National Tax Journal, June 2015, 68 (2), 281–318 http://dx.doi.org/10.17310/ntj.2015.2.03

ACCOUNTING FOR INCOME CHANGES OVER THE GREAT RELATIVE TO PREVIOUS : THE IMPACT OF TAXES AND TRANSFERS

Jeff Larrimore, Richard V. Burkhauser, and Philip Armour

Using decomposition analysis together with March CPS data, we consider the relative importance of factors accounting for changes in post-tax, post-transfer income during each of the last four recessions. Unlike the double dip recession of the , employment drops rather than falling wage earnings drove income declines during the . Furthermore, taxes and transfers played a much greater role in offsetting market income losses — a result largely missed in analyses not accounting for taxes and transfers. This is particularly so among the bottom quintile where lower taxes and increased transfers offset more than one-half of market income declines.

Keywords: income decomposition, recessions, tax credits, transfer income JEL Codes: D31, H24, J3

I. INTRODUCTION edian cash market income fell more over the first three years of the Great Recession Mand its aftermath (2007–2010) than over the first three years of any other recession since the March Current Population Survey (CPS) first enabled annual measurement in 1968. However, an unprecedented governmental response in the form of tax credits, tax cuts, and transfer increases, including both in-cash ( Insurance, etc.) and in-kind (Supplemental Nutrition Assistance Benefits (SNAP) or Food Stamps, etc.) benefits, substantially offset this loss in market income for middle- and lower-income Americans. In this paper, we estimate the relative importance of factors accounting for

Jeff Larrimore: Division of Consumer and Community Affairs, Board, Washington DC, USA ([email protected]) Richard Burkhauser: Department of Policy Analysis and Management and Department of Economics, Cornell University, Ithaca, NY, USA, and Melbourne Institute of Applied Economic and Social Research, University of Melbourne, Melbourne, ([email protected]) Philip Armour: RAND Corporation, Santa Monica, CA, USA ([email protected]) 282 National Tax Journal income trends over the last four recessions, including the Great Recession. In accounting for income trends, we not only consider government tax and transfer policies, but also the importance of demographics, employment, and source-specific incomes. Doing so provides a more complete understanding of the economic crisis presented by the Great Recession and the way that policymakers responded to it. Using a decomposition analysis to compare the factors underlying income trends in each of the past four recessions, we show that the falling real earnings of those who remained employed played a relatively minor role in household income declines dur- ing the Great Recession. Instead, drops in employment primarily drove these income declines, which would have been much greater were it not for the role of tax policies and both in-cash and in-kind transfer policies. Because previous decomposition studies (Burtless, 1999; Daly and Valletta, 2006; Larrimore, 2014) did not account for taxes or in-kind transfers in their analyses, they greatly understate the increasing role govern- ment policies played in mitigating median income declines as well as the resources they made available to Americans over the Great Recession. This is particularly so among the bottom quintile where lower taxes, larger tax credits, and increased transfers offset more than one-half of market income declines. Although we focus our analysis on the relative importance of each income source in accounting for changes in income over the Great Recession and how it differed from previous recessions, our findings also contribute to the ongoing debate over the role and effectiveness of social safety net programs during recessions (e.g., Bitler and Hoynes, forthcoming; Moffitt, 2015; Mulligan, 2015). While we are not able to iden- tify behavioral responses to the tax and transfer programs we consider, we are able to provide guidance on the breadth of government policies that should be included in broader conceptualizations of what constitutes the social safety net and its behavioral and distributional consequences. For example, Bitler and Hoynes (forthcoming) show that the role of the four programs traditionally considered to represent the social safety net — SNAP (food stamps), AFDC/TANF, the EITC, and Unemployment Insurance — were largely consistent in the Great Recession with their role seen in previous recessions. But only focusing on these four programs will understate the full impact of government tax and transfers policies in offsetting declines in market income during recessions, and it will disproportionally do so during the Great Recession. We show that in the Great Recession new tax policies like temporary stimulus tax credits and payroll tax reductions were used effectively to offset income declines. These tax policies are often not considered in safety net research, but during the Great Recession they had an impact on income trends that was not only greater than in previous recessions, but also approaching in magnitude that from traditional safety net programs for low-income households.

II. DATA We primarily use the public-use March Current Population Survey (CPS) supple- mented with cell-means for top-coded incomes from Larrimore et al. (2008) in our Income Changes over the Great Recession Relative to Previous Recessions 283 analyses.1 The public-use CPS data is commonly used for evaluating U.S. income trends, given its high quality income data, including income from both market income sources (e.g., wages, self-employment income, , dividends, rents) and cash transfers (e.g., Social Security, Supplemental Security Income (SSI), Unemployment, Public Assistance, Workers Compensation). The CPS income measure is intended to capture all cash income, excluding capital gains. Based on this data, we calculate cash market income exclusive of all transfers (“market income”), and pre-tax income including cash transfers but excluding taxes and in-kind transfers (“pre-tax income”). Market income mirrors the income concept used by Piketty and Saez (2003), while pre-tax income mirrors the income measure used by most CPS-based researchers (e.g., Gottschalk and Danziger, 2005; Daly and Valletta, 2006; Blank, 2011). Both of these measures, however, are incomplete since they exclude taxes and in-kind transfers. We therefore construct a broader income measure, which includes taxes and several major in-kind transfers (“post-tax income”). In constructing this broader measure, we generally follow the procedures for esti- mating the value of in-kind transfers, tax credits, and tax liabilities from Armour, Burkhauser, and Larrimore (2013). We include the value of SNAP (food stamps), housing subsidies, and school lunches, as estimated by the Census Bureau. Each are now generally recognized as important economic resources, and the Census Bureau now includes them as resources in their Supplemental Poverty Measure (Interagency Technical Working Group, 2010). Using NBER TaxSim 9.0, we also impute the value of tax credits and liabilities, including federal, state, FICA, and SECA taxes based on the tax laws in effect in each year.2 The NBER TaxSim program is widely used for estimating tax liabilities (Auerbach and Feenberg, 2000; Kopczuk, 2005; Eissa, Kleven, and Kreiner, 2008). However, it calculates taxes based on 22 variables that do not reflect deductions or credits that have more specialized eligibility criteria, such as some higher education credits. The NBER program does, however, compute both the EITC and Child Tax Credit, which are important credits for many households. When imputing tax credits and liabilities, we assign individuals to tax units using the same procedure as in Burkhauser et al. (2012), which follows the assumptions regarding potential tax units from Piketty and Saez (2003). While we include the major in-kind transfers that are widely considered as near-cash income, we do not include employer- or government-provided health insurance in our main analysis. This is not because these benefits are unimportant. Instead, our exclusion of these benefits reflects the ongoing controversy over how to value them. Unlike the in-kind transfers we include, the value of access to health insurance is neither included as a resource in the Census Bureau’s Supplemental Poverty Measure nor is it included at its full market value for those in poverty in Census Bureau studies (DeNavas-Walt,

1 Larrimore et al. (2008) observe that results from the internal CPS data used for the Census Bureau’s of- ficial income statistics (DeNavas-Walt, Proctor, and Smith, 2013) can be closely matched by using the public-use CPS data with this cell-mean series. 2 Feenberg and Coutts (1993) provide an overview of the NBER TaxSim program. 284 National Tax Journal

Proctor, and Smith, 2013; Interagency Technical Working Group, 2010; Short, 2012). Recognizing this controversy, and maintaining the primary focus of this paper on the impact of tax policies, we exclude them from our analysis. However, in Appendix Table A2 we illustrate the extent to which the inclusion of employer- and government-provided health insurance at their full market value, as done by the Congressional Budget Office (CBO) (2012a, 2013), Burtless and Svaton (2010), and Burkhauser, Larrimore, and Simon (2013), alters income trends over each recession period that we consider. In each recession, the inclusion of health insurance benefits would further offset or reverse market income declines, particularly at the middle and bottom of the distribution. For all results, we size-adjust household income, which accounts for economies of scale in household consumption, by dividing total income by the square-root of household size and assigning this value to each household member.3 We also exclude individuals in the armed forces or living in group quarters. All income is adjusted for using the Consumer Price Index Research Series (CPI-U-RS) (Stewart and Reed, 1999).

III. LONG-TERM TRENDS IN MEDIAN INCOME Figure 1 shows the trends in the median size-adjusted household income of persons between 1979 and 2012, using our market (dark gray dashed-line), pre-tax (light gray dashed-line with circles), and post-tax (solid black line with squares) income measures. We normalize income to 1.00 in 1979 to show the percentage changes since 1979 for all other years. We denote peaks of each (1979, 1989, 2000, and 2007) with solid vertical lines, and troughs (1983, 1992, and 2004) with dashed vertical lines.4 Regardless of the income measure, median income traditionally has risen over time when compared at equivalent points in the business cycle, although it fluctuates within business cycles. This is the case for both the 1979–1989 business cycle — all three measures rise by at least 8 percent — and the 1989–2000 business cycle — all three rise by over 13 percent. However, over the 2000–2007 business cycle this was no longer the case. Real market income fell by 1.4 percent over the course of the business cycle, while pre-tax income remained at about the same level. In contrast, post-tax median income rose by 2 percent over this period. This was the first business cycle where pre-tax income and post-tax income diverged to a meaningful degree. This divergence began with the passage of the Economic Growth and Tax Relief Reconciliation Act (“EGTRRA”) in 2001, which was intended as a permanent measure, and continued with subsequent tax

3 This size adjustment is commonly used in U.S. and cross-national studies (Gottschalk and Smeeding, 1997; Atkinson and Brandolini, 2001; Burkhauser et al., 2011) as well as by the OECD in its official inequality and poverty measures (d’Ercole and Förster, 2012). It also closely matches the household size adjustment implied by the Census Bureau’s poverty thresholds (Ruggles, 1990). 4 In our analyses, business cycle peaks and troughs are defined by peaks and troughs in median income. Because median income declined continuously in the early 1980s we consider the 1979–1983 double-dip recession as if it was a single continuous recession. Income Changes over the Great Recession Relative to Previous Recessions 285

Figure 1 Trend in Median Size-Adjusted Household Income of Persons with and with- out Taxes and In-Kind Transfers, 1979 Normalized to 1.00 (1979–2012)

1.30

1.25

1.20

1.15

1.10

1.05

1.00

0.95

0.90 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 Market income Pre-tax cash income Post-tax income + in-kind transfers (excluding health ins.) Notes: Peak and trough years of business cycles are based on market income and are denoted with solid black and dashed gray vertical lines respectively. The starting year of the period (1979) also represents a peak business cycle year. Official NBER recession periods are denoted by vertical gray bars. Due to a change in CPS survey collection methods, income trends are not directly comparable between 1992 and 1993 (Ryscavage (1995) and Weinberg (2006) discuss this issue further). Because we assume that the change in the income series in this year is due solely to collection method differences, in Figure 1 we assume there is no change in the income series in this year. This assumption matches that described in Larrimore (2014), which is similar to that used by Atkinson, Piketty, and Saez (2011) and Burkhauser et al. (2012). Source: Authors’ calculations using Public-Use March CPS data (1980–2013)

reform packages that preserved and expanded the divergence. Hence, while each series records slower growth in median income in the early , for the first time accounting for government taxes and in-kind transfers substantially increases the rate of median income growth over a full business cycle. Moreover, the divergence between our post-tax series from both the market income series and the pre-tax income series continues through the Great Recession. Between 286 National Tax Journal

2007 and 2008, while median market income fell by almost 5 percent and median pre-tax income fell by almost 4 percent, post-tax median income fell by less than 1 percent. As we will argue below, this largely reflects the enactment of the Economic Stimulus Act of 2008 and subsequent temporary tax measures enacted in response to the Great Recession. These tax policy interventions substantially cushioned the decline in resources available to the average American during the Great Recession and the use of tax policies for such purposes has increased in the most recent two recessions. Hence, tax policy interventions should be included in any income measure when considering the role of government policy in mitigating income losses, or simply tracking the resources available to American households.

