Overall, the results from nearly thirty years of evaluating job training programs are remarkably consistent -- surprisingly so, given the variation in the kinds of programs that have been supported and the differences in the methods of evaluation. A large number of job training programs lead to increased earnings, and the benefits generally outweigh the costs -- though the increases in earnings are moderate by almost any standards, insufficient to lift those enrolled in such programs out of poverty. Welfare-to-work programs also increase employment and reduce the amount of welfare, but they rarely allow individuals to leave welfare. Furthermore, any benefits probably fade out after four or five years: job training programs do not seem to put many individuals on career trajectories with continued earnings increases, as formal schooling does.[48] For some groups -- youth in particular -- job training programs generally seem to be ineffective (unless perhaps the program is very intensive, as the Job Corps is), and programs appear to be more effective for women than for men -- but otherwise it is difficult to conclude that any particular group benefits more than any other. There are, to be sure, some outliers -- particularly effective programs like the Center for Employment Training and the Riverside GAIN program. But there are spectacular failures as well -- including some experimental programs with carefully-considered program designs, most job training programs for youth, the worst of the GAIN programs, and Project Independence as a whole -- that actually leave those enrolled worse off and thereby violate the first maxim of intervention: "do no harm".
One can assess the modest outcomes of job training programs in either positive terms, as indicating that these programs are worth doing on the average, or in negative terms. In my interpretation, however, the results are very discouraging: thirty years of experimentation with job training programs have created a substantial number of programs whose benefits -- for individuals in dire need of employment and economic independence -- are quite trivial, and are completely inadequate to the task of moving them out of poverty, off of welfare, or into stable employment over the long run. The puzzle is why such well-intentioned efforts have been so ineffective. In this section I present ten possible explanations, based on the results in Section III, direct observations of job training programs, and comparisons with education programs. Such explanations must remain speculative, of course, since there is not yet enough evidence about truly effective job training program to "prove" what works. Nonetheless, such explanations help provide some guidance for recommendations -- the subject of the last section.
[48] These conclusions are consistent with Fischer's (1995) meta-analysis. He concludes that effects sizes are significantly different from zero, but very small: individuals who receive job training average increased earnings of $200 to $540 per year, and a decrease in welfare payments of $200 to $400 -- but that these effects generally decay over time.
The first and most obvious explanation is simply that most job training programs are "small": they last a very short period of time, rarely more than twenty weeks; they often provide a single kind of service -- on-the-job training, or classroom training, or job search assistance -- rather than a variety of complementary services. Job training administrators often take pride in this aspect of their programs: they will say, for example, that they offer "Chevrolet" programs compared to the "Cadillac" programs of educational institutions, by which they mean that they can get to the same destination at much lower cost; they often scorn education programs for being too "academic" and unconcerned with immediate employment.
However, the individuals enrolled in these programs often have multiple problems and several barriers to employment: they often lack job-specific skills, general academic skills, and the kinds of values (including motivation, punctuality, persistence, the ability to work with others) necessary to find and keep employment, and some of them have more serious problems like drug and alcohol abuse, physical handicaps, other health problems, depression and other mental health problems that may be biological rather than experiential. Even when this is not the case, the gap between the needs of those enrolled and the scope of programs is sometimes breath-taking; for example, one job training program I observed was trying to train Spanish-speaking women to be English-proficient secretaries in a 15-week, part-day program. There is, then, a disjuncture between the profound needs of those who have not found stable employment and the small size of job training programs -- and we should not be surprised to find the effects on employment to be trivial.
