Tax refunds and income manipulation: evidence from the EITC

Abstract

Welfare programs are important in terms of reducing poverty, although they create incentives for recipients to maximize their income by either reducing their labor supply or manipulating their taxable income. In this paper, we quantify the extent of such behavioral responses for the earned income tax credit (EITC) in the USA. We exploit the fact that US states can set top-up rates, which means that at a given point in time, workers with the same income receive different tax refunds in different states. Using event studies as well as a border pair design, we document that raising the state EITC leads to more bunching of self-employed tax filers at the first kink point of the tax schedule. While we document a strong relationship up until 2007, we find no effect during the Great Recession. These findings point to important behavioral responses to the largest welfare program in the USA.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5

Notes

  1. 1.

    For surveys, see Hotz and Scholz (2003), Eissa and Hoynes (2006), Meyer (2010) and Nichols and Rothstein (2016).

  2. 2.

    A key result of the existing literature on labor supply reactions to the EITC is that there are positive effects at the extensive margin (Eissa and Liebman 1996; Meyer and Rosenbaum 2001; Grogger 2003; Hotz and Scholz 2006; Gelber and Mitchell 2012). This result, which was found primarily for single mothers, does not hold true for secondary wage earners, for whom Eissa and Hoynes (2004) find a decrease in participation. In contrast to these findings, previous research suggests that there are none or only small effects at the intensive margin (Rothstein 2010; Chetty and Saez 2013).

  3. 3.

    See, e.g., Hausman (1985), Pencavel (1986), Killingsworth and Heckman (1986), Heckman (1993), Blundell and MaCurdy (1999), Meghir and Phillips (2008), Keane (2011), Keane and Rogerson (2012), McClelland and Mok (2012) and Bargain et al. (2014).

  4. 4.

    See Fig. 6 in Appendix A for an illustration. For families with two children, the kink points for 2009 are at $12,570 and $16,420. The maximum tax credit is $5,028, which results in steeper phase-in and phaseout regions compared to the schedule for families with one child.

  5. 5.

    Wisconsin has a top-up rate of zero for childless people, but top-up rates of 4, 14 and 43% for families with one, two and three and more children, respectively.

  6. 6.

    We are aware that DC is technically not a state. However, it has its own EITC.

  7. 7.

    To put these numbers into perspective, in 2009, the total number of people with income from self-employment was 16.8 million, which represents 10.7% of the workforce (Source: Social Security Administration). According to Chetty et al. (2013), the share of self-employed EITC claimants was 19.6%, whereas the share of EITC-eligible filers among all tax filers was 18.9% (Source: Brookings Institution, Characteristics of EITC-eligible tax units 2015). Therefore, the share of filers that were both eligible for the EITC and had income from self-employment was around 3.7%.

  8. 8.

    For this reason, our analysis spans these years, although in the future it would be desirable to have data past 2010, which would allow us to study the effects of the EITC during and after the Great Recession. In Appendix D, we explain in greater detail how we convert zip-code-level information to the county level.

  9. 9.

    As explained in footnote 14 in Chetty et al. (2013), the results are robust to (i) defining the denominator of the bunching measure using only self-employed individuals rather than the full population, (ii) the choice of bandwidth around the kink point and (iii) a measure whereby bunching is measured as the excess mass over a smoothly fitted polynomial within a certain bandwidth.

  10. 10.

    See Feenberg and Coutts (1993) for a documentation.

  11. 11.

    In our main analysis in Sect. 5, these county pairs will be included. We also performed the event study including these cases. The results remain unchanged. The tables are available from the authors upon request.

  12. 12.

    This approach—controlling for leads and lags as well as year fixed effects—is similar to that used by Jäger (2016) and Fuest et al. (2018).

  13. 13.

    Similar approaches have been used by Dube et al. (2010) to evaluate changes in minimum wages in the USA, and by Lichter et al. (2015) to estimate the impact of government surveillance in East Germany.

  14. 14.

    This result is also consistent with the findings of Castanheira et al. (2012) for income tax reforms in Europe and Foremny and Riedel (2014) for local business taxes in Germany. Both studies show that tax setting is driven by political factors rather than the business cycle.

  15. 15.

    While these two dummies are multicollinear, it is possible to include these interactions in the regression because we do not include the dummies on their own.

  16. 16.

    Sources: minimum wages: St. Louis Fed, welfare benefits: welfare rules databook, tax revenue: Annual Survey of State Government Tax Collections, consumer price index: St. Louis Fed, marginal income tax rates: NBER TAXSIM.

  17. 17.

