Public Economics Program Meeting

November 3-4, 2011
Raj Chetty of Harvard University and Emmanuel Saez of University of California at Berkeley, Organizers

Timothy Dowd, Joint Committee on Taxation, and Robert McClelland and Athiphat Muthitacharoen, Congressional Budget Office
The Tax Elasticity of Capital Gains in the 21st Century

There has been wide variability in estimates of the elasticity of capital gains realizations to their marginal tax rate. In 1994, Burman and Randolph offered evidence that the variation was partially due to the difference between permanent and transitory tax changes: using panel data for 1979-83, they estimated a very low elasticity of "permanent" tax rates, -0.18. However, their estimate was not significantly different from both 0.00 and -1.00. Dowd, McClelland and Muthitacharoen use the same model with a panel of taxpayers during 1999-2008. They concentrate their analysis on personal capital gains realizations, but they also examine total realizations, from both pass-through entities and mutual funds. They improve the precision of their methodology by adding variables to the equation that models the decision to realize gains, and they conduct several sensitivity analyses. The resulting elasticity estimates vary from -0.814 to -1.10, although the elasticity of capital gains from pass-through entities is approximately -1.70 while the elasticity of capital gains from mutual funds is closer to -0.20.

Peter J. Brady, Investment Company Institute, and Kevin Pierce, Statistics of Income, Internal Revenue Service
Using Panel Tax Data to Examine the Transition to Retirement

The data used in the study by Brady and Pierce are derived from the SOI 1999-2008 Edited Panel of individual income tax returns. Additional data are obtained by linking individual taxpayers in the SOI sample to information returns. Specifically, taxpayers are linked to any Form W2, Form 1099-SSA, Form 1099-R, and Form 5498 that are filed on their behalf. With this information, individual taxpayers are observed both before and after they first receive Social Security retirement benefits. Because individual taxpayers are linked to information return data, all individuals in the sample can be followed even if they no longer file a tax return. The data are used to examine the relative importance of various income components such as wage and salary compensation, Social Security benefits, pension benefits, and investment income both before and after individuals claim Social Security. In addition, the data are used to determine the relationship of total income after claiming to total income prior to claiming. On average, taxpayers in the sample do not experience a drop in income after claiming Social Security retirement benefits. In addition, the incidence and the amount of pension, annuity, and IRA distributions are considerably higher in the administrative tax data than reported in survey data.

Nicholas Turner, Department of the Treasury
Do Students Profit from For-Profit Education? Estimating the Returns to Postsecondary Education with Tax Data

Turner uses administrative panel data on earnings and educational attendance from the Internal Revenue Service to estimate the differential earnings effect of attending a not-for-profit post-secondary institution, relative to a for-profit institution. The results suggest that there is a substantial differential increase in earnings after enrollment, relative to years prior to enrollment, for individuals who attend not-for-profit institutions, compared to individuals who attend for-profit schools. The results also suggest that earnings decrease during enrollment, compared to years before enrollment, but that this decline is relatively smaller for individuals who attend not-for-profit institutions. Given that for-profit institutions cost more on average than not-for-profit institutions, these results imply that students may realize better long-run financial outcomes from attending not-for-profits institutions.

Raj Chetty and John Friedman, Harvard University and NBER; Emmanuel Saez; and Nathaniel Hilger and Danny Yagan, Harvard University
The IRS Databank: Developing a Population Panel Dataset for Tax Policy Research

Using population-wide U.S. administrative tax data since 1996, Chetty and his co-authors develop a new method of estimating responses of earnings to taxation by exploiting differences across neighborhoods in knowledge about the Earned Income Tax Credit (EITC). ) Using self-employment income, the authors first proxy knowledge about the EITC with the fraction of its beneficiaries bunched at the first kink of the EITC (which maximizes their tax refund). Because bunching is primarily attributable to non-compliance and appears to be driven by knowledge, bunching spreads spatially over time and is higher in high-EITC-density places. EITC recipients who move to high bunching places also start bunching; recipients moving out of high bunching places continue to bunch. Wage earnings--which cannot be misreported easily -- also respond to the EITC in high knowledge places. Wage earnings densities have more mass around the EITC plateau in high knowledge places. Also, using the fact that the EITC increases sharply after a first birth, the researchers find that wage earnings after a birth are more concentrated around the EITC plateau in high knowledge places (relative to low knowledge places). This labor supply response comes both from the extensive margin (higher likelihood of working) and from the intensive margin (increase in earnings for those starting in the phase-in and decrease in earnings for those starting in the phase-out). The total elasticity of earnings with respect to the net-of-tax rate in high bunching places is high but still relatively modest on average countrywide.