It is useful to compare the intensity of job training programs to those of education programs, measuring intensity by expenditures. The average JTPA program for adult men and women cost about $2,200 in 1987-89, for a period of enrollment of 20 weeks (Bloom et al, 1994, Exhibit 2); the cost per person in the GAIN program was about $2,300 in 1993 dollars (Riccio, Friedlander, and Freedman, 1994, p. 75). (Of course, many of the experimental programs are much more expensive, and the current cost per person in Job Corps of about $15,000 is the highest of all.) In contrast, one year of full-time enrollment in a community college -- approximately 30 weeks -- averaged $5,700 (Digest of Educational Statistics, 1994, Table 325), which is considerably more than the limited JTPA and GAIN programs. On the other hand, a one-year certificate program increased earnings by about 15 percent over those of high school graduates, and over a long period of time in the labor force (Grubb, 1995b, Table 3), rather than increasing earnings only over a four or five-year period. Of course, this comparison is not especially fair because the characteristics of JTPA and GAIN clients are quite different from those of students enrolling in community colleges. These comparisons indicate once again that the typical job training program provides less services at substantially lower cost per person, to individuals with less education and (often) more personal problems, compared to typical postsecondary occupational programs-- and it should not be surprising that the effects of job training are so small.
A second possibility is that the basic strategy of many job training programs, and virtually all welfare-to-work programs, is simply the wrong one. Most programs -- including the successful Riverside program -- have stressed moving individuals into employment quickly, using job search assistance, work experience placements, and on the job "training"; they provide relatively little actual training, despite their name. The underlying assumption is that the basic problem of the unemployed is one of job-finding, and that once individual get jobs they will remain employed. Welfare-to-work programs are particularly insistent on the value of getting any kind of job, and the current political rhetoric in the United States about "ending welfare as we know it" concentrates on pushing welfare recipients into work. This tactic assumes that there are plenty of jobs available to those who want to work and that the appropriate motivation to work -- from either the "stick" of reduced welfare benefits or the "carrot" of increased incentives to work -- will be sufficient. There has been much less attention paid to the problem of enhancing the basic competencies -- cognitive, vocational, and personal -- of job trainees, except in a limited number of intensive and experimental programs.[49]
The success of this strategy is confirmed in the widespread finding that job training programs increase earnings by increasing the amount of employment, rather than by increasing the wage rates (and presumably the productivity) of individuals. But this strategy ignores the reality that the low-skilled labor market for which job training programs prepare individuals is so unstable that -- without the increase in basic skills that would enable individuals to escape the secondary labor market -- they will continue to suffer intermittent employment, low earnings, and the kind of discouragement that leads them over the long run back to marginal employment or welfare. (This is consistent with the finding in Table 18 of benefits declining in years 4 and 5.) Furthermore, the kinds of jobs that individuals leaving such programs can typically get are so dreadful -- with repetitive, boring work, few prospects for advancement, and often harsh and demeaning supervisions -- that it is no wonder that individuals leave after short periods of time. Ethnographic and journalistic accounts have sometimes stressed the difficulty of job-keeping rather than job-finding (Quint, Musick, and Ladner, 1994), and a current experiment is intended to assess the value of continuous services intended to help individuals keep the jobs they find. But whether this is possible without changing the nature of low-wage work is an open question.
The implication of this argument is that, in the interests of greater long-run effects, there should be more attention to the enhancements of skills -- both basic education and job skills -- and less attention simply to getting individuals into employment.[50] Indeed, the long-run evidence on the effects of welfare-to-work programs, summarized in Section III.6 and Table 18, tends to confirm this: the only program without long-run decay in earnings was the Baltimore Options program, which was distinguished from the others by more intensive education and training (Friedlander and Burtless, 1995, p. 144). The most powerful evidence, however, is the contrast between the typical job training benefits -- which decay over 4 or 5 years -- and the age-earnings profiles associated with different levels of education, where the benefits of education expand over time.
[49] The current evaluation of the JOBS program in 7 states will provide a test of the "get-a-job" approach versus skills training, since individuals will be randomly assigned to the two different service approaches.
