The Workforce Training Grant

Executive Summary

The American education system and labor market are heavily biased toward college graduates. High schools function predominantly as college-prep academies, state and federal governments subsidize higher education by more than $200 billion annually, and those with the academic aptitude to succeed on a college pathway are likely to find their skills in higher demand and applicable to higher-paying jobs.

However, for many Americans—most do not hold even a community college degree—and for many career paths, positions that combine immediate on-the-job experience with employer-sponsored training offer the best opportunity to enter the workforce and build valuable skills. Such positions receive little to no public support, and employers often have little incentive to create them. A neutral approach for public policy toward workforce preparation would recognize that employers, not universities, often provide the most socially valuable form of training and would redirect public resources accordingly.

This report proposes how this could be accomplished:

1. Define a “trainee” status, akin to the status of a student enrolled in a college, for any worker whose wage is below an established threshold and whose time is divided evenly between on-the-job experience and formal training.
2. Allow for diverse forms of formal training, including programs operated by an employer; by a consortium of employers, an industry association, or a partnership between employers and organized labor; or by a technical high school or community college when its curriculum is designed in partnership with an employer.
3. Provide a substantial grant to private-sector employers who create trainee positions and provide for their training, on the order of $10,000 per year per trainee. Fund this grant by redirecting subsidies provided to traditional higher education.

This paper expands upon a proposal first published by Mr. Cass at the Manhattan Institute in 2019.

I. The Training Problem

U.S. public policy relies almost exclusively on college to prepare young men and women for productive employment. Within the education system, high schools operate primarily as college-prep academies, and waves of reform have focused ever more intensively on “college readiness.” Federal and state spending on higher education totals more than $200 billion annually and has more than doubled in recent decades. K–12 spending has surged as quickly, but the share of federal dollars allocated to career and technical education (CTE) has fallen from 11% to 3%. The share of high school students earning CTE credit, the share of credit-earners qualifying as CTE “concentrators,” and the average number of CTE credits earned per student have all declined.

This overwhelming emphasis on college as the path to productive employment is a mistake. Only one-third of Americans earn a bachelor’s degree by age 25, and that figure has changed little in the past two generations; only half of young Americans now attain even a community college degree. And among recent college graduates, 41% hold jobs that do not require a degree. All told, fewer than one in five Americans move smoothly from high school to college to career. The disconnect only widens for workers wanting or needing to begin a new career path later in life, at a stage when returning to campus with a knapsack full of books is not only ineffectual but also implausible.

One reason for these meager results is that colleges themselves are not designed to play the role of career preparation for the masses. By overwhelming margins, Americans’ top priority for higher education is employment opportunity, but that’s not the role that institutions see for themselves. As Harvard University’s Claudia Goldin and Lawrence Katz have observed, “The business of colleges and universities is the creation and diffusion of knowledge.” In a 2017 survey by Gallup of more than 700 college and university presidents, only 1% strongly agreed with the statement that “most Americans have an accurate view of the purpose of higher education”; four times as many disagreed as agreed. Nor do most educators have up-to-date experience with relevant technical skills or in industries outside the field of education.

A second problem with pumping students through college campuses has been the labor market. The popular narrative holds that, thanks to globalization and technological change, the modern labor market is rapidly creating good jobs that require college degrees, and the education system needs to keep up. In fact, America has been producing new college graduates at twice the rate its labor market produces new jobs for them. The result of this misalignment is that millions of young Americans are sinking years into education that has little economic value, often accumulating debt in the process. Employers looking for skillsets unrelated to college coursework, meanwhile, find a shortage of trained applicants.

Appropriate pathways for most workers entering new careers, whether after high school or later in life, would focus on technical training coupled with time on the job. This is the approach taken by virtually every developed economy besides the United States. For most developed countries, according to the Organisation for Economic Co-operation and Development (OECD), 35%–55% of high school students are enrolled in career pathways that emphasize technical training, often with a significant on-the-job component. “All countries except the United States have some students enrolled in vocational upper secondary education,” the OECD reports. The United States is excluded from the analysis because it has “no distinct vocational path at the upper secondary level.”

Despite widespread consensus that job-relevant training on a noncollege pathway would benefit millions of Americans, neither the standard private- nor public-sector models are capable of delivering the needed investment.

