Printer Friendly

Careers and Contingency.

Disagreement among legal scholars over the phenomenon of "contingent employment"--work having limited hours, duration, or security--has led to disparate prescriptions for legal reform. For some, the best solution would be to either leave the market alone, or eliminate existing regulations that drive employers to create contingent jobs. Others believe current regulations do not go far enough and advocate reforms ranging from expanding mandatory benefits and protections to facilitating collective bargaining among contingent workers in order to restore such benefits as long-term security, training, and career advancement. The debate about law reform has centered partly on disputes over the size, growth, and characteristics of the contingency phenomenon. Workers variously described as "contingent" are so eclectic as to render broad-brush claims on both sides of the debate misleading. As an alternative, Professor Gillian Lester offers the concept of underemployment--a failure of the market to match workers with jobs that fully exploit their human capital and preferences--as a superior explanation of the labor market problems that ought to concern policy makers. More fundamentally, however, opponents in the debate disagree over how labor markets work. Orthodox neoclassical economists, who generally oppose regulation, believe workers are matched with jobs in accordance with their human capital, preferences, and employers' needs. Reform-minded "strong segmentationists," by contrast, argue that contingent jobs tend to be dead-end "secondary" jobs, often involuntary and alienating. Finding both orthodox and strong segmentationist accounts incomplete, Professor Lester turns to "New Keynesian" explanations of labor markets, originally developed to explain equilibrium unemployment. She argues that this third approach provides important insights into contingent employment that have largely eluded contemporary debates on contingency. Finally, she discusses the policy implications of New Keynesian accounts, identifying directions for further research.

INTRODUCTION

Perhaps the most celebrated "crisis of work"(1) of the past decade is the perceived replacement of career employment with "contingent" jobs of limited duration, hours, or security. Journalists, academics, and legislators have alternately expressed alarm and enthusiasm about what is widely seen to be an important labor market trend.(2) Enthusiasts maintain that contingent work is a symptom of a well-functioning labor market that matches individuals' skills, preferences, and aptitudes with the needs of employers. Skeptics argue, in contrast, that contingent work epitomizes a dangerous phenomenon of increasing job insecurity and inequality of labor market opportunity in the United States.(3) They advocate labor market reforms to assist a rising class of workers who, despite a desire for permanent employment, oscillate between unemployment lines and short-term jobs that under-utilize their education, skills, and availability. Permanent workers, skeptics argue, may develop firm-specific skills over time, increasing their productivity and thereby opening the door to promotions and higher compensation. Contingent workers, by contrast, never develop the long-term attachments necessary to command such advantages. Moreover, job changes experienced by contingent workers are stigmatizing, making it more difficult to find a new job, especially for older workers, whose opportunities may be particularly limited.

Compounding these harms, reformists further argue, is the fact that many contingent workers fall outside the scope of mandated employment benefits and protections, which are geared toward the traditional archetype of the full-time, long-term employee. Indeed, contingent workers may fail to meet the legal definition of a covered "employee" under both statutory and private rules.(4) Moreover, even those contingent workers who qualify as employees may be denied benefits or protections under any number of bright-line "gatekeeper" tests. Unemployment insurance (UI), for example, is available only to workers who have worked some minimum number of hours, earned some minimum level of income from wages, or both, over a particular period; these factors may exclude contingent workers at a disproportionate rate.(5) This lack of coverage by benefits and protections reduces the quality of contingent jobs, assuming workers who lack but desire coverage will find it difficult or impossible to bargain for these terms.(6) At the same time, these attenuated de jure or de facto responsibilities towards workers with less continuity, permanence, or hours of work are undoubtedly an important reason why employers cultivate contingent relationships.

In this article, I explore the debate on contingent employment and argue that neither opponents nor advocates of reform offer a fully satisfactory account of the labor market problems at stake. First, the descriptive category of "contingent employment," as currently cast, is either under- or overinclusive of the class of workers who require market reforms. Second, even if we agree on a functional definition of the "problem class," assessing policy solutions requires grappling with radically divided views of the causes and implications of contingency. The result is that the opponents in the debate often talk past each other as well as past the underlying problem.

