Skip to main content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.


The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

The Low-Wage Labor Market: Challenges and Opportunities for Economic Self-Sufficiency. Job Creation for Low-Wage Workers: An Assessment of Public Service Jobs, Tax Credits, and Empowerment Zones

Publication Date
Nov 30, 1999

By Burt S. Barnow



In the current budget and policy environment, it is more crucial than ever to make sure that low­wage workers are participating in the labor market to the fullest extent possible. Eligibility for some safety net programs, such as Temporary Assistance for Needy Families (TANF), is restricted to five years at most over a lifetime, and other programs such as General Assistance have been cut back severely. The TANF program also includes strict work requirements for participants and for states administering the programs.

There are a variety of approaches that can be used to increase employment by low­wage workers, and the appropriate strategy depends on labor market conditions, policy goals, budget constraints, and the situation of the individual. Broad policy options to increase employment include:

  • Occupational training and education. Classroom training can be used to provide low­skill workers with additional skills that will help them compete in the labor market. Appropriate training can increase the number of jobs to which a person has access, the wages they can receive, or both. Under some circumstances, the training can also be provided as on­the­job training by employers. Education is a more general form of training, and educational attainment is highly correlated with earnings.
  • Assessment, counseling, labor market information, and labor exchange services. The new Workforce Investment Act (WIA) refers to these as "core services," and authorizes their provision to the entire labor force. These services do not enhance the vocational skills of workers or create new jobs, but they should reduce the friction in the labor market, resulting in a better match between workers and jobs and less frictional unemployment.
  • Tax incentives for workers and employers. Incentives in the form of tax credits can be provided to workers to increase their interest in working and/or to employers to reduce the cost of labor (or other factors) to encourage hiring additional labor. Additionally, the tax incentives can be restricted to workers with particular characteristics (e.g., welfare recipients) and employers that meet specified criteria (e.g., locating in a designated empowerment zone).
  • Public employment programs. In these programs, the government provides jobs in the public sector or arranges for partially or fully subsidized jobs in the nonprofit sector for eligible persons. The jobs can either be human capital­oriented, where the objective is to increase the work skills of the person, or countercyclical, where the goal is to provide useful work when there are insufficient suitable jobs available in the economy.

The basic approaches listed above differ in terms of how they achieve additional employment for low­wage workers, and their suitability will depend on the cause of the lower­than­desired employment and society's goals. Occupational training and education will lead to increased employment if the trained workers move from a sector with an oversupply of labor to one that can absorb them.(1)  The second category of services, providing information to job seekers about themselves and the labor market, can increase employment by reducing search time for jobs. Tax incentives for workers, under certain conditions, can result in some individuals deciding to seek employment rather than remain on welfare or otherwise out of the labor force. Tax incentives for firms can reduce the price of all or certain types of labor, resulting in an increase in demand for labor. Public employment programs can directly lead to job creation if the newly created jobs do not displace or substitute for jobs that would have been filled without the program.

In this paper, the research findings dealing with public employment programs and several types of tax credits are reviewed, including the Earned Income Tax Credit (EITC), the Targeted Jobs Tax Credit (TJTC) and its recently renamed reincarnation the Work Opportunity Tax Credit (WOTC), and geographically targeted tax credits used in empowerment zones and related programs. Job creation, as used here, means creating new jobs, not simply placing people into existing positions. As structurally oriented employment­generating programs, we are interested in evidence on the extent to which the programs increase employment, any undesired effects (such as creating jobs in one area by taking them from adjoining areas), and what the costs are per job created for the target group of interest.(2)

The focus of the paper on job creation and tax credit policies does not necessarily imply there is no utility associated with training and programs intended to reduce labor market frictions. There is currently some debate in the policy community about whether training or labor force attachment programs are more effective for various target groups, but both strategies have been shown to have positive impacts on employment and earnings, at least in the short run.(3)  The next section of the paper summarizes the literature on public job programs. The following section discusses tax credits, and the final section presents conclusions. Because public job creation programs are the most likely approach to produce new jobs, more attention is devoted here to that approach. This does not imply that tax credits are not as important, simply that job creation is not generally their central function.

Public Job Creation Programs

Public service employment (PSE) programs in the United States have had a checkered history, and interest in their use for countercyclical and structural purposes has had cycles as well. This section provides a review of the U.S. experience with these programs, emphasizing the use of PSE for disadvantaged groups rather than for the general population.

A public service employment program is defined here as a program where government funds are used to hire individuals with certain characteristics on a temporary, time­limited basis with the goals of providing income maintenance and/or human capital development for the participants and valuable services for the community that would not have been produced in the absence of the program. Although somewhat lengthy, this definition captures the key aspects of PSE programs and distinguishes PSE from related programs with similar means and objectives. The specific aspects of the definition and their implications are discussed below.

Eligibility Requirements. PSE programs are sometimes intended to serve as countercyclical programs, and thus provide employment opportunities to individuals who are unemployed because of business cycle conditions. They have also been used as structural programs, and thus serve individuals in need of additional human capital. Appropriate eligibility requirements for structural programs include such factors as unemployment, income, and receipt of transfer payments. In practice, identifying cyclically unemployed workers that can be helped by the program is not a simple matter, and proxies must be used.(4) 

Temporary Positions. PSE jobs are temporary in nature. Time limits can be imposed on the length of time individuals may participate, the length of time that a particular job can exist, or both. Any inclination by the units of government operating the program to use PSE funds to substitute for regular funds will be reduced by limitations on the service of participants or the time the position can be filled.

Salaries. Although the definition of PSE programs used here is restricted to programs that pay a wage, this has been done primarily because unpaid work is generally not classified as employment. Thus, the definition used here excludes "workfare" or community work experience programs (CWEP) operated under the Family Support Act and other welfare programs where welfare recipients were required to "work off their grants," but did not receive salaries for doing so.