I.A Public-Sector Limitations

The federal government operates numerous job-training programs with reliably unreliable results. For example, a 2011 report by the U.S. Government Accountability Office (GAO) identified 47 programs operated across nine agencies at an annual cost of $18 billion, all but three of which overlapped with at least one other program. “Little is known about the effectiveness of the employment and training programs we identified because only 5 reported demonstrating whether outcomes can be attributed to the program through an impact study,” explained GAO. “The five impact studies generally found that the effects of participation were not consistent across programs, with only some demonstrating positive impacts that tended to be small, inconclusive, or restricted to short-term impacts.”

More recently, the U.S. Department of Labor (DOL) released its “Gold Standard” evaluation of the Workforce Investment Act (WIA) Adult and Dislocated Worker programs, which are among the nation’s largest publicly funded employment and training programs, serving more than 6 million people, at a cost of $2 billion in 2015. Participants in 28 of the programs’ local areas were randomly assigned for 15 months to groups eligible for either “core services” (e.g., workshops and online assessments), more “intensive services” (e.g., case management and job-search assistance), or intensive services plus “training.” Their behaviors and labor-market outcomes were then tracked for that period plus an additional 15 months.

The study was not able to identify positive labor-market effects associated with training. This happened in part for the surprising reason that, despite being made eligible for more training services, participants in the “training” group were not that much more likely to actually obtain training. Insofar as they did, this had the effect of reducing their work and earnings during the initial 15-month period. During the subsequent 15-month period, earnings were indistinguishable from the group that received intensive services but no training. Fewer than half of those who did pursue training that was linked to a particular occupation ended up with a job in that occupation. Overall:

Though not conclusive, our findings suggest that providing WIA-funded training represented a net cost to both customers and taxpayers during the follow-up period. This cost arose mainly from the earnings forgone when the customers were in training. Early in the follow-up period, when full-WIA customers were more likely than core-and-intensive customers to enroll in training, they worked and earned less. Their quarterly earnings caught up to those of the core-and-intensive group in the latter half of our follow-up period, but these increases did not offset the earnings losses customers incurred while in training. Positive impacts on earnings would have to materialize after the three-year follow-up period for WIA-funded training to be a net benefit.

Likewise, a 2018 DOL evaluation of Job Corps, which spends as much as $45,000 per participant, found that the program “could not demonstrate beneficial job training outcomes.” Median annual earnings for a sample of 231 participants who found placement after training were only $12,105—less than the wages of a full-time, minimum-wage job and less than half the median earnings for all individuals lacking even a high school diploma.

Answering the question, “Why Is the U.S. So Bad at Worker Retraining?” The Atlantic summarized the view of scholars that “programs are too divorced from employers’ needs, too unrelated to workers’ interests, too light-touch, and too limited in their reach, among other flaws.” Or, according to a bipartisan group convened by Opportunity America, the American Enterprise Institute, and the Brookings Institution, “Employers, educators, scholars and policymakers agree: there can be no effective career education without employers.… [T]hat’s the only way to ensure that students are learning skills in demand in today’s job market.”

I.B Private-Sector Limitations

Employers would play a central role in an effective system of job training, but they are not playing that role today.

Estimates of private-sector training investment vary widely and are hampered by a lack of public data; the Bureau of Labor Statistics (BLS) last conducted a survey on the issue in 1995. In 2015, Georgetown University’s Center on Education and the Workforce applied the distribution of training dollars identified in that 1995 survey to U.S. census data on the composition of the workforce in 2013 to create an updated picture of how training dollars are likely spent in the modern economy. The study estimated that $177 billion was being spent in formal employer-provided training and an additional $47 billion for certificate programs and apprenticeships, as compared with $407 billion in spending at two- and four-year colleges. Importantly, the vast majority of formal employer-provided training goes toward workers who have already earned college degrees; only 17% benefits workers holding a high school diploma or less. Thus, “formal employer-provided training typically complements, rather than substitutes for, a traditional college education; employer-provided training should not be viewed as a substitute for college or K–12 coursework.”

While the Georgetown study offers a partial picture at best, relying heavily on a description of business practices last documented more than 20 years ago, it gives at least a sense of magnitudes and relative weights. If anything, broader economic trends are likely to have skewed investment further toward more educated workers in the intervening decades.

A number of factors help to explain the low levels of private-sector investment in training of less educated workers and prevent such investment from accruing to their benefit.

