"Household Incomes in Tax Data: Using Addresses to Move from Tax Unit to Household Income Distributions," with Jeff Larrimore and David Splinter -- January 2017. Finance and Economics Discussion Series: 2017-002. Divisions of Research & Statistics and Monetary Affairs, Federal Reserve Board, Washington, D.C.
Tax return data are increasingly the standard for tracking income statistics in the United States. However, these data have traditionally been limited by their inability to capture non-filers and to identify members of separate tax units living in the same household. We overcome these obstacles and create household records directly in the tax data using mailing address information included on tax forms. We then present the first set of tax-based household income and inequality measures for the entire income distribution. When comparing household income inequality results in the tax data to those using the March CPS, we confirm previous findings that the March CPS understates the inequality of household income. However, we also find that the previous approach of using tax units in the IRS data to proxy for households leads to an overstatement of household income inequality. Finally, using households in the IRS tax records, we illustrate how focusing on tax units rather than households alters the observed distribution of tax programs such as the Earned Income Tax Credit.
“Bunching to Maximize Tax Credits: Evidence from the U.S. Tax Schedule,” with Andrew Whitten -- November 2016. [Coverage from the Wall Street Journal here.]
We document bunching at tax kinks using a panel of 258 million income tax returns in the United States from 1996 to 2014. Though filed tax returns increased by only 20%, the number of bunchers at kinks in the ordinary income tax schedule grew by 700% – from 104,000 in 1996 to 834,000 in 2014. The strongest bunching occurs at kinks maximizing tax credits, particularly at the kink that maximizes taxpayer refunds. Though the self-employed are disproportionately likely to bunch, a substantial number of wage earners also bunch. However, wage-earner bunching is driven exclusively by misreporting wage and tip income.
“The Effect of Required Minimum Distribution Rules on Withdrawals from Individual Retirement Accounts,” with Heidi Schramm and Andrew Whitten -- November 2016. [Online appendix here.]
Traditional Individual Retirement Accounts (IRAs) are a substantial source of retirement savings. In 2013 individuals age 60 or older held $3.8 trillion in wealth in IRAs. Current law requires that some fraction of these funds be withdrawn each year beginning the year one turns 70 and one half years of age, with the fraction increasing in age. We study the effects of these Required Minimum Distribution (RMD) rules on the asset decumulation behavior of retirees using a panel of administrative tax data from 1999 to 2014. This period encompasses a one-year suspension of the RMD rules in 2009, which we exploit for identification. Though the RMD rules are modest – leaving one third of the original balance intact by age 90 even if investments generate zero returns – they have substantive effects on behavior. We estimate that 52% of individuals subject to the rules would prefer to withdraw less than their required minimum but that over one third of these RMD-constrained individuals do not adjust their withdrawals in response to temporary changes in the rules. As further evidence that paying attention to the rules is costly, we find that individuals newly subject to RMD rules are disproportionately likely to close their IRAs. This paper is the first to study this effect of RMD rules and to show that the rules represent a binding constraint for the majority of IRA holders.
"All Income is Not Created Equal: Cross-Tax Elasticities in the US," -- Draft as of March 2016.
In the United States, marginal tax rates on capital income and ordinary income (e.g. wages and salaries) can vary substantially. The same household might face a federal tax rate on their wage income that is 20 percentage points higher than the tax rate on long-term capital gains. This differential complicates welfare and revenue analyses of tax reforms, as individuals might alter their capital income decisions in response to an ordinary income tax change. In this paper, I estimate cross-tax elasticities for capital and ordinary income, using a large non-public panel of federal income tax returns from 1997-2007. I find two key results, though both are sensitive to a variety of specification decisions. First, capital gains respond to both the capital gains tax rate and the ordinary income tax rate. Second, I find mixed evidence that ordinary income responds to the ordinary income tax rate and the long-term capital gains tax rate. These results suggest cross-tax responses between these two bases are important, and should not be ignored when estimating taxpayer responses in the United States.
"Income and Earnings Mobility in U.S. Tax Data," with Jeff Larrimore and David Splinter, Board of Governors of the Federal Reserve System: 2015 Community Development Research Conference, pages 482-516. [Previous version: Fed Finance and Economics Discussion Series 2015-061.]
"Recent Income Trends for Top Executives: Evidence from Tax Return Data," National Tax Journal, December 2013, 66 (4), 913–938, with Seth H. Giertz.
"Attaching the Left Tail: A New Profile of Income for Persons who do not Appear on Federal Income Tax Returns," in the 2009 National Tax Association Proceedings, with James Cilke, Jonathan Zytnick, and Michael Udell. [Coverage from Forbes here, and Tax Policy Center here.]
"The Effect of Recent Tax Changes on Taxable Income: Correction and Update,'" with Bradley Heim, Journal of Policy Analysis and Management.