[50] There is even a small amount of evidence for this proposition. In Fischer's (1995) meta-analysis, the effects of job search increase in quarter 2 and then decay essentially to zero in quarter 4, while the effects of basic education programs are initially negative but then increase through quarters 2, 3, and 4. In addition, the effects of "staged" job search -- in which individuals are screened through job search and then proceed to other activities including education -- also increase, presumably both from screening and human capital effects. See Table B.1.
When they do not emphasize pushing individuals into employment, job training programs sometimes provide (as their name implies) some training in job-specific skills. Sometimes this takes place in classroom settings, and sometimes in work settings or on-the-job training.[51] However, a study of on-the-job training revealed that in a large fraction of these programs (55 percent), there was little or no explicit training going on: employers viewed the program as a source of subsidized labor and used individuals in routine, unskilled work, without any attention to providing either job-specific or more general skills (Kogan et al., 1989). This approach among employers to on-the-job "training", which occurs in a variety of job training and apprenticeship programs,[52] is particularly likely to occur where employers are small and marginal and are pressed for resources. In many local programs, JTPA agencies seem to act as a screen to provide such employers with a steady source of relatively stable, low-cost labor, and can therefore come up with jobs for JTPA trainees -- but these placements have very little training and few long-run prospects.
The quality of classroom-based job skills instruction in job training programs has not, to my knowledge, ever been closely examined. However, here too there are likely to be serious problems. Keeping up with technological changes is difficult enough in the more sophisticated, longer-term programs offered in community colleges and technical institutes, but in short-term job training programs with little funding for capital outlays it must be nearly impossible. Similarly, the problem of finding instructors from industry is a difficult issue for postsecondary educational institutions, and this problem must even more serious in local job training programs with their intermittent offerings and therefore unstable employment of instructors. Many job training programs are operated by community-based organizations, which typically pay low wages. And because such organizations are often principally involved in other activities -- promoting the rights of black Americans, for example, or of recent immigrants, or advocating on behalf of the disabled -- their experience in job training and education and their connections to employers may not be strong. While the quality of job-related instruction merits further investigation, the conditions in many job training programs are not conducive to high-quality training.
[51] There is a long-running debate in the United States about whether classroom-based or work-based training is more effective. The recent infatuation with the work-based apprenticeship systems of Germany and other European countries has led to greater interest in work-based learning even though the pedagogical problems are virtually the same in both settings, as Berryman (1995) has pointed out.
[52] For a description of the generally excellent co-operative education programs in Cincinnati, distinguishing those who view co-op programs as an educational experience versus those who view it as a source of well-trained short-term labor, see Villenueve and Grubb (1995).
In educational institutions, there is currently a great deal of debate about the most effective pedagogies, and a concerted effort by reformers to replace conventional, didactic methods of teaching -- what I have called "skills and drills" -- with other approaches to teaching associated with a very different tradition of "meaning making" that enable students to be more active in learning, that are student-centered rather than teacher-directed, and that use a wider variety of activities and motivation in the classroom.[53] In adult education too, there has developed an orthodoxy of good practice that advises programs to tailor instruction to the interests and goals of adults and to use a variety of instructional methods including more active techniques. Unfortunately, none of this discussion has affected the world of job training programs, where even the existence of a debate about pedagogy is unknown. Job training programs virtually universally use conventional pedagogical techniques based on "skills and drills", with instructors breaking reading, writing, and mathematical skills into tiny sub-skills and drilling endlessly on a series of inherently meaningless sub-skills (Grubb et al., 1992; Grubb and Kalman, 1994).[54] The instruction is particularly bad in programs that have adopted computer-based instruction: while administrators are often quite proud of their computer programs, the existing programs are the worst examples of "skills and drills" converted to the computer screen, with even shorter reading passages, less writing, and more trivialized arithmetic examples than standard textbook instruction (Weisberg, 1988). Here too there is often a pride among job training administrators in distinguishing themselves from what educators do: they will say, for example, that they are "trainers" rather than "educators", and those in charge of computer-based programs well describe themselves as "managers" of the program rather than "teachers". But this pride masks their deep ignorance of pedagogical issues, and results in instruction that is quite horrifying to see.[55]
While there is evidence that conventional didactic approaches are the least effective methods for teaching many individuals,[56] this approach is likely to be particularly ineffective for the individuals in job training programs. Most of them by construction have not done well in many years of conventional schooling, using conventional didactic instruction; why they should suddenly be able to learn from this approach in very short programs with bad teaching is completely unclear. The ineffectiveness of conventional approaches to teaching may be inferred by a study of remedial education in the GAIN program: the only county with an increase in test scores was San Diego, which developed an innovative program to avoid the problems with the "school-like" adult education system. As one administrator described their efforts: "These people had an unproductive experience in school and were not able to benefit. We wanted to avoid the perception that they were going back. We wanted to make it different and make it work for them (Martinson and Friedlander, 1994, p. 41).