    We found splitting the number of claimants evenly between counties the most transparent way of converting zip-code-level data to county-level data. It would also be possible to (dis-)aggregate the numbers based on population measures. However, without further assumptions, this would only be possible for disaggregation (one zip code contains more than one county), but not for aggregation (one county contains more than one zip code).

  18. 18.

    This procedure follows Kennedy (1995) and Chetty et al. (2009).

References

  1. Agrawal, D. R., & Hoyt, W. H. (2017). Commuting and taxes: Theory, empirics, and welfare implications. The Economic Journal. https://doi.org/10.1111/ecoj.12550.

    Article  Google Scholar 

  2. Bargain, O., Orsini, K., & Peichl, A. (2014). Comparing labor supply elasticities in Europe and the United States new results. Journal of Human Resources, 49, 723–838.

    Article  Google Scholar 

  3. Bargain, O., & Peichl, A. (2016). Own-wage labor supply elasticities: Variation across time and estimation methods. IZA Journal of Labor Economics, 5, 10.

    Article  Google Scholar 

  4. Bastian, J. (2017). Unintended consequences? More marriage, more children, and the EITC. Working Paper.

  5. Bastian, J., & Michelmore, K. (2018). The long-term impact of the earned income tax credit on children’s education and employment outcomes. Journal of Labor Economics. https://doi.org/10.1086/697477.

    Article  Google Scholar 

  6. Bertrand, M., Duflo, E., & Mullainathan, S. (2010). How much should we trust differences-in-differences estimates? The Quarterly Journal of Economics, 119, 249–275.

    Article  Google Scholar 

  7. Bitler, M., Hoynes, H., & Kuka, E. (2017). Child poverty, the great recession, and the social safety net in the United States. Journal of Policy Analysis and Management, 36, 358–389.

    Article  Google Scholar 

  8. Blau, F. D., & Kahn, L. M. (2007). Changes in the labor supply behavior of married women: 1980–2000. Journal of Labor Economics, 25, 393–438.

    Article  Google Scholar 

  9. Blundell, R., & MaCurdy, T. (1999). Labor supply: A review of alternative approaches. Handbook of Labor Economics, 3, 1559–1695.

    Article  Google Scholar 

  10. Castanheira, M., Nicodème, G., & Profeta, P. (2012). On the political economics of tax reforms: Survey and empirical assessment. International Tax and Public Finance, 19, 598–624.

    Article  Google Scholar 

  11. Chetty, R., Friedman, J. N., & Saez, E. (2013). Using differences in knowledge across neighborhoods to uncover the impacts of the EITC on earnings. American Economic Review, 103, 2683–2721.

    Article  Google Scholar 

  12. Chetty, R., Looney, A., & Kroft, K. (2009). Salience and taxation: Theory and evidence. American Economic Review, 99, 1145–1177.

    Article  Google Scholar 

  13. Chetty, R., & Saez, E. (2013). Teaching the tax code: Earnings responses to an experiment with EITC recipients. American Economic Journal: Applied Economics, 5, 1–31.

    Google Scholar 

  14. Correia, S. (2015). Singletons, cluster-robust standard errors and fixed effects: A bad mix. Federal Reserve Board of Governors, mimeo.

  15. Dube, A., Lester, T. W., & Reich, M. (2010). Minimum wage effects across state borders: Estimates using contiguous counties. Review of Economics & Statistics, 92, 945–964.

    Article  Google Scholar 

  16. Eissa, N., & Hoynes, H. W. (2004). Taxes and the labor market participation of married couples: The earned income tax credit. Journal of Public Economics, 88, 1931–1958.

    Article  Google Scholar 

  17. Eissa, N., & Hoynes, H. W. (2006). Behavioral responses to taxes: Lessons from the EITC and labor supply. In J. Poterba (Ed.), Tax Policy and the Economy, 20, 73–110.

  18. Eissa, N., & Liebman, J. B. (1996). Labor supply response to the earned income tax credit. The Quarterly Journal of Economics, 111, 605–637.

    Article  Google Scholar 

  19. Evers, M., Mooij, R. D., & Vuuren, D. V. (2008). The wage elasticity of labour supply: A synthesis of empirical estimates. De Economist, 156, 25–43.

    Article  Google Scholar 

  20. Feenberg, D., & Coutts, E. (1993). An introduction to the taxsim model. Journal of Policy Analysis and Management, 12, 189–194.

    Article  Google Scholar 

  21. Feldstein, M. (1995). The effect of marginal tax rates on taxable income: A panel study of the 1986 tax reform act. The Journal of Political Economy, 103, 551–572.

