One of the great policy questions of the 1990s is how to increase youths' skills to meet the needs of the current workforce. National blueribbon panels have written about the need for young workers with academic skills but complain of youths' poor academic skills (Committee for Economic Development 1985; National Academy of Sciences 1984; National Center on Education and the Economy 1990). Business leaders receive prominent coverage in the mass media for placing much of the blame for young workers' inadequate skills on schools. Yet critics have been skeptical about whether employers really need workers with high academic skills (Ray and Mickelson 1993).

Economists and sociologists offer conflicting views of employers' needs. Neoclassical economists assume that employers seek to maximize productivity, shop around for the best workers, and pay workers according to their skills and productivity (Becker 1964; Heckman 1994). In contrast, sociologists contend that employment structures constrain access to jobs and wages. In a wellknown study, Berg (1971) showed that access to jobs is constrained by educational credentials and that employers often do not need or have jobs that use the amount of education they state in their formal requirements. Subsequent research supported this argument (Attewell 1987; Levin and Rumberger 1987; Shaiken 1984; Squires 1979).

This dispute remains unresolved partly because of difficulties in providing suitable evidence. Economists say that employers need workers with skills, but needs and skills are not easily measured, so empirical studies have tended to examine indirect proxies. Years of education is the usual measure of academic skills, yet it is clearly a poor indicator, since many high school graduates lack eighth-grade mathematics and reading skills (National Assessment of Educational Progress, NAEP, 1985).

Economists have measured employers' "needs" by the wages paid for a job or workers' education or test scores (Murnane, Willett, and Levy 1995). They have rarely distinguished between stated requirements and actual needs. Sociologists like Berg have stressed this distinction: They have measured "needs" by rating the skills required to perform the tasks of jobs and have found that employers pay excessive wages for workers with "unneeded" education. Although some sociologists concede that wages bear some relation to employers' needs, the relationship is far from perfect, since wages are also affected by pay hierarchies; compensation systems; and norms of age, gender, status, and so on (Althauser and Kalleberg 1981; Jacobs 1989; Rosenbaum 1984), including job-evaluation systems that explicitly constrain pay levels by the imputed value of particular credentials (Bellak 1984; Cappelli 1991; Rosenbaum 1984). If wages are a poor indicator of employers' needs and years of education is a poor indicator of skills, then studies of the relationship between wages and years of education are not good tests of economists' contention that employers need good academic skills or, conversely, sociologists' contention that educational needs are overstated.

This article takes an alternative approach: It uses qualitative evidence to examine the concepts of skills and needs in greater detail. First, rather than quantitatively measure academic skills by the usual crude proxy (years of education), we qualitatively describe employers' reports of the specific academic skills they say they have difficulty getting and under what conditions these needs arise. If employers are able to identify specific needed academic skills and to indicate the relevance of these skills to job conditions, then Berg's contention of excessive credential requirements would be unsupported.

Second, rather than identify employers' needs through wages, we examine two other indicators of needs: (1) whether some employers take costly actions to recruit or retain workers with skills or, conversely, to adapt to workers' low skill levels and (2) whether employers who state that they need workers with high academic skills are more likely to take such costly actions than are other employers. We are particularly interested in actions that are discretionary and are taken at employers' own initiatives, for we expect that such actions are a less ambiguous indication that employers actually need these skills than are wages. If employers take costly actions to recruit or retain workers or to adapt to workers' skills and if these actions are taken by employers who report difficulty getting skilled workers, then Berg's contention that employers do not actually need additional skills would be unsupported.

Third, regardless of whether analyses support the foregoing contentions, sociologists may still be correct about labor market structures-that employers limit access to jobs through restricted channels. If employers actually take special actions to recruit and retain skilled workers, their special actions may introduce labor market structures that stratify workers and may limit workers' access to jobs. The free and unfettered market assumed by economists ignores these realities.

Our findings contradict aspects of each model. Contrary to Berg's model, we found that some employers do have clear needs for specific academic skills, that these needs occur under relevant job conditions, and that employers engage in costly actions to get those skills. We also found that these costly actions are more common among employers who express the need for skilled workers. For instance, our findings suggest that employers sometimes use unskilled entry jobs to screen people for higher jobs ("on-the-job screening"). When Berg (1971) and Squires (1979) found workers overqualified for their present job tasks, it might have been because employers used these jobs for on-the-job screening for higher jobs that would actually use these skills.

However, contrary to the economic model of unstructured labor markets, we found that some employers' efforts to respond to their need for skills create labor market structures. Employers increase supervision, decrease the complexity of jobs, instigate special job accommodations, and construct recruiting linkages with some schools. In the process, labor markets become more stratified. Of particular interest, some employers create linkages with schools that constrain hiring, so individuals outside these networks have difficulty gaining access to jobs. Although our findings support economists' premise that employers need workers with academic skills, they also support sociologists' contention that employers' actions sometimes create labor market structures that channel access to jobs.

While much sociological research has shown how workers use informal personal networks to learn about possible jobs, few studies have investigated employers' use of networks (Granovetter 1995:152). Some employers in our sample said they used networks for recruitment. In this article, we describe these networks and employers' reasons for participating in them. The Conclusion explores some distinctions between personal and institutional networks.

