Did the Work Opportunity Tax Credit Cause Subsidized Worker Substitution?

Abstract

This article questions whether the implementation of the Work Opportunity Tax Credit (WOTC) created an incentive for employers to substitute subsidized workers for incumbent workers. To see if this substitution occurs, the author uses a differences-in-differences methodology to test whether the implementation of the WOTC caused both an increase in employment from a representative target group and a decrease in employment of a group that is a close substitute for members of the target group. The author finds no evidence that subsidized worker substitution occurred in the period after the WOTC was implemented. There is evidence that the WOTC is effective in increasing the employment rates of long-term welfare recipients.

Keywords

Work Opportunity Tax Credit (WOTC)subsidized worker substitution differences-in-differences welfare reform

Job creation strategies have been at the forefront of policy debates recently (Stewart, 2011). Policymakers argue whether the appropriate policy tool for reducing unemployment is through spending policy or tax policy. What is omitted from the debate is how to reduce unemployment for specific demographic groups. Research has shown that unemployment rates are consistently higher for African Americans and Hispanics (Austin, 2010) and for the young and low skilled. One particular group that has difficulty gaining employment is welfare recipients (Danziger et al., 2000). Research has found that although firms did have unmet labor demand, it may be unlikely that these jobs would be available to welfare recipients (Holzer, 1999). This provides motivation for an employer tax subsidy such as the Work Opportunity Tax Credit (WOTC). The WOTC is a tax credit that acts as a subsidy to employers to hire individuals who are deemed “unemployable.” Employer tax subsidies for the disadvantaged are a popular option since they can be targeted (Katz, 1996).

As with most employer tax subsidies, there are concerns about the incentive structure and the design of the tax credit (Bartik, 2001). There is a concern that employers would engage in churning and subsidized worker substitution to maximize the benefit they would receive from using the WOTC (General Accounting Office [GAO], 2001). The precursor to the WOTC was the Targeted Jobs Tax Credit (TJTC), which was plagued by these incentive issues (Lorenz, 1995). Another issue with the TJTC was that it led to stigma effects for those who receive employment through the program (Hollenbeck & Willke, 1991). Additionally, the TJTC was deemed ineffective since most firms who took the credit were planning on hiring those individuals in the absence of the credit (Bishop & Montgomery, 1993). Another concern is with the effectiveness of these programs. Although the majority of studies have found that employer tax subsidies are not effective,¹ some studies have outlined conditions for success (Jaenichen & Stephan, 2011). The question, thus, is whether the WOTC has improved on the TJTC, since there is still a need for targeted employment programs.

This article provides an evaluation of the incentive effects of the WOTC using quasi-experimental methods. Although the literature has shown that employer tax subsidies can be effective (Hooper, 2003), there are other questions that can provide further evaluation of these programs. The specific research question is whether the implementation of the WOTC created an incentive for employers to practice subsidized worker substitution with welfare recipients. The focus on welfare recipients for this study is due to the large push for individuals to be removed from welfare. The main thrust of the 1996 Personal Responsibility and Work Opportunity Reconciliation Act was to decrease reliance on welfare (Blank, 1997). Subsidized worker substitution occurs when employers fire employees who are not eligible for the tax credit and replace them with workers who are eligible. To see if this substitution occurs, I use a differences-in-differences estimator to test whether the implementation of the WOTC caused both an increase in employment of long-term welfare recipients and a decrease in employment for short-term welfare recipients. The results show that the WOTC does not cause subsidized worker substitution. Although the WOTC increases the employment rate of those in the target group, there is no significant effect on the employment rate of those who would experience unemployment as a result of the credit. The format of the article is as follows: The next section describes the WOTC and gives a review of the literature on the WOTC. I then provide a theoretical exploration of subsidized worker substitution. The next section describes the differences-in-differences methodology, followed by the description of data from the Current Population Survey (CPS). I then place the differences-in-differences estimation within a regression framework. The conclusion ends the article.