Emphasis on Production of Services of Value. This component of the definition is important because it distinguishes PSE programs from public works programs and work experience programs. Public works programs, particularly those operated during the Great Depression of the 1930s, are generally similar to PSE programs in that they are based on the temporary employment of unemployed workers, but unlike PSE programs the primary objective of public works programs is building infrastructure; with that said, defining the precise cutoff between public works and public service employment is somewhat arbitrary, particularly if the state and local governments operating the programs are allowed to supplement the federal funds that are provided.

Employment by Government or a Nonprofit Organization. A PSE program that did not require employment by a government or nonprofit organization would obviously be misnamed. In the traditional Keynsian macroeconomic framework, any expansion in government procurement would generate additional employment, but the program would not be a PSE program.

Production of Services that Would Not Have Been Produced in the Absence of the Program. To have an impact on aggregate employment, a program must lead to an expansion in the government employment beyond what would have occurred in the absence of the program. When program funds are used to substitute for state or local funds that would have been spent in any case, "fiscal substitution" is said to take place.

Public Service Employment and Related Programs in the U.S.A.

Direct job creation programs have been used on a limited basis in the U.S. since the Great Depression. These programs have been controversial throughout this period, with the programs often being the butt of jokes and long periods when such programs were not operated. The Job Training Partnership Act (JTPA), the nation's former major federal employment and training program, specifically prohibits PSE. This section reviews the use of such programs, and the next section summarizes the evidence on the effectiveness of PSE and related programs.

During the Great Depression eight major work relief and public works programs were initiated.(5)  Under President Hoover, the Reconstruction Finance Administration provided loans to state and local governments for welfare and public employment. Although this program was in operation for less than a year, $300 million was spent on work relief, and at its peak nearly 2 million people were employed through the program. When Franklin Roosevelt assumed office, a number of PSE-like programs were enacted, and as many as 4.3 million workers were employed in PSE-like jobs in the largest program.

There were no significant federal PSE programs between the end of the Great Depression programs and the early 1970s. In fact, there was little government involvement in employment or training programs between 1943 and 1961. Federal involvement in training increased with enactment of the Manpower Development and Training Act (MDTA) of 1962.

The first significant PSE program in recent times was the Public Employment Program (PEP), which was authorized by the Emergency Employment Act of 1971.(6)  PEP was a countercyclical program with a two-year life span. Because of the virtual lack of restrictions on eligibility and wages, it was not surprising that the PEP participants resembled the public­sector workforce more than the participants served under MDTA; this is not necessarily an undesirable result for a countercyclical program.

In 1973, the Comprehensive Employment and Training Act (CETA) was enacted. As the title implies, CETA was intended to be a comprehensive program, replacing the numerous categorical programs that existed at the time. The program was intended to replace the training programs provided through MDTA and PSE funded under PEP. CETA reflected President Nixon's strong interest in federalism through block grants and revenue sharing by passing most of the money through to units of state and local government who were to act as "prime sponsors" for the program. The prime sponsors, in turn, funded service providers for training programs and administered the PSE programs.

The original CETA legislation authorized classroom and on­the­job training and work experience under Title I and PSE under Title II. A significant increase in the unemployment rate led to enactment of the Emergency Jobs and Unemployment Assistance Act of 1974, which added a new countercyclical PSE program as Title VI of CETA. In the original CETA, PSE was not the dominant part of CETA that it later became. In FY 1974, for example, $620 million was appropriated for PSE, compared to $1,010 for training activities under Title I; by 1978, $4,684 million was appropriated for PSE, and $1,880 was appropriated for the training title.(7) 

The CETA program was amended significantly three times over its life.(8)  The 1978 amendments had the greatest impact on the PSE programs, but the other amendments also modified the program. The modifications were made to improve targeting and reduce perceived problems with fiscal substitution. Characteristics of the PSE program in the original program and after the amendments are described below.

Eligibility and Targeting. Eligibility for PSE under CETA varied considerably over the life of the program. The major eligibility and targeting requirements have been summarized by Mirengoff et al. (1980), and they are reproduced in table 1. To be eligible for PSE under the original Title II program, an individual had to reside in an area of substantial unemployment (an area having an underemployment rate of at least 6.5 percent) and be either unemployed for at least 30 days or underemployed. Preference was to be given to the most severely disadvantaged in terms of length of unemployment, former training program participants, and Vietnam veterans.

Table 1.
Eligibility and Targeting in CETA Public Service Employment Programs

Date Title Eligibility Targeting
12/28/73 II (Areas of substantial unemployment) 1. Unemployed for 30 or more days or underemployed. 1. Consideration for the most severely disadvantaged in terms of length of unemployment and prospects for obtaining a job; Vietnam veterans; and former training program participants.
12/31/74 VI (Countercyclical PSE) 2. Unemployed for 30 or more days or underemployed; if unemployment rate 7% or higher, unemployed 15 days or more. 2. Same as 1 above plus UI exhaustees, those ineligible for UI except new entrants, people unemployed at least 15 weeks, and recently separated veterans.
10/01/76 VI (Countercyclical PSE) 3. For half of vacancies in regular positions above June 1976 level, same as in 2 above. 3. For half of vacancies in regular positions above June 1976 level, same as in 2 above.
    4. For the remaining half of regular vacancies and for new project positions: member of low­income family and either received UI for 15 or more weeks, was ineligible for UI and was unemployed for 15 or more weeks, exhausted UI entitlement, or was an AFDC recipient. 4. For the remaining half of regular vacancies and for new project positions: same as in 2 above, plus equitable allocation of jobs among members of low­income families who received UI for 15 or more weeks, was ineligible for UI and was unemployed for 15 or more weeks, exhausted UI entitlement, or was an AFDC recipient.
10/27/78 IID (PSE for the economically disadvantaged) 5. Unemployed at least 15 weeks, unemployed at time of determination, and member of low-income family; or member of family receiving AFDC or SSI. 5. Intended for most severely disadvantaged in terms of length of unemployment and prospects for obtaining employment. Consideration to be given to Vietnam­era veterans, public assistance recipients, groups facing labor market disadvantages including offenders, persons with limited English proficiency, people with disabilities, women, single parents, displaced homemakers, youth, older workers, dropouts, and others identified by the Secretary.
  VI (Countercyclical PSE) 6. Unemployed 10 of last 12 weeks and unemployed at time of determination; and an AFDC or SSI recipient or a member of a low­income family. 6. Same as in 5 above.