The group of workers variously described as "contingent" is eclectic: it includes teenagers holding part-time jobs after school, retirees earning extra cash on the side, single mothers holding down multiple temporary jobs in a struggle to accommodate family and market demands, "second earners" from affluent families, "downsized" executives working as independent contractors, and elite professionals whose fast-paced careers move them from one project to the next. Different definitions give conflicting estimates of the number of contingent workers, ranging from 2 to 30% of the workforce, and studies attempting to measure changes in job stability have produced equivocal results. These wide variations in interpretations of the scope and magnitude of the problem make it difficult to assess competing law reform proposals. A central task of this article is to illustrate how greater precision in specifying both harms and remedies is essential to any coherent program of reform. I argue that the concept of underemployment--a failure of the market to match workers with jobs that fully exploit their human capital and preferences--may better capture the labor market problems at issue than many existing definitions.

A prior question in thinking about any kind of labor market reform is whether the underlying labor market theory informing a proposal itself offers a satisfactory account of contingency. A second task of this article is thus to investigate two dominant theoretical accounts of contingent employment, which I label as "orthodox neoclassical" and "strong segmentationist." Orthodox neoclassicists tend to deny the existence of "involuntary" contingent employment (and thus deny the need for government intervention unless it is designed to eliminate regulations). However, they offer only weak explanations for the wage gap between contingent and noncontingent workers which persists even after controlling for conventional human capital measures. In the absence of a stronger explanation of these persistent gaps, we are left to speculate that there is a nontrivial class of workers who, despite genuine motivation and capabilities, are unable to secure stable employment, and have lower pay, benefits, and opportunities for advancement than other workers with the same preferences, human capital, and endowments.

Strong segmentationists, by contrast, argue that orthodox neoclassical observers misunderstand and understate the degree and extent of coerciveness of contingent employment. They use this claim to defend proactive government interventions to mandate wider availability of benefits and opportunities for collective bargaining. The challenge of defining and measuring labor market coercion, however, is daunting, and I argue that strong segmentationists have fallen short. While involuntary employment in bad jobs with poor benefits, security, and promotion opportunities is a real and troubling problem, "coerced" contingency may be less pervasive than segmentationists imply. At the same time, assuming quite plausibly that workers' choices are more constrained than orthodox observers acknowledge, it is not obvious that elimination of such harms is a task for labor market reform per se. Parallel difficulties plague strong segmentationists' argument that contingent workers experience "alienation."

Finding neither of these approaches entirely satisfactory, I turn to a third theory--which I label as "New Keynesian"--arguing that it offers a more tailored explanation of the contingency phenomenon at issue. New Keynesian "efficiency wage" and "insider-outsider" models, originally developed to explain persistent unemployment, may be adapted to the problem of contingent employment. These accounts explain how perfectly rational labor market behavior may lead to persistent underemployment. Remarkably few commentators have mentioned New Keynesian labor market models in contemporary debates about the rise of contingent employment, and to my knowledge, no legal scholars have marshaled these accounts in evaluating different law reform proposals. The principal contribution of this article is to introduce these models into the legal debate about contingency.

In summary, I argue that workers variously described as "contingent" are so heterogeneous that the more discrete concept of underemployment may better capture the labor market problem that concerns, or ought to concern, reformists. Moreover, although the dominant paradigms offer concrete reform proposals, their explanations of the causes and consequences of contingency leave troubling gaps that may render their respective proposals a poor fit. New Keynesians synthesize elements of both dominant paradigms, offering a more nuanced account of contingency. However, New Keynesian accounts are intrinsically less compatible with across-the-board legal solutions. This article recasts the entire debate about contingent employment as sitting at a difficult crossroads. The challenge ahead is to reassess the possibilities for, and limitations of, legal intervention.

The remainder of the article is organized as follows. Part I reviews and critiques competing attempts to define contingent employment. I discuss why I think existing definitions have led to confusion, and propose a more helpful way of identifying the problem. Part II assesses how effectively each faction in the current debate explains contingent employment. The soundness of their explanations, in turn, informs my assessment of their correlative policy proposals. I close with speculations on the future directions the discussion might take.