    Article  Google Scholar 

  22. Feldstein, M. (1999). Tax avoidance and the deadweight loss of the income tax. Review of Economics & Statistics, 81, 674–680.

    Article  Google Scholar 

  23. Foremny, D., & Riedel, N. (2014). Business taxes and the electoral cycle. Journal of Public Economics, 115, 48–61.

    Article  Google Scholar 

  24. Fuchs, V. R., Krueger, A. B., & Poterba, J. M. (1998). Economists’ views about parameters, values, and policies: Survey results in labor and public economics. Journal of Economic Literature, 36, 1387–1425.

    Google Scholar 

  25. Fuest, C., Peichl, A., & Siegloch, S. (2018). Do higher corporate taxes reduce wages? Micro evidence from Germany. American Economic Review, 108, 393–418.

    Article  Google Scholar 

  26. Gelber, A. M., & Mitchell, J. W. (2012). Taxes and time allocation: Evidence from single women and men. Review of Economic Studies, 79, 863–897.

    Article  Google Scholar 

  27. Grogger, J. (2003). The effects of time limits, the EITC, and other policy changes on welfare use, work, and income among female-headed families. Review of Economics & Statistics, 85, 394–408.

    Article  Google Scholar 

  28. Hargaden, E. (2015). Taxpayer responses over the cycle: Evidence from Irish notches. University of Tennessee, mimeo.

  29. Hausman, J. A. (1985). Taxes and labor supply. Handbook of Public Economics, 1, 213–263.

    Article  Google Scholar 

  30. Heckman, J. J. (1993). What has been learned about labor supply in the past twenty years? American Economic Review, 83, 116–121.

    Google Scholar 

  31. Heim, B. T. (2007). The incredible shrinking elasticities married female labor supply, 1978–2002. Journal of Human Resources, 42, 881–918.

    Article  Google Scholar 

  32. Hotz, J. V., & Scholz, J. K. (2003). The earned income tax credit. In R. A. Moffitt (Ed.), In means-tested transfer programs in the United States (pp. 141–198).

  33. Hotz, J. V., & Scholz, J. K. (2006). Examining the effect of the earned income tax credit on the labor market participation of families on welfare. NBER Working Paper 11968.

  34. Hoynes, H. W., Miller, D., & Simon, D. (2015). Income, the earned income tax credit, and infant health. American Economic Journal: Economic Policy, 7, 172–211.

    Google Scholar 

  35. Hoynes, H. W., & Patel, A. J. (2015). Effective policy for reducing inequality? The earned income tax credit and the distribution of income. NBER Working Paper 21340.

  36. Jäger, S. (2016). How substitutable are workers? Evidence from worker deaths. Discussion Paper, Harvard University, mimeo, job Market Paper.

  37. Jones, M. R. (2014). Changes in EITC eligibility and participation, 2005–2009. CARRA Working Paper 2014-04.

  38. Keane, M. P. (2011). Labor supply and taxes: A survey. Journal of Economic Literature, 49, 961–1075.

    Article  Google Scholar 

  39. Keane, M. P., & Rogerson, R. (2012). Micro and macro labor supply elasticities: A reassessment of conventional wisdom. Journal of Economic Literature, 50, 464–476.

    Article  Google Scholar 

  40. Kennedy, P. E. (1995). Randomization tests in econometrics. Journal of Business and Economic Statistics, 13, 85–95.

    Google Scholar 

  41. Killingsworth, M. R., & Heckman, J. J. (1986). Female labor supply: A survey. Handbook of Labor Economics, 1, 103–204.

    Article  Google Scholar 

  42. Kleven, H. J., & Schultz, E. A. (2014). Estimating taxable income responses using Danish tax reforms. American Economic Journal: Economic Policy, 6, 271–301.

    Google Scholar 

  43. LaLumia, S. (2009). The earned income tax credit and reported self-employment income. National Tax Journal, 62, 191–217.

    Article  Google Scholar 

  44. Lichter, A., Löffler, M., & Siegloch, S. (2015). The economic costs of mass surveillance: Insights from Stasi spying in east Germany. IZA Discussion Paper 9245.

  45. McClelland, R., & Mok, S. (2012). A review of recent research on labor supply elasticities. CBO Working Paper 43675.

  46. Meghir, C., & Phillips, D. (2008). Labour supply and taxes. IFS Working Papers 8.

  47. Meyer, B. D. (2010). The effects of the earned income tax credit and recent reforms. In J. R. Brown (Ed.), Tax Policy and the Economy, 24, 153–180.