First, firms face a serious problem in attempting to capture a return on their investments in training. Insofar as such training increases the productivity of their workers, those workers can command a higher wage within the firm or by leaving for a competitor. This problem is discussed frequently in the economic literature, going back at least to Gary Becker’s seminal 1962 article, “Investment in Human Capital: A Theoretical Analysis.”

From the firm’s perspective, several potential solutions exist. One, emphasized by Becker, is that firms can invest in the “specific” human capital of workers—skills that are valuable only within the particular firm. If other firms don’t value his training, the worker can’t command a higher wage in the market and the employer can capture the training’s value. A second argument, advanced by proponents of training investments today, holds that workers are more loyal to a firm that invests in them, so the training boosts retention even when the workers might be able to obtain a higher wage by leaving. LinkedIn, for instance, reports that 94% of employees say that they “would stay at a company longer if it invested in their career.”

Whether or not this retention story is correct, it highlights a fallacy at the heart of most discussions about training investment: for purposes of public policy, the goal is not to create value for the employer. In making the case to employers that they should invest in training, an emphasis on the potential return is paramount. But the public interest in training is precisely that workers will develop skills through which they do command a higher wage. Patterns of training investment through which employers capture the training’s value are directly at odds with this objective, so demonstrating that they can do so solves little.

A second problem is scale. Most private-sector workers are employed at firms with fewer than 500 employees, and the smallest firms tend to have the most job churn—most new hires occur at firms with fewer than 50 employees. Most small firms lack the critical mass of workers in a particular role and a particular career stage to support an ongoing formal program of training and skill development. In some industries with widespread unionization, unions play a central role in achieving the necessary scale. North America’s Building Trades Unions, for instance, operate 1,600 joint labor-management training centers, funded by more than $1 billion in annual dues and employer contributions. But only 6% of private-sector workers are union members.

A third problem is international competitiveness. Especially in the manufacturing sector, even firms with the scale to operate major training programs face the obstacle that their overseas competitors benefit from systems where such training is aggressively subsidized. The result is that domestic firms have the incentive to compete by pursuing business models and production processes that require the lowest skill level possible, accepting lower productivity in pursuit of lower cost.

Individuals might pursue and finance training themselves—certainly, they have the incentive to do so—and, in some instances, this happens. The entire higher education system operates on this premise. But where the employer is the appropriate center of the training process, the mechanism breaks down. Are employees supposed to pay employers for training, or perhaps accept submarket wages while receiving it? In one sense, an apprenticeship can be just that: a worker accepts a lower wage than he might attain elsewhere on the understanding that he will develop skills that will subsequently entitle him to a higher wage. But many firms are hesitant to take on apprentices who offer little initial value; and in many industries, the apprenticeship relationship may not be the right one. Many workers, especially those starting with fewer resources, cannot afford to “invest in themselves” this way; and unsecured loans to fund on-the-job training for younger and lower-income workers without a strong credit history are unlikely to be an appealing proposition for financial institutions.

In short, employer-based job training has substantial value to individuals beyond what firms or individuals are likely to pursue, and raising the skill level and incomes of less educated workers has substantial social value as well. Rationales along these lines are, of course, precisely why society invests heavily in traditional education. Yet economic outcomes suggest that policymakers have their priorities out of order. Higher education receives the overwhelming majority of funding, even though more than 40% of college graduates take jobs that do not require their degrees and even though those who do succeed are well positioned to cover the costs they have incurred.

The model that delivers such poor job preparation in a traditional higher education sector not designed for that task—large pools of money attached to learners, paid to the institutions that they choose to guide their skill development—would be better targeted at private-sector employers.

II. A Flexible Solution

The solution proposed here is called the Workforce Training Grant: an open-ended government stipend attached to eligible private-sector workers and payable to any employer placing a worker in a program of combined on-the-job experience and formal skill development.

