gains than what Auten( and Splinter 2018 the share of other untaxed capital income earned obtain after) adding various income components( ) by the top 1 percent declines linearly from 30 that enlarge the denominator income base by 46 percent in 1960 to 10 percent in 2016. Hence, percent from $10.26Tr to $14.98Tr. we assume that this share declines dramatically As shown on Figure 1, Auten and Splinter so that nontaxable capital income outside of also find that the top 1 percent share of pretax pension funds is now more equally (distributed national income has barely increased since 1960 than labor income.) This assumption is therefore 2.8 points instead of 9.4 for fiscal income . extreme and amounts to assuming that capital For(+ this to be true, it must+ be the case that the) income and wealth are now extremely equally 36 percent of national income which are not distributed in the United States. reported on individual tax returns have become Figure 4 shows that our simplified meth- enormously less concentrated over time. This odology combined with these two alternative equalization process must have been so powerful as to offset the upsurge in the concentration of the much larger flow of reported fiscal income. 3 The classical reference on the distribution of under-re- Piketty,( Saez, and) Zucman 2018 find that the ported income Johns and Slemrod 2010 finds that adding ( ) under-reported income does not affect( the) distribution of fis- 36 percent of national income not in tax returns cal income their Table 5 . Johns and Slemrod 2010 also has become slightly less concentrated over time. find that the( fraction of Schedule) C business income( evaded) Is it conceivable that it has in fact become dra- by the top 0.5 percent is 55 percent and almost identical to matically less concentrated? the full population average of 57 percent their Table 4 . This is consistent with PSZ methodology but (in sharp contrast) to What assumptions on the evolution of the dis- Auten and Splinter 2018 who attribute a disproportionately tribution of the untaxed income categories are large and growing fraction( ) of under-reported income to the needed to recover the Auten and Splinter 2018 bottom 90 percent. ( ) 294 AEA PAPERS AND PROCEEDINGS MAY 2019

25% 40% SCF including Forbes 400 20% ( ) 35%

15% Capitalized incomes Saez-Zucman 30% ( ) 10% 25%

5% Simplified Auten and Splinter 20% Auten and Splinter 0% 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Figure 5. Top 1 Percent Wealth Share in the United States: Capitalized Incomes and SCF Figure 4. How to Recover Auten and Splinter Top 1 Percent Income Share Series Using Simplified Notes: This figure displays the top 1 percent wealth share Computations obtained by capitalizing incomes and obtained from the Survey of Consumer SCF . The wealth of the Notes: This figure displays the top 1 percent pretax national Forbes 400 richest Americans which( by) design are excluded income share from Auten and Splinter 2018 in solid line from the SCF is added to the (wealth of the top 1 percent in ( ) and using a simplified computation in dashed line. The sim- the SCF. The )unit of observation is tax units for capitalized plified computation needs to use unrealistic assumptions to incomes and for the SCF. reproduce closely the Auten and Splinter series both in lev- Sources: Saez and Zucman 2016 , updated, and Bricker et els and trends. We need to assume that the concentration al. 2017 . Series are reproduced( )from Zucman forthcom- of nontaxable labor income and capital income on pension ing,( figure) 2 . ( funds is stable and low even though the concentration of tax- ) able labor income increases sharply. We also need to assume that the concentration of other nontaxable capital income distributing all sources of household wealth declined sharply from 1960 to 2016 to unrealistically low from financial accounts. They also find a large levels by 2016. Sources: Auten and Splinter 2018 and authors’ increase in wealth concentration. These find- computations. ( ) ings are illustrated in Figure 5 reproduced from Zucman forthcoming . Figure (5 displays the top assumptions reproduces closely the Auten and 1 percent wealth share) estimates based on Saez Splinter 2018 estimates both in levels and and Zucman 2016 series capitalizing income trends. However,( ) these alternative assumptions tax returns and( the )top 1 percent wealth share are extreme and hence unrealistic. In particular, combining the official SCF estimates of Bricker assumption 2 goes starkly against a body of evi- et al. 2017 and the Forbes 400 wealth share as dence showing that the concentration of wealth the Forbes( )400 are excluded by definition from( in the United States has in fact increased sharply the SCF . Both series show a sharp increase in as summarized in Zucman forthcoming . As far the top )1 percent wealth share. In 2016, both as( we can see, Auten and Splinter do not )provide series show that about 40 percent of total house- any corroborating evidence which could justify hold wealth is owned by the top 1 percent. The assumption 2, while there is ample evidence to SCF estimates for 2016 show a slightly higher justify assumption 1 as a benchmark hypothesis. top 1 percent wealth share 40.7 percent than The share of total household wealth owned by the Saez-Zucman capitalized( income estimates) the top 1 percent of households in the Survey 38.9 percent . These 2016 estimates are up of Consumer Finances SCF which excludes from( 30.8 percent) in 1989 using the SCF series the Forbes 400 has increased( ) from( 29.7 percent and up from 23.6 percent in 1980 using the cap- in 1989 to 38.8) percent in 2016 Bricker et al. italized income tax series. 2017 . The share of wealth owned( by the Forbes Therefore, our simplified methodology can 400 )has been multiplied by more than three show very simply that one needs extreme and since 1982 see Zucman forthcoming growing hence unrealistic assumptions on equalization from less than( 1 percent in the early) 1980s to of income components not in fiscal income to over 3 percent in the 2010s. Saez and Zucman reverse the large increase in income concen- 2016 created wealth inequality series using tration obtained from fiscal income series of the( capitalization) method and systematically­ Piketty and Saez 2003 . ( ) VOL. 109 SIMPLIFIED DISTRIBUTIONAL NATIONAL ACCOUNTS 295