In sum, we sought to discover elements of the hiring process through qualitative methods. Despite the many quantitative studies of hiring outcomes (Bills 1992, Gamoran 1994), few studies have examined the hiring process, per se (Bills 1988a). This study sought to discover what actions, if any, employers take in pursuit of their stated needs for skills and to assess whether employers initiate actions that seem to be motivated by these needs. Open-ended interviews are a good way to obtain such information, and they provide opportunities to discover previously unidentified actions that have not been in the answer sets included in surveys of employers.

Using these methods, we examined the following propositions:

1. Some employers report the specific academic skills they need and indicate job conditions that make these skills necessary.

2. Some employers take costly actions to recruit and/or retain young workers with academic skills or adapt to workers with low-level skills.

3. Employers who state their need for skills tend to take such costly actions.

4. Some employers limit access to jobs through social networks.

Previous research did not allow us to be specific about what skills, job conditions, actions, or social networks we would find, so these analyses must be seen as an exploration, not a test, of the four propositions.

EMPLOYERS' COMPLAINTS AND BEHAVIOR

Although there is little doubt that many high school graduates lack strong academic skills (NAEP 1990), sociologists have raised some doubts about whether employers really need bettereducated workers. Berg (1971) contended that employers' reliance on education in making hiring decisions is not based on the actual skills needed for jobs. Rather, he argued, it fulfills an organizational desire for some kind of sorting criteria. Berg, Squires (1979), and Collins (1971) reviewed numerous studies that showed that educational attainments are unrelated to workers' productivity, turnover, or absenteeism.1 Expanding on this argument, Collins (1971:1018) concluded that "employers tend to have quite imprecise conceptions of the skill requirements of most jobs."

More recent multivariate studies, however, found relationships between academic achievement and productivity. Many studies in personnel psychology have demonstrated that cognitive ability is the strongest predictor of on-the-job performance in many occupations (Hunter and Hunter 1984). Similarly, econometric analyses have shown associations between test scores and performance (Barrett and Depinet 1991; Bishop 1993; National Research Council 1989) and between course work (and academic skills) and wages and employment (Cameron and Heckman 1993; Daymont and Rumberger 1982; Gamoran 1994; Kang and Bishop 1986; Murnane et al. 1994). After an extensive review, Gamoran (1994) concluded that the preponderance of evidence suggests that there is a positive relationship between academic schoolwork and labor market outcomes.

Yet it is not clear whether employers act on these relationships. Research has found that employers fail to reward high school graduates for academic skills in terms of hiring, better jobs, or better pay (Bills 1988b; Crain 1984; Griffin, Kalleberg, and Alexander 1981; Kang and Bishop 1986; Rosenbaum and Kariya 1991). Using data from the National Longitudinal Study of the High School Class of 1972 (NLS:72), Griffin et al. (1981) found that aptitude, class rank, and other school-performance measures have small and often insignificant effects on unemployment and job attainments of high school graduates who directly enter the workforce. Meyer and Wise (1982) showed that rank in the high school class of 1972 had insignificant effects on wage rates two years after graduation (1974), and Willis and Rosen (1979) observed that increased mathematics and reading scores of high school graduates slightly lowered the wages of their first jobs. Despite their claims that they need workers with academic skills, employers do not offer immediate rewards to high school graduates with better academic performance. It is interesting that although grades did not improve the wages of new high school graduates in the 1982 cohort of High School and Beyond, they had a strong payoff for these graduates' earnings 10 years later (Rosenbaum and Roy 1996).

Ray and Mickelson (1993) indicated that studies need to consider employers' actions, which sometimes convey messages that contradict executives' speeches. As if responding to this point, some recent research has looked at actions that employers take to get better workers. Studies of employers' responses to school-work programs have found that employers have a limited commitment to such initiatives, offer few positions to students, and have low perseverance in these programs (Bailey 1994; Lynn and Wills 1994; Pauly, Kopp, and Haimson 1995). Bailey concluded that employers' behaviors raise some doubts about their commitment to skills. But these were studies of employers' responses to special programs, not the ordinary actions that employers routinely take to hire workers.

Nor do employers respond to their purported problems in obtaining workers with skills by providing academicskills training. Zemsky (1994) noted that employers express highly negative views of the academic skills of high school graduates, but they do not use training programs to redress these deficiencies. In a survey of 2,800 employers, Boesel (1994, Table 2) discovered that 71 percent provided training to their employees, but less than 3 percent provided basic training in academic skills. Another survey of 3,000 employers found extensive complaints by employers about academic skills, but little indication of tuition benefits or remedial programs to address academic shortcomings (U.S. Bureau of the Census 1994). Commenting on these findings, Cappelli (1995:1) noted, "There has been a fair amount of noise in the business community about establishments having to provide remedial training. But it does not appear that they're doing it. It appears to be just noise." Cappelli raised the same issue as Berg (1971) did: Do employers really need academic skills, or are they complaining for other reasons?

Bowles and Gintis (1976) interpreted findings like those just mentioned to mean that employers do not really have a need for better academic skills and that employers' calls for academic skills are, at best, unwarranted and, at worst, a means of accomplishing some kind of social sorting, social control, or reduction in pay. Although empirical tests have not supported Bowles and Gintis's contention that employers need compliance, not academic skills (Cappelli 1992; Olneck and Bills 1980), these studies suffered from the same problems of measuring "needs" and "skills" noted earlier.