IV. MEDIAN INCOME TRENDS DURING ECONOMIC DECLINES Researchers commonly compare business cycle peaks to peaks or troughs to troughs, as we did above, to strip income trends of business-cycle variations. But it is also pos- sible to focus on the relative severity of economic downturns by examining similar periods in each business cycle. This can be done by comparing a fixed length of time after each peak year. Since two of the last four economic declines lasted just three years, we examine the recessions by comparing the first three years of economic decline fol- lowing each business-cycle peak.5 The first column of Table 1 reports the percent change in median household size- adjusted market income, exclusive of all taxes and transfers. Using this income measure, the severity of the most recent recession is evident as real median income fell by over 10 percent from 2007 through 2010. This is almost 3 percentage points greater than the decline seen over the severe 1979–1982 recession. When we include transfer income but exclude taxes in Column 2, as is done by the Census Bureau in their official economic reports (DeNavas-Walt, Proctor, and Smith, 2013), the income decline during the Great Recession remains the most severe of the four recessions considered. However, as Column 3 shows, this is no longer the case once we include taxes and in-kind transfers. In particular, in both the 2000–2003 recession and the Great Reces- sion, sizeable tax policy responses occurred. As a result, median post-tax income rose from 2000 through 2003 despite the recession, and the drop in median post-tax income during the Great Recession was smaller than that observed on a pre-tax basis. Since no

5 While we focus on symmetric three-year periods for each recession, there are other reasonable approaches for the comparison. In particular, one may want to examine just the NBER recession periods, examine the period from the peak to the trough in pre-tax income, or examine the period from the peak to the trough in market income. Using the last of these approaches would extend the early 1980s analysis to include 1983 and would extend the Great Recession analysis to include 2011. In Table 7 and Appendix Table A3, we present results for each individual year during these two recessions to allow for comparisons over alternate periods. When considering just the initial years of rapid income decline in the , the results are largely consistent with those that we present here. As is discussed further in Section VIII, the sources of income changes are more variable over the course of the Great Recession, partially due to the changing nature of the stimulus programs from year to year during this period. Income Changes over the Great Recession Relative to Previous Recessions 287 1.85 –3.12

–3.63 –2.78 Post-tax + in-kind –4.24 –2.43 –4.01 –1.38 Pre-tax Top Quintile (Mean) Top –4.60 –2.58 –4.19 –1.85 Market –4.10 –4.62 –4.23

–13.23 Post-tax + in-kind –12.26 –7.69 –7.87 –13.96 Pre-tax Bottom Quintile (Mean) Table 1 Table –30.17 –19.82 –18.13 –25.48 Market 0.35 –4.10

–3.49 –6.57 Post-tax + in-kind Three Years of Economic Downturns since 1979 since Downturns of Economic Years Three –6.97 –2.59 –4.01 –5.79 Pre-tax Median Income –3.70 –5.75 –7.56 –10.34

Market Percent Change in the Median Size-Adjusted Household Income and Income Inequality of Persons during the First Change in the Median Household Income Size-Adjusted during and Income the First Inequality of Persons Percent 2007–2010 2000–2003 1989–1992 Note: Median and quintiles are defined based on each series separately (i.e., the quintiles for the pre-tax income are based on pre-tax income while the quintiles for post-tax + in-kind income are based on that broader income Asdefinition). a result, the individuals in each quintile are not necessarily the same across income definitions. calculations using Public-Use March CPS Data (1980–2011) Authors’ Source: 1979–1982 288 National Tax Journal similar tax response occurred during earlier recessions, the decline in post-tax income during the Great Recession was substantially less severe than the post-tax income decline seen in the 1979–1982 recession. This is consistent with a comparison of pre-tax and post-tax income trends in supplemental data from the Congressional Budget Office (2013), which illustrated that including taxes reduced the severity of income declines in both the 2000–2003 recession and the Great Recession.6 This pattern is even starker for income changes among the bottom quintile of the distribution (Columns 4–6). Mean market income for the bottom quintile fell by over 30 percent in the Great Recession — exceeding that seen in each of the other periods. In contrast, using a traditional pre-tax measure of income, the decline in bottom quintile income was slightly greater over the 1979–1982 recession (–13.96 percent) than over the Great Recession (–12.26 percent). Once taxes and in-kind transfers are included, this gap widens (–13.23 percent vs. –4.10 percent) substantially. The decline in the post-tax income of the bottom quintile of –4.10 percent during the Great Recession is now no greater than what they experienced in the previous two recessions. This find- ing is consistent with Bitler and Hoynes’ (forthcoming) observations regarding trends in poverty by income definition. When they include taxes and additional transfers in their measure of income, their poverty rate falls by less during the Great Recession compared to their traditional pre-tax, cash poverty rate. Furthermore, similar to our finding, they report this moderating effect was greater during the Great Recession than for earlier recessions. Although transfers are primarily directed towards those in the bottom of the dis- tribution, the increased use of tax responses to recessions means that policy changes can impact the upper quintile as well. In the 2000 recession, the passage of EGTRRA in 2001 and the Job Growth and Tax Relief Reconciliation Act (“JGTRRA”) in 2003 reduced tax liabilities for higher-income households and pushed their income trends positive during this recession (Columns 7 through 9). (Appendix Table A1 provides a summary of major tax policy changes enacted in each year of economic down- turns, including those from EGTRRA and JGTRRA.)7 Reductions in tax liabilities also improved mean income trends for the top quintiles in the 1989 recession and the Great Recession. Hence focusing only on market income or pre-tax income will understate the resources of Americans and overstate their decline relative to broader measures that incorporate taxes and in-kind transfers during these recession years.

6 In Appendix Table A2 we add employer-provided health insurance to market income, and all subsequent series. We also add a fourth column of values that includes the market value of government-provided health insurance based on methods consistent with those used by the CBO (2012a) and Burkhauser, Larrimore and Simon (2013). In all four recessions, but particularly in the most recent two recessions, increases in the value of government-provided health insurance further mitigate declines in median income. Once health insurance gains are included, there is almost no change in median income over the Great Recession. This is consistent with the results in the supplemental data from CBO (2013). This pattern is similar for the bottom and top quintiles. 7 See Tax Policy Center (2011a, 2011b) for a more detailed description of tax policy changes in each year. Income Changes over the Great Recession Relative to Previous Recessions 289

In what follows, we focus solely on post-tax income and more precisely measure what sources of income are accounting for changes in median post-tax income and changes in the mean income of the bottom and top quintiles of the post-tax income distribution.

V. METHOD OF ACCOUNTING FOR SHIFTS IN INCOME DISTRIBUTIONS To isolate the factors that account for these changes over the last four recessions, we use a decomposition analysis similar to Burtless (1999), Iceland (2003), Daly and Valletta (2006), and Larrimore (2014). The decomposition analysis used here does not demonstrate causality, but it is a common method for accounting for changes in household income via changes in demographic and source-level income. It does so by allowing the demographic composition of the population (age, race, and marital status) and the level and distribution of income sources within the population to change, one factor at a time, thus separately accounting for changes in income. To avoid double counting, we consider the impact of each factor sequentially, and also do so conditional on the demographic factors considered. Embedded within this approach are four distinct techniques for decomposing income distribution changes. The first considers changes in the fraction of the population in population subgroups, including demographic groups or employment statuses. The second considers changes to source-level income distributions within these popula- tion subgroups, but keeps the rank-correlation of income sources constant. The third considers changes in the correlation of income sources over time. The fourth, and most important for this paper, analyzes tax changes by separating the policy-neutral tax liability changes that occur as a result of changes to underlying taxable income in a progressive tax system from those that occur from new tax policies. The first three of these decomposition techniques are employed in Burkhauser and Larrimore (2014). Larrimore (2014) also provides a further discussion of these aspects of the decomposition approach, including discussions of robustness to order of analysis and variants to the sharing unit definition.8

A. Changes in Population Characteristics Based on Atkinson (1998) and Burtless (1999), this technique accounts for changes in the frequencies of categorical characteristics in the population — including demo-

8 As with all similar decomposition analyses, a potential concern is that the order of analysis may impact the results due to the interaction between factors. While this concern cannot be eliminated without analyzing all possible analysis orders, it is mitigated here for several reasons. First, in a similar analysis of inequal- ity changes over the past 30 years, Larrimore (2014) analyzed effects in the order of those presented here and the reverse and found that the results were largely consistent. Second, since interaction effects should increase with longer time periods, this concern is smaller for our analysis of just the recession periods. Additionally, since our primary analysis is comparing effects in the same way across different recessions, the comparison will be impacted only if interaction effects differ substantially from one recession to the next. Since there is no reason to expect this to be the case, we do not expect the order of analysis to impact our findings greatly. 290 National Tax Journal graphic trends. For example, it considers how an increase in the share of Hispanics in the total population will change the overall income distribution, given that the income distributions of whites, blacks, and Hispanics are held constant. This technique re-weights observations from the base year, t, such that the weighted fraction of the population in each population group matches that in future years, t '. We increase the weight of individuals with characteristics (e.g., Hispanic) that are more prevalent in year t ' than in year t, while holding the incomes of all individuals fixed. This allows us to change the prevalence of those with this characteristic without altering the underlying income distributions of those within each characteristic group. In all cases, the base year, t, is the starting year of the business cycle and t ' is the comparison year during the economic downturn.

B. Changes in Source-Level Income Distributions within Population Groups The second decomposition technique is based on Burtless (1999) and Daly and Valletta (2006). It incorporates changes to individual incomes from each specific income source. t Each individual’s income, Yik , can be represented as the sum of their incomes from t t each income source, f1ik through fNik t t t ... t . (1) Yfik =+12ik ffik ++Nik

We assign individuals a percentile rank, pfik , for each income source based on the rank of their source-level income within their population subgroup k. Note that an individual’s percentile rank in the distribution of a specific income source will not, in general, be the same as that individual’s percentile rank in the distribution of overall household income.