The inability or outright refusal of many job training programs to understand pedagogical issues is exacerbated by the problems of hiring instructors. There has been little research about those who teach in job training programs; but the conditions of providing short-term and intermittent programs, often in community-based organizations with low pay, are not conducive to hiring good teachers. Typically, instructors in job training programs are given little preparation in teaching -- a further indication of how little attention is given to teaching. In contrast, within the schooling system in this country the issue of preparing teachers well, and of paying them enough to attract a stable, experienced, and dedicated teaching force, are widely discussed. The fact that job training programs have typically not even raised this question is yet another sign of the unimportance of teaching -- and another contributor to the low quality of instruction.
[53] The issues in "skills and drills" versus "meaning making" as approaches to teaching are complex, since each involves many different assumptions about the nature of learning, the roles of student and teachers, the appropriate competencies to be taught, and the like; for summaries to these differences see Grubb et al. (1992) and Grubb and Kalman (1994).
[54] There are almost surely a few exceptions since the job training world is so large and varied; for a very brief description of one of them, in San Diego, see Martinson and Friedlander (1994).
[55] In the course of observing job training and remediation programs, we saw several -- in particular a STEP program -- that almost caused us to violate good research protocol by complaining to senior administrators about the cruel and unusual practices we observed. Even in the highly-regarded CET program the teaching was completely conventional.
[56] Some direct evidence based on learning outcomes of the superiority of alternatives to conventional teaching is available for elementary students (Knapp, Shields, and Turnbull, 1992); a meta-analysis of writing has shown that the presentational (or didactic) mode and the conventional teaching of grammar are the least effective approaches (Hillocks, 1986); and some specific practices have been confirmed superior, like cooperative learning (Slavin, 1987). However, there is relatively little evidence based on learning outcomes for adults taught in different ways partly because there is relatively little empirical research of any kind about adult education, developmental education in community colleges, and basic skills within job training programs and partly because the efforts to evaluate outcomes of different instructional methods have used inconsistent conceptions of instruction and therefore inconsistent observations of classrooms; see Romberg and Carpenter (1986) for math and see Hillocks (1986)for writing. Another problem is that different approaches to teaching generally emphasize different goals: Advocates of teaching in the meaning-making tradition usually stress "authentic" tasks and higher-order competencies, notoriously difficult to assess reliably, while those following skills and drill are more likely to be content with standardized tests.
Many job training programs, particularly those operated by JTPA and JOBS (which often works through JTPA programs), are highly local. A local decision-making authority, the Private Industry Council or PIC, makes important decisions about the nature of services to be provided, to which groups of individuals, and then establishes the methods for subcontracting with other groups (typically, community-based organizations, educational institutions, and proprietary schools) to provide the various services. The local nature of these decisions is certainly necessary in light of the fact that low- and middle-skilled labor markets are themselves quite local, and therefore programs must adjust themselves to local conditions. However, this also make job training programs vulnerable to local political influence as well. This usually operates to direct funds to particular providers of services, regardless of whether they are effective or not, and makes it difficult for local programs to shift resources from ineffective providers to more effective organizations.
