Russell Sage Foundation, Chris Tilly, the Good, the Bad, and the Ugly: Good and Bad Jobs

<!DOCTYPE HTML PUBLIC "-//SQ//DTD HTML 2.0 HoTMetaL + extensions//EN">

THE GOOD, THE BAD, AND THE UGLY: GOOD AND BAD JOBS IN THE UNITED STATES AT THE MILLENNIUM

Chris Tilly
Visiting Scholar
Russell Sage Foundation
112 E. 64th Street
New York, NY 10021
USA

and

Department of Policy and Planning
University of Massachusetts
Lowell, MA 01854
USA

June 25, 1996

Introduction

Conceptual issues

The state of the art in measuring job equality

Seven dimensions of job quality

Wages

Fringe benefits

Due Process

Hours flexibility

Permanence

Mobility

Control over the work process

Summing up the seven components

Why, so what, and what to do about it

References

Paper prepared for presentation at conference on "Jobs and Justice: Towards Improving the Employment Relationship," Pennsylvania State University, June 26-29, 1996. I thank the Russell Sage Foundation for support during the writing of this paper.

Introduction

Twenty-five years ago, in 1971, Peter Doeringer and Michael Piore published their classic Internal Labor Markets and Manpower Analysis, which, among other things, explored the distinction between primary and secondary labor markets. Twenty-five years ago, we in the United States thought we knew the difference between good and bad jobs--a difference conceptualized in the primary/secondary distinction. Good jobs were well-paid, secure, and connected to paths of upward mobility. Bad jobs were low-paid, unstable, and dead-end. This distinction was hitched to the powerful post-World War II optimism about economic growth. In this optimistic framework, bad jobs were peripheral and to a large extent vestigial--in the process of being swept away by U.S. economic progress.

Today, this picture of good and bad jobs has been altered in three ways. First, people place more value on issues such as autonomy and fulfillment on the job, and on balancing work and family, reducing the dominance of bread and butter issues such as wage levels. Second, the business press, along with some scholars, have challenged the notion that a good job is a stable one, proposing instead that good jobs are stops--often quite short ones--in an upward career trajectory. Third, and most important, bad jobs can no longer be described as vestigial. They form a central, and by many measures growing portion of employment in the United States. This "ugly" fact motivates this paper's title.

The core of this paper, in fact, dwells on the last point, reviewing what is known about a variety of job characteristics in the United States to build a case that jobs have, on average, gotten worse. But before proceeding to that review, I start by discussing conceptual issues, and assessing the state of the art in measuring good and bad jobs. And after the review of job characteristics, I briefly discuss the causes of changes in job quality, and possible policy responses.

Conceptual issues

Jobs represent one particular way of organizing work. Work, in turn, includes any human effort adding use value to goods and services. However much their performers may enjoy or loathe the effort, conversation, song, decoration, pornography, table-setting, gardening, housecleaning, and repair of broken toys all involve work to the extent that they increase satisfactions their consumers gain from them. Prior to the 20th century, most of the world's workers performed the bulk of their work in other settings than salaried jobs as we know them today. Only a prejudice bred by western capitalism and its industrial labor markets fixes on strenuous effort expended for money payment outside the home as "real work", relegating other efforts to amusement, crime, and mere housekeeping.

Nonetheless, analysis of jobs directs our attention to this zone of "real work." Conceptually, we can build up from transactions, to contracts, via networks, to jobs (Tilly and Tilly 1994). Although solitary workers certainly exist, in general work depends on transactions among parties, notably between producers and immediate recipients of use value added by work. Transactions consist of interpersonal transfers of information and/or goods of which the parties are aware; they become work transactions when the effort of at least one party adds value to the element transferred. Organized, durable transactions cluster into implicit or explicit work contracts stipulating parties, rights, obligations, and sanctions to the transactions in question. Work contracts differ from mere accumulations of work transactions in featuring enforceable agreements that govern durations, limits, enforcement mechanisms, and relations among transactions.

Work contracts are typically embedded within a variety of social networks, including production networks. Social networks in general are sets of relations among persons, organizations, communities, or other social units. Most commonly we think of a single network as the set of relations among specified actors of a given type defined by a certain kind of tie: for example, the interlocking of corporate boards of directors or shared involvement among supporters of a social movement. A production network consists simply of a connected set of work contracts linking multiple producers and recipients. The social structure of work centers on concatenated contracts; jobs, occupations, careers, firms, unions, labor markets, discrimination, and inequality all appear as special cases or outcomes of that concatenation.