To estimate the impact from changes to source f1’s distribution on overall income distributions, each individual’s income from the source f1 in year t is replaced with the income of the individual at the same percentile rank of the source f1 income distribu- tion in year t '

ˆt'() t' ()t ... t . (2) Ypik 11ik =+fpik 12ik ffik ++Nik

This preserves the conditional earnings rank of each individual from source f1 and the rank-correlation of earnings from source f1 with other income sources, while capturing changes in the source-level income distribution of source f1 within each population group. For now, the correlations of individuals’ positions in the distribution of source- level incomes (rank-correlations) within each population subgroup are assumed to be constant, which allows us to separate the importance of changes to a specific income source from changes to its relationship with other sources. Since this procedure com- bines income across years, prior to the analysis all income is adjusted for inflation using the CPI-U-RS. Income Changes over the Great Recession Relative to Previous Recessions 291

C. Changes in Income-Source Rank Correlations within Population Subgroups The previous techniques hold the rank-correlation of income sources constant. That is, if the male and female at percentile-ranks p1ik and p2ik in their conditional earnings distributions are married to each other in one year, we assume the same rank pairing continues in all future years. By performing these rank-preserving income exchanges for sources f1 and f2 separately, we analyze the impacts of the separate earnings distribu- tions without impacting the correlation between the two

ˆt'(, )(t' )(t' ) t ... t . (3) Ypik 12ik pfik =+1ik pf12ik ik pf23ik ++ik fNik

Once again, note that the percentile rank of the individual in the two source-level distributions, p1ik and p2ik , neither have to be the same nor do they have to match the individual’s rank in the overall household income distribution. To update the correlation between sources f1 and f2 , we combine these two sources before the rank-preserving income exchange rather than after. As a result, we have (N – 1) sources such that g1 = f1 + f2 while f3 through fN are unchanged. Calling each individual’s percentile-rank in the g1 distribution qfik , we calculate estimated incomes as

ˆt ' () t ' () t ... t . (4) Yqik 11ik =+gqik 13ik ffik ++Nik

This updates the correlation between sources f1 and f2 along with their income distribu- tions. The impact of the changing correlation between sources f1 and f2 is obtained by comparing the results in the case for the equation where only their separate income distributions change (3) with the case where their joint distribution changes (4).

D. Changes in Tax Policies Even in the absence of new tax policies, individuals’ tax liabilities change over time in response to income changes. If taxable income increases, then tax liabilities generally should as well, partially offsetting the increase in resources from those pre-tax income gains. Conversely, if taxable income declines then the reduction in tax liabilities partially offsets the fall in pre-tax income.9 However, tax liability changes may also result from new tax policies. Since this distinction is important for interpreting the relationship between taxes and income trends, we separate these two effects. We first consider a policy-neutral tax change. In the sequential analysis, changes to tax liabilities are recomputed separately from changes to the underlying incomes that feed into tax computations. The policy-neutral tax change uses the NBER TaxSim pro- gram with the tax laws from year t but the incomes from year t ' to observe how much

9 This is not universally the case, however, as phase-ins of credits may result in negative effective marginal tax rates. 292 National Tax Journal tax liabilities would have increased or fallen if tax policies were left unchanged over the period of analysis, given the income changes previously observed in the analysis. This method for isolating policy-neutral tax changes is similar to those previously employed in the by Clark and Leicester (2004), in European countries by Bargain and Callan (2010), and in the by Kasten, Sammartino, and Toder (1994), and Bargain et al. (2011).10 In all cases, constant-policy taxes are based on inflation-adjusted tax brackets, where the brackets from year t are adjusted using the CPI-U-RS up to price levels from year t ' (Kasten, Sammartino, and Toder (1994) illustrate how failing to do so dramatically increases policy-neutral tax liabilities due to inflation pushing individuals with unchanged real incomes into higher tax brackets).11 We then rerun the NBER TaxSim program using the actual tax laws in effect in year t ' to determine the portion of tax changes attributed to new tax policies. The increase or decline in tax liabilities from this second NBER TaxSim analysis relative to that observed in the policy-neutral analysis is the change attributed to the new policies.

VI. ACCOUNTING FOR SHIFTS IN MEDIAN INCOME The values in Row 1 of Table 2 repeat values first reported in the third column of Table 1 — the decline in median post-tax income during the first three years of each economic downturn since 1979. The rest of the rows in Table 2 show how much each factor accounts for these trends using the decomposition method described above. This analysis starts with four major demographic trends: an aging population, a more racially and ethnically diverse population, shifts in the number of secondary adults living in households (e.g., cohabiting partners, adult children, etc.), and the decline in the rate of marriage in the population.12 Rows 2 through 5 of Table 2 illustrate the

10 This approach approximates a “constant policy” baseline where tax changes scheduled to take effect in the future are not part of current policy. For example, under the policy-neutral approach the Economic Growth and Tax Relief Reconciliation Act (EGTRRA) and the Jobs and Growth Tax Relief Reconciliation Act (JGTRRA) were assumed to continue to follow the policies from 2007 and would not expire after 2010. Had they expired, this would have been treated as a policy change even though it would have been due to inaction rather than new legislation. 11 This is important prior to 1985 when brackets were not indexed to inflation — that is, when government inaction allowed inflation to lower the “bend points” in the real income distribution where marginal tax rates increased, effectively increasing average and marginal taxes for many individuals. Our approach to inflate tax brackets using price inflation for the policy-neutral environment matches that of Bargain et al. (2011) and Clark and Leicester (2004). Kasten, Sammartino, and Toder (1994) and Bargain and Callan (2010) also inflate tax brackets, but base inflation on income growth rather than price growth. We are not aware of any previous research that does not inflation-adjust tax brackets when isolating policy- neutral tax effects. 12 Aging patterns are considered using four categorical age groups: children (0–18), young adults (19–44), older adults (45–64), and the aged (age 65 and older). Races and ethnicities considered are white non- Hispanic, black, and Hispanic. Other races are included with white non-Hispanics because the small size of these groups prevents analyzing them separately. Marital status is the marital status of the household head, who can either be married, a single male, or a single female. Income Changes over the Great Recession Relative to Previous Recessions 293

Table 2 Factors Accounting for Changes in Median Size-Adjusted Household Income of Persons During the First Three Years of Economic Downturns Since 1979

1979–1982 1989–1992 2000–2003 2007–2010 (1) Percent change in median –6.57 –3.49 0.35 –4.10 income Change accounted for by: (2) Age 0.15 0.10 0.26 0.06 (3) Race –0.30 –0.38 –0.98 –0.63 (4) Presence of other adult 0.29 –0.04 –0.04 0.25 household members (5) Marriage –0.06 –0.06 0.02 0.02

(6) Male-head employment1 –2.02 –1.14 –1.23 –2.38 (7) Male-head earnings1 –3.90 –2.60 0.46 –1.76 (8) Female-head employment1 0.44 0.57 –0.65 –0.87 (9) Female-head earnings1 0.36 0.61 1.29 0.13 (10) Spouse’s earnings correlation –0.29 0.07 0.28 –0.25

(11) Earnings of others –1.78 –1.43 –1.62 –1.68 (12) Private non-labor income 0.33 –1.15 –0.64 –0.76

Subtotal, lines 2 through 12 –6.79 –5.44 –2.83 –7.88

(13) Cash public transfers 0.51 0.96 0.23 1.53 (14) In-kind public transfers –0.13 0.04 –0.06 0.11

(15) Policy-constant tax liability 6.74 1.05 0.13 0.98 changes (16) Tax liability changes due to –6.85 –0.06 2.96 1.20 new tax policies Notes: 1 Figures include household heads and the spouses of household heads. Source: Authors’ calculations using Public-Use March CPS Data (1980–2011) 294 National Tax Journal change in median income accounted for by changes in the demographic makeup of the United States (for summary statistics on these demographic trends, see Appendix Table A4). These estimated effects focus exclusively on changes in the number of people in the demographic groups and not on changes in the income gaps between these groups over the various recession years. Shifts in age and racial demographics generally do not change substantially over the two- to four-year period of a recession, and are included largely as controls for underlying demographic shifts. These demographic trends, especially with respect to the growing Hispanic population who typically have lower mean incomes, provide a baseline level of median income decline in all recessions. However, with the exception of the , these factors account for less than 15 percent of median income declines in each recession and generally cannot explain differences in the sever- ity of income declines across the recessions.13 The “doubling up” of households, with adults who would otherwise live separately combining into a single household for economic reasons, has been observed in pre- vious research to be a mechanism for low-income persons to weather the economic storm during recessions (Kochhar and Cohn, 2011; Mykyta and Macartney, 2011). The positive coefficient for the change accounted for by the addition of other house- hold members during the early 1980s recession and during the Great Recession (Row 4 of Table 2) provides evidence that doubling up slightly boosted median income trends during these more severe recessions. Our final demographic factor, marriage, is the least important of these demographic factors over these short-run recession periods. In looking at specific income sources, which are more consequential for short-run trends, we first focus on the primary members of a household. We separate them into male household heads or the male spouses of household heads, which for short we call “primary males,” and female household heads or the female spouses of household heads, which for short we call “primary females.” (In the CPS data, individuals are defined as household heads if they are the primary owners or primary renters of the dwelling.) This separates the impact of working children and other secondary earners from that of primary males and primary females in our analysis. Together, the labor earnings of these primary household members represent the bulk of all labor earnings in the United States (86 percent in 2010). When considering the impact of employment changes on household income, we consider three classifications of employment for each primary worker: full-time work (35+ hours per week for 50 weeks in the year), part-time work, or not employed. During each recession since 1979, while the part-time employment of primary males increased, their full-time employment declined. Combined, this overall decline in male

13 See Burkhauser and Larrimore (2014) for a further discussion of the impacts of changing demographics and the large racial income gap on long-run income trends, including their expected impacts in future decades. Income Changes over the Great Recession Relative to Previous Recessions 295 employment lowers median household income in each recession (Row 6 of Table 2). However, the fall in median income from male employment declines was greater dur- ing the 2007–2010 recession (2.38 percent) than that seen in any of the previous three recessions. In contrast, the decline in median income accounted for by changes in the earnings distribution of primary male workers during the 2007–2010 recession (Row 7 of Table 2) was substantially smaller than that seen from 1979–1982. Thus, in contrast to the early 1980s recession, a declining male employment rate — not declining male wage earnings — was the most important factor accounting for the severity of the first three years of the Great Recession. The first four columns of Table 3 provide context for the changing importance of pri- mary male earnings and employment on household income trends. Over the 1979–1982 recession, the decline in primary male full-time employment (5.51 percent) was much smaller than over 2007–2010 (6.85 percent). Consistent with findings of substantial underemployment in the Great Recession by Sum and Khatiwada (2010), there was also a concurrent increase in part-time work from 2007 to 2010. However, the mean earnings of full-time primary male workers actually rose by 0.93 percent from 2007– 2010, a sharp departure from the 3.90 percent decline from 1979–1982. This may in part be because layoffs and reduced hours in the Great Recession impacted low-wage workers to a greater extent than that seen in earlier recessions. But it also may reflect more real wage stability in the Great Recession than in earlier periods of economic decline. One potential explanation for the surprisingly stable earnings of full-time male work- ers in the Great Recession relative to the early 1980s recession is the level of inflation over the two periods. Over 2007–2010, inflation was at historic lows (1.6 percent annually) while 1979–1982 saw substantial inflation (9.4 percent annually). Consistent with theories of downward nominal wage rigidity, where employers are hesitant to cut nominal wages (Kahn, 1997; Card and Hyslop, 1997; Altonji and Devereux, 2000), low inflation may result in fewer real wage declines during the recession. Daly and Hobijn (2013) observe this during the Great Recession, as a large fraction of the population report no wage change from one year to the next while fewer than expected report nominal wage declines.14 Daly and Hobijn (2013) also model how this wage rigidity increases unemployment in periods of low inflation by changing firm behavior. They observe that in periods of low inflation, firms’ labor market adjustments in a recession disproportionately occur on the unemployment margin while in periods of high inflation these adjustments occur through wage declines. Our results are consistent with this model as the early 1980s period, with high inflation, saw income declines occur disproportionately through

