This paper shows the effects of adjusting for technical tax issues and the sensitivity to alternative assumptions for distributing missing income sources. Our results suggest that top income shares are lower than other tax-based estimates, and since the early 1960s, increasing government transfers and tax progressivity resulted in little change in after-tax top income shares.
These results destroy the 2003 Piketty and Saez paper that found greatly increasing after tax income inequality. The PS paper was and continues to be the basis for the Left's claim of increasing income inequality.
Here is a key chart from the AS paper.
Once again, the Alarmists have been proved wrong.
The paper's introduction follows.
-------------------------------------------
Based on the results of studies using income tax data (Piketty and Saez, 2003; Piketty, Saez, and Zucman, 2018), the idea that U.S. income inequality has increased dramatically since the 1960s has become one of the most powerful narratives of our time. Broad acceptance of this view has fueled concerns that increasing inequality could indicate greater concentration of political power and increased rent-seeking (Stiglitz, 2012; Lindsey and Teles, 2017) or increased bargaining power of top earners for compensation (Piketty, Saez, and Stantcheva, 2014). In turn, these concerns have led to speculation that inequality could lead to problems such as decreasing institutional accountability, reduced economic efficiency, and stagnating middle-class wages due to shifts in relative bargaining power.
These profound implications emphasize the importance of correctly measuring income inequality. Estimating the income distribution over long time periods, however, is complicated by major challenges. These include changes in social conditions (marriage rates, household size and composition) and demographics (age distribution). Rising education standards and increased college attendance resulted in higher earnings but later entry into the labor force. Retirement incomes have changed due to expanded Social Security benefits and the shift from defined benefit to defined contribution plans. Periods of high inflation have distorted the measurement of income, and business cycles had differential effects on income groups.
Compared to survey data, tax data better represent top income groups, but using tax data presents additional challenges. Tax rules and incentives for reporting income have changed over time as the result of tax legislation. Declining marriage rates and changing household structures can lead to biased results when tax units are the unit of observation. While many adults do not file tax returns, many returns are filed by individuals under age 20, other dependents, and nonresidents. Important sources of income are missing in tax data, including government transfer payments and non-taxable employer-provided benefits. The share of income missing in tax data has increased over time, such that market income on tax returns accounts for only about 60 percent of national income in recent years. In addition, there are many technical issues with respect to differences between what is reported on tax returns and what economists regard as current-year economic income. Prior studies may have been misleading as a result of failure to adequately account for these challenges.
This paper presents new estimates of the levels and trends of U.S. top income shares that address these challenges. We start with income as reported on tax returns and develop an improved measure of market income—also referred to as fiscal income—that corrects for tax reforms and technical tax issues as well as social issues such as declining marriage rates. We then account for total national income with estimates of pre-tax and after-tax income. While our results are similar to several other recent studies, they suggest lower top income shares and less upward trend than Piketty and Saez (2003) and Piketty, Saez, and Zucman (2018, hereafter PSZ). Finally, we discuss why our results differ from PSZ and the implications for considering the distribution of economic growth and tax burdens.
Using income reported on individual tax returns, the highly influential paper by Piketty and Saez (2003 and updates) estimated that the top one percent share more than doubled from 9 to 20 percent between 1962 and 2015. About 40 percent of this increase, however, occurred in the years just before and after the Tax Reform of 1986 (TRA86). This major reform lowered statutory tax rates and broadened the tax base, thereby substantially changing tax rules and incentives for reporting income and organizing businesses. The potential for TRA86 to affect measures of U.S. inequality was noted by Feenberg and Poterba (1993) and Gordon and Slemrod (2000). Basing income groups on tax units produced additional upward bias in top income shares over time because marriage rates decreased disproportionately more among lower-income groups.
PSZ addressed some of these issues and made several important methodological improvements. Their recent paper uses an expanded measure of income that targets total national income and incorporates Social Security benefits. It addresses declining marriage rates by basing income groups on the number of adults age 20 and over. Despite changes that might be expected to reduce top income shares, PSZ still concluded that the top one percent share doubled since 1980 and increased by about two-thirds since 1962.