METHODS

To discover whether employers actually value academic skills and act in accordance with their valuations, researchers must go to the right informants and use methods that are sensitive to the task. Rather than rely on the views of corporate executives (as blue-ribbon panels do), we interviewed the plant or office managers who actually hire entrylevel workers. In the less frequent cases in which human resources departments handled the hiring of high school graduates, we interviewed managers in these departments.

To gain detailed information about the hiring process, we conducted one-hour interviews in employers' offices. Although observational methods would have yielded even better information, it is difficult to conduct observations in many organizations. Employers' detailed statements about their behaviors are a useful way to get at the intervening processes in a large number of organizations.

To obtain good qualitative data about the nature of job conditions and employers' behaviors, face-to-face interviews were required. Although the decision to conduct interviews reduced the number of respondents we could contact, our sample size (N = 51) was larger than those of some other qualitative studies (Bills 1988a; Borman 1991) or those in than any city in surveys by Lynn and Wills (1994) or Kirschenman, Moss, and Tilly (1995). Our sample was drawn from the roster of members of two local Chambers of Commerce, one in Chicago and one in Chicago's western suburbs. While membership in these associations might have introduced uncertain distortions on generalizability, the associations' support gave us a response rate of over 90 percent in both sites (better than the 65 percent and 55 percent in the surveys by Lynn and Wills, 1994, and Boesel, 1994, respectively).

Since we were interested in how employers' attitudes limited youths' access to jobs in the primary labor market, we excluded businesses that are likely to offer only "youth jobs" without opportunities for advancement (such as restaurants and small stores). The sample cannot be considered a random sample of all employers, so statistical analyses would have been inappropriate. Yet it includes employers that have a wide variety of the kinds of entry-level jobs that are potentially available to high school graduates in graphics, manufacturing, skilled trades, financial services, office work, and so forth. These are mainly ordinary firms, not leading-edge, best-practice ones that are known for their high-skill demands (Murnane and Levy, 1996), but they may offer the possibility of access to the primary labor market. They are appropriate for our purposes of discovering the range of academic skills that employers may need, the job conditions they may provide, and the range of costly actions they may take to obtain these skills.

Twenty-three of the 51 employers in this sample were located in the city of Chicago, and 28 were located in suburbs west of Chicago. The sample was composed mostly of small or medium-size firms, but it included a few large ones; the companies ranged in size from fewer than 10 to more than 80,000 employees.

FINDINGS

This section probes the sociological contention that employers overstate their need for skilled workers. It investigates whether employers report a need for workers with specific academic skills and whether they indicate the relevance of specific academic skills to specific job requirements. It then examines if employers take any costly actions to realize their stated needs and, if so, what strategies they use and whether these actions are taken by those who report difficulty finding workers with academic skills.

Skills Needed for Entry-Level Jobs

Though one should be skeptical about the needs voiced by the top executives on American blue-ribbon panels, our first-line managers expressed similar concerns. Of the 51 urban and suburban managers who were interviewed, 35 stated that basic academic skills in mathematics and English are needed for the entry-level jobs they are seeking to fill, and many described specific job conditions or tasks that demand these skills. They were most certain about the requirements of their jobs after they had experience with workers with poor skills.

Mathematical skills. Thirteen of the 35 managers described the tasks that require mathematics skills in their entrylevel jobs. Most reported that the jobs require workers who can do simple arithmetic and sometimes add fractions and that some jobs require algebra and trigonometry. The general manager of one Chicago steel manufacturing company said that the general-labor jobs in his plant require "the concept of adding or reading a ruler or tape measure." He added: "We would like to hire people who know eighth-grade math, such as knowing the difference between a fraction and a decimal, but kids aren't getting that from high school, and we generally don't see that level of knowledge." A manager at another Chicago manufacturer complained that many workers have fallen short in basic mathematics. When asked what experiences she has had with young employees, she recounted the story of a young man she just hired to work full time in the shipping department:

He came in my office and said, "You know, down in that shipping room you've got a lot of numbers out there." I said, "We've got a lot of numbers out there? Well yes, I guess we do." And he said, "I don't know a lot about numbers." I said, "Oh, do you want to learn?" He said, "Yeah, I think so." I said, "Have you noticed that there are periods between some of the numbers?" He said, "Yeah, what's that all about?" I said, "Where were you when they learned decimals in school?" He said, "I must have been absent."

English skills. Another 10 of the 35 managers reported that reading, writing, and communication skills are needed for their jobs and that their employees who are high school graduates do not even have simple skills. One office manager in a small suburban graphics company told of a secretary "who tried to spell quick with a 'w.' She didn't know that . . . all words that have 'q' need a 'u'!" Echoing an often-reported condition, a recruiter for a Chicago insurance company said that applicants for data-entry and claims-adjustment jobs "can't understand some of the questions on the job application" and that some men bring girlfriends to fill out the applications.

Another Chicago manager unhappily stated that many of his applicants for unskilled labor jobs "can't read and write beyond, I suspect, a fifth- or sixth-grade level, and when they can read, they certainly can't comprehend what they have read." This is often a serious shortcoming because many employers report that their jobs require better than eighth-grade reading and writing skills. A manager for a suburban manufacturer observed that "today's high school graduates don't comprehend as much. It takes them longer to catch on to instructions, and they can't read manuals for instructions as well as they used to, which are written at the 12th grade level." Another manager noted the need to find workers who can "put together two or three sentences in a complete thought."