14 Elsby, Shin, and Solon (2013) offer an opposing view on nominal wage rigidities, observing in British administrative data the fraction of workers reporting unchanged wages year-over-year is substantially smaller than that in CPS data in the United States. 296 National Tax Journal 333 –33 2.34 5.95 –595 1,117 –0.29 –2.94 11,429 11,395 18,778 19,895 14,246 14,579 20,196 19,601 Earnings Mean PT Mean PT

PT 0.46 29.59 –1.23 23.81 23.39 –0.42 –1.85 27.74 27.01 25.78 21.84 22.30 Percent Employed Percent Employed 946 834 995 2.61 6.18 2.75 2.18 2,616 30,374 42,352 44,968 31,209 36,196 37,142 45,690 46,686 Earnings Mean FT Mean FT Female Household Heads 1.34 1.05 26.99 40.35 39.06 –1.29 28.04 33.92 35.26 40.95 38.22 –2.73 Full-Time Percent Employed 666 2.14 –9.14 –9.35 26,811 26,687 31,132 31,798 24,191 –15.40 24,360 –2,450 33,290 28,164 –5,126 –2,496 Table 3 Table (2010-$) Earnings Mean PT Mean PT

1.44 3.29 3.01 19.36 14.20 15.36 1.16 22.37 17.77 19.21 14.62 17.91

Part-Time (PT) Part-Time Percent Employed 618 0.93 –916 –2.86 –1.34 –3.90 55,459 68,345 67,428 53,299 59,487 57,788 –1,699 66,485 67,103 –2,161 Earnings Mean FT Mean FT Male Household Heads 63.42 64.60 61.60 –3.00 57.91 62.39 59.36 –3.03 62.76 55.91 –6.85 –5.51 Full-Time (FT) Full-Time Percent Employed Employment and Earnings of Household Heads by Gender During the First Three Years of Economic Downturns Downturns of Economic Years Three Gender and Earnings During of Household Heads by the First Employment 1979 Percent Change 2000 2003 Change 1982 Percent Change Percent Change 1989 1992 Change 2007 2010 Change Percent Change Note: Household heads refers to both the household head as defined by the Census Bureau and their spouse, so a married couple is considered to have one male household head and one female head. calculations using Public-Use March CPS Data (1980–2011) Authors’ Source: Change Income Changes over the Great Recession Relative to Previous Recessions 297 reduced wages while the Great Recession, with low inflation, saw income declines disproportionately through reduced employment.15 A similar picture emerges when considering the employment and earnings of pri- mary females (Rows 8 and 9, Table 2). During the 1979–1982 and 1989–1992 periods, female full-time employment grew despite the recessions, offsetting otherwise declining household incomes. The strength of the long-term secular movement of women into the work force during the 1970s and 1980s overwhelmed any cyclical employment declines during recession years. However, by the 2000s, the secular movement of women into the work force slowed and, during the Great Recession, female employment fell.16 As a result, female employment declines in the Great Recession accounted for a 0.87 percent drop in median income — a reversal in sign from 1979–1982. The combined decline in primary male and female employment in rows 6 and 8 of Table 2 accounted for a 3.25 percent reduction in median household income during the Great Recession (2.38 plus 0.87), compared to a combined 1.59 percent reduction from these factors in the early 1980s recession. Thus, while falling median income in the 1979–1982 reces- sion was driven by a substantial drop in real wage earnings of working primary men, falling median income in the Great Recession was driven by declines in employment of primary men and women. Row 11 of Table 2 looks at the earnings of all other workers in the household. Dur- ing the Great Recession, declines in the earnings of these workers accounted for a 1.68 percent reduction in household median income. However, there is much less variation in the importance of this source of income over the four recessions. Declines in the wage earnings of other labor earners reduced household income in all four recession periods. Row 12 of Table 2 looks at private non-labor income. It declines in all recession periods except the recession of 1979–1982, where an increase in this income source offsets the decline in other sources of private market income. Part of the explanation for this increase in the early 1980s may be the high real rate of return on bonds and interest-bearing accounts over this period due to the high rates of inflation experienced in the early 1980s. As we discussed earlier, market income is only part of the story when considering income trends during recessions. Tax policies along with public in-cash (e.g., Unemploy- ment Insurance, etc.) and in-kind (e.g., SNAP, etc.) transfers are important components of many households’ incomes — especially among those who experienced earnings declines or job losses during the economic downturn.

15 Akerlof et al. (1996) estimates that a decrease in inflation from 3 to 2 percent increases the natural rate of unemployment from 5.8 to 6.1 percent, and a decrease of inflation to 1 percent increases it to 6.5 percent. These estimates may guide a reasonable lower bound on the impact that low inflation (1.6 percent) had on employment during the Great Recession, suggesting that low inflation contributed to approximately a 0.5 percent increase in unemployment due to nominal wage rigidities. This represents about 10 percent of the increase in unemployment that occurred from 2007 through 2010. 16 Blau and Kahn (2007) document the slowdown in female labor supply growth in the 1990s. More recent statistics from Macunovich (2010) and the Bureau of Labor Statistics (2012) indicate that female labor force participation for adults age 16 and over peaked in 2000 and fell over the past decade. 298 National Tax Journal

To demonstrate the magnitude of public transfers during the Great Recession rela- tive to earlier recessions, Table 4 reports the mean value of public in-cash and in-kind transfers by their source during each recession. Reflecting the extension of Unemploy- ment Insurance (UI) benefits to 99 weeks and the loosening of UI eligibility criteria (Farber and Valletta, 2013), unemployment, workers’ compensation, and veterans’ benefits more than doubled between 2007 and 2010. This compares to just a 30 percent increase in these benefits during the early 1980s recession. Similarly, in-kind transfers increased by 72 percent during the Great Recession, well exceeding the growth seen in earlier recessions. This likely reflects the relaxing of SNAP eligibility criteria in the Food, Conservation, and Energy Act of 2008 and increasing its maximum benefits in the American Recovery and Reinvestment Act of 2009 (Andrews and Smallwood, 2012). Overall, mean public transfers increased by 24.80 percent from 2007 to 2010, which is more than twice the 10.33 percent increase during the 1979–1982 recession (Column 6). These trends illustrate the scope of stimulus activities in the Great Recession, which Burtless (2010) estimated to represent 2.5 percent of the national economy in 2010. However, this does not necessarily mean that the measures were large relative to the size of the recession or necessarily targeted where unemployment was the greatest. Bitler and Hoynes (forthcoming) observe that relative to increases in state-level unemployment, increases in UI and other social safety net programs during the Great Recession were within the historically expected range. Given the size of the recession and the fact that the fall in median income was driven more by unemployment than a fall in real wage earnings, the increase in UI and other social safety net program benefits may be in line with unemployment rates, while also being quite large in absolute terms. In particular, the safety net from UI is designed to cushion income declines from lost employment, but not real wage declines. As a result, since income losses in the Great Recession were disproportionately due to unemployment, a larger portion of the income declines in the Great Recession than in earlier recessions fit into the category of economic hardship that this program is designed to offset. Additionally, to the extent that stimulus measures were broad-based and not well targeted at states with the highest levels of unemploy- ment, the state-level responses to elevated unemployment may appear weaker even if the national response was substantial. Even though these transfer programs were primarily targeted at low-income groups, their increase during the Great Recession had a noticeable impact on median household income trends (Rows 13 and 14 of Table 2). Cash public transfers mitigated median income declines by 1.53 percent and the growth in in-kind transfers contributed an additional 0.11 percent to median income growth. The positive impact of in-kind transfers on median income trends is consistent with Moffitt’s (2013) finding that SNAP expenditures doubled between 2007 and 2010 and Ziliak’s (2013) observation that SNAP successfully operated as an automatic stabilizer during the Great Recession. This is in contrast to the 1979–1982 recession where a fall in the real value of in-kind transfers reduced the much smaller increase in median household income accounted for by cash public transfers. Income Changes over the Great Recession Relative to Previous Recessions 299 49 0.65 –561 7,637 7,686 8,375 7,814 –6.70 9,029 9,302 8,129 10,620 –1,591 –14.98 –1,173 –12.61 Mean Tax Tax Mean Liabilities 271 307 227 778 7.71 11.26 2,620 2,891 10.33 2,723 3,030 2,939 3,165 3,137 3,915 24.80 Income, All Sources Income, Mean Public Transfer Transfer Mean Public –5 42 16 210 205 181 223 141 156 174 299 125 11.04 –2.55 22.94 71.97 In-kind Transfers 426 554 127 336 483 147 337 468 132 341 756 414 29.84 43.80 39.06 121.45 UC, WC, and UC, Veterans’ Benefits Veterans’

Table 4 Table 82 69 164 206 9.97 4.31 3.10 8.53 1,650 1,814 1,892 1,974 2,241 2,310 2,412 2,617 Income Social Security 2 95 97 26 18 25 119 145 158 175 176 201 2.13 SSI 11.14 21.54 14.15 Income

Mean Public Transfer Income, by Disaggregated Source Transfer Mean Public 8 10 63 55 –8 35 43 239 221 –18 195 206 5.30 –7.35 23.25 –12.31 or Welfare or Public Assistance Public Mean Size-Adjusted Tax and Transfer Income during the First Three Years of Economic Downturns (2010-$) Downturns of Economic Years Three Income during the First Transfer and Tax Mean Size-Adjusted 1979 1982 Change Percent Change 1989 1992 Change Percent Change 2000 2003 Change Percent Change 2007 2010 Change Percent Change Source: Authors’ calculations using Public-Use March CPS Data (1980–2011) Authors’ Source: 300 National Tax Journal