Note:  UI = unemployment insurance
Source:  William Mirengoff, Lester Rindler, Harry Greenspan, Scott Seablom, and Lois Black. 1980.  The New CETA:  Effect on Public Service Employment Programs.  Washington, D.C.:  National Academy Press.

At the end of 1974, Title VI was added to serve as a countercyclical PSE program; the program was originally scheduled to last for only one year until the unemployment rate receded.(9)  Eligibility was the same as for Title II except that eligible individuals needed to be unemployed for only 15 days if they resided in an area where the unemployment rate was at least 7 percent. The 1976 amendments added low­income eligibility criteria for some Title VI positions.

In 1978, Congressional concern with creaming among the eligible population, enrollment of significant numbers of ineligible participants, and substitution of federal funds for state and local funds led to major amendments to the CETA PSE programs. Eligibility for the structural program, now under Title IID, required unemployment of at least 15 weeks and low family income or receipt of Aid to Families with Dependent Children (AFDC) or Supplemental Security Income (SSI). The countercyclical program now required that participants be unemployed for at least 10 out of the last 12 weeks and unemployed at the time of determination as well as meeting the same income/welfare requirements as Title IID participants. In addition, prime sponsors were required to establish independent monitoring units to assure compliance, and they were required to repay the federal government for all wages paid to ineligible participants.

Wages and Supplementation. The original 1973 legislation restricted the federal contribution to Title II wages to a maximum of $10,000 per year, but employers were permitted to supplement the wages without limit. The Department of Labor was permitted to make general recommendations to maintain an average annual wage of $7,800. The requirements for Title VI were similar.

Amendments passed in 1978 reduced the national average wage for participants entering the program after March 1979 to $7,200. The average for specific areas was indexed to relative wages in the area. The average was adjusted annually for changes in average wages for unsubsidized jobs. Maximum PSE wages were restricted to $10,000 in areas with below­average wages and $12,000 in areas with average wages above the national average. Wage supplementation by employers was prohibited for new Title IID participants, and supplementation for Title VI participants was generally limited to 10 percent of the maximum CETA wage for the area.(10) 

Limits on Individual Participation and Job Length. There were no limits on either individual participation or position duration in the original Title II program and the Title VI program added in 1974. In 1976, Title VI was amended to require that all net new positions be funded in special projects that were to last no more than 12 months. Under the 1978 amendments, projects were permitted to last up to 18 months and be extended for an additional 18 months. The amendments also limited individual participation in both Title II-D and Title VI to a maximum of 18 months in a five­year period, but the Department of Labor was permitted to grant waivers for up to 12 additional months to areas with an unemployment rate of at least 7 percent and hardship in finding unsubsidized jobs for participants.

PSE's role as a significant part of federal employment and training programs ended with the enactment of JTPA in 1982. Johnson and Lopez (1997) note, however, that several state and local programs and some demonstration projects have instituted public employment on a limited scale. These programs include youth corps programs operating in 37 states, the AFDC Homemaker­Home Health Aide Demonstrations (which included a short period of training followed by up to a year of subsidized employment), and the New Hope demonstration project in Milwaukee.

Effectiveness of Public Service Employment Programs

The most important criteria for judging PSE in terms of economic impacts are effects on earnings and employment of participants, net job creation, and value of goods and services created.(11)  These goals are sometimes at cross purposes. Increasing earnings, for example, generally results from emphasizing human capital development rather than maximizing the value of program output, net job creation, or targeting participation opportunities.

Impact on Participants' Earnings. Estimates of earnings impacts are available primarily for the CETA program. Human capital development was an important PSE concern for part of CETA's history, but certainly not for the entire period. The addition of Title VI as a countercyclical program changed priorities significantly, however: "To encourage rapid implementation, Congress relaxed the requirement that sponsors attempt to find jobs for participants in unsubsidized employment. Placement was to be considered only as a goal that could be waived; indeed, more than 90 percent of all sponsors requested and received waivers."(12) 

Several evaluations of the impact of PSE and other CETA activities on earnings were undertaken in the 1980s. All of the studies consider participants who enrolled between 1975 and 1977, which was during the period when job placement was not emphasized; thus, they may yield lower estimates than would be obtained if an earlier or later cohort were analyzed. In addition, the studies used nonexperimental methods, so the estimates may be biased.

Table 2 provides a summary of the earnings impacts of PSE from the CETA evaluations reviewed in Barnow (1987). In general, the studies found moderately positive, statistically significant impacts on earnings for PSE programs. Many of the evaluations found annual impacts of $1,000 or more for women, but the estimates for men were generally lower and often lacked statistical significance. In comparison with other CETA activities, most authors found PSE to have smaller impacts than on­the­job training (OJT), about the same or slightly larger impacts than classroom training, and greater than work experience.

Table 2.
The Impact of CETA Public Service Employment on Earnings for Various Groups

  Overall White women White men Minority women Minority men Women Men
Westat (1981) $250* $950* $100 $650* ($50)
Westat (1984) FY 76 117
Westat (1984) FY 77 654*
Bassi (1983) 614*-701* 259-815* (213)-(23)
Bassi et al. (1984) nonwelfare disadvantaged adults 1,049*-1,229* 302-303 1,605*-1,623* 8-161
Bassi et al. (1984) welfare 1,558*-1,563* 1,218*-1,307* 1,648*-1,673* (32)-274
Bassi et al. (1984) youth 882*-990* (180)-(81) 1,125*-1,196* (396)-(314)
Dickinson, Johnson, West (1984) adults $464* ($836)*
Dickinson, Johnson, West (1984) youth 52 (403)
Geraci (1984) 1,121* (217)

*  Statistically significant impact.
Source:  Burt S. Barnow, "The Impact of CETA Programs on Earnings: A Review of the Literature." The Journal of Human Resources. Spring (1987).