As illustrated in Figure 1, our results are quite different, especially since 1979. We both estimate that the top one percent shares of pre-tax and after-tax incomes declined slightly between the early 1960s and 1979, although our top share levels are slightly lower. Between 1979 and 2014, however, PSZ estimated that pre-tax top one percent shares increased by 9.0 percentage points while our estimates suggest that they increased by only 3.2 percentage points. For after-tax income, which also includes transfers, PSZ estimated that top one percent shares increased by 6.5 percentage points between 1979 and 2014.
Our estimates suggest that they increased by only 1.4 percentage points. Over the longer period since the early 1960s, our estimates suggest that the top one percent share increased less than half as much as PSZ and that after-tax income shares were nearly unchanged. Our estimates also differ for the bottom half of the distribution. PSZ estimated that the bottom 50 percent share of pre-tax and after-tax shares decreased by about 7.5 and 6.2 percentage points since 1979. In contrast, we estimate that pre-tax and after-tax income shares decreased by 5.6 and 2.2 percentage points over this period. Our estimates suggest that taxes and transfers offset most of the recent changes in both bottom and top income shares. In contrast with the PSZ estimate that average real pre-tax incomes of the bottom 50 percent remained virtually unchanged, we estimate that they increased by nearly one-third. For pre-tax/after-transfer income (which includes Social Security benefits) and after-tax income, we estimate a real increase for the bottom half of the distribution of nearly two-thirds. Our estimates are similar to those of the Congressional Budget Office (2018) for the bottom two quintiles.
Why are our results so different from Piketty and Saez (2003) and PSZ? There are many methodological differences, and this paper provides a detailed decomposition of their effects. Our estimates correct the tax sample to remove dependent and non-resident filers, as well as filers under age 20. We account for increases in the share of single-parent households and changing family size, as well as for falling marriage rates. We also correct for many special features of how income is reported on individual and corporate tax returns and how this has changed over time. While many improvements have only small or offsetting effects on top income shares, their cumulative effects can be significant.
Different treatments of business losses and pension income prove to be particularly important. Our approach corrects for the large tax shelter losses prior to TRA86 and adds back net operating loss carryovers from prior years, which are not current-year income. Our approach also accounts for business losses when allocating underreported income because detailed IRS audit studies show that returns with business losses account for a significant share of underreported income. PSZ, however, ignore losses and allocate underreported income only by positive reported income. Our retirement income allocation methodologies also produce quite different results, in part because PSZ mix taxable retirement income flows with non-taxable amounts, which are largely rollovers of assets.
We are not alone in finding lower levels and smaller increases in U.S. top income shares when using broad measures of income. Combining tax return and Census data, Fixler, Gindelsky, and Johnson (2019) estimated a top one percent share of personal income in 2012 of 13 percent, compared to 21 percent in Piketty and Saez (2003 and updates, hereafter PS). Using Survey of Consumer Finance data, Bricker et al. (2016a) found that the top one percent share increased 3 percentage points between 1988 and 2012, compared to 6 percentage points in PS. Using tax return and Census data, the Congressional Budget Office (2016) found that the top one percent share of before-tax income increased 6 percentage points from 9 to 15 percent between 1979 and 2013, compared to the PS estimate of a 10 percentage point increase from 9 to 19 percent. Our pre-tax income share increases by 4 percentage points from about 10 to 14 percent over this period. Using internal Census data to overcome top-coding issues, Burkhauser et al. (2012) estimated that the top one percent share only increased 2 percentage points from 10 to 12 percent between 1967 and 2004.
Our paper makes several important contributions to this emerging “consistent income inequality” literature on the distribution of income in the U.S. First, we provide new estimates of top income shares using administrative data to address major challenges in measuring the income distribution over long periods, while accounting for major tax reforms and other technical issues in using tax data. Second, we address the uncertainty created by the need to impute components of national income not reported in tax data by showing our step-by-step adjustments and imputations as well as sensitivity tests of less certain assumptions. This allows other researchers to see the effect of each adjustment and consider alternative estimates based on different combinations of assumptions. Third, we compare our methodology with PS, PSZ, and the Congressional Budget Office so that readers will have a better understanding of why our results (and those of other researchers) differ.
No comments:
Post a Comment