Both mathematical and English skills. Another 12 managers cited high school graduates' problems with both mathematics and English, saying that both types of skills are needed in today's entry-level jobs. A manager at a Chicago metal-parts manufacturer complained that even though his entry-level jobs require only seventh-grade reading and mathematics skills, he has a "terrible time getting even a 10 percent yield" for these skills in the applicants he interviews.

Of the 16 managers who stated that they had no need for employees with academic skills, 11 said that their entrylevel jobs simply require no skills, and 3 more said that their jobs require occupational skills, but not academic skills. A suburban production manager noted: "There aren't a lot of qualifications other than wanting to work." A suburban plant manager observed that academic skills are actually counterproductive; for such people, "within a year or so, they'll get bored and move on." A manager at a Chicago custom-gearing manufacturer reported that his company needs people with "technical ability . . . [for] grasping the skills needed for machine jobs."

Promotion to Higher-Level Jobs

Some managers noted that although academic skills are not needed in their entry-level jobs, these skills are needed for higher jobs in their companies that entry-level workers can move into. Even though these entry-level jobs are largely in the secondary labor market, in that they offer minimal pay, benefits, and job security, they can sometimes lead to the bottom rung of a career ladder to better jobs. Doeringer and Piore (1971:167) referred to such jobs as "secondary jobs . . . attached to internal labor markets."

Employers note that upward movement is possible from such entry-level jobs only if workers possess adequate basic skills at the time of entry. Of the 35 managers who stated that they need workers with academic skills, 17 (48.6 percent) reported that although their entry-level jobs are undemanding, they allow some workers to move into higherskilled jobs, and those jobs do require academic skills. Indeed, some employers prefer to recruit for their skilled jobs from entry-level workers who can learn the firm's procedures and techniques by observation.

When asked if entry-level unskilled employees could advance into higherskilled jobs in the future, a manager at a Chicago manufacturer answered yes, but only if they have basic academic skills to begin with:

They also have to have the ability, then, to understand math pretty well. Have to be able to add and subtract and multiply. They have to be able to subtract dimensions.... They have to be able to measure something like .236 plus or minus .005. They have to understand what that means. They could be someone in the flamecutting department, who would be cutting steel plate. In that case, they'd have to have the ability to run a small computer and the read-out on it and so forth. . .. Some of that certainly can be learned by someone without a high school education, but a high school education is very helpful in those areas.

But the ladder cannot always be climbed. When workers in the lowest-level jobs lack basic academic skills, as they often do, they are stuck in these jobs. In a Chicago commercial printing firm, workers with deficient skills are fine in entry-level jobs, but they cannot be promoted to better jobs. As the manager of that firm put it:

There are some [reading and math] deficiencies, but none that would probably keep them from performing their entrylevel jobs. [But] it may affect them, you know, when it comes time to move up the ladder.... Once they get on to a piece of equipment, once they're performing a job that requires regular reading of customer job tickets . . . or written instructions from their supervisor, then it could become more of a problem.

Some managers reported that they prefer to fill their more advanced jobs from within by promoting entry-level workers. A manager for a Chicago manufacturer stated that this firm has to recruit skilled workers from within, since the workers need highly specific "training-and usually that training has to be in-house because it's a specialized type of training." In effect, while few skills are required for the tasks in these entry-level jobs, these managers expected workers to be able to learn from the experience, since that is the main way they can fill their skilled jobs. On-the-job training is a central part of these jobs (cf. Doeringer and Piore 1971).

On-the-Job Screening

Other employers used their entry-level jobs for what may be called on-the-job screening. A manager at a printing plant reported that his firm uses entry-level jobs as a way to discover if workers have the ability to advance in the future:

Well, one of the questions I ask the pressman [the employee's supervisor] is, "Are they fast learners?" I ask, "How are they doing? Can they read and write? Are they picking up on the math and the instructions fine?" If that's true, there's no problem. I ask [their supervisors], "Hey, are these our future pressers?" They say, "Yeah." So that's the answer I'm looking for. That's what I'm looking for when we're hiring somebody: Are they going to be able to go up the ladder and become the feeder and the second man, then up to the first-man spot?

In entry-level jobs that offer on-the-job training or screening, if workers do only what is demanded by their simple daily tasks, they will not need academic skills. But if they can do only these daily tasks and lack academic skills, then they will not advance to more demanding jobs, which employers expect to fill from these positions.

Jobs that offer on-the-job screening may illustrate Berg's (1971) findings on the artificially high premium placed on academic skills, but they suggest a possible limitation of his study. That is, Berg may have found workers "overqualified" for the task demands of their present jobs because employers were preparing these workers to advance into higher jobs or were testing their capabilities. To the extent that employers use some jobs to train or screen workers for higher jobs, their job requirements will include skills needed for the higher jobs, but not for the entry-level jobs. Indeed, if reformers want employers to offer more career ladders, then increased entry criteria may be required, even for entrylevel jobs with simple tasks. Although we must be skeptical about employers' claims and cannot examine their actual rates of promotion from entry-level jobs, our study suggests that a complete test of Berg's contentions about excessive qualifications requires researchers to obtain measures of the probability of advancements and the academic skills required for higher-level jobs.