While we are unable to separate the cause of transfer increases into new policies versus automatic stabilizers, recent research can help inform the relative magnitudes of these effects. Considering UI growth, the CBO (2012b) estimates that of the $123 billion increase in UI payments between 2007 and 2010, $84 billion was paid through Emergency Benefits and Federal Additional Compensation. Since these Emergency Benefits were permitted specifically in response to the recession, this implies that approximately 68 percent of increased UI benefits are due to new policies. In contrast, the majority of the increase in SNAP benefits during the Great Recession can be attributed to the economic downturn rather than policy changes. Ziliak (2013) estimates that just 29 percent of the increase in food stamp usage during the Great Recession is due to food stamp policy changes. Ganong and Liebman (2013) offer a similar, but slightly lower estimate of the impact of new policies on the growth of SNAP during the Great Recession, observing that 18 percent of the growth is due to relaxed eligibility rules and an additional 3 to 11 percent is due to increased take-up from the higher benefit levels. Of course, there may also be secondary effects of these programs that partially coun- teract or enhance the increased incomes from the direct effects observed here.17 However, at least in the short-run, the direct effects of increases in public transfers — especially the growth and extension of in-cash unemployment benefits and the increased eligibility for means-tested in-kind transfer benefits like SNAP — mitigated a substantial fraction of the decline in market income during the Great Recession on median post-tax income. In addition to transfers, tax reductions and increases in tax credits are increasingly used as a countercyclical tool to offset market income declines. As can be seen in Column 7 of Table 4, despite the substantial decline in market income during the 1979–1982 recession, tax liabilities rose. In contrast, over the last two recessions, lower incomes as well as changes in tax policies resulted in double digit declines in tax liabilities. During the Great Recession, mean household tax liability fell by 12.61 percent. Changes in tax liabilities during a recession come from two distinct sources. The first occurs in a policy-neutral environment. To the degree that a recession lowers personal income, tax liabilities also generally fall. The second occurs through legislated tax policy changes that increase or reduce tax liabilities for individuals. During the early 1980s recession, real taxable income fell and as a result, tax liabilities fell when considered in a “policy-constant” setting. We account for this policy-constant component of tax liability in Row 15 of Table 2. In 1979–1982, liability changes due to falling real incomes would have reduced the decline in post-tax median income by 6.74 percent with no change to tax laws.

17 Mulligan (2012), for example, argues that increases in unemployment compensation and other public transfers could have delayed a return to work and hence partially contributed to the drop in employment. However, Chetty (2008) notes that a substantial portion of the extended length of unemployment may be due to a welfare-enhancing alleviation of liquidity constraints rather than a welfare-reducing induction of . For an early review of the literature on the relationship between increasing unemployment compensation and the duration of unemployment, see Danziger, Haveman, and Plotnick (1981). Income Changes over the Great Recession Relative to Previous Recessions 301

But this was a period of substantial inflation and Congress opted not to adjust tax brackets for inflation. (See Appendix Table A1 for a description of tax policy changes enacted during each year of the four economic downturns.) Automatic indexing of tax brackets for inflation was not implemented until 1985, and, since the Federal Income Tax during this period was highly progressive, this resulted in substantial bracket creep and increased real tax liabilities (Auerbach and Feenberg, 2000). Although the Act of 1981 reduced tax rates in 1981 and 1982, this was not suf- ficient to offset the lack of indexing. The “policy-change” components of tax liability are accounted for in Row 16 of Table 2, including both the lack of indexing and the newly legislated policies in the Economic Recovery Act of 1981. In 1979–1982 this increase in taxes on real income increased the decline in real income by 6.85 percent. As a result, the combined impact of the policy-neutral liability changes and those due to policy changes accounted for an additional 0.11 percent decline in median incomes. Contrast this with either the 2000–2003 recession or the Great Recession. In the 2000–2003 recession, EGTRRA and JGTRRA substantially reduced tax rates through- out the income distribution. These policy changes accounted for an increase in median income of 2.96 percent which, when combined with the policy-constant tax changes, accounted for a 3.09 percent increase in median incomes. This, in large part, explains why post-tax median income increased slightly during the first three years after the 2000 recession despite the decline in pre-tax income over the same period reported in Table 1. Similarly, tax changes during the Great Recession were used as a tool to offset fall- ing market incomes. These changes include the refundable tax rebate in the Economic Stimulus Act of 2008, the Making Work Pay credit and other tax changes in ARRA in 2009, and the 2 percent payroll tax holiday in the Tax Relief, Unemployment Insur- ance Reauthorization, and Job Creation Act of 2010. Tax policy changes since 2007 accounted for a 1.20 percentage point offset (Row 16, Table 2) of median income declines during the Great Recession that added to the offset that would have occurred in a policy-constant environment. As a result, while increases in tax liabilities reduced median income by 0.11 percent in the early 1980s recession, decreases in tax liabilities in the Great Recession increased median income by 2.18 percent. When we combine these median income increases from tax changes with those from public transfers, we estimate that government taxes and transfer policies accounted for a 3.82 percent increase in median post-tax income during the Great Recession compared to a 0.27 percent offset during the 1979–1982 recession. This helps explain the surpris- ing finding in Table 1 that when government taxes and in-kind transfers are added to pre-tax income, the decline in median post-tax income was smaller during the Great Recession than during the recession of 1979–1982.

VII. ACCOUNTING FOR SHIFTS AT THE BOTTOM AND TOP OF THE DISTRIBUTION We use this same decomposition procedure to consider the factors accounting for trends in the mean income of the bottom and the top of the post-tax income distribution. Table 5 provides such information for the bottom quintile of the post-tax, post-transfer 302 National Tax Journal

Table 5 Factors Accounting for Changes in the Mean Income of the Bottom Quintile of the Size-Adjusted Post-Tax Household Income Distribution Over the Course of the Past Four Recessions

1979–1982 1989–1992 2000–2003 2007–2010 (1) Percent change in mean –13.24 –4.23 –4.62 –4.10 bottom quintile income Change accounted for by: (2) Age 0.27 0.06 0.10 –0.05 (3) Race –0.52 –0.64 –1.19 –0.68 (4) Presence of other adult 0.33 –0.09 –0.22 0.52 household members (5) Marriage –0.33 –0.30 –0.40 –0.10

(6) Male-head employment1 –3.18 –2.03 –2.09 –4.56 (7) Male-head earnings1 –6.09 –2.19 0.53 –2.68 (8) Female-head employment1 –0.05 0.15 –1.48 –2.66 (9) Female-head earnings1 –0.05 0.00 0.34 –0.90 (10) Spouse’s earnings correlation –0.99 –0.24 –0.43 –0.88

(11) Earnings of others –2.88 –1.54 –2.37 –2.76 (12) Private non-labor income 0.36 –1.48 –0.72 –0.45

Subtotal, lines (2) through (12) –13.14 –8.30 –7.92 –15.19

(13) Cash public transfers –0.53 1.08 0.77 3.23 (14) In-kind public transfers 0.18 1.09 0.42 2.97

(15) Policy-constant tax liability 2.68 1.02 –0.08 1.49 changes (16) Tax liability changes due to –2.30 1.07 2.48 3.51 new tax policies Notes: 1 Figures include household heads and the spouses of household heads. Source: Authors’ calculations using Public-Use March CPS Data (1980–2011)

household income distribution while Table 6 does so for the top quintile of the post-tax, post-transfer distribution. Many of the same patterns seen in Table 2 emerge, although the magnitudes differ. At both the top and bottom of the distribution, decreases in primary male earnings account for the largest part of mean post-tax income declines in the early 1980s recession. In Income Changes over the Great Recession Relative to Previous Recessions 303

Table 6 Factors Accounting for Changes in the Mean Income of the Top Quintile of the Size-Adjusted Post-Tax Household Income Distribution Over the Course of the Past Four Recessions

1979–1982 1989–1992 2000–2003 2007–2010 (1) Percent change in mean top –2.78 –3.63 1.85 –3.12 quintile income Change accounted for by: (2) Age 0.04 0.23 0.46 0.14 (3) Race –0.16 –0.22 –0.65 –0.37 (4) Presence of other adult 0.23 0.04 0.00 –0.05 household members (5) Marriage –0.01 0.03 0.02 0.11

(6) Male-head employment1 –1.02 –0.72 –0.69 –1.05 (7) Male-head earnings1 –2.35 –3.09 –1.44 –0.53 (8) Female-head employment1 0.27 0.31 –0.05 –0.39 (9) Female-head earnings1 0.93 0.76 3.35 0.43 (10) Spouse’s earnings correlation 0.67 0.38 –0.29 0.27

(11) Earnings of others –1.16 –1.03 –0.39 –0.01 (12) Private non-labor income 2.03 –1.65 –2.16 –2.32

Subtotal, lines (2) through (12) –0.53 –4.96 –1.83 –3.78

(13) Cash public transfers 0.24 0.02 –0.01 0.35 (14) In-kind public transfers –0.01 –0.01 0.01 –0.08

(15) Policy-constant tax liability 10.87 1.67 0.07 0.43 changes (16) Tax liability changes due to –13.30 –0.25 3.77 0.03 new tax policies

Notes: 1 Figures include household heads and the spouses of household heads. Source: Authors’ calculations using Public-Use March CPS Data (1980–2011)

the Great Recession however, the decline in primary male earnings is less important than the decline in their employment. In the bottom quintile, the drop in primary male employment accounts for the largest part of the decline in mean post-tax income. But in the top quintile, reflecting the importance of investment income, the fall in non-labor income accounts for the largest part of the fall in mean income. 304 National Tax Journal

There are also dramatic differences in the importance of tax and transfer income on mean income of the bottom and top of the distribution. In the bottom quintile of the distribution, increases in public transfers are the single most important factor in mitigat- ing income declines. During the Great Recession, increases in transfer income offset over 6 percentage points of the fall in income. This is almost three times that seen in any of the previous recessions. In contrast, public transfers have very little effect for the top quintile. Similarly, while tax legislation enacted during the Great Recession was primarily targeted towards those near the bottom, it also accounted for a smaller increase at the top. For example, the $800 refundable Making Work Pay tax credit represents 6.2 per- cent of the income for tax units reporting income at or below $12,903 per year, but tax units making above $12,903 got no additional credits, and the $800 credit is phased out at higher income levels. As a result, for the bottom quintile, reduced tax liabilities due to tax policy changes accounted for more than a 3.5 percent increase in mean income. But for the top quintile, tax policy changes only accounted for a 0.03 percent increase in their mean income. This is in marked contrast to the 2000–2003 recession, where lower marginal tax rates in EGTRRA and JGTRRA offered substantial tax reductions for individuals at the bottom, middle, and top of the distribution and increased household incomes throughout the income scale accordingly.18

VIII. MEDIAN INCOME OVER THE COURSE OF THE GREAT RECESSION AND ITS AFTERMATH In Table 2 we compared the 4.10 percent decline in median income over the first three years of the Great Recession and the sources that accounted for it with the first three years of earlier recessions, but the declines in income across these Great Reces- sion years were not uniform, nor were the sources of income that accounted for them. In Table 7 we repeat our analysis of the sources of income that accounted for median income declines over the Great Recession, but now do so on a year-by-year basis.19 A comparable table for the early 1980s recession is provided in Appendix Table A3. Row 1 of Table 7 shows that median post-tax income declined 0.82 percent during the first year (2007–2008) of the Great Recession, before falling by over 2.09 percent between 2008 and 2009. These two years represent the official NBER time period of the Great Recession and the deepest drops in median income, but median income is a