Political interference appears to take place in several different ways.[57] In some cases, providers of training services are represented on the PIC, and seem to be able to direct contracts to their own organizations. In other cases, influential local politicians can effectively threaten to create trouble for a job training program if they do not support a favorite local provider. In still other cases a local job training administration, anticipating political problems, will arrange to allocate resources through non-competitive processes[58] which are essentially rigged so that particular community-based organizations receive funding. In many cases, this kind of interference takes place on behalf of certain groups with particular racial identities -- for example, a local group representing the black community or the Hispanic community (or particular parts of the Hispanic community like Haitians or Puerto Ricans). In other cases the community-based organizations represent women, or older women trying to re-enter the workforce, or the handicapped, or some other particular group. They operate simultaneously as advocates for "their" groups, as sources of guidance and counseling for members of these groups, and as providers of education and training services through public funding. In all these cases, constituents served by such CBOs can exert considerable political pressure. Unfortunately, some of the worst job training programs seem to take place in cities with well-organized minority community-based organizations with political influence; in these cases the difficulty of detecting political interference is compounded by racial tensions, which make it difficult for white administrators at the local or state level to challenge the allocations of funds to ineffective black or Hispanic groups. Conversely many of the best programs seem to take place in rural and suburban areas that are relatively free of such political interference.[59]
The effects of local political interference on the effectiveness of job training programs is difficult to assess. In many cases, it is clear that ineffective organizations are given resources because of political interference, and effectiveness suffers. However, community-based organizations with political influence also include many extremely effective organizations, highly dedicated to the group they serve. My own hunch is that -- as is generally true of the private sector compared to publicly-provided programs -- the community-based organizations providing job training services include some of the best as well as some of the worst providers. The problem with local political influence is that it becomes difficult to eliminate some of the worst providers -- and those who enroll in them are the ones who suffer.
[57] The question of local political interference is widely acknowledged by observers of job training programs, but almost no one has acknowledged the problem in writing -- probably because this problem has racial dimensions that are highly controversial in the United States; for an attempt to describe the problem in one community (Fresno) see Grubb and McDonnell (1991). Other efforts to cope with this problem arise in cases where particular JTPA programs have been investigated for fraud, mismanagement, or ineffectiveness. There has been to my knowledge no real analysis of the local politics of job training programs, however.
[58] It is common in the job training world for local programs to use a competitive RFP (request for proposal) process where the SDA invites organizations to bid on proposals for providing services; the competitive RFP process is intended to increase the number of organizations bidding on the basis of low cost and high quality. However, there appear to be a number of ways to manipulate the RFP process so that certain organizations will be chosen; for example, RFPs can be written so they apply only to specific organizations, or the knowledge that an RFP is rigged for a particular organization will prevent others from applying. One individual commenting on an early draft of this monograph reported that a charismatic PIC director had his life threatened for changing contracting procedures.
[59] For example, in a study of efforts to coordinate job training, welfare-to-work programs, and vocational education, none of the exemplary coordination efforts were in cities because of the dominance of purely political allocation of resources there; see Grubb et al. (1991).
One of the conventions in vocational education and job training is that the labor market value of job-specific education and training is likely to be quite low if individuals are unable to find jobs for which they have been prepared. While there has been relatively little analysis of the consequences of job-related versus unrelated education and training, the little evidence that exists indicates that job-related vocational education does have higher economic benefits than unrelated education (Grubb, 1995b; Rumberger and Daymont, 1984).