Finally, a job is a cumulation of work contracts assigned durably and formally to a single person; it therefore does not constitute a network as such, but a junction of two or more networks. A person may hold more than one job at a time, two persons may have the "same" job, and jobs always entail interaction with other persons, but the basic matching of roles with individuals marks off jobs from other ways of organizing work. A lathe operator in a shop has a job, but a freelance writer does not. An actor typically holds a sequence of jobs punctuated by unemployment, but a street mime has no job.

Having specified what a job is, we still face conceptual difficulties on the way to defining a good job. The term "good job" implies that certain jobs are good regardless of who holds them. In effect, we are assuming that the ranking of jobs is conceptually prior to the sorting of people among these jobs (Granovetter and Tilly 1988). But in fact, this is an oversimplification. A job's holder brings particular expectations to a job, and the employer's view of the job is colored by the worker holding it. Thus, job-holders shape jobs. In some cases they set out to mold their jobs in individualized and conscious ways (Miller 1988). In other cases, the talent, skills, or productivity of the incumbent are decisive: established stars get paid more for acting in a movie than unknowns. But the largest impact of job-holder characteristics on jobs occurs when there is large-scale occupational or industrial segregation, such as job segregation by gender. Paula England and colleagues (1994) have demonstrated that in the United States, the percent female in an occupation lowers the pay it offers to both women and men. Though gender segregation itself is a constant across societies, there is enormous variation in which jobs are defined as male and female (Reskin and Hartman 1986, p.7). Despite this variation, women's jobs are in general valued less--indicating that job-holder characteristics, rather than the intrinsic characteristics of the job itself, contribute powerfully to the valuation process. Having noted this difficulty, I will proceed to ignore it for much of the paper, treating jobs as independent of their holders to simplify the analysis.

Yet another conceptual issue is the extent to which various job characteristics cluster. According to a labor market segmentation perspective, particular sets of characteristics or governing rules tend to be found together: "It is not possible, as it were, to pick a rule from each category and establish a stable set of employment relationships," remarks Paul Osterman (1985: 58). "Rather, only certain configurations of rules fit together." This points to multiple, qualitative distinctions between good and bad jobs. Labor market segmentation, in turn, is typically held to link production networks to other social networks of recruitment and supply--once more eroding the distinction between job and job-holder.

A neoclassical perspective, in contrast, adopts a "hedonic" view of job quality: each job is a bundle of characteristics that may be varied at will (within certain limits) by employers; workers with varying valuations of these characteristics vote with their feet by choosing to apply for particular jobs, clearing the market. There is no reason to expect that a job that is bad in one way will also be bad in other ways; indeed, the theory of compensating differentials predicts that undesirable aspects of jobs will on average be offset by higher wages (Rosen 1986). This does not mean that all jobs are equally good, but it does imply that (1) the distinction between good and bad jobs is a finely graded continuum, rather than a qualitative break, and (2) there is no reason to expect that all good jobs, nor all bad jobs, will have particular features in common.

Empirical research has left this choice between segmented and hedonic versions of job quality at a standoff. Job characteristics do cluster--and, in particular, good jobs tend to be good along many dimensions--but the clustering is far from perfect. Compensating differentials are found in a few cases (particularly in compensating for the risk of death), but only a few. In this paper, I will draw most heavily on segmentation-based analysis, but will nonetheless consider major job characteristics separately, neoclassical-style.

Another issue is a familiar one: aggregation, and the use of measures of central tendency to summarize distributions, may obscure important variation. For example, based on mean characteristics, part-time jobs are bad jobs. They offer lower pay and fewer benefits, and have higher turnover than other jobs. But while most part-time jobs do conform to this profile, a small subset are good jobs--in some ways better than the full-time jobs that surround them. These are part-time jobs, often at a professional level, created to meet the needs of valued employees. Such part-time jobs typically offer equal hourly wages and the standard benefit package (amounting to more benefits on a per-hour basis), long-term attachment to the business, and added schedule flexibility (Tilly 1992, 1996). Similarly, tracking the mean or median wage over time as an indicator of job quality is inadequate if wage inequality is growing over time--as it has been.

A final conceptual/empirical issue deserving of particular attention is that the definition of a "good job" is itself a moving target. Specifically, there is evidence that succeeding generations of workers have placed decreasing emphasis on material rewards, and more on enjoyment of the job and a feeling of making a meaningful contribution--what Daniel Yankelovich (1993) calls "expressive" values. Yankelovich documents that there has indeed been a seachange in this direction in the United States. When a 1962 Gallup poll asked about the "formula for success in today's America," only six percent mentioned having a job that one enjoys doing; in a 1983 poll with similar wording the percentage rose to 49 percent. Among 30- to 40-year-old men and women surveyed in 1986, 60 percent stated that they placed much more emphasis on "pursuing satisfaction in a career" than their parents did; only 13 percent felt they placed less emphasis on such satisfaction.