18 The early 2000s recession also exhibited a negative policy-constant tax coefficient for bottom quintile income. This partially reflects the difference between tax-units and households, since we calculate taxes for tax-units and then aggregate to households — so the tax-base of tax-units does not always match household income. It also may reflect that for some income levels individuals’ tax liabilities fall as their incomes decline due to the phase-ins of the EITC and other credits. This is consistent with the Bitler, Hoynes, and Kuka (2014) findings of mixed effects for the EITC as an automatic stabilizer. 19 To offer comparability to Table 2, this analysis is always based on the 2007 base-year. Thus, changes described between 2008 and 2009 represent the change from 2007 through 2009 minus the changes that had occurred between 2007 and 2008. This allows the individual year results in Table 7 to sum to the three-year results from Table 2. Income Changes over the Great Recession Relative to Previous Recessions 305

Table 7 Accounting for the Annual Trend in Median Size-Adjusted Post-Tax Household Income of Persons During the Great Recession

2007–2008 2008–2009 2009–2010 2010–2011 2011–2012 (1) Percent change in median –0.82 –2.09 –1.19 –1.80 0.54 income

Change accounted for by: (2) Age 0.01 –0.02 0.07 –0.06 –0.11 (3) Race –0.24 –0.16 –0.23 –0.43 –0.14 (4) Presence of other adult 0.11 0.07 0.06 0.02 0.02 household members (5) Marriage 0.16 0.06 –0.20 0.03 0.19

(6) Male-head employment1 –0.92 –1.50 0.05 0.48 0.28 (7) Male-head earnings1 –1.06 –0.16 –0.54 0.06 0.03 (8) Female-head employment1 –0.21 –0.38 –0.27 –0.14 0.27 (9) Female-head earnings1 –0.67 0.63 0.17 –0.42 0.17 (10) Spouse’s earnings –0.24 –0.13 0.12 –0.14 0.21 correlation

(11) Earnings of others –0.92 –0.61 –0.14 –0.25 –0.27 (12) Private non-labor income –0.50 –0.58 0.32 –0.33 0.28

Subtotal, lines (2) through (12) –4.50 –2.79 –0.59 –1.19 0.93

(13) Cash public transfers 0.26 1.42 –0.15 –0.37 –0.31 (14) In-kind public transfers 0.00 0.11 0.00 0.06 0.00

(15) Policy-constant tax 1.12 –0.17 0.03 0.30 –0.20 liability changes (16) Tax liability changes due 2.31 –0.65 –0.45 –0.60 0.09 to new tax policies Notes: 1 Figures include household heads and the spouses of household heads. Source: Authors’ calculations using Public-Use March CPS Data (1980–2011)

lagging indicator of economic recovery. So while median income did not fall as much during 2009–2010 as it did during 2008–2009, it still fell by 1.19 percent. The sum of these three years of decline equals the decline of 4.10 percent for the three-year period reported in Table 2, as all changes are calculated using the 2007 base year. The last two columns extend our analysis to 2010–2011, the fourth consecutive year of median income decline (1.80 percent) and 2011–2012, the first year since the beginning of 306 National Tax Journal the Great Recession during which our median post-tax income measure increased (0.54 percent). The remainder of Table 7 shows how each of our demographic and economic variables account for these annual declines in median income and provides some insight for the continuing decline in median income we observe past the official NBER end of the Great Recession. As can be seen in Table 7, the relative importance of primary male employment varied greatly over the four years of median income decline. In 2007–2008 a drop in primary male earnings was the most important factor, while in 2008–2009 the drop in primary male employment accounted for, by far, the largest part of the decline in median income. After these official NBER recession years, primary male employment began to increase and hence accounted for modest increases in median income over each of the next three years. However, it was not until 2010–2011 that primary male earnings began to increase as well. Additionally, female employment, and the earnings of other household members, were even slower to recover. It was only in 2011–2012 that the sum of all employment and earnings factors in Table 7 accounted for an increase in median income. The annual decomposition in Table 7 also illustrates the short-term nature of stimulus programs as they played out in the tax code and in public transfers. During the first official NBER recession year of 2007–2008, increased transfers accounted for a 0.26 percent increase in median income growth. More importantly, in 2008, the launch of the Economic Stimulus Act provided up to a $1,200 credit for married couples plus additional credits for each dependent child (Joint Committee on Taxation, 2008). Largely resulting from this credit, new tax policies in 2008 mitigated median income declines by 2.31 percent (Row 16). In contrast, while tax policies in 2009 were still more generous to the median house- hold than in 2007 before the recession, 2009 represented the start of the withdrawal of tax stimulus for the middle of the distribution. Although the implementation of the Making Work Pay credit in 2009 provided a maximum benefit of $800 to a married couple, this was a less generous credit than the one offered by the Economic Stimulus Act the year before. As a result, changes in tax policy in 2009 reduced median income growth by 0.65 percent between 2008 and 2009. These declines were offset, however, by substantial increases in public transfers in 2009 which mitigated median income declines by 1.53 percent (Rows 13 and 14). Thereafter, decreases in tax-based stimulus programs and decreases in public transfers both worked in the same direction, each offsetting some of the growth in private market sources of income as economic stimulus is withdrawn. From 2009 to 2010 transfers fell from their 2009 peak, and, as a result of ending the partial exclusion of unemployment benefits from taxable income along with some state tax increases, tax liabilities around the median grew. This reduction in public transfers from the previous year and these increased tax liabilities accounted for a decline in after-tax median income. From 2010 to 2011, tax and transfer stimulus measures were further scaled back, accounting for additional declines in after-tax median income that offset the growth in private market income in that year. While public transfers abated in 2012, in part because Income Changes over the Great Recession Relative to Previous Recessions 307 of the continued improvement in the private-sector economy, it was the first year since 2008 that tax policy changes did not account for a year-over-year reduction in median income growth. This development, along with the continued employment and earnings gains of both men and women, allowed 2012 to be the first year since the start of the Great Recession that post-tax median income produced a year-over-year improvement. In contrast to the Great Recession, where the year-by-year results show that the expan- sion of tax credits offset income declines in the early years and the scaling back of these programs slowed income growth in the recovery, changes in taxes and transfers in the early 1980s recession (Appendix Table A3) played a much smaller role in the income trend. As a result, median incomes fell more sharply in the initial years of the recession — but then turned slightly positive by the third year. Additionally, when considering just the initial years of the 1980s recession, our finding that income declines in that reces- sion were largely driven by declining earnings rather than falling employment is even stronger. This suggests that while the dynamics of the Great Recessions are sensitive to the periods of analysis, this is less true when considering the early 1980s recession.

IX. CONCLUSIONS Using a decomposition analysis to compare the factors underlying income trends in each of the past four recessions, we show that the falling real earnings of those who remained employed played a relatively minor role in median post-tax household income declines during the Great Recession. Instead, employment declines primarily drove these income declines, which would have been much greater, except for the unprecedented role of public tax and both in-cash and in-kind transfer policies. Because previous decomposition studies have not included the role of either tax policies or in-kind transfers, they greatly understate the increasing role that government policies have played in mitigating median post-tax household income declines and understate the resources that were available to the bottom half of the distribution of Americans over the Great Recession. However, something that cannot be drawn from our analysis is whether the tax and transfer policies had secondary effects such as altering incentives for the unemployed to look for work (Mulligan, 2012), improving welfare by reducing the liquidity constraints of unemployed workers (Chetty, 2008), or resulting in stimulus multipliers that short- ened the length or severity of the recession (CBO, 2014; Wilson, 2012). Furthermore, both tax reductions and increased transfers come at the cost of increased public debt, which is not included in our analysis since it does not impact short-term economic resources. What can be concluded is that the substantial importance of temporary tax and transfer policies for supporting median and bottom-quintile income during the recession means that their withdrawal — as policymakers shift their focus to deficit reduction and the scaling down of stimulus measures — is likely to result in short-term headwinds toward achieving growth in post-tax median income. Hence growth in post-tax median income over the remainder of the current business cycle will depend on the ability of currently 308 National Tax Journal underemployed or unemployed individuals to find full-time jobs in a growing economy when these temporary public-transfer programs, which limited median income declines during the Great Recession, are scaled back.

ACKNOWLEDGEMENTS AND DISCLAIMERS Support for this research from the Russell Sage Foundation is cordially acknowl- edged. We also thank Marianne Bitler, Gary Burtless, Sheldon Danziger, John Logan, and Jeff Thompson for their helpful comments and suggestions on earlier versions of this paper. All opinions are those of the authors and do not represent the concurrence of the Federal Reserve Board, the Federal Reserve , or their staff.

DISCLOSUREs Burkhauser and Larrimore received funds in excess of $5,000 from the Russell Sage Foundation for work on earlier versions of this research. In addition to these funds, Burkhauser over the past 12 months has received funding in excess of $5,000 from the National Institute on Disability and Rehabilitation Research and the Employment Policies Institute. In addition he received funding not in excess of $5,000 from: The American Enterprise Institute, The Brookings Institution, the Federal Reserve Board, and the Pew Charitable Trusts. In addition, Armour over the past 12 months has received funding in excess of $5,000 from the Disability Research Consortium, and funding not in excess of $5,000 from the Association for Convenience and Fuel Retailing.

REFERENCES

Akerlof, George A., William T. Dickens, George L. Perry, Robert J. Gordon, and N. Gregory Mankiw, 1996. “The Macroeconomics of Low Inflation.”Brookings Paper on Economic Activ- ity 27 (1), 1–76. Altonji, Joseph G., and Paul J. Devereux, 2000. “The Extent and Consequences of Downward Nominal Wage Rigidity.” In Polachek, Solomon W. (ed.), Worker Well Being, Research in Labor Economics, Volume 19, 383–431. Elsevier, Amsterdam, The Netherlands. Andrews, Margaret, and David Smallwood, 2012. “What’s Behind the Rise in SNAP Participa- tion?” Amber Waves 10 (1), 1–6. Armour, Philip, Richard V. Burkhauser, and Jeff Larrimore, 2013. “Deconstructing Income and Income Inequality Measures: A Crosswalk from Market Income to Comprehensive Income.” American Economic Review 101 (3), 173–177. Atkinson, Anthony B., 1998. Poverty in Europe. Blackwell, Oxford, UK. Atkinson, Anthony B., and Andrea Brandolini, 2001. “Promises and Pitfalls in the Use of Sec- ondary Data Sets: Income Inequality in OECD Countries as a Case Study.” Journal of Economic Literature 39 (3), 771–799. Income Changes over the Great Recession Relative to Previous Recessions 309

Atkinson, Anthony B., Thomas Piketty, and Emmanuel Saez, 2011. “Top Incomes in the Long Run of History.” Journal of Economic Literature 49 (1), 3–71. Auerbach, Alan J., and Daniel Feenberg, 2000. “The Significance of Federal Taxes as Automatic Stabilizers.” Journal of Economic Perspectives 14 (3), 37–56. Bargain, Olivier, and Tim Callan, 2010. “Analysing the Effects of Tax-benefit Reforms on Income Distribution: A Decomposition Approach.” Journal of Economic Inequality 8 (1), 1–21. Bargain, Olivier, Mathias Dolls, Herwig Immervoll, Dirk Neumann, Andreas Peichl, Nico Pestel, and Sebastian Siegloch, 2011. “Tax Policy and Income Inequality in the U.S., 1978–2009: A Decomposition Approach.” IZA Discussion Paper No. 5910. IZA, Bonn, . Bitler, Marianne, and Hilary Hoynes, forthcoming. “The More Things Change, the More They Stay the Same: The Safety Net, Living Arrangements, and Poverty in the Great Recession.” Journal of Labor Economics.