In the world of job training, some services convey general competencies -- for example, remediation should enhance basic academic skills that are useful in virtually every job, and on-the-job training and work experience that enhance the personal characteristics required at work should have the same effects. But a good deal of job skill training and on-the-job training is job-specific, and may not benefit individuals much if they fail to find related employment. In one particular data set -- the Survey of Income and Program Participation (SIPP) -- individuals are asked if they have received different forms of job training, and whether this training is related to their current employment. Among those who were in JTPA programs, 49 percent of men and 46 percent of women reported that they used training on their current job; for those reporting having enrolled in CETA, 42 percent of men and 46 percent of women reported that their job training was related (Grubb, 1995b). And -- consistent with the hypothesis that related training has a much higher economic return -- the earnings of men with related JTPA training were on the average 55 percent higher than those with unrelated training, with the comparable figure for women 42 percent; for those with CETA training, men earned 21 more if their training was related to their current job, while women earned 6 percent more.[60] Thus the average economic benefit of job training programs -- the figures that show up in formal evaluations -- are as low as they are partly because they combine the much lower benefits (presumably near zero) of those who failed to find related employment with the more substantial benefits of those with employment related to their job training. The conclusion in the SIPP data that only a minority of individuals are in jobs related to previous job training suggests that, even with some placement efforts, job training programs have not done a good job of placement. Alternatively, if individuals are placed in related jobs with little future, then they would normally shift over time into different occupational areas with greater prospects for mobility. Under either explanation, however, job training programs appear to do a mediocre job of placing individuals in appropriate jobs.
[60] These are results for calendar year 1987. Of course, the quality of the data entering these calculations is much poorer than the data in random assignment experiments: the SIPP data are entirely self-reported and are retrospective, generating problems of whether the reports of having been enrolled in job training are accurate, as well as assessments of job-relatedness. The other independent variables available in the SIPP data are inadequate to control for the various characteristics of those in job training programs; therefore all the coefficients on JTPA and CETA job training programs are negative, reflecting negative selection.
One of the basic characteristics of job training programs in the United States is that, since they enroll individuals with substantial barriers to employment and provide relatively limited training, they aim to place individuals in jobs with relatively low levels of skill and pay. The limited ambitions of these programs -- confirmed by the finding that increases in earnings are modest, and probably decay after 4 or 5 years -- may be "realistic" given their resources, but they still do not help individuals move out of poverty or off welfare. And the emphasis on quick placement into employment, while it generates modest benefits in the short run, reinforces the notion that job training should be a short, one-shot event, with individuals then leaving the job training system.
Typically, job training programs are not linked to other programs, either to other job training programs or to education programs.[61] On occasion, job training programs refer individuals to other programs -- for example, they may refer those in need of remediation to adult education programs -- but in these cases there is little effort to identify which such programs might be most effective, or to follow individuals to make sure they enroll and complete other programs (Grubb et al., 1991). Even in welfare-to-work programs, where there is a caseworker assigned to each individual to make sure that he or she can navigate the array of services offered, individuals often become "lost" when they are referred to services but then never enroll, or never complete the program, or fail to enroll in subsequent programs. As one GAIN administrator in California commented about individuals referred for remedial education, the lack of information about progress means that many clients "fall into the black hole of adult basic education", staying in adult education for long periods of time without much progress and without caseworkers knowing where they are (Grubb and Kalman, 1994). When individuals complete job training programs they are usually referred to employment, not to subsequent education; and of course the need for most trainees to earn a living generally precludes them from immediate enrollment in other education or training. The consequence is that the possibilities for expanding resources to those in job training programs -- for gaining them access to additional training that might provide them with access to jobs of increasing skills levels and pay -- are virtually non-existent,[62] and it is not surprising that the benefits of job training seem to decay over time.
But an alternative way to view education and training is the notion -- sometimes encapsulated in the overworked phrase "lifelong learning" -- that a low-level job training program would be only a start back into the labor market. If job training programs were linked to one another and to education programs, then an individual could enroll in a low-level program, complete it and enter low-skilled and low-paid employment, and then -- at such a point where time and resources permitted -- could continue in a more advanced job training program, or a credential program in a community college, in order to gain access to higher-skilled and better paying jobs. This kind of "ladder" of training and education opportunities would therefore be able to take an individual at any level of skill and then -- in a series of short-term programs rather than a single, one-shot program -- provide access to a range of jobs with better long-run prospects, a possibility to which I turn in the Conclusion.[63]
[61] On the problems in the current "system" of education and training, see especially Hansen (1994); Grubb and McDonnell (1992); McDonnell and Grubb (1991); Grubb et al. (1992); and Grubb et al. (1991).