One possible concern is that PSE was more of a countercyclical program during the period that was evaluated while any future PSE program is likely to be more structurally oriented. Bassi et al. (1984) looked exclusively at economically disadvantaged individuals, and they found that PSE had an impact over $1,000 for nonwelfare women and both men and women on welfare. They found no statistically significant impact, however, for nonwelfare men.

Finally, it should be kept in mind that increasing earnings was not a goal of PSE during the period being evaluated. As noted above, most prime sponsors obtained waivers from placement goals. In conclusion, it is reasonable to assume that under CETA PSE worked about as well as classroom training in increasing earnings, even when that was not an explicit goal of the program.

Job Creation Effects. The job creation goal is most important when PSE is used as a countercyclical program, where the primary goal is to add jobs to the economy.(13)  Job creation is still important in a structural PSE program because if the program is not creating additional jobs, then other workers are being displaced. Additionally, if there is no job creation, the federal government is paying for existing state and local services, which may be more appropriate under revenue sharing than a PSE program.(14) 

Evaluations of the job creation effects of PSE programs have involved both the use of econometric models of government employment and field surveys of state and local governments. In the econometric modeling approach, researchers develop equations to predict state and local government employment (or the wage bill) and then determine the impact PSE slots have on the number of regular employees. A study by Johnson and Tomola (1977) concluded that although there was no substitution in the first 3 months of funding CETA PSE positions, by the end of 18 months the PSE funds substituted for regular state and local funding so that no jobs were created by the program. Using similar models, Mirengoff and Rindler (1978) found that over the first 10 months of funding in the CETA PSE program, an average of 65 percent of the positions funded represented net new employment. Later work by Bassi and Fechter (1979) also found substantial substitution, although Bassi and Fechter were less sanguine about their point estimates. Cook et al. (1985) found that each dollar of PSE funding led to an increase in state and local expenditures on salaries and wages of $.28 in 1977 and over $.76 to $.78 in 1978 and 1979 when the eligibility rules were modified to restrict the program to the economically disadvantaged.(15) 

The fragility of the underlying econometric models was illustrated by Borus and Hamermesh (1978). In reanalyzing Johnson and Tomola's data, they found that by making reasonable alternative assumptions they could obtain estimates of substitution ranging from 0 to 100 percent. Borus and Hamermesh concluded that the data and models available to use are simply too crude to reliably estimate the substitution effects of the CETA PSE program.

The alternative to estimating quantitative models is to have researchers conduct field studies and record how much net job creation results from PSE funds. The utility of this approach depends on the ability of the field researchers to accurately assess what would have happened in the absence of PSE funding. A major field evaluation of PSE was carried out in 40 sites selected to be representative of the national program; the research was conducted in four rounds between 1977 and 1980.(16) 

This study found that between 80 and 90 percent of all PSE positions funded in the sites studied constituted job creation rather than job displacement. A controversial aspect of the findings was that a significant portion of the job creation was "program maintenance," defined as "cases in which PSE employees were used to maintain existing services that would have been curtailed without PSE funding."(17)  The researchers concluded that displacement would have been higher in later years had the restrictions on wages, eligibility, tenure, and projects not been added. They found that displacement was highest in rural areas (31 percent), but not especially high in large distressed cities (18 percent), where the researchers classified many of the PSE positions as program maintenance.

It is difficult to reach a firm conclusion about the level of displacement that occurred in the CETA PSE program.(18)  Some observers have found the field study estimates of 10 to 20 percent substitution lack credibility, but the quantitative estimates follow no consistent pattern, so they are not of much help. On the other hand, later studies indicate that CETA amendments added to reduce displacement by increasing targeting of participants more toward the economically disadvantaged, reducing wages, limiting participation time of individuals, and requiring positions to be used in special projects of limited duration all were successful in reducing the problem.(19)

Value of Output. Unlike classroom training programs, PSE programs are frequently judged in part on the basis of the value of the output produced by participants. In this section, studies on the value of output produced in PSE and related programs are discussed. There have been no general evaluations of the value of the output in CETA programs, so we must rely on studies of other programs and special PSE programs.

A fundamental problem in assessing the value of output from government programs is that there is no market mechanism to assign a value to the output. Usually the programs are used to produce services rather than goods, and the outputs are usually not sold in a free market. Thus, we cannot observe the "value" of the output as economists would usually use that term. Instead, researchers can observe or estimate potential proxies such as what the cost would be of producing the output from the PSE program (referred to as supply­side estimates) by using regular government workers or private­sector workers, and what society would be willing to pay for the output (demand­side estimates). While imprecise, such measures can provide some reassurance that the projects are not simply "makework."

The National Supported Work Demonstration was conducted between 1975 and 1979 to test the utility of providing selected target groups — disadvantaged out­of­school youth, ex-offenders, former substance abusers, and AFDC recipients — with work experience under conditions of graduated stress to increase their employment and earnings.(20)  The cost­benefit analysis for the evaluation used the costs for alternative suppliers to estimate the value of the output produced by participants during their enrollment.(21)  The value of output per participant ranged from $2,973 for ex-offenders to $4,520 for AFDC recipients. Although these values were close to the wages paid to the participants, the projects also involved substantial costs for material and overhead. From the perspective of society as a whole, the researchers found that the value of the output defrayed between one­half and two­thirds of the costs of the project, depending on the target group.

The Youth Incentive Entitlement Pilot Projects (YIEPP) demonstration was a large­scale intervention that determined the feasibility, costs, and impacts of guaranteeing part­time school­year jobs and full­time summer jobs to all youth who remained in high school or returned to school in the 17 program sites.(22)  Participating youth were assigned to public­sector or private­sector jobs for their work experience. Although a formal analysis of the value of output produced was not undertaken, a random sample of 250 projects were visited to assess quality of the work being performed by YIEPP participants. The researchers concluded that "the quality of work in the demonstration was, on the whole, adequate or better, with some 86 percent of the worksites falling into this category."(23)  A comparison of youth performance in public­sector, nonprofit­sector, and private­sector YIEPP jobs indicated that performance was more similar than different across sectors.