Background and Literature Review

The WOTC began in 1996 as part of the Personal Responsibility and Work Opportunity Reconciliation Act,² as an incentive for employers to hire eligible workers from eight different target groups that are normally considered “unemployable.” The eight groups are the following:³

Individuals in families currently or previous receiving welfare benefits under the Temporary Assistance for Needy Families for at least 9 months

Veterans in families currently or previously receiving assistance under a food stamp program

Food stamp recipients—aged 18 through 24 years—in families currently or previously receiving assistance under a food stamp program

Youth—aged 18 through 24 years—who live within an empowerment zone or enterprise community

Youth—aged 16 and 17 years—who live within an empowerment zone or enterprise community and are hired for summer employment only

Ex-felons in low-income families

Individuals currently or previously receiving Supplement Security Income

Individuals currently or previously receiving vocational rehabilitation

The tax credit can decrease an employer’s federal tax liability up to $2,400 per worker, depending upon the type of worker and how many hours the employee works. For example, if a qualifying employee works between 120 and 399 hours, the employer can claim a credit of 25% on the first $6,000 the employee earns. If a qualifying employee works more than 400 hours, the employer can claim a credit of 40% on the first $6,000 the employee earns. The tax credit first expired in July 1999, but has been reauthorized biannually with the most recent reauthorization extended through December 2011.

Although there have been a number of studies on different employment tax credits (Heckman, Lalonde, & Smith, 1999), there have been relatively few studies of the WOTC. The GAO conducted a survey of the major firms that have used the WOTC to understand the characteristics of the type of firms who use the WOTC and to see if these employers practiced churning and subsidized worker substitution. Analyzing the survey data, the GAO found that the majority of firms that took the tax credit were concentrated in the states of California and Texas. Also, although they were unable to make any definitive statements as to the practice of churning and subsidized worker substitution, the results of the survey cast doubt on the practice of churning. Hamersma (2003) provided an analysis of worker participation using a unique sample of administrative data from Wisconsin. She also analyzed the employment outcomes from the WOTC and finds there are short-term benefits to the program (Hamersma, 2008). Devaro (2008) developed a theoretical model on employer recruitment. He performed a simulation on the potential impact of a tax credit such as the WOTC, which showed that the credit would lead firms to recruit informally, like through current employees and word of mouth rather than formally. There is little research specifically on the WOTC, and it is my hope to add to the literature on the evaluation of the WOTC.

Theoretical Exploration

In a study of how welfare reform would affect labor markets, Bartik (2000) laid out a theoretical exploration of the issue of displacement. Using a simple model of labor supply and demand, he showed that welfare reform should boost labor supply, which would cause a decrease in wages. In a New York Times article about the entrants of welfare recipients, the author said, “Normally, they [employers] raise wages to lure people who would not otherwise be willing to take these jobs. But with the injection of so many welfare recipients as workers, the wage pressure is dulled.”⁴ Any potential effect of the WOTC would be on the number of workers in the labor market rather than any effect on wages. Figure 1 shows the impact of the implementation of WOTC on labor markets within the context of welfare reform.

In Figure 1A, the effect of welfare reform would be to increase the supply of labor shifting the curve from L_S to L′_s The effect of the WOTC would be to increase the demand for workers who are eligible for the credit, shown as the shift from L_D to L′_D. For those not eligible, the labor demand will fall from L_D to L″_D.

Taking all the factors into account, I find the loss of employment by ineligible workers will decrease, whereas the gain in employment by WOTC-eligible workers will increase from the increase in labor supply. The increase in employment by WOTC-eligible workers is due to this boost in labor demand (as measured by the distance between n₀ and n₁). The fall in the demand for the ineligible workers is countered by the rise in labor supply from welfare reform. Therefore, any substitution that would occur due to the implementation of WOTC should be muted in the presence of welfare reform (as shown by the distance between n′₀ and n′₁ ).