I. THE ELUSIVE CONCEPT OF "CONTINGENT" EMPLOYMENT

One need not venture far into the debate over contingent work to find an empirical landscape riddled with disagreements. In this Part, I explore these disputes at greater length, concluding that there is credible evidence of an emerging labor market phenomenon worthy of serious concern and policy evaluation. At the same time, I will argue, participants in the debate have either misidentified the locus of the problem, or chosen proxies that are a poor fit for the class of workers whose labor market problems seem most pressing. I call this class of workers the "underemployed," and offer a tentative definition.

A. Weak Affiliation

The most widely cited definition of contingent employment emerges from Richard Belous's important 1989 book, The Contingent Economy.(7) Belous defined contingency as "weak affiliation" with a firm, which in turn was determined by a range of factors, including job security, level of employee commitment to and identification with the firm, and the strength of the employer's incentive to invest in the human capital of a particular worker.(8) Although a few scattered occupations (e.g., seasonal farm labor, acting, construction work, and substitute teaching) have always been associated with weak affiliation, Belous estimated that somewhere between 25 and 30% of American workers in 1988 held contingent jobs, and that contingent employment was growing at a faster rate than employment generally.(9) Belous argued that the costs of this "dramatic growth" in contingency included greater insecurity and risk of job loss for millions of workers, a potential for underinvestment in human capital, higher unemployment rates in recessionary periods, a potential decline in equal employment opportunity, and a more difficult environment for unions. Not surprisingly, his striking estimates of the size and growth of the contingent workforce catalyzed a surge of interest in the topic.

Despite the impressive figures, there are problems with Belous's definition and the empirical estimates it drives. Belous arrived at his numerical estimates by combining the number of part-time, temporary, "business services" (various kinds of subcontracted workers, such as consultants and leased workers), and self-employed workers. Yet each of these groups is internally heterogeneous and may not fully embody Belous's criteria of contingency. Take part-time workers, who comprise more than half of Belous's contingent workforce.(10) Many part-timers possess more "core" than "contingent" characteristics. For example, they may have a significant stake in the company, work there for many years, and have positions as secure as traditional, permanent workers.(11) Moreover, about 80% have chosen part-time work, a factor Belous does not consider to be disqualifying.(12) For reasons I will discuss momentarily, voluntary labor market choices of workers may be an important (though not always decisive) factor in determining whether there is a labor market problem in need of solution. Similar arguments--regarding both heterogeneity and voluntariness of choice--can be made for the workers who fall within Belous's other, smaller, categories. Consequently, Belous's own estimate exaggerates the number of workers whose labor market circumstances fit within the spirit of the problems he seeks to address.

Nevertheless, subsequent policy observers have made generous use of Belous's estimates to anchor their own policy recommendations, at best paying little heed to the aforementioned critique, and at worst using the estimates inaptly. Jeremy Rifkin, who laments the simultaneous rise of technology and demise of decent jobs in America in his well-known book, The End of Work, is not unusual in these respects.(13) He cites data demonstrating rapid growth in temporary help employment, which is the smallest subgroup (a liberal estimate of 3%) in Belous's contingent class. He then quotes Belous's estimate that more than 25% of workers are "temporary, contract, and part-time workers" and that "upwards of 35% of the U.S. workforce will be contingent workers by the year 2000.(14) Rifkin later misleadingly asserts that "temporary workers and outsourcing make up the bulk of today's contingent workforce--millions of Americans whose labor can be used and discarded at a moment's notice and at a fraction of the cost of maintaining a permanent workforce."(15) By conflating the high growth rate of a tiny subgroup with the absolute size of Belous's larger group, Rifkin projects a deceptive message.