The final and most relevant value of output study relates to the Employment Opportunity Pilot Projects (EOPP) to test job search assistance and subsidized employment and training activities to welfare recipients and other low­income employable adults. An important aspect of the program involved placing participants who could not find a job on their own into PSE positions or training programs through the CETA program. EOPP participants in PSE positions were highly disadvantaged in the labor market — over 98 percent were eligible for CETA, 73.5 percent were receiving AFDC, 38 percent had less than a high school education, 82 percent were women, and 65 percent were members of minority groups.(24) 

A review of 68 EOPP PSE projects found them to be quite successful from a value­of­output perspective:

  • With respect to the relative productivity measures, participants were, on average, about 77 percent as productive as the alternative suppliers in terms of the number of production (nonsupervisory) hours alone. If the need for supervision is taken into account, however, the productivity ratio falls to about 73 percent. These measures show that, although EOPP participants produced, on average, at a rate consistent with the minimum wage, they were only about three­fourths as productive as the workers who would normally perform the work — workers whose wages were, on average, well above the minimum wage. (Whitebread 1983, p. 77.)

Thus, the experience from the EOPP demonstration indicates that economically disadvantaged participants were not quite as productive as regular government workers, but their low wages compensated for their lower productivity.

In summary, previous research indicates that PSE workers are generally slightly less productive than unsubsidized workers, but the lower wages they receive partially or fully offset this lower productivity. Although the estimates are admittedly crude, studies from work relief programs during the Great Depression and more recently from CETA PSE programs found workers in subsidized employment programs to be between 70 and 80 percent as productive as unsubsidized workers. The productivity findings are not limited to highly qualified workers hired in some of the early CETA countercyclical projects. Programs targeted on disadvantaged youth, AFDC recipients, and other disadvantaged workers all found such workers to be highly productive.

What of the frequent allegations of fraud and waste in CETA programs?  Mucciaroni (1990) repeats some of the more intriguing stories reported in Time, the Reader's Digest, and other sources.(25)  He also notes that the problems included CETA training programs as well as PSE and that the prevalence of such incidents may have been exaggerated in the press.(26) 

Conclusions on the Effectiveness of PSE Programs. This review of the literature focused on how well PSE programs have performed along three dimensions: increasing human capital, net job creation, and value of output. On two of these yardsticks, PSE has performed fairly well — certainly better than its reputation would lead one to believe. The consensus from evaluations of CETA PSE is that it increased participant earnings by several hundred dollars per year for women, at least initially. Regarding net job creation, the evidence is mixed and inconclusive. PSE programs have generally been given high marks in terms of the value of their output. Evaluations that have been conducted have generally found PSE participants to be about 75 percent as effective as regular employees. By producing output of value, a PSE program's net costs are significantly reduced, thus adding to the attractiveness of such a program for specific target groups.

Tax Incentive Programs

In this section the literature on the job creation potential of tax credits is reviewed. As noted above, the focus of the paper is on direct job creation programs, so the review presented here is brief.

Earned Income Tax Credit

The Earned Income Tax Credit (EITC) is a wage subsidy program that began modestly in 1975, but has increased in recent years to be the largest cash­transfer program for nonelderly low­income families.(27)  The program seeks to encourage labor force participation by "making work pay" for potential low­wage workers. The program has increased in generosity since 1975, with the maximum benefit increasing from $400 in 1975 to $3,656 in 1997 for a family with two children. The size of the benefit varies by the number of children, but the benefits are quite small for families without children. Attention in this paper is restricted to EITC's job creation potential, but the program also plays a significant role in reducing poverty.

The structure of the EITC is straightforward. In 1997, for a family with two children, the first $9,140 of earnings entitles the family to a refundable tax credit of 40 percent of earnings for a maximum credit of $3,656. The credit remains at $3,656 until earnings reach $11,950. The credit is then reduced by 21 percent of all earnings above $11,950 until the credit is phased out entirely at earnings of $29,290.

The EITC has no direct effect on the number of jobs available in the economy. Instead, it can create employment by inducing individuals to work who would have remained out of the labor force in the absence of the EITC. For a single­parent family, the labor supply effects of the EITC vary depending on how much the person would have earned without the EITC. For individuals out of the labor force or in the phase-in range (where the wage rate is increased), the higher post­EITC wage rate provides an incentive to work, but the extra income generated by the credit could reduce hours of work; thus, the overall effect of the credit for those with very low earnings is ambiguous. For individuals who would receive the maximum credit, there is no wage increase for additional hours worked and economic theory predicts that the extra income from the tax credit would lead to a reduction in hours. For individuals in the phase­out range, where the credit is reduced for each dollar earned, the EITC actually reduces the after­tax wage even though the family still receives some income from the credit; for families in this range, economic theory predicts a decrease in labor supply. Overall, for a single­parent family, economic theory suggests that some individuals would enter the labor market, but some of those already working would be expected to reduce their hours. The situation is more complex for two­parent families, and it is difficult to make predictions on how labor supply will be affected by the EITC.

Caution must be exercised in interpreting the findings. Evaluations of the EITC require strong assumptions about what factors lead to changes in labor market behavior over an extended period where the EITC changes. Studies generally make use of families without children as a "control group" to purge time trends of factors that affect all potential workers. In addition, comparing findings across studies is difficult because the studies vary in terms of the time period studied, the aspects of the EITC studied, and the estimating methods. Nonetheless, many of the recent studies show a consistent pattern of EITC effects.

A study which conducted simulations to estimate the impact of the 1993 round of EITC increases on labor supply concluded that the 1993 expansion of the EITC would lead to an increase of 3.3 percentage points in the proportion of single parents working. For married couples, they projected that primary earners would increase labor supply by .7 percentage points and that secondary earners would decrease their labor supply, but they could not accurately project by how much.(28) 

Another study examined the effects of the 1987 EITC expansion of the EITC on the labor supply of single women with children in 1991.(29)  The authors used several years of Current Population Survey (CPS) data before and after the change and data on single women without children as a control group to eliminate other factors that may have affected labor supply. They concluded that the 1987 changes in the EITC increased labor force participation for single women with children by 2.8 percentage points, from 73.0 to 75.8 percent. The authors also analyzed the impact of the 1987 changes on hours worked for those already in the labor force and, contrary to what theory predicts, they found no impact on hours worked by single women with children.