II.A Principles

The Grant’s structure proceeds from three principles:

1. Policy plus culture. Developing effective noncollege pathways is as much a cultural challenge as a technical one—there is no shortage of effective models in other developed economies, and even in the United States. Better policy will facilitate widespread adoption; but ultimately, there has to be a change of cultural mindset among policymakers, educators, employers, and families about how young people can best spend their time and about what “success” even means. You can’t legislate social change, but you can legislate with an eye toward how policy is likely to affect perception. For instance, it’s important not only to add funding to new priorities but to actively shift it from college to alternatives. It’s important to reallocate authority and responsibility toward those who need to have skin in the game. And it’s important to develop and report metrics that emphasize the goal of full-time employment with rising wages over time.
2. Employers lead. Employers have to be the nexus of the training effort. They are best positioned to understand what skills are required and what training works, to provide on-the-job opportunities, and to coordinate between the various participants. Government should encourage their participation, give them resources to deploy, and then ensure that they are investing alongside so that they feel accountable for results. In addition to delivering the best outcomes, this will contribute to an important cultural shift in how employers view their role and (hopefully, over time) how they design their own operations.
3. Simplicity and flexibility. A funding program should be as simple, open-ended, and flexible as possible, even at the expense of some waste at the margin. Many different groups will potentially benefit from a noncollege pathway—high school students, postsecondary students and entry-level workers, and even mid-career job changers—and the system will ideally accommodate all of them. Likewise, the numerous possible providers of training—high schools, community colleges, employers, employer-created consortia, unions—all need to be eligible. No one yet knows what will work best, and that answer will likely vary by type of worker, industry, and so forth, so the program needs to accommodate and encourage experimentation. Better to monitor and address inappropriate uses where they emerge and accept some misuse initially, than attempt to prescribe precisely the desirable format and preempt all others in advance.

II.B Structure

The Grant should be structured as a per-worker payment that employers receive for employing someone under the conditions defined as workforce training. Other programs exist that provide tax credits for the employment of particular classes of workers (e.g., the Work Opportunity Tax Credit and the proposed ELEVATE Act). But these programs typically target narrow groups with specific formulas. The better approach is to define broadly the circumstances of someone who is employed while in training and designate that person as a “trainee,” essentially the equivalent of a “student” as we recognize someone enrolled in college.

For example: a trainee might be defined as any person who is employed at least 15 hours per week and also engaged in a certified training program for at least 15 hours per week—regardless of the trainee’s personal characteristics and regardless of where the training program is provided. Department of Labor regulations already establish these types of standards for existing registered apprenticeship programs.

An employer would receive a $10,000 per-year payment (prorated) for employing a trainee, disbursed directly to the employer, just as traditional tuition loans and grants are paid directly to schools. Employers would initially register employees as program participants, with verification from employees themselves of their trainee status, after which that status would be tied directly to the tax identification data through which payroll and other taxes are reported and withheld each pay period. As a condition of program participation, employers could be required to participate in a reporting program that tracks the employment status and long-term earnings of employees who begin as trainees.

Employers would control the offering of jobs and related training, but workers would control what program/employment they want to accept. An employer could hire trainees and receive the Grant only if their employment/training offer were the one most attractive to the trainee—if some other firm wanted to offer better training, or a higher wage, or a more attractive career path, the trainee could go there instead. Employers could potentially even employ trainees for “free” but would want to do so only if they saw the trainees as adding some value to the business. (The Grant would be sufficient to pay the worker $13 per hour for 15 hours per week, though the employer would still need to provide a training program or pay the cost of attendance at a third-party program.)

In many cases, community colleges might provide the site for training. Critically, though, colleges could no longer attract public funding only by enrolling a student—rather, their customers would now also include employers, and their success would depend on offering programs that appeal to employers’ needs. The employer would likewise have a greater incentive to engage with the community college in designing a relevant and integrated program of study. In other cases, employers might operate training programs themselves or through industry associations or union partnerships.

Employers should be limited in how long an employee can remain in “trainee” status and eligible for the Grant; the intention is to subsidize the on-ramp stage of employment as someone moves toward well-paying, full-time employment, not to create a class of permanently subsidized jobs. But a trainee should not be limited in the Grant funding that might ultimately accompany him to an employer—the Grant is not a fictional “savings account” to be used once. For instance, a trainee moving to another employer might re-enter a training program; later in life he might transition to a new career and require training anew. The Grant value (say, $10,000 per trainee per year) establishes what the employer can receive for providing a year’s trainee employment, not the amount the government might ever spend on behalf of a particular person. While it is hypothetically possible that a trainee would “abuse” this opportunity and consume excessive training, in practice it is unlikely to ever be in his interest to do so, or to be in an employer’s interest to hire or retain someone behaving in that manner.