  48. Meyer, B., & Rosenbaum, D. (2001). Welfare, the earned income tax credit, and the labor supply of single mothers. The Quarterly Journal of Economics, 116, 1063–1114.

    Article  Google Scholar 

  49. Moffitt, R. A. (2013). The great recession and the social safety net. The Annals of the American Academy of Political and Social Science, 650, 143–166.

    Article  Google Scholar 

  50. Neumark, D., & Williams, K. E. (2016). Do state earned income tax credits increase participation in the federal EITC? Discussion Paper.

  51. Nichols, A., & Rothstein, J. (2016). The earned income tax credit. In R. A. Moffitt (Ed.), Economics of Means-Tested Transfer Programs in the United States, 1, 137–218.

  52. Pencavel, J. (1986). Labor supply of men: A survey. Handbook of Labor Economics, 1, 3–102.

    Article  Google Scholar 

  53. Rothstein, J. (2010). Is the EITC as good as an NIT? Conditional cash transfers and tax incidence. American Economic Journal: Economic Policy, 2, 177–208.

    Google Scholar 

  54. Saez, E. (2010). Do taxpayers bunch at kink points? American Economic Journal: Economic Policy, 2, 180–205.

    Google Scholar 

  55. Saez, E., Slemrod, J., & Giertz, S. H. (2010). The elasticity of taxable income with respect to marginal tax rates: A critical review. Journal of Economic Literature, 50, 3–50.

    Article  Google Scholar 

Download references

Acknowledgements

We would like to thank the editor, two anonymous referees, David Agrawal, Jacob Bastian, Manasi Deshpande, as well as audiences at IZA, ZEW, the IIPF 2017 in Tokyo, and the NTA 2017 in Philadelphia for their helpful comments.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Benjamin Elsner.

Appendices

Appendix A: The EITC tax schedule

Figure 6 illustrates the EITC tax refund schedule for families with one and two children. The refunds refer to 2009, the last year in our sample.

Fig. 6
figure6

The EITC schedule in 2009. Notes: This graph displays the relationship between the tax refund and household labor income according to the 2009 federal EITC schedule. Tax units with adjusted gross income above the earned income threshold are not eligible. First EITC kink point for families with one child: USD 8950; for families with two children USD 12,570. Second kink point at USD 16,420

Appendix B: Predicting EITC expansions

Our identification strategy relies on the assumption that the top-up rate in a state is uncorrelated with county or state characteristics. A central concern with this assumption is that the generosity of the state EITC is driven by the business cycle, state-level fluctuations in tax revenue, or changes in minimum wages. To address this concern, we follow Bastian and Michelmore (2018), and predict the level of the state EITC based on current and lagged state-level economic variables in a panel regression. If any of the variables turned out to be statistically significant, this would be reason for concern, as it would cast doubt on the validity of the identifying assumption.

For this purpose, we collected state-level data on the welfare state (top marginal income tax rate, level of minimum wage, monthly welfare benefits), as well as tax revenues, which can be seen as a measure of the business cycle. The data span the years 1995–2009.Footnote 16 The regression results are shown in Table 4. Given that statistically insignificant results are more likely when standard errors are clustered, we report here conventional standard errors. None of the regressors is statistically significantly different from zero, which we interpret as strong evidence that changes in the state EITC are not driven by state-level fluctuations in the economy.

Table 4 OLS results: predictors of State EITC top-up rates

Appendix C: EITC claimants before, during and after the Great Recession

Figure 7 displays the number of EITC claimants around the time of the Great Recession, between 2007 and 2012. This number has been increasing throughout, although the increase was strongest during the Great Recession, between 2008 and 2009.

Fig. 7
figure7

Source: IRS

EITC claimers, 2007–2012. Notes: This graph displays the number of people in the USA claiming the EITC in a given year

Appendix D: Converting zip-code-level data to county-level data

The dataset by Chetty et al. (2013) provides data at the level of three-digit zip codes. Because the border pair design requires information at the county level, we convert the data from the zip code to the county level. The dataset mainly consists of absolute numbers, such as the number of EITC claimants in a given zip code. If a zip code comprises more than one county, we divide the absolute numbers evenly across all counties within a zip code. For example, if there are 1000 claimants in zip code A and A consists of two counties, we assign each county 500 claimants. If, on the other hand, a county is part of more than one zip code, we assign this county the sum of the absolute numbers. If the zip code that cuts through a county also covers another county, we split the absolute numbers between these countries before adding up within counties. For example, if zip codes A (1000 claimants) and B (500 claimants) are completely contained in county X, we assign county X 1500 claimants. If, however, zip code A also covers another county while B is fully contained in X, we assign county X 500 claimants from A and 500 claimants from B.Footnote 17

For the 3141 counties in our dataset, we apply the first method—split the numbers between counties within a zip code—to 1179 counties. For another 1960 counties, we apply both methods, namely we split numbers between counties as well as aggregate numbers within counties. The remaining two counties coincide with the zip codes.