An evaluation of the impact of the EITC on married couples found that because married couples tend to have more income than single­parent families, 70 percent of the EITC recipients have pre­tax income that puts them in the range where the EITC is at its maximum or is declining and the incentive is for less work.(30)  The researchers discuss the complexities involved in modeling the labor force behavior of two­adult units, and they analyze the data several ways to account for the difficulties. In one analysis, they find that the EITC increases labor force participation by .9 percentage points for men and decreases participation by 3.1 points for women in married couples with children. They also perform some simulations, and in those analyses they find smaller effects — an increase of .1 percentage points for men and a decrease of .5 points for women.

A complex study examined the impact of EITC, along with changes in the Aid to Families with Dependent Children (AFDC) program, Medicaid, child care, and job training on the labor supply of single mothers over the 1984 to 1996 period and the 1992 to 1996 period.(31)  They conclude that the EITC accounts for 52 percent of the increase in weekly labor force participation between 1984 and 1996 and 70 percent of the increase in annual labor force participation over this period. To put these figures in perspective, the labor force participation rate for single mothers increased by 8 percentage points between 1984 and 1996, from 73 to 81 percent. Since the EITC was responsible for 70 percent of this increase, the EITC is estimated to have increased the labor force participation rate by 5.6 percentage points over this period.

In summary, the EITC appears to have been effective in increasing labor market participation among single mothers. In addition to increasing labor force participation among the poor, the EITC helps provide income to poor children, and the program is extremely popular across the political spectrum.(32)  The major problems with the EITC are that it provides work disincentives to married couples and to some single parents. In addition, the refundable nature of the credit creates potential for fraud.(33)  Finally, it is important to recognize that unlike public service jobs, EITC cannot create new jobs. Thus, it is not an appropriate tool to use if the underlying problem is too little aggregate demand.

Conclusions and Policy Implications

In this paper, the evidence on several approaches to employment generation have been reviewed. In spite of their reputation from the alleged abuses that took place during CETA, public job creation programs appear to be a successful means of generating new jobs for low­skill workers. For such programs to work, however, care must be taken to keep the scale modest and the jobs attractive to both the workers and the employing organizations. The Earned Income Tax Credit is an effective tool for increasing employment of single­parent families, but the evidence is more ambiguous for two­parent families. The EITC does not create new jobs as a public employment program does, but instead provides incentives for low­skill individuals to work; the EITC is also an important means of reducing poverty among the working low­skill population. The literature on programs that affect employer tax liabilities, both individually targeted programs and geographically targeted programs, indicates that, although such programs often sound good in theory, in practice they are characterized by windfalls to employers with little employment generation.

These research findings have several implications for policy. If there is insufficient demand for low­skill workers, a carefully designed public employment program can be used to increase employment. Earned income tax credits, on the other hand, can provide an effective means of increasing the supply of the low­skill segment of the population and can help make work pay without increasing wage costs to employers. The evidence to date offers little support for using employer tax credits to increase employment. There is scant evidence supporting the efficacy of tax credits, such as the Targeted Jobs Tax Credit and the Work Opportunity Tax Credit, that offer employers a credit for hiring workers with certain characteristics. As Ladd speculates, the current form of empowerment zones in the U.S. may generate additional employment, but there is no evidence yet that this has occurred, and prior versions have proven very ineffective.


Adams, Charles F., Jr., Robert F. Cook, and Arthur J. Maurice. 1983. "A Pooled Time­Series Analysis of the Job Creation Impact of Public Service Employment Grants to Large Cities." Journal of Human Resources 18 (2, Spring).

Baily, Martin Neil, and James Tobin. 1978. "Inflation­Unemployment Consequences of Job Creation Strategies." In Creating Jobs: Public Employment Programs and Wage Subsidies, edited by John L. Palmer. Washington, D.C.: Brookings Institution.

Barnow, Burt S. 1987. "The Impact of CETA Programs on Earnings: A Review of the Literature." Journal of Human Resources 22 (2, Spring).

------. 1989. "Government Training as a Means of Reducing Unemployment." In Rethinking Employment Policy. Edited by D. Lee Bawden and Felicity Skidmore. Washington, D.C.: Urban Institute Press.

Barnow, Burt S., Amy B. Chasanov, and Abhay Pande (1990). Financial Incentives for Employer­Provided Worker Training: A Review of Relevant Experience in the U.S. and Abroad. Washington, D.C.: Urban Institute.

Bassi, Laurie, and Alan Fechter. 1979. The Implications for Fiscal Substitution and Occupational Displacement under Expanded CETA Title VI. Office of the Assistant Secretary for Policy, Evaluation, and Research Technical Analysis Paper No. 65, Washington, D.C.: U.S. Department of Labor, March.

Bassi, Laurie J., Margaret C. Simms, Lynn C. Burbridge, and Charles L. Betsey. 1984. "Measuring the Effect of CETA on Youth and the Economically Disadvantaged." Final report prepared for the U.S. Department of Labor under Contract No. 20-11-82-19. Washington, D.C.: Urban Institute.

Bishop, John H., and Mark Montgomery. 1993. "Does the Targeted Jobs Tax Credit Create Jobs at Subsidized Firms?" 32 (3).

Borus, Michael E., and Daniel S. Hamermesh. 1978. "Estimating Fiscal Substitution by Public Service Employment Programs." Journal of Human Resources 13 (4, Fall).

Burtless, Gary. 1985. "Are Targeted Wage Subsidies Harmful? Evidence from a Wage Voucher Experiment." Industrial and Labor Relations Review 39.

Cook, Robert F., Charles F. Adams, Jr., and V. Lane Rawlins. 1985. Public Service Employment: The Experience of a Decade. Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research.

