ability? What are the consequences for workers who cannot provide these signals?
To help provide answers to these questions, I recently administered a new survey to some 3,200 employers in four large metropolitan areas: Atlanta, Boston, Detroit, and Los Angeles. (This survey is part of a larger Multi-City Study of Urban Inequality [MCSUI], funded by the Ford and Russell Sage foundations in these four metropolitan areas.) The survey was administered over the phone between June 1992 and May 1994 to individuals in the establishments who were identified as being responsible for the hiring of noncollege workers in general or workers in specific occupations. The sampling process used here generated a stratified random sample of employers who are distributed across establishment size categories in roughly the same proportions as the overall labor force.16 Because of this sampling property, the characteristics of a sample of newly filled jobs at these firms can be considered, which should fairly well represent the jobs that are now being filled by employers and which new job seekers will currently face in the labor market.
Survey respondents were asked a variety of questions about all currently filled or vacant jobs at the firm as well as the last job that they filled and the last worker hired into this job, including the skills needed, recruitment and screening methods used, and demographic characteristics of hired workers. Data regarding the latter are analyzed below.17
Table 2-1 presents data on the cognitive and social tasks that need to be performed daily on these jobs and the personal skill credentials required of people who are hired. The tasks are measured as responses to questions of how frequently each task in the list had to be performed (i.e., daily, at least once a week, once a month, or never). The credentials are based on a question of whether or not each is "absolutely necessary," "strongly preferred," etc. (The credentials are counted as "required" if they are listed as "absolutely necessary" or "strongly preferred.'') These results are presented for all jobs together and separately for jobs that do and
TABLE 2-1 Skills and Credentials Required for New Jobs (percentage)
do not require college degrees. The latter category is also divided by occupational category: white-collar versus blue-collar and service jobs.18
The results show that credentials requirements and task performance on newly filled jobs are extensive. Almost three-fourths of these jobs involve daily customer contact (either in person or on the phone); over two-thirds require daily reading or writing of paragraph-length material and the use of arithmetic; and well over half involve the use of computers.19 Furthermore, almost two-thirds require not only general work experience but also specific experience related to the particular type of work, and over 40 percent require some type of certification of previous training.
These requirements vary with the educational requirements of the job and by occupation. Thus, jobs that require a college education generally require a higher degree of daily task performance and more experience or training than jobs that do not require college; among the latter, white-collar jobs require more than blue-collar and service jobs. But even for blue-collar and service jobs, requirements are quite extensive. Over half of these jobs require daily customer contact, reading and writing, arithmetic, general work experience, or specific work experience. Indeed, only about 5 percent of all jobs in central-city areas require none of these tasks, and the percentage requiring none of these credentials is the same (Holzer,
1996). Comparable statistics for suburban areas are just a few percentage points higher.
Of course, some questions remain about whether or not the requirements are hard and fast—in other words, whether employers really do as they say. But no such questions even seem relevant for the task performance data, which report current employee activities, and the responses to questions on hiring requirements are largely confirmed by information, where it is available, on the characteristics of the workers actually hired for these jobs.20
Thus, these data cast considerable doubt on whether there are enough jobs available with very low skill requirements to absorb all potential workers with very low skills, especially if welfare recipients are soon required to seek work in large numbers (Holzer, 1996).21
As argued above, the tasks represent some of the actual skills needed for job performance, while the credentials are signals of a candidate's potential ability to perform a job. But employers appear to use a wide range of other screens as well to gauge these signals, and they also use a variety of attitudes about various personal characteristics in making their hiring decisions.
Table 2-2 provides data on these other screens and attitudes. It shows that written applications, personal interviews, and requests for references are almost universally used, even though there are some major doubts about their predictive power or objectivity (Karren, 1980). Nonphysical tests are now used in 29 percent of all cases, and work samples are reviewed in 21 percent; over 40 percent use at least one or the other.22 On the other hand, checks on educational or criminal backgrounds are made less than one-third of the time.
The data presented here on employer attitudes toward various personal characteristics are striking. For noncollege positions (the only ones for which these questions were asked), employers would not hire people who had been unemployed for over a year in about 30 percent of the cases; they would not hire anyone with only short-term or part-time work experience in about 50 percent of the cases; and they would not hire anyone with a criminal record almost 70 percent of the time.