[62] There are only a few exceptions to the general pattern of job training occurring in isolation from other education and training programs. In some welfare-to-work programs, caseworkers emphasize "self-initiated placement", where an individual can put together a series of individualized education and training programs. And in some cases job training programs contract with community colleges to provide short-term job training; in such cases administrators claim one of the benefits to be that trainees can more easily enroll in regular community college programs -- though the frequency of movement from short-term job training to longer-term educational programs in such situations is unknown.
[63] Part of the history of education in the United States is the development of a coherent system, with clearly-articulated links among different levels of education -- though this system was not complete until the 1920s or so. One could argue that the job training "system" is still in its infancy, since it is barely 30 years old -- and that it will take a longer period of time for a system to emerge.
A possible explanation for the mediocre employment effects of job training programs is simply that there are not enough jobs for unskilled and semi-skilled workers, and the labor market is unable to absorb those who complete job training programs.[64] For example, the finding that welfare-to-work programs in West Virginia were the least effective (see Table 8) was generally attributed to weak economic conditions in those regions. However, in other cases it has proved impossible to blame labor markets. For example, the outcomes across the six GAIN counties of California could not be explained by variation in labor market conditions; and the analysis of JTPA after 18 months found no significant effect of the local unemployment rate on earnings, and only a minimally significant effect of urban location on youth (but not adult) outcomes (Bloom et al., 1993, Exhibit 7.12). In addition, as I pointed out in Section II, labor market conditions have contradictory effects: although high unemployment or low employment growth may make job placement more difficult, it may also cause more job-ready individuals to enroll in job training programs, making placement easier than in boom times when those enrolled in job training programs have the greatest barriers to employment. In general labor market explanations have not been popular among those examining job training programs.
However, it is still possible that labor market conditions may explain the mediocre effects of job training programs. In the first place, there have been relatively few systematic examinations of labor market effects, aside from that undertaken for the JTPA evaluation, and the examination of labor market effects on JTPA may have been marred by insufficient variation in labor market conditions.[65] Second, it is possible that the weak condition of labor markets for modestly-skilled work explains the pervasively mediocre results of job training programs, even if it fails to explain the cross-section variation in outcomes. Finally, it is worth noting that job training programs represent a supply-side solution to the problem of underemployment and poverty -- the assumption that if the skills of the labor force can be improved, or if individuals out of the labor force can be induced to enter it, then employment and earnings will improve without intervention on the demand side. The alternative tactic, of course, is that a supply-side policy should be coordinated with a demand-side policy to increase the demand for modestly skilled workers. Indeed, something like this took place in the last years of the CETA program, when public service employment (PSE) created additional jobs for CETA trainees in governmental and non-profit community-based organizations. Such efforts were denounced as "un-American" since they might substitute public employment for private employment, and were quickly abolished by the Reagan administration. But the notion that demand-side policy in labor markets should be coordinated with the more common supply-side policy remains an attractive idea, and one that might improve the mediocre effects of job training programs.
[64] Equivalently, it may be that individuals completing job training programs find employment at the expense of others who do not -- though, as mentioned in Section II, this kind of displacement cannot be detected with conventional evaluation methods. In economic terms, a shift outward in the supply function for a particular kind of labor (e.g., for modestly-skilled employment) along a stationary and inelastic demand function will result in very little additional employment and in a fall in the wage rate -- and so placement rates will be low, displacement high, and the increase in earnings modest. In effect, job training programs assume that the demand for labor is relatively elastic.
[65] Because there were only 16 sites, there was not much variation in labor market conditions. In addition, the results were carried out only for the 18-month outcomes, not the 30-month results.