Bitler, Marianne, Hilary Hoynes, and Elira Kuka, 2014. “Do In-Work Tax Credits Serve as a Safety Net,” NBER Working Paper No. 19785. National Bureau of Economic Research, Cambridge, MA.

Blank, Rebecca M., 2011. Changing Inequality. University of California Press, Berkeley, CA.

Blau, Francine D., and Lawrence M. Kahn, 2007. “Changes in the Labor Supply Behavior of Married Women: 1980–2000.” Journal of Labor Economics 25 (3), 393–438.

Bureau of Labor Statistics, 2012. Household Data Annual Averages: Employment Status of the Civilian Noninstitutional Population 16 Years and Over by Sex, 1973 to Date. U.S. Department of Labor, Washington, DC, http://www.bls.gov/cps/cpsaat02.pdf.

Burkhauser, Richard V., Shuaizhang Feng, Stephen P. Jenkins, and Jeff Larrimore, 2011. “Esti- mating Trends in United States Income Inequality Using the March Current Population Survey: The Importance of Controlling for Censoring.” Journal of Economic Inequality 9 (3), 393–415.

Burkhauser, Richard V., Shuaizhang Feng, Stephen P. Jenkins, and Jeff Larrimore, 2012. “Recent Trends in Top Income Shares in the USA: Reconciling Estimates from March CPS and IRS Tax Return Data.” Review of Economics and Statistics 94 (2), 371–388. Burkhauser, Richard V., and Jeff Larrimore, 2014. “Median Income and Income Inequality: From 2000 and Beyond.” In Logan, John (ed.), Diversity and Disparities: America Enters a New Century, 105–138. Russell Sage Foundation Press, New York.

Burkhauser, Richard V., Jeff Larrimore, and Kosali Simon, 2013. “Measuring the Impact of Valuing Health Insurance on Levels and Trends in Inequality and How the Affordable Care Act of 2010 Could Affect Them.” Contemporary Economic Policy 31 (4), 779–794. Burtless, Gary, 1999. “Effects of Growing Wage Disparities and Changing Family Composition on the U.S. Income Distribution.” European Economic Review 43 (4–6), 853–865. Burtless, Gary, 2010. “Crisis No More: The Success of Obama’s Economic Stimulus Program.” Pathways Magazine (Summer), 24–28. 310 National Tax Journal

Burtless, Gary, and Pavel Svaton, 2010. “Health Care, Health Insurance, and the Distribution of American Incomes.” Forum for Health Economics & Policy 13 (1), 1–41. Card, David, and Dean Hyslop, 1997. “Does Inflation Grease the Wheels of the Labor Market?” In Romer, Christina D., and David H. Romer (eds.), Reducing Inflation: Motivation and Strat- egy, 71–122. National Bureau of Economic Research Studies in Business Cycles. University of Chicago Press, Chicago, IL. Chetty, Raj, 2008. “Moral Hazard versus Liquidity and Optimal Unemployment Insurance.” Journal of Political Economy 116 (2), 173–234. Clark, Tom, and Andrew Leicester, 2004. “Inequality and Two Decades of British Tax and Benefit Reform.” Fiscal Studies 25 (2), 129–158. Congressional Budget Office, 2012a.The Distribution of Household Income and Federal Taxes, 2008 and 2009. Congressional Budget Office, Washington, DC. Congressional Budget Office, 2012b.Unemployment Insurance in the Wake of the Recent Reces- sion. Congressional Budget Office, Washington, DC. Congressional Budget Office, 2013. The Distribution of Household Income and Federal Taxes, 2010. Supplemental Data. Congressional Budget Office, Washington, DC, http://www.cbo.gov/ publication/44604. Congressional Budget Office, 2014.Estimated Impact of the American Recovery and Reinvestment Act on Employment and Economic Output in 2014. Congressional Budget Office, Washington, DC. Daly, Mary C., and Robert G. Valletta, 2006. “Inequality and Poverty in the United States: The Effects of Rising Dispersion of Men’s Earnings and Changing Family Behavior.” Economica 73 (289), 75–98. Daly, Mary C., and Bart Hobijn, 2013. “Downward Nominal Wage Rigidities Bend the Phillips Curve.” FRBSF Working Paper 2013-08. Federal Reserve of San Francisco, San Francisco, CA. Danziger, Sheldon, Robert Haveman, and Robert Plotnick, 1981. “How Income Transfer Programs Affect Work, Savings, and the Income Distribution: A Critical Review.” Journal of Economic Literature 19 (3), 975–1028. DeNavas-Walt, Carmen, Bernadette D. Proctor, and Jessica C. Smith, 2013. U.S. Census Bureau, Current Population Reports P60-245, Income, Poverty, and Health Insurance Coverage in the United States: 2012. U.S. Government Printing Office, Washington, DC. d’Ercole, Michael M., and Marco Mira Förster, 2012. “The OECD Approach to Measuring In- come Distribution and Poverty: Strengths, Limits and Statistical Issues.” In Besharov, Douglas J., and Kenneth A. Couch (eds.), European Measures of Income and Poverty: Lessons for the U.S., 27–58. Oxford University Press, New York, NY. Income Changes over the Great Recession Relative to Previous Recessions 311

Eissa, Nada, Henrik Jacobsen Kleven, and Claus Thustrup Kreiner, 2008. “Evaluation of Four Tax Reforms in the United States: Labor Supply and Welfare Effects for Single Mothers.” Journal of Public Economics 92 (3), 3–4. Elsby, Michael, Donggyun Shin, and Gary Solon, 2013. “Wage Adjustment in the Great Reces- sion.” NBER Working Paper No. 19478. National Bureau of Economic Research, Cambridge, MA. Farber, Henry S., and Robert G. Valletta, 2013. “Do Extended Unemployment Benefits Lengthen Unemployment Spells? Evidence from Recent Cycles in the U.S. Labor Market.” FRBSF Working Paper 2013-09. of San Francisco, San Francisco, CA. Feenberg, Daniel, and Elizabeth Coutts, 1993. “An Introduction to the TAXSIM Model.” Journal of Policy Analysis and Management 12 (1), 189–194. Ganong, Peter, and Jeffrey Liebman, 2013. “The Decline, Rebound, and further Rise in SNAP Enrollment: Disentangling Business Cycle Fluctuations and Policy Changes.” NBER Working Paper No. 19363. National Bureau of Economic Research, Cambridge, MA. Gottschalk, Peter, and Sheldon Danziger, 2005. “Inequality of Wage Rates, Earnings and Family Income in the United States, 1975–2002.” Review of Income and 51 (2), 231– 254. Gottschalk, Peter, and Timothy M. Smeeding, 1997. “Cross-National Comparisons of Earnings and Income Inequality.” Journal of Economic Literature 35 (2), 633–687. Iceland, John, 2003. “Why Poverty Remains High: The Role of Income Growth, Economic Inequality, and Changes in Family Structure, 1949–1999.” Demography 40 (3), 499–519. Interagency Technical Working Group, 2010. Observations from the Interagency Techni- cal Working Group on Developing a Supplemental Poverty Measure. U.S. Bureau of the Census,Washington, DC. Joint Committee on Taxation, 2008. Technical Explanation of the Revenue Provisions of H.R. 5140, The “Economic Stimulus Act Of 2008” as passed by the House of Representatives and the Senate on February 7, 2008. Joint Committee on Taxation Report JCX-16-08. Joint Committee on Taxation, Washington, DC. Joint Committee on Taxation, 2013. General Explanations of Tax Legislation Enacted in the 112th Congress. Joint Committee on Taxation Report JCS-2-13. Joint Committee on Taxation, Washington, DC. Kahn, Shulamit, 1997. “Evidence of Nominal Wage Stickiness from Microdata.” American Economic Review 87 (5), 993–1008. Kasten, Richard, Frank Sammartino, and Eric Toder, 1994. “Trends in Federal Tax Progressivity, 1980–93.” In Slemrod, Joel (ed.), Tax Progressivity and Income Inequality, 9–50. Cambridge University Press, Cambridge, MA. 312 National Tax Journal

Kochhar, Rakesh, and D. Cohn, 2011. “Fighting Poverty in a Tough Economy, Americans Move in with Their Relatives.” Pew Research Center, Washington, DC. Kopczuk, Wojciech, 2005. “Tax Bases, Tax Rates and the Elasticity of Reported Income.” Journal of Public Economics 89 (11–12), 2093–2119. Larrimore, Jeff, 2014. “Accounting for United States Income Inequality Trends: The Changing Importance of Household Structure and Male and Female Labor Earnings Inequality.” Review of Income and Wealth, 60 (40), 683–701. Larrimore, Jeff, Richard V. Burkhauser, Shuaizhang Feng, and Laura Zayatz, 2008. “Consistent Cell Means for Topcoded Incomes in the Public Use March CPS (1976–2007).” Journal of Economic and Social Measurement 33 (2–3), 89–128. Macunovich, Diane, 2010. “Reversals in the Patterns of Women’s Labor Supply in the United States, 1977–2009.” Monthly Labor Review 133 (11), 16–36. Moffitt, Robert A., 2013. “The Great Recession and the Social Safety Net.”Annals of the American Academy of Political and Social Science 650 (1), 143–166. Moffitt, Robert A., 2015. “The U.S. Safety Net and Work Incentives: The Great Recession and Beyond.” Journal of Policy Analysis and Management 34 (2), 458–466. Mulligan, Casey, 2012. The Redistribution Recession: How Labor Market Distortions Contracted the Economy. Oxford University Press, New York, NY. Mulligan, Casey, 2015. “The New Employment and Income Taxes.” Journal of Policy Analysis and Management 34 (2), 466–473. Mykyta, Laryssa, and Suzanne Macartney, 2011. “The Effects of Recession on Household Com- position: ‘Doubling Up’ and Economic Well-Being.” U.S. Census Bureau SEHSD Working Paper 2011-4. U.S. Department of Commerce, Washington, DC. Piketty, Thomas, and Emmanuel Saez, 2003. “Income Inequality in the United States, 1913–1998.” Quarterly Journal of Economics 118 (1), 1–39. Ruggles, Patricia, 1990. Drawing the Line: Alternative Poverty Measures and their Implication for Public Policy. Urban Institute Press, Washington, DC. Ryscavage, Paul, 1995. “A Surge in Growing Income Inequality?” Monthly Labor Review 118 (8), 51–61. Short, Kathleen, 2012. U.S. Census Bureau Current Population Reports P60-244. The Research Supplemental Poverty Measure: 2011. U.S. Government Printing Office, Washington, DC. Stewart, Kenneth J., and Stephen B. Reed, 1999. “CPI Research Series Using Current Methods, 1978–98.” Monthly Labor Review 122 (6), 29–38. Income Changes over the Great Recession Relative to Previous Recessions 313