The aversion of employers to hiring those with spotty work histories has been noted by Ballen and Freeman (1986), who argue that previous employment
TABLE 2-2 Other Screens and Employer Attitudes for New Jobs (percentage)
|
|
|
No College Required |
|
|
All Jobs |
College Required |
White-Collar |
Blue-Collar/Service |
Other screens: |
|
|
|
|
Written application |
82% |
86% |
83% |
80% |
Interview |
88 |
91 |
89 |
85 |
References |
76 |
86 |
75 |
69 |
Test (not physical) |
29 |
25 |
34 |
25 |
Work sample |
21 |
29 |
16 |
21 |
Education check |
30 |
42 |
30 |
23 |
Criminal record check |
32 |
34 |
33 |
29 |
Would hire applicant with |
|
|
|
|
Unemployment for entire past year |
— |
— |
70 |
72 |
Short-term or part-time work experience only |
— |
— |
50 |
49 |
Criminal record |
— |
— |
31 |
34 |
instability sends a very negative signal to employers. Evidence on the possible "scarring" effects of early unemployment more generally is mixed and is consistent with the smaller negative effects this trait has on hiring. (See Ellwood, 1982, and Meyer and Wise, 1982, for earlier analyses of this issue and Rich, 1994, for a more recent analysis.)
The large fraction of employers who will not hire people with criminal records stands in sharp contrast to the much smaller fraction who actually check criminal records. As noted above, the costs of checking criminal records are quite high in many places, and therefore many employers forgo this information. Unfortunately, some probably infer such activity from the race, age, and gender of job applicants; and, given the high rates of crime among less-educated and unemployed black males, many employers will likely be especially suspicious of young black males who lack employment for significant periods of time (Bushway, 1995). Thus, developing more effective ways of signaling one's work-related experiences could be especially important for youth in these situations.
Table 2-3 provides information on how many of these firms have actually experienced rising skill needs in the past 5 to 10 years and which skill needs have risen. The data indicate that 42 percent of establishments report rising skill needs, with somewhat higher rates in the jobs requiring college and somewhat lower ones in blue-collar and service jobs. These data may well understate the extent that skill needs have increased in this period.23
TABLE 2-3 Increases in Skills Needed for New Jobs in Past 5 to 10 Years (percentage
|
|
|
No College Required |
|
|
All Jobs |
College Required |
White-Collar |
Blue-Collar/Service |
Skill needs higher |
|
|
|
|
Yes |
42% |
50% |
43% |
32% |
No |
58 |
50 |
57 |
68 |
Of those with increases, primarily which skills? |
|
|
|
|
Read/write/arithmetic |
33 |
26 |
33 |
42 |
Social/verbal |
36 |
44 |
34 |
30 |
Computer |
31 |
30 |
33 |
28 |
The data also indicate that the demands for basic numeracy/literacy, social/verbal, and computer skills have risen more or less equally, even across educational and occupational categories. Interestingly, the rising need for basic skills has been highest among noncollege blue-collar and service jobs, where they have always been in relatively less demand; this confirms the earlier finding that even within these categories fewer jobs are available that do not involve substantial need for basic cognitive functions. On the other hand, the need for social and verbal skills has risen the most for jobs requiring a college education.
Do any of these skill needs and employer attitudes affect who actually gets hired, especially once obvious characteristics such as educational attainment are controlled for? Table 2-4 presents data that speak to this issue. Percentages are given for newly hired workers in these jobs who are either black or Hispanic. This is done for all job categories as well as for subsets of jobs in which each cognitive or social task or each credential is either required or not required.
The results clearly indicate that blacks and Hispanics are hired less frequently for jobs for which any of these tasks or credentials is required. In percentage terms the differences are not small. For example, the gap of 12 percentage points between those hired into jobs that require computer use and those hired into jobs that do not constitutes a difference of 32 percent (with the second group used as the base).
In separating blacks and Hispanics, some interesting differences emerge. Basically, Hispanics are more disadvantaged by requirements for high school diplomas and direct customer contact, while blacks are more disadvantaged by previous experience and training requirements (Holzer, 1996).24
Separating by gender also indicates that black and Hispanic males are the
TABLE 2-4 Blacks and Hispanics Hired Into New Jobs by Skills and Credentials Required (percentage)
|
|
|
No College Required |
|
|
All Jobs |
College Required |
White-Collar |
Blue-Collar/Service |
All jobs |
31% |
20% |
28% |
44% |
Jobs requiring daily performance of |
|
|
|
|
Customer contact |
29 |
24 |
27 |
41 |
Reading or writing |
28 |
19 |
25 |
45 |
Arithmetic |
25 |
16 |
24 |
39 |
Computer |
26 |
22 |
26 |
42 |
Jobs requiring credentials |
|
|
|
|
High school diploma |
27 |
20 |
26 |
39 |
GED |
— |
— |
26 |
39 |
No GED |
— |
— |
24 |
38 |
General work experience |
29 |
21 |
24 |
44 |
Specific work experience |
28 |
18 |
25 |
44 |
Previous training |
29 |
23 |
24 |
42 |
Jobs not requiring daily performance of |
|
|
|
|
Customer contact |
36 |
— |
34 |
48 |
Reading/writing |
38 |
22 |
35 |
44 |
Arithmetic |
43 |
31 |
39 |
52 |
Computers |
38 |
09 |
35 |
45 |
Jobs not requiring credentials |
|
|
|
|
High school diploma |
47 |
— |
41 |
51 |
GED |
— |
— |
42 |
55 |
No GED |
— |
— |
35 |
38 |
General work experience |
38 |
14 |
39 |
45 |
Specific work experience |
37 |
24 |
35 |
46 |
Previous training |
33 |
17 |
31 |
46 |
groups most limited by these requirements. In fact, the rising skill needs of employers might well account for most or all of the relative deterioration in wages as well as employment that these groups have experienced in the past 10 to 20 years (Holzer, 1996).