II.C Launch and Expansion

An important benefit of the Workforce Training Grant is its scalability. Unlike the typical training program that requires a bureaucracy to design, staff, and run it, employers are responsible for its development and adoption. This means uptake will be gradual at first and then accelerate in proportion to the program’s effectiveness. Policymakers can also place boundaries around the launch, for instance focusing on a limited number of states or metropolitan areas, or certain industries or occupations. But in general, leaving the opportunity open-ended and allowing those employers most enthusiastic to take the first steps is likely the best formula for success.

States should play a role as well. Roughly half of current higher education spending comes from state and local governments, as does most oversight of existing noncollege pathways, community colleges, and public training programs. Insofar as the Workforce Training Grant provides an alternative to subsidized higher education, its funding should likewise come from both state and federal governments. Thus, the best mechanism for any effort to scale program growth gradually may be to establish the program structure with funding for half of the Grant’s value and allow implementation only in states willing to provide matching funds. This would ensure that the program takes root first in the most hospitable environments and that its objective of cultural change is achieved in part through reforms adopted at the state and local level.

III. Key Questions

Within the Workforce Training Grant’s basic structure, numerous policy choices remain.

III.A Training Definition

Trainee. In the context of higher education, the definition of “student” is straightforward: someone enrolled in an accredited program. A workforce-based definition must be more flexible but might start with a basic requirement for substantial time spent each week in both training and employment. This would rule out useful training modes that benefit from concentrated periods, such as a three-week boot camp, followed by periods of more on-the-job and less in-the-classroom experience. An alternative definition might require that at least one-third of paid time each month be spent in training and at least one-third in the workplace.

Some policymakers may be tempted to impose demographic constraints on eligible trainees—for instance, require some minimum level of education, or exclude college graduates, or focus on younger people. But these constraints are likely to do more harm than good. The 25-year-old who never graduated high school, the college graduate who finds few opportunities requiring his degree, and the middle-aged worker laid off from one job and needing to switch careers are all potential beneficiaries. Better to allow employers, through their hiring decisions, to screen who is or is not an appropriate trainee.

Program. Programs fulfilling the requirement for time spent in training would require some form of registration, likely from DOL, but that process should impose minimal constraints; trainees can vote with their feet if training programs are not valuable. A presumptive approval with, for instance, a review after the first two years might be preferable to overly detailed regulation ex ante. In exchange for the generous Grant, employers can be required to provide transparent data on outcomes for participants. Particularly in the program’s early stages, policymakers should err on the side of accepting some waste over preempting models that might prove viable.

The appropriate agency, whether DOL or a state agency to which DOL delegates authority, should require that a program specify at least the following parameters:

• Responsible entity (firm, community college, consortium, school district, etc.)
• Program duration (hours per week/month/year, etc.; total time to completion)
• Overview of curriculum
• Standards for completion
• Expected outcomes (qualifying positions at participating firms, entry-level salary ranges, etc.)
• Any formal certifications that trainees will earn

Consortium. An antitrust exemption may help employers partner to create training programs in particular locations. Likewise, reforms to labor law could allow partnerships between employers and worker organizations outside the confines of traditional unions. Conversely, as a condition of certification, large firms creating programs internally or—especially—in partnership could be required to open participation to smaller firms as well.

High school. The proposal described here envisions that coursework in a technical high school could qualify for certification as a training program. Such certification would offer an excellent opportunity to promote that pathway, encourage states and districts to reciprocate by allowing time on the job to count toward graduation, and encourage employers to hire people still in high school on a part-time basis.

Any approach chosen for the definition of the eligible training relationship will suffer problems of both over- and under-inclusivity. Training relationships that exhibit high potential will inevitably face exclusion. Training relationships that plainly deserve exclusion, whether because of ineffectiveness or active gaming of the system, will slip through. The best policymakers can hope to do is strike a balance that limits errors of both kinds.

But the relevant measure of success is not the elimination of errors; it is the reduction of errors relative to the status quo. Today, no workforce-based programs are eligible for support while tens of billions of dollars flow to ineffective classroom-based programs on college campuses. Introducing a Workforce Training Grant with imperfect initial guidelines, which moves prospective workers out of the broken system and toward programs more likely to work, can expand the number of quality programs eligible for support and reduce the number of ineffective programs receiving funding. Once in place, the program can be refined. But creating it is the necessary first step.