Methodology

To test whether subsidized worker substitution occurs, this study uses a modified version of a differences-in-differences methodology. This method is a popular procedure to evaluate the behavioral response to changes in tax rates or tax credits. It has been used to estimate the effect of tax reform changes (Eissa, 1995), Earned Income Tax Credit (Eissa & Liebman, 1996), and the Working Families Tax Credit (Blundell, Brewer, & Shepard, 2005). The methodology treats the implementation of the tax credit or change in the credit as a natural experiment. The purpose is to isolate the effect of the tax credit change on an affected population. The question evaluated is how the change in taxation affects a given outcome, such as employment. To perform the analysis, a treatment group is created that includes individuals who are eligible for the credit and a control group that includes individuals who are not eligible, but who have similar characteristics to an individual within the treatment group. The differences-in-differences estimator measures the increase in employment of the treatment group relative to the control group. Formally, the differences-in-differences estimator (α) is given in Equation 1:

α_{D I D} = ({\bar{Y}}_{t_{1}}^{t} - {\bar{Y}}_{t_{0}}^{t}) - ({\bar{Y}}_{t_{1}}^{c} - {\bar{Y}}_{t_{0}}^{c}),

where ${\bar{Y}}_{t_{1}}^{t}$ is the average outcome measure for an individual in treatment group t in time period t₁ and ${\bar{Y}}_{t_{0}}^{t}$ is the average outcome measure for the same individual in time period t ₀. The pre-WOTC period and post-WOTC period are represented by t₀ and t₁, respectively. ${\bar{Y}}_{t_{1}}^{c}$ and ${\bar{Y}}_{t_{0}}^{c}$ are the average outcome measures for an individual in the control group.

The usual procedure for the differences-in-differences methodology creates one treatment group and one control group. Since the focus of this article is on substitution between those eligible for the WOTC and those not eligible for the WOTC, there needs to be two treatment groups created since it is hypothesized that the WOTC will have an effect on the two groups. Using the CPS, a profile of a member of one treatment group is crafted and a member of a treatment group that represents a close substitute is also crafted. The first treatment group created is a member of a target group that is eligible for the WOTC. The second treatment group created is an individual who is very similar to an individual in that same target group, but is ineligible for the WOTC. If substitution occurs, the WOTC would affect one treatment group in a positive manner and affect the second treatment group in a negative manner.

It is insufficient to compare the average outcomes for the treatment groups. There are other factors that may affect employment, such as demographic characteristics. To control for these factors, the differences-in-differences estimation is placed within a regression framework. Using a probit model, the probability of a person being employed is regressed on a person being a member of the target group or a substitute group, as well as a number of demographic factors. To show the effect of WOTC on each group, interactive terms are included. A dummy variable is created to delineate the period prior to the credit’s implementation and the period after the credit’s implementation. If substitution were to occur, it is expected that the coefficient on the target group variable interacted with a post-WOTC variable is positive and the coefficient on the substitute group variable interacted with a post-WOTC variable is negative. That is, if a person was a member of the target group and it is the period after WOTC was implemented, this increases the probability that they would be employed. Likewise, if a person was a member of the substitute group and it is the period after WOTC was implemented, this decreases the probability that they would be employed.