B. Absence of Long-Term Employment Security

An alternative and substantially narrower definition of contingent employment was recently devised by the federal Department of Labor's Bureau of Labor Statistics (BLS). The BLS identified job insecurity as the essence of contingency. In an effort to capture the phenomenon as cleanly as possible, they attempted to count only those workers lacking an explicit or implicit promise of long-term employment in their current jobs. To determine whether a worker met these criteria, survey questions were designed to expose involuntary transitoriness.(16) A special supplement to the February 1997 Current Population Survey (CPS) estimated that somewhere between 1.9 and 4.4% of the labor force worked in contingent arrangements so defined.(17)

The BLS definition includes only a subset of the workers within each category identified by Belous. For example, the BLS definition excludes most part-time workers.(18) Still, part-time workers, along with employees of temporary help agencies, on-call or day laborers, and people who work in "contract firms" (similar to temp work) are more likely than workers as a whole to be contingent under the BLS definition.(19) Many of these workers also experience Belous's "weak affiliation" with a firm, but this is not decisive under the BLS definition. For example, independent contractors are not especially likely to be contingent under the BLS definition because, contrary to popular belief, they tend to have very stable long-term relationships with the firms for whom they perform work, even if they maintain "independent" status.(20)

Do contingent workers thus defined exhibit the explosive growth observers claim exists using Belous's definition? The BLS has only begun to track long-term trends in the prevalence of contingent employment as measured by the CPS supplement.(21) The brief review to follow of the empirical literature reveals that there is considerable controversy over whether we are witnessing an era of declining job security.(22)

One popular approach to measuring changes in job security has been to concentrate on an easily identified category of workers who are widely thought to have low job security. For example, the practice of "outsourcing," in which firms rely on individual contractors and contract firms that tend to provide less secure jobs, appears to have mushroomed in recent years.(23) Furthermore, as mentioned earlier, there has been a rapid increase in employment through temporary agencies and the like.(24) Yet these methods fall prey to the same criticism I leveled against Belous's broad estimations. Growth in visible forms of employment that are likely to be contingent, such as temp jobs, may create a misleading impression about overall changes in job security. First, as mentioned earlier, not all subcontracted or temporary help services workers are "insecure": Some of these workers (e.g., many professionals) may in fact be better diversified and face less insecurity than similar workers bound to a particular firm.(25) Second, in the case of temporary workers, their small absolute numbers can make modest figures appear dramatic in terms of "percentage growth."(26) Finally, firms that previously hired and trained "in-house" temporary workers may have externalized these administrative burdens to temporary agencies. It may well be that workers previously hidden from public scrutiny are now part of the celebrated growth phenomenon in temporary help services.(27)

Others have attempted to get at the phenomenon of declining job security by measuring changes in overall job stability. In order to assess trends one might consider any number of proxy measures, including changes in median job tenure, retention rates of workers, and job turnover. These lines of research, however, have not been much more helpful. Analysts disagree both on which proxy best measures changes in job stability, and on how to characterize recent trends.(28) At best we can say that there is a credible empirical controversy about whether overall job stability has declined in recent years, with some agreement among studies (using a variety of proxies for stability) that there have been particular declines in stability among older, less educated and minority men.(29)

Even assuming that we could resolve some of these uncertainties by monitoring changes in the size of the contingent workforce as defined by BLS, the BLS definition has its own limitations. For example, it arbitrarily treats an expectation of job retention for one year or less (with variations on the theme) as the mark of "contingency." In the words of a technical note to the survey, "being able to hold a job for a year or more could be taken as evidence of at least an implicit contract for ongoing employment."(30) It is far from clear, though, that a worker who expects to retain her job for, say, 18 months is a "career" employee. This is a minor criticism, however, as the same attack could be leveled at any expectation an econometrician might select to distinguish "secure" from "insecure" workers. A second concern is that this may not be an adequate proxy for an implicit contract.(31) Although self-reported lack of expectation of extended tenure may be consistent with the absence of an implicit contract, it may not be sufficient evidence. Third, the methodology used to measure long-term employment prospects relies on workers' perceptions of job security, even though these perceptions may not correspond with actual rates of job loss. More systematic evidence of a positive relationship between worker perceptions of employment duration and actual job turnover would increase confidence in this proxy.(32)