In sum, most of the managers (35 of 51) reported the need for workers with specific academic skills and indicated the relevance of specific academic skills to specific job requirements. They stated that either their entry-level jobs need those skills or that their entry-level jobs allow access to higher-level jobs that require those skills. Employers who stated that they need academic skills tended to have jobs that demanded specific academic skills or did on-thejob screening for advancement. The relationship to the potential for advancement is of particular interest. Many of these entry-level jobs fit Doeringer and Piore's (1971) model of secondary jobs attached to internal labor markets. The managers reported that they seek to fill these jobs with workers with adequate academic skills who have the potential to be trained for more demanding positions in the future.

Employers Demonstrate Their Need for Skills

While most employers state a need for academic skills, we still do not know if these needs remain merely at the level of discourse or if they produce real changes in employers' recruitment activities and on-the-job procedures and policies. If employers loudly cite their needs but ultimately make no real efforts to act on them, then one must be skeptical about their stated needs. It may be assumed that they are just sounding off about "needs" that pose no real problems to them. However, if employers actually take actions that entail extra efforts or costs to ensure the selection of better academic skills at the recruitment stage, or if they compensate for present employees' shortcomings in skills (such as by actions that are expensive or that require additional supervision), then one would be more inclined to believe that they have some commitment to these concerns.

In-depth data allowed us to investigate these questions. The following sections explore whether such actual timeconsuming and/or expensive recruitment and retention activities occur. We found that employers respond to their need for skills by taking four kinds of actions not noted in previous research: They increase supervisors' responsibility to explain tasks; adjust jobs to match workers' skills; offer accommodations to hire and retain workers with valued skills; and invest in linkages with potential sources of labor, like schools, to get prescreened workers. The last action investing in institutional linkages - is of particular interest, for it indicates that in seeking to achieve the productivity goals that economists describe, employers take actions that structure the labor market and limit access to jobs to individuals from certain institutions.

Strategies to Increase Retention

When faced with workers who fall beneath a certain level of academic skills on the job, do companies undertake expensive actions that are designed to compensate for their workers' poor skills? If they do, we could infer that these companies have some commitment to their professed need for academic skills. We found that employers take three types of compensatory actions: increasing supervisors' responsibilities to assist and supervise less skilled workers, simplifying job tasks to match workers' poor skills, and accommodating good workers when they come along.

Increasing supervisors' responsibilities to assist and supervise less skilled workers. Many employers assign more experienced-and expensive-workers the task of assisting less skilled workers in performing their jobs, explaining the tasks in minute detail, and supervising their performance more closely. Often this is an additional task for a supervisor or manager. This kind of strategy occurred in 11 companies in our sample-in 9 of the 35 (25.7 percent) whose managers stated that their companies needed workers with academic skills, but in only 2 of the 16 (12.5 percent) who stated that their companies did not. Apparently, those who said that their companies had jobs that required such academic skills were more likely to increase supervisors' responsibilities.

A manager at a Chicago manufacturer noted that her firm compensates for high school graduates' lack of basic reading and mathematics skills by repeatedly spelling out each task that must be done:

We find if they can't understand the reading, they have to have an illustration, like we'll take a gauge and show them where it has to be. Instead of the workers being able to say to themselves, "This part has to be made within thousandths of an inch" and taking the part and measuring it to those thousandths, the foreman has to go over to the employee, pick up the part, pick up the gauge, set the gauge, and say, "If it does this, OK. And if it doesn't, not OK" and has to set up a scenario each and every time.

The plant manager of a small company succinctly summed up this strategy when he observed that he "basically just spend[s] time with them on the floor with a ruler and show[s] them the basic marks, and show[s] them how it works."

Simplifying job tasks to match workers' poor skills. This strategy often requires additional costs. Of the 51 managers we interviewed, 23 (45.1 percent) reported that they must adapt job tasks to make up for the basic academic skills that their entry-level employees lack. Although 20 of the 35 (57.1 percent) managers who stated that the jobs at their companies need academic skills adapt job tasks to respond to entry-level workers' shortcomings in skills, only 3 of the 16 (18.8 percent) who did not mention such needs take such actions. Apparently, those who said that they need employees with academic skills are more likely to take these actions.

Many of those who were interviewed were matter-of-fact about the need to make jobs easier, as when one plant manager in a Chicago manufacturing firm said, "Yes, [simplifying] is a must. What I've done is gone all the way down to the grammar-school level to get them to understand simple, simple math." The plant manager at a suburban metal fabricator stated that his company has taken the ultimate step in bypassing workers' poor reading skills: "We have eliminated a need for math and reading skills altogether; every instruction we give workers is now verbal." A plant manager at a Chicago manufacturer agreed: "[We] only give people instructions for a certain amount of instructions at a time. Spoon-feed them a little bit." Both these managers thought that such "spoon-feeding" is an additional and costly burden that they would prefer to avoid, but they are struggling with problems of poor skills and find it is an unanticipated cost. There is much concern about the low-paid, low-skilled jobs in the work world and much criticism of employers for not offering better jobs. In these examples, however, we see employers who feel compelled to reduce the skill demands of entry-level jobs because of workers' limitations.

Altering job conditions or rules to accommodate good applicants or workers. Such accommodations can occur either at the time of hiring (such as when an employer changes the hiring schedules to meet the needs of a valued applicant) or as a way to retain valued employees (such as when an employer allows an employee to work at home, rather than at the job site).

Of the 51 managers who were interviewed, 16 said that their companies try to accommodate valued employees in some way-13 of the 35 (37.1 percent) who reported the need for workers with academic skills, but only 3 of the 16 (18.8 percent) who did not. Again, those who stated the need for academic skills were more likely to take these actions.