Cross-national surveys show similar trends at work in most industrialized countries (Yankelovich et al 1985). Particularly intriguing is the case of Sweden, whose economy's dominant sector has rapidly shifted from primary (agriculture, forestry, fisheries) to secondary (manufacturing) to tertiary (services). This shift is reflected in the attitudes toward work of successive age cohorts: the oldest cite survival and sustenance as the main reasons for work, Swedes in their middle years cite material success, and young Swedes are more likely than their elders to cite instead the realization of expressive values.

The move toward expressive values raises two important measurement issues. First, measurements of job characteristics along these dimensions are inevitably far more subjective than measures of--say--the purchasing power of a dollar of wages. Whether one can even imagine a job serving as a source of personal fulfillment is largely determined by social context--the shared values embedded in production networks and other social networks--as well as economic well-being. And second, a shift in values implies that the "weights" of different characteristics contributing to job quality are not themselves fixed. Both of these render assessment of changes in job quality over time more difficult, and indeed, no consistent time series of fulfillment on the job exist.

However, from the viewpoint of those of us struggling to measure job quality, it is comforting to note that the shift in values has not yet become a complete reversal. In no country has the emphasis on expressive values eclipsed material measures of success. In 1991, 40 percent of Americans still rated material success as most important, compared to only 22 percent putting intangibles at the top (Yankelovich 1993). Even among the Swedes, the nationality most likely to rate expressive values the most important aspect of work, fewer than one in four embraced this prioritization (Yankelovich et al 1985). So while flagging this as a central problem for future research, I will pay little attention to jobs as a source of fulfillment in the empirical section of the paper.

The state of the art in measuring job quality

The last significant advance in measuring job quality in the United States was accomplished by Christopher Jencks, Lauri Perman, and Lee Rainwater (1988). They conducted a Survey of Job Characteristics in 1980, asking each worker to rate his or her job on a ratio scale in which an average job scores 100 (so a job twice as good as the average job would be rated 200). Jencks and colleagues regressed these scores on 48 job characteristics, including earnings, whether one gets dirty on the job, the degree of repetitiveness, and so on. They selected the 14 variables whose standardized coefficients fell above a certain cutoff, and proposed the predicted job score based on this set of 14 variables and coefficients as an Index of Job Desirability. Essentially, this empirically implements a hedonic model of job quality: each characteristic is assumed to independently contribute to the overall quality of the job.

What makes this Index of Job Desirability (IJD) an advance? As Jencks and co-authors point out, in most work on job quality economists use a single measure--earnings; sociologists use another single measure--occupational status. Such single-measure choices are understandable, since data are readily available to construct these measures. However, each of these measurement strategies has flaws. Earnings omits the impact of other job characteristics; Jencks et al. demonstrate that variation in non-monetary elements of the IJD is more than twice as large as variation in earnings (see also Rosenthal 1989). Occupational categories combine vastly disparate jobs (consider the wide range of jobs falling within "salaried manager: business services"). Most variation in the 14 components of the IJD takes place within three-digit occupational categories. The IJD does a better job of accounting for the impact of race, sex, education, and labor market experience on labor-market success than do earnings or standard occupational status measures.

Examining the IJD teaches us two main lessons. First, despite their shortcomings, earnings and occupational status are not bad as quick-and-dirty measures of job quality. Both are strongly correlated with the IJD index itself. And both have the relationship with most of the components of the index that would be predicted by a segmentation perspective: a negative correlation with getting dirty on the job, a positive correlation with the number of vacation weeks, and so on. Second, both bread-and-butter job traits and issues of fulfillment at work contribute strongly to job satisfaction. The index includes earnings (+), weeks of vacation (+), and the risk of job loss (-), but also includes task repetitiveness (-), frequency of supervision (-), the average education of people in this job (+), and the ability to decide on one's own hours (+). Third, however,

The problem with the IJD, of course, is that it was based on a one-time survey, rendering broader applications or tracking of change over time problematic. Furthermore, it finesses the issue of defining what a "job" is, taking workers' definitions of jobs for granted rather than distinguishing between jobs and less formalized and durable work contracts and less formalized and durable work contracts. For both of these reasons, I retreat from this approach in the rest of the paper, instead relying on a series of individual indicators of job quality. However, workers' valuations of various job characteristics within the IJD inform the remainder of the discussion.