Thomas Piketty, Paris School of Economics, and Emmanuel Saez
A Theory of Optimal Capital Taxation

Piketty and Saez develop a realistic, tractable normative theory of socially-optimal capital taxation. They present a dynamic model of savings and bequests with heterogeneous tastes for bequests to children and for wealth accumulation per se. They derive formulas for optimal inheritance tax rates expressed in terms of estimable parameters and social preferences. The long-run optimal inheritance tax rate increases with the aggregate steady-state flow of inheritances to output, decreases with the elasticity of bequests to the net-of-tax rate, and decreases with the strength of preferences for leaving bequests. For realistic parameters, the optimal inheritance tax rate should be as high as 50-60 percent -or even higher for top bequests- if the government has meritocratic preferences (that is, puts higher welfare weights on those receiving little inheritance). In contrast to the Atkinson-Stiglitz result, bequest taxation remains desirable in this model even with optimal labor taxation because inequality is two-dimensional: with inheritances, labor income is no longer the unique determinant of lifetime resources. In contrast to Chamley-Judd, optimal inheritance taxation is desirable because these preferences allow for Â…finite long-run elasticities of inheritance to tax rates. Finally, the authors discuss how capital market imperfections and uninsurable idiosyncratic shocks to rates of return can justify shifting one-off inheritance taxation toward lifetime capital taxation, and can account for the actual structure and mix of inheritance and capital taxation.

Philippe Aghion and William Kerr, Harvard University and NBER; Ufuk Akcigit, University of Pennsylvania and NBER; and Julia Cage, Harvard University

Taxation, Corruption, and Growth

Aghion, Akcigit, Cage, and Kerr study the impact of different types of taxation on aggregate productivity growth and firm dynamics. To analyze the growth-tax relationship, they construct a model that emphasizes two sides of taxation: the disincentive effect of taxes for effort (both labor supply and entrepreneurial undertakings) versus the provision of public goods necessary for growth. They consider two types of policy instruments: corporate profit tax and household income tax. They show that growth has an inverted-U relationship to taxation because of these competing disincentive and public good effects. Moreover, for any given level of infrastructure, while entrants always respond negatively to taxation, incumbents may respond positively because of the reduction in the entry rate induced by the tax increase. The authors then calibrate their model using U.S. Census Bureau data covering 1982-2007. The model suggests that increases in personal income taxation from the current level would be growth enhancing. To test these predictions, they consider panel variation in state-level taxation rates and growth outcomes over 1982-2007. They find that forward growth shows a concave relationship to taxation in estimations that include or exclude state fixed effects. Moving from the 10th to 30th percentile of this panel variation is associated with higher growth of 0.2%-0.6%; this rate decreases until shifts from the 70th to the 90th percentile show very minimal or no gains. The fitted relationship suggests that the potential gain from increasing the tax level from its 2002-2007 (pre-crisis) level would be on the order of increasing growth from 2 percent to 2.1 percent. On the other hand, declines in taxes from their current level could reduce growth more dramatically because of the fitted relationship's concavity. The empirical analysis also examines predictions the model makes regarding entrant shares and average firm size. It most notably verifies out-of-sample two of the key model predictions: that higher taxation is associated with larger average establishment size and reduced entrant shares in employment.

Nikolaos Artavanis, Virginia Polytechnic Institute, and Adair Morse and Margarita Tsoutsoura, University of Chicago
A New Method to Estimate Tax Evasion Using Financial Institution Lending: The Case of Greece

Artavanis, Morse, and Tsoutsoura introduce a new methodology to estimate tax evasion based on households' ability to borrow. In many developed economies, a large percentage of the population earns at least some income in actions unreported to tax authorities. Pervasive tax evasion implies that banks adapt credit models, adjusting households' reported income to reflect their perception of true effect income. The authors use detailed individual loan application data in Greece to quantify the extent of tax evasion country-wide. Their estimates suggest that 18 to 26 billion euros in taxable income goes unreported and that the foregone revenues for the tax authorities amount to 27 to 41 percent of the deficit. They then document the incidence of tax evasion by income class and occupation. They find that the upper income accounts for a large portion of tax evasion implied by bank adaptation. Doctors and many professional service occupations are the largest offenders. Finally, they discuss the occupational distribution in light of paper trail mandates and theories as to the optimal level of tax subsidies.