To accommodate skilled employees, the manager of a Chicago printing firm said that he has sometimes created part-time positions to hire workers with good academic skills, even when he really needs full-time workers and the part-time arrangement requires extra efforts to cover uncovered hours. He has accommodated many high school seniors this way, giving them part-time jobs they can do during the school year and biding his time until they graduate. As he put it:

There have been times in February, March, April when we may have a full-time position [open], but try to get by with a part-time [student], knowing that [the student] will be graduating in May and that we can then make the position full time. We'll do that to accommodate a high school student with good academic skills.

Meanwhile, the supervisor of the pressroom at a suburban print shop said that he tries to hire applicants with good academic skills, even when the company has put a cap on full-time hiring. He does so by giving part-time jobs that turn into full-time jobs later under the guise of "summer help." He is willing to find "loopholes in the system" when he finds someone with good skills because such applicants are rare.

Once a worker proves to be a valuable employee, supervisors are loath to lose them. To retain valued workers, some employers will accommodate their workers by allowing them to work flexible hours, at home, or in jobs they ordinarily would not hold. When the manager of a manufacturing firm in Chicago received a call from a valued former young worker who was moving back to the city, he enthusiastically asked, "How soon can you start?" Although it was a small firm and it did not have a job vacancy, he added that the company would "reorganize to have her back." The manager of a suburban metal shop explained: "When you see a guy or girl [is working] out well, has a good work ethic, you hate to let [him or her] go. I'll find a place for them. Because somewhere down the line I'm going to need that person again." This manager said that even in periods of slack demand, he will accommodate a skilled employee.

In sum, employers use various methods to compensate for their workers' low levels of skills, such as assigning oversight and training responsibilities to supervisors or by adapting job tasks to match the workers' poor skills. Moreover, employers accommodate employees who possess valued capabilities. Of the 51 managers we interviewed, 30 (58.8 percent) used at least one of these methods.

Recruitment Strategies

Employers participate in several different types of contacts with schools and other institutions to seek skilled high school graduates for their entry-level positions and use some of these linkages to provide information about prospective workers. Some activities, like apprenticeship programs, do not involve schools, but they are sometimes related to the need for workers with academic skills. School-related activities range from minimal involvement (job fairs) to high involvement (long-term links with teachers). Some of these linkage activities seem to be a form of investment in prescreening.

Apprenticeships. Although apprenticeships have been praised as a type of training (Hamilton 1989), in Germany, apprenticeship programs also have a screening function (Faist 1992; Rosenbaum 1992). The managers of four of the companies use apprenticeships, and all four stressed that apprenticeship programs are a dependable way to screen their applicant pool for academic skills. Asked whether his company has trouble finding high school graduates with sufficient mathematics and reading skills, one manager of a Chicago manufacturing firm referred to his use of an apprenticeship program for hiring young, entrylevel machinists-in-training: "They won't be in here if they don't have [academic skills]." Even before we asked about academic skills, this manager explained that the apprenticeship test of academic and job skills creates an "up-or-out" situation. For entry-level workers to become apprentices at his firm, they either have to pass the test that the apprenticeship program "administers and then they are enrolled in the apprenticeship program, or they [fail] the test and they are out of a job."

When asked if his company is willing to hire employees who have not passed the apprenticeship test, the manager at a suburban manufacturer responded, "Oh yeah, we'll put them out here and give them the work experience. But of course, it would be a lesser-degree job." Both these employers place a value on basic skills, and both have found a means of ensuring that their better entry-level positions are filled by high school graduates with these skills: They use apprenticeship linkages. Three of the four managers at manufacturing plants that use apprenticeship referrals were among those who said that their firms need workers with academic skills. But while apprenticeships are a means of selecting applicants with academic skills, they are not common.

Close links with schools. Employers' contacts with schools are far more common. Indeed, 42 of the 51 employers (82.4 percent) in our study have minimal contacts with schools-through job fairs at schools, talks to classes, or notices of openings for part-time jobs. But employers view these activities as community services to help students learn about work; they rarely regard them as leading to full-time jobs.

Some employers, however, have longterm, close linkages with school staff. Those who do frequently devote a substantial amount of time to these relationships; they serve on school advisory boards and try to hire recommended graduates of certain vocational programs. Unlike school-business partnerships, which are often short-lived (Lynn and Wills 1994), these contacts usually carry long-term obligations and last many years. When these employers have job openings, they ask school staff to nominate students. Employers see these longterm linkages as a way to get the straight scoop on young graduates' skills.

Of the 51 employers in our study, 13 have long-term contacts with teachers and counselors.3 These linkages are much more common among employers who said they need academic skills; 12 of the 35 (34.3 percent) managers who said their firms need employees with academic skills used a close form of school linkage, compared to only 1 (6.3 percent) of the 16 who said his firm does not.

Despite the costs, several managers suggested that the investment of time involved in the closer form of contacts (trusted links with school staff) pays off in employees with better academic skills. One executive at a family-owned printing company in Chicago said that his company is still able to get workers with good academic skills from a local high school on the basis of the personal relationship his father established with a teacher from that school more than 10 years ago. Although this relationship is not without its costs (volunteered time on the school's advisory council), its benefits have been worth the sacrifices:

I think we've had success because we have been able to reach inside the schools and talk specifically to certain teachers. . . . And the positive [aspect] of those contacts for us is that we're getting the straight scoop. And there [are] many times [when] we've called, and the teachers, the instructors, have told us, "Look, I've got a classroom full of kids, but there's no way I would send any of them to you." So, I don't know if the schools are doing a great job, but our success has been good because we've been handed, I think, a few of the better ones.

