This website uses cookies.

By clicking the "Accept" button or continuing to browse our site, you agree to first-party and session-only cookies being stored on your device to enhance site navigation and analyze site performance and traffic. For more information on our use of cookies, please see our Privacy Policy.

  1. Home
  2. ASSA Annual Meeting
  3. 2021
  4. Preliminary Program
  5. Poverty: Tax Credits and Tax Data
« Back to Results

Poverty: Tax Credits and Tax Data

Paper Session

Sunday, Jan. 3, 2021 3:45 PM - 5:45 PM (EST)

Hosted By: Society of Government Economists
  • Chair: Liana Fox, U.S. Census Bureau

The Antipoverty Impact of the EITC: New Estimates from Survey and Administrative Tax Records

Maggie R. Jones
,
U.S. Census Bureau
James P. Ziliak
,
University of Kentucky

Abstract

Evaluations of the EITC, including its antipoverty effectiveness, are based on simulated EITC benefits using either the Census Bureau’s tax module or from external tax simulators such as the National Bureau of Economic Research’s TAXSIM or Jon Bakija’s model. Each simulator utilizes model-based assumptions on who is and who is not eligible for the EITC, and conditional on eligibility, assumes that participation is 100 percent. However, recent evidence suggests that takeup of the EITC is considerably less than 100 percent, and thus claims regarding the impact of the program on measures of poverty may be overstated. We use data from the Current Population Survey Annual Social and Economic Supplement (CPS ASEC) linked to IRS tax data on the EITC to compare the distribution of EITC benefits from three tax simulation modules to administrative tax records. We find that significantly more actual EITC payments flow to childless tax units than predicted by the tax simulators, and to those whose family income places then well above official poverty thresholds. However, actual EITC payments appear to be target efficient at the individual tax unit level, whether correctly paid or not. We then compare the antipoverty impact of the EITC across the survey and administrative tax measures of EITC benefits. We find that in the full CPS ASEC the tax simulators overestimate the antipoverty effects of the EITC by about 1.8 million persons in a typical year. Restricting to a harmonized sample of filers, we find that the antipoverty estimates derived from the TAXSIM and Bakija models align more closely to actual EITC payments compared to the CPS, suggesting a discrepancy in assignment of tax filers between the tax simulators.

Do Older Americans Still Have More Income Than We Think? Evidence from the CPS ASEC Redesign

Adam Bee
,
U.S. Census Bureau
Joshua Mitchell
,
Welch Consulting

Abstract

The Current Population Survey Annual Social and Economic Supplement (CPS ASEC) is the source of the nation’s official household income and poverty statistics. In 2012, the CPS ASEC showed that median household income was $33,800 for householders aged 65 and over and the poverty rate was 9.1 percent for persons aged 65 and over. When we instead use an extensive array of administrative income records linked to the same CPS ASEC sample, we find that median household income was $44,400 (30 percent higher) and the poverty rate was just 6.9 percent. We demonstrate that large differences between survey and administrative record estimates are present within most demographic subgroups and are not easily explained by survey design features or processes such as imputation. Further, we show that the discrepancy is mainly attributable to underreporting of retirement income from defined benefit pensions and retirement account withdrawals. Using archived survey and administrative record data, we extend our analysis back to 1990 and provide evidence of underreporting from an earlier period. We also document a growing divergence over time between the two measures of median income which is in turn driven by the growth in retirement income underreporting. Turning to synthetic cohort analysis, we show that in recent years, most households do not experience substantial declines in total incomes upon retirement or any increases in poverty rates. Our results have important implications for assessing the relative value of different sources of income available to older Americans, including income from the nation’s largest retirement program, Social Security. We caution, however, that our findings apply to the population aged 65 and over in 2012 and cannot easily be extrapolated to future retirees.

Incorporating Administrative Data in Survey Weights for the Survey of Income and Program Participation

Jonathan Eggleston
,
U.S. Census Bureau
Ashley Westra
,
U.S. Census Bureau

Abstract

Declining survey response rates in federal surveys has led to concerns of increasing nonresponse bias in key government statistics. A potential solution is to leverage administrative data from federal agencies when constructing survey weights. This project performs initial research on incorporating administrative data into the weighting algorithm for the Survey of Income and Program Participation (SIPP). Specifically, we match income data from IRS tax forms and demographic data from the Social Security Administration and the Decennial Census to both respondents and nonrespondents. We then use this matched data in the household nonresponse adjustment of the SIPP weighting algorithm, which adjusts the weights of respondents to account for differential nonresponse rates among subpopulations and reduce nonresponse bias in survey estimates. We show how these new weights affect estimates of wealth, income, poverty, health insurance coverage, and participation in government assistance programs and their impact on nonresponse bias compared to the traditional weights. Overall, the new weights are associated with a small increase in estimated economic wellbeing.

Presence and Persistence of Poverty in Tax Data

Jeff Larrimore
,
Federal Reserve Board
David Splinter
,
Joint Committee on Taxation
Jake Mortenson
,
Joint Committee on Taxation

Abstract

This paper presents new estimates of the level and persistence of poverty among U.S. households since the Great Recession. We build annual household data files using U.S. income tax filings between 2007 and 2018. These data allow us to track individuals over time and measure how tax policies affect poverty trends. Using an after-tax household income measure, we estimate that while roughly 1 in 10 people are in poverty in any given year, over 4 in 10 people spent at least one year in poverty between 2007 and 2018. This implies substantial mobility in and out of poverty—for example, 41 percent of those in poverty in 2007 were out of poverty in the following year. Others spend multiple years in poverty or escape poverty only to fall back into it. Of those in poverty in 2007, one-third were in poverty for at least half of the years through 2018.
Discussant(s)
Jonathan L. Rothbaum
,
U.S. Census Bureau
Charles M. Hokayem
,
U.S. Census Bureau
JEL Classifications
  • H0 - General