Over the past two decades, the results of evaluation youth programs have been especially dismal. Many of the early CETA programs for youth had negative results (see Table 3), with the exception of the expensive Job Corps program (Table 4), and the minimal or negative effects were replicated in the more sophisticated JTPA evaluation (Table 5 - 7). Several of the experimental programs for youth, like JOBSTART and STEP, have proved ineffective despite careful planning and higher costs. These results are particularly discouraging because of the hope that programs aimed at youth could be preventive, by steering young people into paths that would be beneficial in later years.
There are several explanations for the particularly poor results in youth programs. One has to do with labor market conditions: many employers will not hire young people, so under the best of conditions they have a tendency to mill around until reaching their early or mid-twenties. In addition, for many of the moderately-skilled jobs in the sub-baccalaureate labor market, employers will not consider individuals without a high school diploma, effectively condemning drop-outs to completely unskilled positions. Other explanations depend on the special characteristics of youth culture in the United States: this culture, with its rejection of school and discipline and the premium placed on "coolness", may work against job training training programs in ways that do not affect the performance of adults with greater maturity and sense of responsibility. Still other explanations point out that adolescents are still entangled in their families, which may be disorganized and destructive rather than supportive (Quint, Musick, and Ladner, 1994).
Unfortunately, job training programs cannot do much about these factors, However, there may be some systematic failings in the job training programs applied to young people (Granger, 1994; Doolittle, undated). Programs devised for adults may not be developmentally appropriate for adolescents; and the conventional "skills and drills" pedagogy of most job training may be especially abhorrent to them, because of their recent and negative experiences with school. Because of the complex conditions of their lives, young people may need a greater variety of support services -- especially guidance and counseling about such issues as drugs, alcohol, sex, and sexually-transmitted diseases -- that are not usually part of job training programs designed for adults. To be sure, the minimal results of programs designed specifically with a range of services -- like New Chance and JOBSTART -- are not especially encouraging, but most of the efforts to devise better youth programs still concentrate on specifying a broader array of services to cope with the more complex conditions of adolescents' lives. This is essentially the vision I offer in the Conclusion, in which work-based placements would be combined with school-based activities (as well as a variety of supportive services) to provide a variety of learning opportunities for those individuals who have not done well in conventional educational institutions of the "first chance" system.
A final possibility is that the very idea of providing "second chance" programs is flawed. Given the universality of "first-chance" programs -- the elementary-secondary education system in the United States, with its many remedial and compensatory programs, along with a higher education system that virtually guarantees a place for everyone (in open-access community colleges if not in four-year colleges) -- one could argue that the individuals who fail to use the educational system to increase their skills and gain access to employment are by definition those with such serious intellectual, personal, and motivational barriers to employment that no "second chance" system of reasonable cost could possibly help them enter stable employment. Under this argument, the resources that currently go into job training programs should be diverted into improving the "first chance" programs; alternatively, one could argue (as some have[66]) that there has been too much spent already within the first-chance educational programs on preventing school failure, and the education and training system should devote less of its resources to the bottom tenth, or those that have sometimes been demeaned as the "leftovers" (Taggart , 1981).
But this view -- the abandonment of second-chance opportunities, though job training programs and such institutions as adult education and community colleges -- is a distinctly un-American idea. Even now, when the country has swung toward the political right, with many proposals for dismantling important aspects of our welfare and regulatory systems, there are almost no calls to dismantle the education and training system. One reason is that even the most ineffective efforts at training for employment are more palatable than the alternative -- allowing individuals to live at public expense without working. And the effort to build second-chance programs is an expression of the enduring American commitment to equity in some form -- even if that form is not particularly effective. For these reasons this country is unlikely to abandon its current efforts at job training. Daunting as it may be, the appropriate task is that of improving job training programs, rather than abolishing them.
[66] See, for example, Traub (1995), who complains about the inappropriateness of providing so much remediation in colleges.