Asked to speculate why these links with schools work, one supervisor at a suburban metal company observed: "If someone is willing to put their name behind somebody and say, 'I think that this guy would work out good,' well, most people nowadays will not personally endorse anybody unless they're pretty confident they won't wind up with egg on their face." Because he expects the teachers he knows to care about their reputations, this supervisor trusts them to tell him the truth about applicants' skills. Through this link, he can efficiently and effectively prescreen applicants and hire only those who come highly recommended.

But employers' actions to recruit employees through selected networks has adverse implications for alternative channels, even from the same school. The managers of several companies that maintain long-term contacts with trusted teachers said that they are not willing to make an open channel of recruitment from the entire school, such as through co-op placement programs or the counseling office. When asked whether he would consider using other school programs to recommend potential workers, the manager at a Chicago printing plant with strong links to teachers reported:

We bypass the typical channels.... I think the placement offices in . . . schools . . . they're out there, God bless them for it, but they're out there trying to get kids jobs, and I think they probably have less sensitivity to the workplace. I think the best of them probably have the idea that if they get the [kids] placed into [jobs], then they've done their job. And that's good [for their kids]. That probably is their role. But I don't know that it means that they fully understand what [my job requirements are] or whether they fully understand the students either. The teachers on the front line, I think they have their finger on the pulse a lot more.

Similarly, a manager at a suburban printing firm said that the principal of a local high school has done a laudable job in referring qualified applicants to him, but the counselors have disappointed him:

You get a very strong distaste for working with high school counselors. I have very little use for high school counselors altogether because I don't know if their background is sociology with a specialty in psychology, or whatever it is, but the only applicants I've ever gotten from high school counselors have been absolutely worthless. . . . [They] have all just been really a waste of time.

In addition to distrusting conventional school channels for recommendations, employers who seek teachers' assistance in hiring students are also wary of putting too much stock in high school grades. Of the 13 employers with strong links to teachers or counselors, 8 reject the idea of using grades to measure employability. When asked if he thinks that school grades predict anything about work performance, the manager of a Chicago firm with such a link responded:

Not in all cases. And that works both ways. There's a lot of people that come out of high school that have some great credentials and just never do anything, and then there are other young people who come out of high school with below-average grades that just have the will to succeed. For whatever reason, high school didn't grab them. So, actually, at this point in time, I wouldn't put too much weight on grades. The [teacher's] comments I would put a lot of weight on if I know the teacher, but not typically the objective measures.

These comments help to clarify a puzzle in past research. Many employers (both in our sample and in the society at large, cf. Crain 1984; Griffin et al. 1981; Rosenbaum and Kariya 1991) choose not to use grades as a hiring criterion, even when they complain about not getting skilled workers. These findings suggest that the skills students learn in school are valued by some employers, but the traditional ways of reporting those skills-grades-are perceived as untrustworthy. Given this situation, employers' investments in long-term linkages may be the one way of getting trusted information instead of using the grades they do not trust (cf. Miller and Rosenbaum in press).

These findings also indicate that although employers pursue the productivity goals that economists assume, some employers' means of obtaining trustworthy information about applicants' potential productivity actually restrict access to those jobs to only students with the right contacts. Even students who have good grades but who lack access to these contacts will have their qualifications mistrusted and may not be considered by some employers. These links provide a good channel of access for students who have classes with linked teachers, and in inner-city schools, these students' best chance of getting a job is through their teachers. But only a small proportion of teachers have such contacts (Rosenbaum and Jones 1995), and these links represent structural barriers that prevent the labor market from operating in the unfettered way envisioned by economic theory, even though the links are created in pursuit of that theory's goals.

CONCLUSION

This study examined whether employers really need workers with academic skills, as business panels contend. Despite our skepticism about the claims of top executives, we found that plant and office managers take actions that indicate their serious commitment to finding such workers. Our findings confirm economists' assertions, although we found that employers' actions are different from the ones economists suggest and often occur through social structures. For instance, employers rarely hire on the basis of school grades (Crain 1984), but some employers use similar information when it comes from trusted linkages.

We must note that our data come from employers' perceptions and may be distorted as a result. Yet employers' stories are so specific and detailed as to be highly credible. Moreover, we found that the employers who complain the most are not just whining. Their efforts to cope with deficiencies in skills often involve significant efforts or expense, which suggests a commitment to overcome these difficulties. These employers screen people in entry-level jobs and through linkages with schools, increase supervisors' responsibility to explain tasks when workers do not understand, adjust jobs to match workers' skills, and offer accommodations to retain workers with valued skills. These actions are often costly, and these costly actions are more common among employers who report that they are seeking workers with academic skills. In short, employers are generally putting their efforts where their complaints are. These actions have not been noted in previous research, and they warrant further study.

Although motivated by economic interests, employers' actions create social structures with important implications. They lead to changes in authority, preferential treatment, opportunities for hiring and advancement, and institutional networks. In particular, we found that some entry-level jobs, which largely have attributes of the secondary labor market (in that they offer minimal pay, benefits, and job security), are used as a form of screening to detect which workers are good bets for advancement.