Saurabh Bhargava, University of Chicago, and Dayanand S. Manoli, University of Los Angeles and NBER

Why Are Benefits Left on the Table? Assessing Incomplete Take-up with an IRS Field Experiment

Why do some individuals fail to collect the social or private benefits for which they are eligible? Bhargava and Manoli attempt to identify concrete strategies to improve take-up with a unique field experiment in collaboration with the IRS. Specifically, they simultaneously test the role of program information (about benefits, rules, and costs), informational complexity, and stigma on the response to a set of experimental IRS mailings that notify eligible individuals of over $26 million in owed benefits. The researchers distribute these mailings -- that is, eligibility notices, claiming worksheets, and envelopes -- to 35,050 California taxpayers who filed their taxes but did not claim their TY 2009 EITC despite presumed eligibility. Among the eligible filers whose baseline take-up is 90 percent, the authors find residual increases in take-up attributable to: the mere receipt of an experimental mailing (a response of 0.14); a reduction in complexity (+0.09 relative to 0.14); and the display of benefit information (+0.08 relative to a 0.23 baseline). They find no evidence to implicate the misperception of claiming costs or audit likelihood, or program stigma, as causes of incomplete take-up in this context. A novel survey of program-eligible tax filers confirms pervasive low awareness and misunderstanding of program incentives, while psychometric evidence illuminates possible underlying mechanisms. The authors project that the tested interventions, if applied to all filing non-claimants, could reduce incomplete take-up among filers from 10 percent to 7 percent, and overall from 25 percent to 22 percent.

Casey Rothschild, Wellesley College, and Florian Scheuer, Stanford University and NBER

Optimal Taxation with Rent-Seeking (NBER Working Paper No. 17035)

Recent policy proposals have suggested taxing top incomes at very high rates on the grounds that some or all of the highest wage earners are engaged in socially unproductive or counterproductive activities, such as externality imposing speculation in the financial sector. To address this, Rothschild and Scheuer provide a model in which agents can choose between working in a traditional sector, where private and social products coincide, and a crowdable rent-seeking sector, where some or all of earned income reflects the capture of pre-existing output rather than increased production. They characterize Pareto optimal linear and non-linear income tax systems under the assumption that the social planner cannot or does not observe whether any given individual is a traditional worker or a rent-seeker. They find that optimal marginal taxes on the highest wage earners can remain modest even if all high earners are socially unproductive rent-seekers and the government has a strong intrinsic desire for progressive redistribution. Intuitively, taxing their effort at a lower rate stimulates their rent-seeking efforts, thereby keeping private returns for other potential rent-seekers low and discouraging further entry.

Ilyana Kuziemko, Princeton University and NBER; Ryan Buell and Michael I. Norton Harvard Business School; and Taly Reich, Stanford Graduate School of Business

Last-place Aversion: Evidence and Redistributive Implications (NBER Working Paper No. 17234)

Kuziemko, Buell, Reich, and Norton present evidence from laboratory experiments showing that individuals are "last place averse." Participants choose gambles with the potential to move them out of last place which they reject when randomly placed in other parts of the distribution. In money-transfer games, those randomly placed in second-to-last place are the least likely to costlessly give money to the player one rank below. Last-place aversion suggests that low-income individuals might oppose redistribution because it could differentially help the group just beneath them. Using survey data, the authors show that individuals making just above the minimum wage are the most likely to oppose its increase.

Dina Pomeranz, Harvard University

No Taxation without Information Deterrence and Self-Enforcement in the Value Added Tax

Tax evasion generates billions of dollars of losses in government revenue and creates large distortions, especially in developing countries. A growing, mostly theoretical, literature argues that information flows are central to understanding effective taxation. Pomeranz analyzes the role of information for tax enforcement in the case of the Value Added Tax (VAT) through two randomized field experiments with over 445,000 Chilean firms. Claims that the VAT facilitates tax enforcement by generating a paper trail on transactions between firms have led to widespread VAT adoption worldwide, but there is surprisingly little evidence. She finds that the paper trail leads to spillovers that create important multiplier effects in tax enforcement. The impact of a random audit announcement is transmitted up the VAT chain, increasing compliance by firms' suppliers. A second experiment finds that the paper trail acts as a substitute to a firm's own audit risk. A message announcing increased tax enforcement has a much smaller effect on reporting of transactions that are already covered by a paper trail. These findings confirm that when evasion is taken into account, significant differences emerge between taxes that are equivalent in standard models but generate different information on taxable transactions.

Brian G. Knight, Brown University and NBER

State Gun Policy and Cross-State Externalities: Evidence from Crime Gun Tracing (NBER Working Paper No. 17469)

Knight provides a theoretical and empirical analysis of cross-state externalities associated with gun regulations in the context of the gun trafficking market. Using gun tracing data, which identify the source state for crime guns recovered in destination states, he finds that firearms tend to flow from states with weak gun laws to states with strict gun laws, satisfying a necessary condition for the existence of cross-state externalities in the theoretical model. He also finds an important role for transportation costs in this market, with gun flows more significant between nearby states; this finding suggests that externalities are spatial in nature. Finally, he presents evidence that criminal possession of guns is higher in states exposed to weak gun laws in nearby states.

NBER Videos

National Bureau of Economic Research, 1050 Massachusetts Ave., Cambridge, MA 02138; 617-868-3900; email:

Contact Us