Data and Variables

The data used for this analysis are taken from the Bureau of Labor Statistics’ CPS.⁵ The CPS is a nationally representative monthly survey of roughly 60,000 households. Individual-level data from the March supplement are collected from the years 1989 to 2005. This allows for equal time periods before and after implementation of the WOTC.⁶ The treatment groups represent welfare recipients. I focus on the target group consisting of welfare recipients because they are the largest population that is used by employers for the WOTC. Hamersma (2005) found that more than one third of the certifications go to long-term welfare recipients using administrative data from Wisconsin. The control group consists of the low-income individuals who are not eligible for welfare. These are individuals who may be similar to welfare recipients, but not similar enough to be involved in the occurrence of substitution. To create precise treatment groups, I limit the sample to unmarried females who received welfare benefits (i.e., those who received either AFDC benefits or TANF benefits). To isolate the effects of the credit on welfare recipients, I also restrict the sample to individuals between the ages of 25 and 64 years. Several other target groups include individuals under the age of 25 years. This gives a total number of observations of 14,610. The first treatment group, the eligible population, is composed of individuals who have been on welfare (AFDC or TANF) for at least 9 months. To be eligible for the credit, an individual has to have been on welfare for at least 9 out of the last 18 months. The second treatment group, the near-eligible population, is composed of individuals who have been on welfare for a period between 6 and 8 months. This has been used as a substitute group in other studies (see Hamersma, 2008). It is this second treatment group that serves as a substitute for the eligible population.

The main dependent variable is whether the individual is employed (EMPLOYED). This is a dummy variable taking the value 1 if the individual is employed and 0 otherwise. Two alternative dependent variables are used in a later estimation. The first alternative dependent variable is a dummy variable that takes a value of 1 if the person is unemployed (UNEMPLOYED) and 0 otherwise. The second alternative dependent variable is a dummy variable that takes a value of 1 if the person is unemployed due to being laid off (LAYOFF) and 0 otherwise.

Dummy variables are created to represent the treatment groups, the eligible population (WOTC), and the near-eligible population (Non-WOTC). The remaining population encompasses the control group. Another dummy variable delineates the period before and after the credit is implemented (Year 1997). This variable takes the value 1 if the year is after 1997 and 0 if the year is before 1997. Included as control variables are age (Age), whether the individual has a high school diploma (Diploma), whether the individual is not White (Non-White), the number of children (Number of Children), and the number of children under the age of 5 years (Number of Children Under Five). Whether an individual lived in a city center (Center City) and census regional divisions are included to control for geography. All the variables are dummy variables except for Age, Number of Children, and Number of Children Under Five. A summary of these variables is given in Table 1.

Table 1.

Variable Means

Variable	Target	Substitute	Control
Employed	0.205	0.395	0.470
Unemployed	0.118	0.179	0.149
Layoff	0.005	0.015	0.009
Age	35.1	34.2	33.5
Number of children	2.2	1.9	1.9
Number of children under five	0.51	0.49	0.48
Non-White	0.485	0.430	0.386
Diploma	0.485	0.578	0.637
Center city	0.485	0.421	0.370
Number of observations	11,268	1,259	2,083

Table 1 shows that individuals in the target group are older, have more children, are more likely to be non-White, are the least educated, and more likely to live in the center city. Those in the target group are less likely to be employed, whereas those in the control group are the most likely to be employed. Individuals in the substitute group are the most likely to be unemployed and to be unemployed due to layoff.

Differences-in-Differences Regression

Since the dependent variable is a dichotomous variable, a probit model is estimated. What is being estimated is the probability that a person, given a set of characteristics, will be employed. The empirical specification is given in Equation 2:

\Pr (E m p l o y e d = 1) = Φ (α_{0} + α_{1} * WOTC + α_{2} * Non_WOTC + α_{3} * Year_1997 + α_{4} * (WOTC * Year_1997) + α_{5} * (Non_WOTC * Year_1997) + α_{6} * X) .

The coefficients of interest are α₄ and α₅, which show the effect of the implementation of the WOTC on each group. If substitution did indeed occur because of WOTC, then α₄ should be positive and α₅ should be negative and the marginal effects should be similar. Table 2 shows the results of the probit analysis. X represents the covariates that control for demographic factors, as described in Table 1. The coefficients reported in Table 2 are marginal effects.

Table 2.