Most troubling, however, is that in all likelihood the BLS definition may simply fail adequately to capture the class of workers who suffer the type of labor market problems that reform advocates in the contingent employment debate seek to ameliorate. For example, some part-time workers whose jobs are very secure--and thus who would fall outside the BLS definition of contingent--wish they could work full-time, and are capable of working full-time, but cannot obtain full-time work. They are therefore underutilizing valuable human resources. This, I would argue, surely ought to be seen as a labor market "problem," and indeed would seem to fall within the constellation of problems that Belous and other skeptics have sought to expose in the debates on labor market reform. The BLS definition may also be overinclusive: Even assuming (perhaps heroically) that the prevalence of insecure or short-term jobs is increasing, the effect of such change is ambiguous. It may, for example, mark a shift to a rising class of workers who cycle through "dead-end" jobs offering little in the way of opportunities for long-term career advancement. Or it may mark a shift from a rigid, bureaucratic model of single-firm career employment to a more flexible and cosmopolitan career model driven by inter-firm or even inter-industry job mobility. This brings me to a third, more plausible, way to capture the phenomenon of interest.

C. Underemployment

At this point, it seems helpful to revisit the labor market problems that policy advocates seek to address. At core is the fear that a class of workers is forced outside of the system of traditional jobs, in which workers have job security, benefits, and access to promotion ladders by which they may advance along a career path. The claim also seems to be that these "externalized" workers have worse jobs than they deserve and desire--otherwise, law reformers would not seek improvements through regulation. Low affiliation, transitoriness, and part-time hours of work may all be contributing causes of the underutilization and undervaluation of some workers' capabilities. Herein lies the flaw in using existing categories of contingency as categories for law reform. Various proxies for the problem (BLS's lack of implicit or explicit contract of long-term employment, or Belous's aggregated proxy of independent contracting, part-time, and temporary employment) are each imperfect in their own ways. Yet they have taken on lives of their own as "the problem." The reification of these categories of contingency as the nub of the social problem--regardless of whose definition is used--is a mistake.

One way to proceed, then, would be to disaggregate existing categories into their various subtypes of workers and analyze them as numerous distinct policy questions. I take a different route, making both a positive and a normative claim. I hypothesize that there is in fact a root harm embraced by different analyses of contingent employment, even though various attempts to articulate and find adequate proxies for the problem have led to different results. I will also argue that whether or not I am correct in hypothesizing that analysts latently agree on this root harm, it has strong intuitive appeal and ought to be addressed more explicitly. The problem, as I see it, is the persistence or growth of a class of workers for whom there is a "mismatch" between the jobs they hold, and their human capital, abilities, and desires.

As a threshold matter, let me note that problem of "mismatch" between workers' employment status and their actual abilities and desires is a version of the problem long identified in Keynesian economic theory as the defining characteristic of involuntary unemployment, involuntary unemployment, in this view, is the existence of workers who are willing and able to work but cannot find employment. A small group of observers have extended this conception to include "underemployment," i.e., employment in a job that undervalues one's labor relative to that of other workers with the same abilities, availability, and desires.(33)

While this concept of underemployment may better capture the labor market problem presented by various definitions of contingency, I do not mean to suggest that it is simple to apply. Indeed, disputes over the definition of contingent employment may be testament to this difficulty. At the same time I think it is helpful to search for some unifying problem, even at an abstract level, to start thinking about social policy.

I will now set out some tentative criteria of underemployment:(34) The worker must have (1) inferior quality of employment--in terms of the wage or other characteristics--relative to other workers with similar endowments who have made the same level of investment in education, skills, or experience (human capital);(35) and (2) a desire for employment on equal terms as comparably skilled and able workers who hold superior jobs. Thus the concept of underemployment takes both ability and preferences into account. It seems helpful to map the range of relations between human capital, preferences, and job match, in order to analyze the range of circumstances that might qualify as "problems" and weigh the intuitive appeal of characterizing them as such. This appears in Table I.
TABLE I DEFINING UNDEREMPLOYMENT.

                                             Prefer different
                            Prefer current   employment but
                            employment       unable to obtain

Quality of employment
on par with workers of      (1)              (2)
comparable ability,         Not              Not
training, and experience    Underemployed    Underemployed

Employment inferior to
that of workers of          (4)              (3)
comparable ability,         Possibly         Clearly
training, and experience    Underemployed    Underemployed


Workers falling within the first cell are outside the problem. Here a worker's current job is fully matched to her education, skills, experience, and preferences. One could imagine many independent contractors, part-time workers, and temporary workers who fit into this category: the self-employed computer analyst who moves from firm to firm without desiring long term commitments in any one place, for instance, or the high school graduate who takes a temporary job with a personnel firm in the hopes of acquiring some word processing skills and getting her foot in the door of an established firm.