REFERENCES “Changes in U.S. Family Finances from 2013 to 2016: Evidence from the Survey of Con- Alvaredo, Facundo, Anthony B. Atkinson, Lucas sumer Finances.” Federal Reserve Bulletin 103 Chancel, Thomas Piketty, Emmanuel Saez, 3 : 1–42. and Gabriel Zucman. 2016. “Distributional Congressional( ) Budget Office. 2018. The Distribu- National Accounts DINA Guidelines: Con- tion of Household Income, 2015. Washington, cepts and Methods( Used) in WID.world.” DC: Congressional Budget Office. World Wealth and Income Database Working Johns, Andrew, and . 2010. “The Paper 2016/1. Distribution of Income .” Alvaredo, Facundo, , Thomas National Tax Journal 63 3 : 397–418. Piketty, Emmanuel Saez, and Gabriel Zucman. Piketty, Thomas, and Emmanuel( ) Saez. 2003. 2018. World Inequality Report 2018. Cam- “Income Inequality in the United States, 1913- bridge, MA: Harvard University Press. 1998.” Quarterly Journal of Economics 118 Atkinson, Anthony B., Thomas Piketty, and 1 : 1–39. Emmanuel Saez. 2011. “Top Incomes in the Piketty,( ) Thomas, Emmanuel Saez, and Gabriel Long Run of History.” Journal of Economic Zucman. 2018. “Distributional National Literature 49 1 : 3–71. Accounts: Methods and Estimates for the Auten, Gerald, and( ) David Splinter. 2018. “Income United States.” Quarterly Journal of Econom- Inequality in the United States: Using Tax ics 133 2 : 553–609. Data to Measure Long-Term Trends.” http:// Saez, Emmanuel,( ) and Gabriel Zucman. 2016. davidsplinter.com/AutenSplinter-Tax_Data_ “Wealth Inequality in the United States since and_Inequality.pdf accessed January 5, 2019 . 1913: Evidence from Capitalized Income Tax Bricker, Jesse, Lisa J.( Dettling, Alice Henriques,) Data.” Quarterly Journal of Economics 131 Joanne W. Hsu, Lindsay Jacobs, Kevin B. 2 : 519–78. Moore, Sarah Pack, John Sabelhaus, Jeffrey Zucman,( ) Gabriel. Forthcoming. “Global Wealth Thompson, and Richard A. Windle. 2017. Inequality.” Annual Review of Economics.