Differences-in-Differences Regression Estimates^a

	Pr(Employed = 1)
WOTC	−0.2743*** (0.018)
Non-WOTC	−0.0485** (0.020)
Year 1997 dummy	0.0141 (0.021)
WOTC * Year dummy	0.1263*** (0.027)
Non-WOTC * Year dummy	0.0090 (0.035)
Age	−0.0008 (0.001)
Number of children	−0.0009 (0.003)
Number of children under five	−0.0407*** (0.007)
Non-White	−0.0100 (0.009)
Diploma	0.1071*** (0.009)
Center city	−0.0463*** (0.009)
N	14,610
Wald χ²	1000.89***
Pseudo R²	0.0841

Predicted Probabilities	Pr(Employed = 1)

WOTC	.3402***
Non-WOTC	.2482***

Note. WOTC = Work Opportunity Tax Credit. Standard errors in parentheses; coefficients on region dummies not reported.

A similar analysis was performed with hours as the dependent variable and the results are not different.

***

Significance at the 1% level. **Significance at the 5% level.

The coefficients on the interactive terms represent the differences-in-differences estimates, which show the change in employment on each treatment group due to the implementation of the WOTC. The results show that WOTC has increased the employment for those in the target group and has no significant effect on the substitute group. Controlling for demographic characteristics, there is no evidence of substitution occurring. The predicted probabilities show that an individual who is eligible for WOTC is 0.34 more likely to be employed and an individual who is not eligible is only 0.25 more likely to be employed. Looking back to the theory, it was hypothesized that although the implementation of the WOTC may cause substitution, the increase in labor supply from welfare reform should diminish the amount of substitution that occurred. The rest of the results show that those who have children less than the age of 5 years and live in the center city are less likely to employed, but those who have their high school diploma are more likely to be employed.

Although the results in Table 2 show that subsidized worker substitution did not occur, it is necessary to go one step further and look at the unemployment situation. The results from Table 2 only explain the likelihood of gaining employment. The next estimation is to focus on whether those who would have been substituted were fired from their job. In Table 3, the estimation performed analyzes whether the implementation of the WOTC increased the probability that a member of the substitute group would be unemployed, or specifically that they are laid off. If subsidized worker substitution were to occur, we would expect the coefficient on the substitute treatment group interacted with the year dummy to be positive.

Table 3.

Impact of WOTC on Probability of Being Unemployed and On Layoff

	Pr(Unemployed = 1)	Pr(Layoff = 1)
Non-WOTC	0.0724*** (0.017)	0.0071** (0.004)
Year 1997 dummy	0.0278*** (0.008)	0.0030 (0.002)
Non-WOTC * Year dummy	−0.0400** (0.017)	0.0011 (0.004)
Age	−0.0043*** (0.0004)	−0.0001 (0.0001)
Number of children	0.0068*** (0.002)	0.0008 (0.001)
Number of children under five	−0.0304*** (0.005)	−0.0027*** (0.001)
Non-White	0.0383*** (0.007)	−0.0022 (0.001)
Diploma	0.0054 (0.007)	−0.0010 (0.001)
Center city	0.0054 (0.007)	−0.0014 (0.001)
N	14,610	14,610
Wald χ²	197.89***	46.57***
Pseudo R²	0.0249	0.0335

Note. WOTC = Work Opportunity Tax Credit. Standard errors in parentheses; coefficients on region dummies not reported.

***

Significance at the 1% level. **Significance at the 5% level.

The coefficient on the interactive term (Non-WOTC * Year dummy) is negative and significant, which shows that the WOTC did not lead those in the substitute group to lose their job. In terms of being laid off, the coefficient is positive but not significant. Taking the results from Table 3 along with Table 2, there is no evidence of subsidized worker substitution due to the implementation of the WOTC.