Cell 2 contemplates the worker who is frustrated with his current job because he does not get along with other workers, or perhaps because he finds the work dull, or uninspiring, or difficult. Yet he is matched to a job that allows him to utilize his skills and training on par with other workers. Holding an undesirable job may pose a problem for this worker, but he is not underemployed. A tax analyst whose skills, training, and experience make him well-matched to the demands of his job, but who always wished he could be a professional golfer, would fall within this category.

Workers in Cell 3 are clearly underemployed. Here, there is a discrepancy between the job generally held by workers of given human capital, motivation, and preferences, and the actual (inferior) job the worker holds. The worker who seeks full-time work but can only find a part-time job, the "downsized" worker who has been forced to take a job outside of her area of training or expertise in order to avoid unemployment, the full-time temporary worker who has skills, experience, competence, and preferences identical to those of her permanent counterparts but who is classified as temporary and denied benefits--all of these examples might fit into this category. I do not place in this category the person who, for example, graduates at the bottom of his law school class because he is chronically lazy, but cannot obtain an elite law firm job even though he has a law degree.(36)

Cell 4 is more difficult to interpret. Here, a worker chooses a job which underexploits her talents, training, and experience relative to other workers of similar ability. From the perspective of maximizing labor market productivity, this poses a problem. Yet so long as we are satisfied that the worker's decision is freely made, the decision not to work in a more demanding job should be the prerogative of the individual, even though she is, in a sense, underutilizing her talents. Thus the definition of underemployment, as I have cast it, requires that the worker herself desire an alternative better matched to her abilities and training.

For reasons I will discuss more fully in Part II, I realize that this definition of underemployment may be controversial. First, to the extent that a worker adjusts his preferences according to local options, preferences may be "adaptive."(37) A worker unable to secure jobs on equal terms as others with comparable talents and desires may, in path-dependent fashion, develop an impoverished sense of his own worth and abilities and stop searching for the best jobs. Additionally, he may stop honing his skills if they are being undervalued. As the worker's motivation and skills erode, what once was a discrete inefficiency--the underutilization of existing manpower--becomes something less quantifiable: an underrealization of human potential.(38) A key point here is that underemployment, in the narrow technical sense of underutilization of a worker's current skills and preferences, may lead in the longer nm to the deeper labor market pathology of underrealization of potential. Although I target only the narrower problem in my definition of underemployment, I am interested in both short-run inefficiencies and longer-run underrealization and related distributional problems. Eliminating the former may help reduce the latter.

To summarize, despite widespread public perception of the "end of career employment," attempts to define the precise group of workers affected have produced sharp disagreements. Studies like Belous's, which aggregate workers within certain labor market categories (e.g., part-time) in which there is a high prevalence of insecure or dead-end jobs, may understate the heterogeneity of workers in each class and thus overstate the problem. The BLS definition of contingency, which focuses on job security, identifies a narrower group, but may be both too narrow and too broad. Many workers who expect their jobs to last for more than one year may be trapped in a bad labor market path that will lead to labor market harms. Moreover, some who work on temporary and other short-term assignments experience no real economic insecurity, expecting to be at least as "employable" as traditional workers. A third approach focuses on the root harms sought to be avoided, rather than the facial characteristics of a particular worker's employment contract. I suggest that this root harm is "underemployment," i.e., when a worker is not matched to a job that exploits her human capital and motivations as fully as for comparably skilled and motivated workers. This creates a labor market problem that policy-makers should try to solve if feasible.