Diaz, William A., Joseph Ball, and Carl Wolfhagen. 1982. Linking School and Work for Disadvantaged Youths, The YIEPP Demonstration: Final Implementation Report. New York: Manpower Demonstration Research Corporation.

Dickert, Stacy, Scott Houser, and John Karl Scholz. 1995. "The Earned Income Tax Credit and Transfer Programs: A Study of Labor Market and Program Participation." Tax Policy and the Economy 9.

Dickinson, Katherine P., Terry R. Johnson, and Richard W. West. 1984. "An Analysis of the Impact of CETA Programs on Participants' Earnings." Menlo Park, Calif.: SRI International.

Eissa, Nada, and Hilary Williamson Hoynes. 1998. "The Earned Income Tax Credit and the Labor Supply of Married Couples." Unpublished manuscript. Berkeley, Calif.: University of California,.

Eissa, Nada, and Jeffrey B. Liebman. 1996. "Labor Supply Response to the Earned Income Tax Credit." Quarterly Journal of Economics CXI (2).

Fishman, Michael E., Burt S. Barnow, Karen N. Gardiner, Barbara J. Murphy, and Stephanie A. Laud. 1998. Job Retention and Advancement among Welfare Recipients: Challenges and Opportunities. Draft report. Fairfax, Va.: The Lewin Group.

Franklin, Grace A., and Randall B. Ripley. 1984. CETA: Politics and Policy, 1973-1982. Knoxville: University of Tennessee Press.

Geraci, Vincent J. 1984. "Short­Term Indicators of Job Training Program Effects on Long­Term Participant Earnings." Report prepared for U.S. Department of Labor under Contract No. 20-48-82-16, Washington, D.C.

Gottschalk, Peter. 1983. "U.S. Labor Market Policies Since the 1960s: A Survey of Programs and Their Effectiveness." Madison, Wis.: Institute for Research on Poverty.

Greenstein, Robert, and Isaac Shapiro. 1998. "New Research Findings on the Effect of the Earned Income Tax Credit." Washington, D.C.: Center on Budget and Policy Priorities.

Johnson, Clifford M., and Ana Carricchi Lopez. 1997. "Shattering the Myth of Failure: Promising Findings from Ten Public Job Creation Initiatives." Washington, D.C.: Center on Budget and Policy Priorities.

Johnson, George, and James Tomola. 1977. "The Fiscal Substitution Effect of Alternative Approaches to Public Service Employment Policy." Journal of Human Resources 12 (1, Winter).

Kemper, Peter, David A. Long, and Craig Thornton. 1981. The Supported Work Evaluation: Final Benefit­Cost Analysis. Volume 5 of the Final Report. New York: Manpower Demonstration Research Corporation.

Kesselman, Jonathan R. 1978. "Work Relief Programs in the Great Depression." In Creating Jobs: Public Employment Programs and Wage Subsidies, edited by John L. Palmer. Washington, D.C.: Brookings Institution.

Ladd, Helen F. 1994. "Spatially Targeted Economic Development Strategies: Do They Work?" Cityscape: A Journal of Policy Development and Research 1 (1).

Lerman, Robert, Burt Barnow, and Philip Moss. 1979. "Concepts and Measures of Structural Unemployment," Chapter 2 in Increasing Job Opportunities in the Private Sector. National Commission for Employment Policy Special Report No. 29.

Lorenz, Edward C. 1988. The Targeted Jobs Tax Credit in Maryland and Missouri: 1982-1987. Research Report No. 88-18. Washington, D.C.: National Commission for Employment Policy.

Manpower Demonstration Research Corporation Board of Directors. 1984. Summary and Findings for the National Supported Work Demonstration. Cambridge, Mass.: Ballinger Publishing Company.

Meyer, Bruce D., and Dan T. Rosenbaum. 1998. "Welfare, the Earned Income Tax Credit, and the Labor Supply of Single Mothers." Unpublished manuscript. Evanston, Ill.: Northwestern University.

Mirengoff, William and Lester Rindler. 1978. CETA: Manpower Programs Under Local Control. Washington, D.C.: National Academy Press.

Mirengoff, William, Lester Rindler, Harry Greenspan, Scott Seablom, and Lois Black. 1980. The New CETA: Effect on Public Service Employment Programs. Washington, D.C.: National Academy Press.

Mucciaroni, Gary. 1990. The Political Failure of Employment Policy, 1945-1982. Pittsburgh: University of Pittsburgh Press.

Nathan, Richard P., Robert F. Cook, and V. Lane Rawlins. 1981. Public Service Employment: A Field Evaluation. Washington, D.C.: Brookings Institution.

Plimpton, Lisa, and Demetra Smith Nightingale. 1999 forthcoming. "Welfare Employment Programs: Impacts and Cost­Effectiveness of Employment and Training Activities." In Improving the Odds: What We Know about Targeted Training Programs, edited by Burt S. Barnow and Christopher T. King. Washington, D.C.: Urban Institute Press.

Westat. 1981. "Continuous Longitudinal Manpower Survey Net Impact Report No. 1: Impact on 1977 Earnings of New FY1976 CETA Enrollees in Selected Program Activities." Report prepared for U.S. Department of Labor under Contract No. 23-25-75-07, Washington, D.C.

Westat. 1984. "Summary of Net Impact Results." Report prepared for U.S. Department of Labor under Contract No. 23-24-75-07, Washington, D.C.

Whitebread, Christine. 1983. The Value of Output in Subsidized Jobs in the Employment Opportunity Pilot Project. Princeton, New Jersey: Mathematica Policy Research, February.


1.  Training may be desirable even if it does not result in increased employment if it improves the distribution of employment.

2.  Baily and Tobin (1978) refer to this as "bang per buck."

3.  For a summary of work on education and training, see Fishman et al. (1998).  They concur with the findings of researchers at the Manpower Demonstration Research Corporation that the most effective strategies (at least for welfare recipients) are programs that emphasize job placements with provision of short­term training when necessary.  Plimpton and Nightingale (forthcoming) conclude that long­term training is a superior strategy.