Of course, these results do not prove that the real skills gap between whites and minorities drives the observed employer behavior; instead, discriminatory perceptions of ability could result in these outcomes as well. But given the gaps that have been documented between average black and white education and experience levels, test scores, and so forth, it seems likely that at least part of the observed results are based on real differences in average skills.25
Other Recent Survey Evidence
In addition to the employer survey in the MCSUI project reviewed above, a number of other surveys of employers have been administered in the past 5 to 10 years that focus on many of these issues. (Descriptive and qualitative data on employers' skill needs can be found in the report of the Secretary's Commission on Achieving Necessary Skills [SCANS], as described in Packer and Wirt, 1992, and also in Carnevale et al., 1990.) Among the best known of these are (1) a survey of over 3,000 employers in 1994 designed by the National Center on the Educational Quality of the Workforce (EQW), administered by the U.S. Bureau of the Census and funded by the U.S. Department of Education; (2) a mail survey of over 2,500 employers administered by the National Federation of Independent Businesses in 1987; and (3) local surveys administered in New York City in late 1993 by the city's Department of Employment and in Milwaukee in October 1993 and 1994 by the Employment and Training Institute of the University of Wisconsin-Milwaukee. Other surveys of employers include an update of the Employment Opportunity Pilot Project of firms by Barron et al. (1994) and a variety of surveys discussed by Kling (1995). The focus here is on those that deal primarily with hiring requirements and methods.
Also, some much more qualitative and in-depth surveys of smaller samples of employers have been administered in Chicago, Detroit, Atlanta, and Los Angeles (Kirschenman and Neckerman, 1991; Moss and Tilly, 1995; Kirschenman et al., 1995). These surveys generally provide little quantitative information but generate interesting descriptive material that would be hard to generate in the less personal formats (i.e., phone and mail surveys) used in the large-sample studies.
Among the many issues that have been addressed in these surveys (and in reports based on them) are (1) the frequency with which certain skills and personal characteristics are considered important by employers for job performance, (2) the weights placed by employers on a variety of screens and hiring requirements as well as their perceptions of applicants they hire, (3) more objective measures of how these skills and screens contribute to worker or firm performance, (4) the extent to which various skill needs are rising, and (5) the ability of employers to hire employees with the relevant skills. Of course, the responses to at least some of these questions will be subjective and hard to interpret clearly. But they still provide some interesting descriptive data on how employers view their skill needs and hiring.
Employers consistently report that basic literacy and numeracy skills are very important for virtually all jobs. In the New York City survey these skills are heavily weighted in every job category, though again somewhat less in the blue-collar and service jobs than in the white-collar ones.26 Listening and speaking
skills are also considered important in white-collar and service jobs but less so in blue-collar jobs.
A variety of personal qualities, such as responsibility, integrity, and self-management, are considered almost as important as the basic skills described above, and in blue-collar and service jobs they are even more important.27 These results are consistent with those of Moss and Tilly (1995), who stress the importance of what they refer to as "soft skills"—that is, communication skills and attitudes/motivation. (See also Cappelli, 1995a.) In addition, the New York City survey found that thinking skills, such as decision making, reasoning, problem solving, interpersonal skills, and working with technology are all relatively important in white-collar jobs and (in some cases) in service jobs as well.28
These results are consistent with the notion that growing numbers of companies are reorganizing jobs and work in ways that require more problem-solving ability on the part of employees, though this is not true universally. For instance, in the EQW survey 37 percent of establishments use some type of total quality management, which usually requires workers to have a higher degree of analytical ability; fewer establishments use a variety of other high-performance organizational strategies, such as teams (used by just 12 percent). (See National Center on the Educational Quality of the Workforce, 1995.)
As for screens and hiring requirements, both the New York City and the EQW surveys found that employers rank attitudes and communication skills as the most important characteristics they look for when screening candidates. Interviews are also given a lot of weight, despite the findings (noted above) from the human resources literature that question the objectivity and predictive power of information gained in interviews.29 Somewhat more objective measures of skill, such as previous experience and work history, are given almost as much weight in these studies as the attitude and communication measures. References and recommendations score highly as well, as do industry-based credentials in the EQW and previous training for some occupations in the New York City survey.30
The evidence on measures of academic background is somewhat more mixed. On the one hand, educational requirements do exist on most jobs. For instance,