III.B Employment Parameters

Maximum wage. The Workforce Training Grant should not subsidize employment and training for workers already earning high wages, particularly those who have already completed a course of higher education. Thus, society need not support newly hired law-firm associates studying for the bar. The parameters for exclusion would need to be based on some combination of prior education and current earnings.

Minimum wage. Firms could be allowed to pay less than minimum wage to workers enrolled in training, or to apply a worker’s earnings toward funding the training program. Conversely, firms could be prevented from requiring out-of-pocket spending by the worker (beyond wages earned) for costs associated with training, both to ensure that the employer has skin in the game beyond the Grant received and to reduce concerns of worker exploitation that might require more aggressive program review and certification.

Duration. The Workforce Training Grant may also require a limit for the number of years during which a firm can receive it for a specific worker. Neither employer nor trainee would likely want a part-time, training-based relationship to last indefinitely, but a cap after several years may be appropriate to prevent abuse.

Role. Some nexus should exist between roles offered within the firm and those for which the worker is training. For example, a convenience store whose cashier is attending nursing school should not be eligible.

III.C Financing

Grant size. $10,000 is suggested here as an order of magnitude because it approximates the annual support provided to college students, and it would cover fully the wages associated with many part-time, entry-level jobs. While this amount is substantially larger than typical tax credits associated with hiring and training objectives, it offers savings as compared with the amount spent per participant in a federal program like Job Corps.

Grant format. Policymakers frequently default to “tax credits” for programs intended to create incentives for employer behavior. However, direct spending offers several significant benefits: it is easier to prorate so that employers can receive payments through their regular payroll/tax withholding cycle; it retains equal value regardless of a firm’s tax rate or liability; and it resides appropriately among budgetary outlays. The Grant should be understood as a payment from the government to the employer, “earned” by virtue of employing the trainee; not as a fixed amount of money held and then spent by a trainee from a personal account.

Source. Funding for the Workforce Training Grant should be redirected from within the more than $200 billion spent by federal and state governments on higher education each year. The allocation should shift gradually and predictably: if half this total were shifted over ten years (roughly a $10 billion cut to college and a $10 billion increase to noncollege each year), colleges and their students would have time to adjust while states, districts, community colleges, and employers would have to plan for standing up alternatives. Likewise, ample funding for mid-career trainees exists within the $18 billion that GAO found that the federal government spends ineffectually each year—an amount that could potentially support 1.8 million Workforce Training Grants annually. While the federal government can shift only a portion of higher education funding itself, states should be allowed (or, as a condition of participation, required) to supplement the Workforce Training Grant’s value with their own funding.

Both the new funding for employers and the transfer of funding from the higher education system are necessary for a more effective system. As noted, community colleges may ultimately play an active role in this new system, but their attention must turn from enrolling students directly toward partnering with employers. One source of funding will need to decline alongside the other’s increase if a significant change in behavior is to occur.

III.D Accountability

After a program is registered and active, accountability is critical. The standards that responsible entities set in registering their programs, and the results they achieve, should be public. The agency with oversight should require data on trainees enrolled, completion rate and time, and post-completion outcomes. Based on data submitted in registering a program and registering trainees, an agency should be able to monitor whether the program is operating as designed—for instance, what share of trainees are retained and how long they take to complete the program, what jobs at what wages they move into, what credentials are earned, and so forth. Tracking and public reporting of (aggregated) payroll data for participants should persist at the five- and ten-year marks.

Regulators should also look to trainees themselves: a mechanism equivalent to the Occupational Safety and Health Administration (OSHA)’s whistleblower policy should allow a trainee to raise concerns that a program is not operating as promised and prompt follow-up scrutiny to compare outcomes with initial commitments made during the registration process. Regular surveys of trainees could provide an additional check. Importantly, trainees should be aware that they have been registered as trainees, and their paychecks should show the amount of Grant money paid on their behalf to the employer each period.

Conclusion

A rebalancing is in order. Shifting funding from colleges and universities to employers may appear unappealing at first, but it is best understood as a reallocation from one training provider to another. All are entities that might hypothetically equip less educated workers with valuable skills that will accrue to their own benefit in the form of higher wages. None will do so for free. Of the providers, available evidence suggests that the latter (employers) can do a better job than the former (colleges); to pay only the former is backward in principle and has yielded poor outcomes in practice.