4.  For a discussion of the difficulties in distinguishing structurally unemployed individuals from those who are cyclically unemployed, see Lerman, Barnow, and Moss (1979).  The U.S. Department of Labor is developing predictors of which unemployed workers are most likely to benefit from its training programs (referred to as worker profiling).

5.  See Kesselman (1978) for a detailed history and analysis of programs during the Great Depression.  The material in the text on these programs is based on Kesselman's work.

6.  See Mucciaroni (1990), Franklin and Ripley (1984), and Cook et al. (1985) for a discussion of PEP.  Information on the PEP in this section is based primarily on material in Cook et al. (1985).

7.  Because PSE positions are more costly per unit of time and generally last longer than training positions, the discrepancy in the number of participants is greater.

8.  The Emergency Jobs and Unemployment Assistance Act of 1974 added The Title VI countercyclical PSE program, 1976 amendments tightened requirements for PSE positions, the Youth Employment and Demonstration Projects Act (YEDPA) of 1977 established three special youth training and work experience programs — Youth Employment and Training Programs (YETP), Youth Community Conservation and Improvement Projects (YCCIP), and the Youth Incentive Entitlement Pilot Projects (YIEPP).  The Skills Training Improvement Program (STIP) was added to Title III of CETA in 1977 to serve dislocated workers, the Help through Industry Retraining and Employment (HIRE) was added in 1977 and modified in 1978 to train veterans, and the 1978 reauthorization added the Private Sector Initiative Program (PSIP) as Title VII of CETA to increase private­sector participation in CETA training programs and the Young Adult Conservation Corps (YACC) as Title VIII to provide conservation work experience for youth.  See Franklin and Ripley (1984, p. 21) and Cook et al. (1985).

9.  See Mirengoff and Rindler (1978).

10.  In the few areas where the average wages for regular jobs were more than 25 percent above the national average, CETA PSE wages could be supplemented by 20 percent.  Wages after supplementation could be as high as $11,000 in areas where wages were below average, between $11,000 and $13,200 in most higher­wage areas, and as high as $14,400 in a few areas with the highest wages.  See Mirengoff et al. (1980, p. 78) for details.

11.  Other outcomes of interest include serving target groups of interest and creating a local service delivery capacity.  Providing opportunities to groups of interest has been suggested by many analysts; see, for example, Mirengoff et al. (1980).  Franklin and Ripley (1984, p. 188) specifically cite the development of local delivery capacity as an important legacy of CETA.  For a discussion of other outcomes of interest in employment and training programs, see Barnow (1989).

12.  Mirengoff and Rindler (1978, p. 160).

13.  This paper will not address second­round "multiplier" effects of job creation programs.

14.  Note that a program might be considered successful if it provides employment opportunities to economically disadvantaged individuals who would otherwise not have such opportunities.  See Gottschalk (1983) and Barnow (1989).

15.  See also Adams, Cook, and Maurice (1983).

16.  See Nathan et al. (1981) and Cook et al. (1985) for descriptions of the studies.

17.  Cook et al. (1985).

18.  Another concern is that jobs programs sometimes displace private­sector employment.  If PSE positions pay more than comparable private­sector jobs, workers may remain in PSE positions rather than take private­sector jobs.

19.  Cook et al. (1985) and Mirengoff et al. (1980).

20.  See Manpower Demonstration Research Corporation Board of Directors (1984) for a summary of the supported work program and key findings from the demonstration.

21.  Kemper et al. (1981).

22.  See Diaz et al. (1982).

23.  Diaz et al. (1982).

24.  These figures appear to include participants assigned to on­the­job training in the private sector, but the OJT participants are a minority of the sample.

25.  Some of the more notorious examples cited include a "nude sculpting workshop" and hiring a former Black Panther leader to "keep an eye on the city."

26.  Mucciaroni (1990, p. 185) summarizes CETA's problems as follows: "The CETA story is one not so much about the failure of a program as it is about the failure of a set of political institutions.  It should not be confused with the notion that CETA failed to achieve, to one degree or another, several of its substantive objectives.  While few of its programs were undisputed and overwhelming successes, few were abject failures."

27.  See Eissa and Hoynes (1998).

28.  Dickert, Houser, and Scholz (1995).

29.  Eissa and Liebman (1996).

30.  Eissa and Hoynes (1998).

31.  Meyer and Rosenbaum (1998).  A strong point of Meyer and Rosenbaum's work is that they explicitly control for more factors that could change labor force participation than the other studies.

32.  Greenstein and Shapiro (1998).

33.  Greenstein and Shapiro (1998) acknowledge that the error rates associated with EITC are a concern, but they conclude that recent legislative and administrative actions have led to reductions in the error rate and should lead to further reductions.

34.  The WOTC target groups were:

  1. Members of families that received AFDC or TANF for at least 9 of the 18 months preceding the month of hire;
  2. Individuals 18 to 24 years old who were members of families receiving food stamps for at least 6 consecutive months prior to the date of hire or for at least 3 of the 5 months before the date of hire and their food stamp eligibility expired;
  3. Veterans who were a member of a family that received food stamps for at least three consecutive months during the 15 months before the date of hire;
  4. Disabled individuals who completed rehabilitative services approved by a state or the U.S. Department of Veterans Affairs;
  5. Residents of one of the 105 federally designated urban or rural empowerment zones or enterprise communities who were 18 to 24 years old;
  6. Residents of empowerment zones or enterprise zones ages 16 or 17 hired as summer youth employees;
  7. Ex-felons who were members of a low-income family; and
  8. Recipients of Supplemental Security Income (SSI).

35.  Bishop and Montgomery (1993).

36.  Barnow, Chasanov, and Pande (1990) review evaluations of other tax credits, including the new jobs tax credit, the research and development tax credit, and the investment tax credit.  They conclude that, although these credits sometimes have modest effects on the outcome of interest, they are generally inefficient policies because they produce large windfalls to firms that would have undertaken the actions subsidized without a tax credit.

37.  Lorenz (1988).

38.  Burtless (1985).

39.  Ladd (1994).

Low-Income Populations