Conclusion

This article examined whether firms engaged in the practice of subsidized worker substitution due to the incentives present with the WOTC. Using a differences-in-differences methodology, the results showed that there was not any evidence that subsidized worker substitution occurred. Yet the implementation of WOTC did lead to an increase in employment for the target group, as measured by long-term welfare recipients, taking into account labor market and demographic factors. There was evidence that WOTC was effective in giving employment to those who had previously experienced difficulty in gaining employment. Martin and Grubb (2001), in a review of the evaluation literature, find that wage subsidies can work but that it requires careful targeting and specific controls to gain maximum effectiveness. Brown, Merkl, and Snower (2011) show that targeting to individuals with longer unemployment spells rather than targeting to individuals with low skills is more effective. This is important especially given the length and duration of the current recession.

In the only study specifically evaluating the WOTC, Hamersma (2008) finds a significant positive impact on employment, though it is only for the short run. She argues that due to the low participation rate of firms, it is not expected that the effects would persist in the long run. In a study of a German wage subsidy program, Jaenichen and Stephan (2011) find that employment of those in the targeted group is 25% to 42% greater than those in the substitute group. The results in this study show the employment of the targeted group to be 36% greater than those in the substitute group. There was no significant impact of the WOTC on the substitute group, which was hypothesized since any loss in demand for this labor would be mitigated by the increase in labor supply from welfare reform (see Figure 1B).

The contribution of this article is that it provides further proof that employers do not abuse the credit, a finding consistent with the literature (GAO, 2001; Hamersma, 2011). This is important because a concern is the perverse incentives that can occur with these types of subsidy programs (Katz, 1996). This article also shows that the WOTC can be effective in improving the employment prospects for those leaving welfare. This is an important result, especially given the current context of the current downturn. Once the economy turns around, the rate at which those recently leaving welfare and other individuals considered “unemployable” will not be comparable to other unemployed individuals. Thus, a program like the WOTC should be encouraged and promoted. The implications of the results suggest that given proper targeting, the WOTC can be an effective program for improving the job prospects of disadvantaged workers.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Bio

Olugbenga Ajilore is an Associate Professor of Economics at the University of Toledo. His research focuses on the impacts of policies on different demographic groups, with a specific focus on state and local government spending.

References

Austin

(2010). Uneven pain—Unemployment by metropolitan area and race. Washington, DC: Economic Policy Institute.

Bartik

(2000). Displacement and wage effects of welfare reform. In Card

Blank

(Eds.), Finding jobs: Work and welfare reform (pp. 72-122). New York, NY: Russell Sage Foundation.

Bartik

(2001). Jobs for the poor: Can labor demand policies help? New York, NY: Russell Sage Foundation.

Bertrand

Duflo

Mullainathan

(2004). How much should we trust differences-in-differences estimates? Quarterly Journal of Economics, 119, 249-275.

Bishop

J. H.

Montgomery

(1993). Does the targeted jobs tax credit create jobs at subsidized firms? Industrial Relations, 32, 289-306.

Blank

(1997). Policy watch: The 1996 welfare reform. Journal of Economic Perspectives, 11, 169-177.

Blundell

Brewer

Shepard

(2005). Evaluating the labour market impact of the working families tax credit using differences-in-differences (HMRC Working Paper No. 4). London, England: HM Revenue and Customs.

Brown

A. J.

Merkl

Snower

D. J.

(2011). Comparing the Effectiveness of employment subsidies. Labour Economics, 18, 168-179.

Danziger

Corcoran

Danziger

Heflin

Kalil

Levin

. . . Tolman

(2000). Barriers to the employment of welfare recipients. In Cherry

Rodgers

W. M.

(Eds.), Prosperity for all? The economic boom and African-Americans (pp. 245-278). New York, NY: Russell Sage.

10.

Devaro

(2008). The labor market effects of employer recruitment choice. European Economic Review, 52, 283-314.

11.

Dickert-Conlin

Holtz-Eakin

(2000). Employee-based versus employer-based subsidies to low-wage workers: A public finance perspective. In Card

D. E.

(Ed.), Finding jobs: Work and welfare reform (pp. 262-298). New York, NY: Russell Sage Foundation.