Sum, Andrew, and Ishwar Khatiwada, 2010. “The Nation’s Underemployed in the ‘Great Reces- sion’ of 2007–09.” Monthly Labor Review 133 (11), 3–15. Tax Policy Center, 2011a. Selected Provisions of Major Tax Legislation by Effective Date 1981–2019. Urban-Brookings Tax Policy Center, Washington, DC, http://www.taxpolicycenter. org/legislation/upload/Legislation_by_date-2-28-11.pdf. Tax Policy Center, 2011b. Selected Provisions of Major Tax Legislation by Act 1981–2010. Urban-Brookings Tax Policy Center, Washington, DC, http://www.taxpolicycenter.org/legisla- tion/upload/legislation_by_act-2-28-11-4.pdf. Weinberg, Daniel H., 2006. “Income Data Quality Issues in the CPS.” Monthly Labor Review 129 (6), 38–45. Wilson, Daniel J., 2012. “Fiscal Spending Jobs Multipliers: Evidence from the 2009 American Recovery and Reinvestment Act.” American Economic Journal: Economic Policy 4 (3), 251–282. Ziliak, James P., 2013. “Why Are so Many Americans on Food Stamps? The Role of the Economy, Policy, and Demographics.” University of Kentucky Center for Poverty Research Discussion Paper Series, DP2013-01. University of Kentucky, Lexington, KY. 314 National Tax Journal

Appendix Table A1 Major Tax Policy Changes by Year

Major Federal Income Tax Policy Changes Enacted in Each Year Year of Economic Decline

1979–1982 Economic Decline 1980 Brackets: Real tax bracket thresholds fell 10% due to non-indexed brackets. Ordinary Rates: None. Other: None. 1981 Brackets: Real tax bracket thresholds fell 8.7% due to non-indexed brackets. Ordinary Rates: Reduced for all brackets (top rate from 70% to 69.125%).1 Other: Top capital gains rate reduced from 40% to 20%.1 1982 Brackets: Real tax bracket thresholds fell 5.7% due to non-indexed brackets. Top 5 tax brackets combined.1 Ordinary Rates: Reduced for all brackets (top rate from 69.125% to 50%).1 Other: AMT Exemption increased.2 1989–1992 Economic Decline 1990 Brackets: None. Ordinary Rates: Increased for top bracket from 28% to 31%.3 Other: None. 1991 Brackets: None. Ordinary Rates: None. Other: EITC Benefits increased and indexed to inflation.3 Personal Exemption Phase-out introduced for high-income individuals.3 1992 Brackets: None. Ordinary Rates: None. Other: None. 2000–2003 Economic Decline 2001 Brackets: Split brackets into 6 brackets from 5.4 Ordinary Rates: Reduced for all brackets (top rate from 39.6% to 39.1%).4 Other: Non-refundable $300(single)/$600(joint) tax rebate issued.4 Child Tax Credit increased and made partially refundable.4 AMT exemption increased.4 2002 Brackets: None. Ordinary Rates: Reduced for middle and upper income levels (top rate from 39.1% to 38.6%).4 Other: End $300/$600 non-refundable tax rebate.4 2003 Brackets: None. Ordinary Rates: Reduced for middle and upper income levels (top rate from 38.6% to 35%).5 Other: Tax rate on capital gains and dividends reduced; AMT exemption increased.5 Income Changes over the Great Recession Relative to Previous Recessions 315

Appendix Table A1, Continued

Major Federal Income Tax Policy Changes Enacted in Each Year Year of Economic Decline

2007–2012 Economic Decline 2008 Brackets: None. Ordinary Rates: None. Other: Refundable tax credit of up to $600(single)/$1200(joint) plus $300 per dependent.6 Capital Gains tax rate reduced from 5% to 0% for low-income taxpayers.5 Refundability of the Child Tax Credit increased.7 Personal Exemption Phase-out reduced for high-income taxpayers.4 2009 Brackets: None. Ordinary Rates: None. Other: End $600/$1200 refundable credit.6 Introduce $400(single)/$800(joint) Making Work Pay credit.7 Increase EITC benefits for taxpayers with 3 or more children.7 Introduction of the American Opportunity Tax credit.7 Partial exclu- sion of unemployment compensation from taxable income.7 2010 Brackets: None. Ordinary Rates: None. Other: Personal Exemption Phase-out eliminated for high income taxpayers.4 End partial exclusion of unemployment compensation from taxable income.8 2011 Brackets: None. Ordinary Rates: None. Other: End $400/$800 Making Work Pay credit.8 Introduce 2% payroll tax holiday.9 Notes: This table only summarizes the major tax policy changes over the periods of our analysis, and does not represent a comprehensive set of all new tax policies. For additional details on all new tax legisla- tion, see the biennial Joint Committee on Taxation Bluebook (e.g., Joint Committee on Taxation, 2013). 1 Economic Recovery Tax Act (ERTA) of 1981 2 Tax Equity and Fiscal Responsibility Act (TEFRA) of 1982 3 Omnibus Budget Reconciliation Act (OBRA) of 1990 4 Economic Growth and Tax Relief Reconciliation Act (EGTRRA) of 2001 5 Jobs Growth and Tax Relief Reconciliation Act (JGTRRA) of 2003 6 Economic Stimulus Act (ESA) of 2008 7 Emergency Economic Stabilization Act of 2008 8 American Recovery and Reinvestment Act (ARRA) of 2009 9 Tax Relief, Unemployment Insurance Reauthorization, and Job Creation Act (TRA) of 2010

Sources: Tax Policy Center (2011a,b) 316 National Tax Journal 3.11

–2.26 –2.60 –2.04 all HI In-kind + Post-tax + Post-tax HI –2.73 –2.85 2.96 –2.99 In-kind + Employer Post-tax + HI –1.40 –3.48 –1.54 –4.03 Top Quintile (Mean) Top Pre-tax + Employer HI –1.80 –3.58 –1.63 –4.42 Market + Employer 1.50 –1.31 all HI –12.16 1.27

In-kind + Post-tax + HI –4.32 –4.82 –4.62 –13.29

In-kind + Employer Post-tax + HI –7.73 –14.21

–7.61 –12.44 Pre-tax + Employer Bottom Quintile (Mean) Appendix Table A2 Table Appendix HI –25.33 –18.73 –19.38 –30.24 Market + Employer 3.11

–5.74 –1.18 –0.21 + all HI Post-tax + In-kind HI –6.58 –2.79 1.90 –3.77 In-kind + Employer Post-tax + HI Median Income –5.95 –3.43 –1.42 –6.72 Including In-Kind Transfers During the First Three Years of Economic Downturns, Downturns, of Economic Years Three During the First Transfers Including In-Kind Pre-tax + Employer Adding Employer- and Government-Provided Health Insurance to the Income Measure to Health Insurance and Government-Provided Employer- Adding Percent Change in the Pre-Tax Cash Household Income and Post-Tax Household Income, Household Income, Household Income and Post-Tax Cash Change in the Pre-Tax Percent HI –8.11 –5.33 –2.41 –9.86 Market + Employer Employer 1979–1982 1989–1992 2000–2003 2007–2010 Source: Authors’ calculations using Public-Use March CPS Data (1980–2011) Authors’ Source: Income Changes over the Great Recession Relative to Previous Recessions 317

Appendix Table A3 Accounting for the Annual Trend in Median Size-Adjusted Post-Tax Household Income of Persons During the Early 1980s Recession

1979–1980 1980–1981 1981–1982 1982–1983 (1) Percent change in median –3.89 –2.92 0.24 0.88 income Change accounted for by: (2) Age 0.05 0.08 0.02 0.04 (3) Race –0.09 –0.14 –0.06 –0.08 (4) Presence of other adult 0.01 0.08 0.20 –0.10 household members (5) Marriage –0.01 –0.11 0.05 –0.09

(6) Male-head employment1 –0.78 –0.26 –0.98 0.12 (7) Male-head earnings1 –2.20 –1.25 –0.45 –0.41 (8) Female-head employment1 0.12 0.18 0.14 0.30 (9) Female-head earnings1 0.07 –0.20 0.49 0.32 (10) Spouse’s earnings –0.49 0.20 0.00 –0.32 correlation

(11) Earnings of others –0.86 –0.34 –0.57 0.03 (12) Private non-labor income –0.02 0.48 –0.13 0.71

Subtotal, lines (2) through (12) –4.21 –1.29 –1.30 0.52

(13) Cash public transfers 0.23 –0.31 0.59 –0.20 (14) In-kind public transfers –0.05 –0.02 –0.07 0.00

(15) Policy-constant tax 3.93 1.87 0.94 0.30 liability changes (16) Tax liability changes due –3.78 –3.15 0.08 0.29 to new tax policies Notes: 1 Figures include household heads and the spouses of household heads. Source: Authors’ calculations using Public-Use March CPS Data (1980–2011) 318 National Tax Journal 1.50 73.09 68.78 –4.31 65.41 –1.15 64.26 62.92 64.42

0.62 64.21 64.83 White Hispanic / Income Ratio –489 1,035 Mean –987

18,349 20,501 –2,153 21,756 20,769 24,599 25,633

26,072 25,583 Income Hispanic

8.43 0.55 0.95 6.29 6.08 0.21

8.97 12.28

14.05 1.76 15.42

16.37 Percent Hispanic 64.08 66.15 –2.07 62.78 62.73 66.29 –0.05 65.63 –0.66 66.10 –1.08 65.02 Income Ratio Black / White Black /

201 Mean Black –607

26,115 17,094 18,555 –1,461 20,879 20,273 25,914 26,840 –1,180 25,660 Income

Appendix Table A4 Table Appendix 0.31 0.25 11.79 11.48 0.11

12.09 12.35 12.55

12.53 –0.02 12.68 12.79 Black Percent by Race(2010-$) Downturn Duringby Each Economic 699 –940

26,677 –1,374 28,051 33,259 32,319 39,092 39,791 40,605 39,464 –1,140 Income White Mean

81.92 –0.52 82.45 79.48 78.68 75.17 –0.80 73.43 –1.75 71.90 70.84 –1.06 White Percent Racial and Ethnic Characteristics of the U.S. Population and Size-Adjusted Household Income and Size-Adjusted (2010-$) Population Racial Characteristics and Ethnic of the U.S. 1982 Change 1979 calculations using Public-Use March CPS Data (1980–2011) Authors’ Source: 1989 1992 2000 Change 2003 Change 2007 2010 Change