Such on-the-job screening clarifies puzzles in theory. Whereas the theory of the segmented labor market is correct that youths are initially confined to jobs in the secondary labor market, it does not explain what keeps youths confined to this segment (Granovetter 1981; Hodson and Kaufman 1982) or how the selection process changes as youths become adults (Hogan and Astone 1986; Parcel 1987). Our findings indicate that employers use some entry-level jobs to screen workers for access to the primary labor market. On-the-job screening helps explain the jump that some workers are able to make from the secondary labor market (where no skills are required) to the primary labor market (where academic skills are necessary). However, this jump depends on young employees having academic skills.

Moreover, our findings indicate that Berg may have been partly correct about employers' overvaluation of education credentials, but for different reasons than he supposed. When Berg (1971) and Squires (1979) found that employers require academic credentials that are not needed for present jobs, it is possible that the employers were using these jobs as screening mechanisms for higherlevel jobs that did need academic skills. Our study found some cases that illustrate this point, and the issue deserves further study.

We also found that employers create institutional linkages to screen applicants. While sociologists have studied how workers use networks to learn about job openings (Granovetter 1995), research has not examined the other direction: how employers use networks to gain information about applicants (Granovetter 1985). Moreover, research has focused on informal networks among individuals, but it has rarely studied institutional networks. Many employers in our sample reported establishing contacts with high schools, community colleges, and apprenticeship programs. Although these linkages are less formal than those in other nations (such as Japan and Germany), some are long-term networks, transcend the original individuals who initiated the networks, and perform some of the same screening functions as do networks in other nations (Rosenbaum, Kariya, Settersten, and Maier 1990).

Like workers' search networks, employers' recruiting networks show the limits of neoclassical economic models, and both networks use nonmarket processes to improve information-an important prerequisite of market models. Indeed, although institutional networks seem inefficient to neoclassical economists, employers report that these networks improve their efficiency in selecting capable employees. Institutional networks are an important extension of sociological network theory, and they deserve further attention (Granovetter 1995, Kariya and Rosenbaum 1995).

After showing that weak ties to schools have an advantage for workers' job searches, Granovetter (1995:63) suggested that formal strong ties have an important role "where there is no easy way for job seekers and recruiters to assess one another's qualities, . . . for example, as students emerge from school, with no previous work experience and therefore no network of contacts from previous jobs." He developed this contention with illustrations of Japanese recruiting networks (Rosenbaum and Kariya 1989), German apprenticeships (Harhoff and Kane 1994), and some findings from the United States, including the present study (Granovetter 1995: 21, 162-65).

Moreover, although weak ties have value in providing job leads to workers, the findings of this study indicate that strong long-term ties give employers trustworthy information about applicants that reduces costly mistakes in hiring. Thus, although employers participate in linkages for productivityenhancing reasons identified by economic theory, the result of employers' actions is best described by sociological theory. Employers' linkages with schools create structures that put information in a trustworthy context, but they constrain access to jobs to individuals in schools that have these linkages.

Implications for Social Policy

These results have important implications for social policy. Training programs for disadvantaged groups have often been ineffective (Betsey, Hollister, and Papageorgiou 1985; Stern, Finkelstein, Stone, Latting, and Dornsife 1994). Therefore, social policies that encourage school-employer linkages could help disadvantaged students and create a more systematic pathway from school to work, similar to (though weaker than) the pathways in other nations that have formal linkages. Research indicates that high school graduates who find their jobs with help from schools get better jobs than those who rely on want ads or direct applications (Holzer 1995). Moreover, for students in vocational programs, help from schools raises their wages, both right after graduation and 10 years later, and it raises wages even more than do family contacts (Rosenbaum and Roy 1996).

Although teachers' biases are a potential concern, any procedure entails some bias, and the solution may be to choose the lesser of the evils. For example, Neckerman and Kirschenman (1991) found that employers who use employment tests are more likely to hire minorities than are employers who do not. Thus, testing may be biased, but the alternatives to testing may be worse. Similarly, the alternatives to teachers' evaluations may be worse. For instance, employers may create more bias when they rely on their own snap judgments in brief interviews (Miller and Rosenbaum in press) than if they rely on teachers' opinions, which are formed by viewing students' performances over an entire year. Indeed, according to reports from some urban teachers, employers who are reluctant to hire females or minorities for skilled crafts jobs have done so if these youths are highly recommended by trusted teachers (Rosenbaum 1995,1996). Moreover, analyses of the High School and Beyond survey indicate that although few young White men (4.6 percent) get their first jobs after graduation from high school through help from schools, young Black men (9.4 percent) are more likely to get jobs that way, and young Black and Hispanic women (17.1 percent and 12.4 percent, respectively) are even more likely to do so (Rosenbaum, Roy, and Kariya 1995). Although research on the effects of school-employer linkages is still meager, preliminary indications suggest that they have the potential to create effective channels of access for disadvantaged students.

Economic theory offers rational conjectures about how employers hire, but this study has described employers' actual behaviors. Any study has limitations, but empirical evidence is a useful check on theoretical conjectures. We found that economic theory describes employers' motivations pretty well, but it predicts behaviors that do not occur (such as hiring based on grades) and ignores some behaviors that have important sociological implications. That is, employers create social structures, selection mechanisms, and institutional linkages that affect hiring and promotion. These social mechanisms may improve information and efficiency, but they also affect whether youths gain access to better jobs and which youths do so. We hope that this study stimulates further research on these issues.