12.

Eissa

(1995). Taxation and labor supply of married women: The Tax Reform Act of 1986 as a natural experiment (NBER Working Paper No. 5023). Cambridge, MA: National Bureau of Economic Research.

13.

Eissa

Liebman

J. B.

(1996). Labor supply response to the earned income tax credit. Quarterly Journal of Economics, 111, 605-637.

14.

General Accounting Office. (2001). Work opportunity tax credit: Employers do not appear to dismiss employees to increase tax credits. Washington, DC: Author.

15.

Hamersma

(2003). The work opportunity and welfare-to-work tax credits: Participation among eligible workers. National Tax Journal, 56, 725-738.

16.

Hamersma

(2005). The work opportunity and welfare-to-work tax credits. Washington, DC: Urban Institute.

17.

Hamersma

(2008). The effects of an employer subsidy on employment outcomes: A study of the work opportunity and welfare-to-work tax credits. Journal of Policy Analysis and Management, 27, 498-520.

18.

Hamersma

(2011). Why don’t eligible firms claim hiring subsidies? The role of job duration. Economic Inquiry, 49, 916-934. doi:10.1111/j.1465-7295.2009.00260.x10.1111/j.1465-7295.2009.00260.x

19.

Heckman

Lalonde

R. J.

Smith

J. A.

(1999). The economics and econometrics of active labor market programs. In Ashenfelter

Card

(Eds.), The handbook of labor economics (Vol. 3a, pp. 1865-2097). Amsterdam, Netherlands: Elsevier Science.

20.

Hollenbeck

Willke

R. J.

(1991). The employment and earnings impacts of the targeted jobs tax credit. Kalamazoo, MI: W. E. Upjohn Institute for Employment Research.

21.

Holzer

H. J.

(1999). Will employers hire welfare recipients? Recent survey evidence from Michigan. Journal of Policy Analysis and Management, 18, 449-472.

22.

Hooper

T. J.

(2003). Employer tax credits: Promise from the demand side. LBJ Journal of Public Affairs, 25, 58-64.

23.

Jaenichen

Stephan

(2011). The effectiveness of targeted wage subsidies for hard-to-place workers. Applied Economics, 43, 1209-1255.

24.

Katz

(1996). Wage subsidies for the disadvantaged (NBER Working Paper No. 5679). Cambridge, MA: National Bureau of Economic Research.

25.

King

Ruggles

Alexander

J. T.

Flood

Genadek

Schroeder

M. B.

. . . Vick

(2010). Integrated Public Use Microdata Series, Current Population Survey: Version 3.0 [Machine Readable Database]. Minneapolis: University of Minnesota.

26.

Lorenz

E. C.

(1995). TJTC and the promise and reality of redistributive vouchering and tax credit policy. Journal of Policy Analysis and Management, 14, 270-290.

27.

Martin

J. P.

Grubb

(2001). What works and for whom: A review of OECD countries’ experiences with active labour market policies. Swedish Economic Policy Review, 8, 9-56.

28.

O’Neill

D. M.

(1982). Employment tax credit programs: The effects of socioeconomic targeting provisions. Journal of Human Resources, 17, 449-459.

29.

Stewart

(2011, April 23). Weekly addresses highlight bipartisan disagreement on job creation. Retrieved from http://politicalticker.blogs.cnn.com/2011/04/23/weekly-addresses-highlight-bipartisan-disagreement-on-job-creation/

30.

Tannery

F. J.

(1998). Targeted jobs tax credit and labor market experience. Washington, DC: Employment Policies Institute.

31.

Uchitelle

(1997, April 1). Welfare recipients taking jobs often held by the working poor. New York Times. Retrieved from http://www.nytimes.com/1997/04/01/us/welfare-recipients-taking-jobs-often-held-by-the-working-poor.html?pagewanted=all&src=pm