Appendix E: Identifying variation

Table 5 displays the amount of variation—measured by the standard deviation—in the most important variables for different samples as well as different fixed effect specifications. Column (1) displays the variation for all counties, whereas Columns (2)–(4) display the variation for border counties only. In the border pair sample, some counties appear more than once if they have more than one neighbor in a different state. Going from left to right, one can see that the amount of variation is reduced as more fixed effects are added. However, even after controlling for pair-by-year fixed effects, there remains substantial variation in top-up rates as well as the outcome variables.

Table 5 Variation in key variables

Figure 8 further illustrates the relationship between state-specific top-up rate (horizontal axis) and the degree of bunching (vertical axis) in a binned scatter with ten equally sized bins on each axis. The graph controls for state-specific characteristics of the EITC—a dummy that equals unity if the refund depends on the number of children, and a dummy that equals unity if a positive refund is given if a person’s tax credit exceeds his/her tax liability—as well as pair-by-year fixed effects. The regression line corresponds to the regression coefficient in Table 2, Panel A), Column (4).

Fig. 8
figure8

Bunching versus top-up rates. Notes: This graph displays the relationship between the share of self-employed at the first kink point of the EITC and the state-specific top-up rates in a binned scatter, whereby each variable is divided in ten equally sized bins. Both variables have been demeaned by pair-by-year fixed effects, and we control for state-level features of the EITC

Appendix F: Assessing inference through permutation tests

While the border design facilitates estimating a causal effect by providing clear treatment and control counties, it also complicates statistical inference. The error terms can be correlated across space as well as within counties over time, which can lead to an underestimation of standard errors, and an under-rejection of the null hypothesis of no effect (Bertrand et al. 2010). Moreover, in the border pair design, some counties are part of multiple pairs, such that their errors are mechanically correlated. As a first step, to account for correlations in the error term, we applied to all estimates a two-way clustering procedure at the county and pair level. However, this may not eliminate all systematic correlations of the error terms.

To assess the statistical significance of our estimates without relying on assumptions about clustering, we additionally perform permutation tests for the four main outcomes. In these tests, we first obtain an empirical placebo distribution of estimates that would occur under the null hypothesis of there being no effect. In a second step, we compare our estimates to the placebo distribution and obtain an empirical p value that describes the probability of obtaining a result that is at least as extreme as ours.Footnote 18 In a conventional case—namely one in which a treatment is assigned once—the placebo distribution is obtained by repeatedly randomizing the treatment across observations and estimating the same model in each replication. The complication in our case is that top-up rates within states are path-dependent. States do not randomly set a top-up rate every year, but rather adjust the rate of the previous year. To account for path dependency, we therefore randomize over 14-year paths in top-up rates. In each replication, we randomly assign each state a path for its top-up rate and estimate the model.

Figure 9 displays the cumulative density function of the placebo distributions based on 5000 replications, as well as the z-scores of our estimates (vertical lines) from Column (6) in Table 2. The horizontal lines describe the 90-th percentile of the placebo distribution. Statistical significance at the 10% level requires that the intersection of both lines is located southeast of the placebo distribution. This is the case for the outcomes displayed in Panels A-C, where the empirical p values are 0.055, 0.014, and 0.027, respectively. For the outcome in Panel D—namely the total number of non-self-employed claimants—the p value is 0.128, which means that this estimate is not statistically significant at the 10% level.

These results confirm the inference drawn from the two-way clustering approach in Table 2. Raises in the top-up rate significantly increase bunching near the kink point, which is the result of an overproportional increase in the number of claimants with an income close to the kink point. As before, we find no statistically significant effect on the total number of non-self-employed EITC claimants.

Fig. 9
figure9

Permutation tests. a Bunching, b self-employed near kink, c total number of self-employed and d total number of non-self-employed. Notes: These graphs display the results of permutation tests for each of the four outcomes. Each panel displays the cumulative density from 5000 replications, as well as the empirical p values

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Buhlmann, F., Elsner, B. & Peichl, A. Tax refunds and income manipulation: evidence from the EITC. Int Tax Public Finance 25, 1490–1518 (2018). https://doi.org/10.1007/s10797-018-9510-7

Download citation

Keywords

  • EITC
  • Bunching
  • Income manipulation

JEL Classification

  • H20
  • H24