II. COMPETING THEORETICAL ACCOUNTS OF CONTINGENT EMPLOYMENT

Disputes about the definition of contingent employment (and other empirical questions), fractious as they may be, are ultimately ancillary to more foundational disputes over how to explain the existence of jobs characterized by relatively low wages, benefits, and job security. Some--whom I identify as "orthodox neoclassicists"--believe that labor markets generally function well, and that contingency is the result of an efficient process that matches production demands with human capital and individual preferences. They believe labor market interventions are misguided, and, if anything, have already gone too far. Others--whom I call the "strong segmentationist" school--argue these characteristics are due to the systematic coercion of workers and result not only in material deprivations (low wages, etc.), but also in the alienation of workers. They have little faith in the self-corrective capability of labor markets, and therefore advocate aggressive regulatory reforms. In this part of the article, I assess the persuasiveness of these competing accounts. I then turn to a third, "New Keynesian," account. Theorists in the third vein argue that, even when market actors behave in a purely rational fashion, workers may fail to be matched with jobs that fully exploit their talents and motivation. This failure of the market to "clear" thus leads to underemployment which persists in equilibrium and for which it is surprisingly challenging to devise across-the-board legal solutions.

A. Orthodox Neoclassical Accounts

Orthodox neoclassical commentators portray contingent work as a relatively insignificant byproduct of efficient market interaction between individuals and firms. Accordingly, they assert that any systematic pattern in wages and benefits among contingent employees (if one exists) reflects underlying differences in human capital, endowments, preferences, and commercial exigencies. Moreover, neoclassicists argue, even if conventional empirical methods cannot explain observed wage and benefit differentials completely, the reason has more to do with statistical imprecision than with segmented labor markets. This Part explains and evaluates the neoclassicists' argument in greater detail. I find their account is intuitively crisp, and almost certainly correct in explaining some portion of the contingency phenomenon. However, for other, troubling, aspects of contingency, the orthodox account fails to rule out the viability of competing hypotheses with competing corollary ideas about law reform.

1. Overview of the model.

Neoclassical labor market models generally assume that labor markets "clear" in the sense that everyone is matched to a job commensurate with his or her human capital and preferences. Most commonly, they reach this market-clearing result on the basis of a familiar set of assumptions. Though many contemporary economists who consider themselves within the neoclassical orthodoxy tend to relax one or more of these assumptions, the commentary I discuss below leans closer to this orthodoxy than to other theories I discuss in later Parts. Specifically, neoclassical orthodoxy generally assumes that: (i) firms attempt to maximize profits, paying wages equal to the marginal productivity of each worker; (ii) workers attempt to maximize utility, and continue to supply labor up to the point where their wages equal their respective marginal rates of substitution between leisure and consumption; (iii) through competition among both firms and workers, unregulated labor markets will "clear" in the sense that at the equilibrium market wage, there is neither an excess supply nor an excess demand for labor;(39) and (iv) the parties' expectations about the future are rational, in that they incorporate all existing information available, accurately weighing the probabilities of various potential outcomes given that information. From these assumptions, orthodox neoclassical models generally predict that, in the long nm, a competitive equilibrium will be a Pareto-efficient one (in other words, no gains will be possible through additional transactions without making at least one other person worse off).

A number of corollaries to this account describe the features of a well-functioning labor market in significantly greater detail. They can be divided roughly into "supply-side" and "demand-side" phenomena (representing workers and owners, respectively). From the supply side, the neoclassical model predicts that a worker's eventual occupation, wages, and so on, reflect both her existing stock of human capital and endowments, and her personal preferences.(40) Over time, of course, workers make investments in building and maintaining their human capital much as they would a set of tools. The tools may depreciate over time, or they may, as the worker hopes, lead to future returns.(41) Not surprisingly, the neoclassical model predicts that the worker will continue to make marginal investments in her human capital until the present value of those benefits just offsets its costs. It also presumes that workers' labor market decisions reflect their preferences. Thus, in making occupational choices, workers will evaluate the multiple personal trade-offs, including work as opposed to leisure, wages versus non-pecuniary considerations (such as benefits or job safety), and so on. Assuming that wages in each type of job respond to other relevant job characteristics, the worker should receive a wage which reflects the marginal worker's valuation of the costs and benefits of contingent employment.(42)

From the "demand side," neoclassical models predict that a firm's hiring decisions will reflect its technological constraints and anticipated costs of hiring another laborer. Firms that have "factor substitutability" (e.g., the ability to use machines instead of laborers, or vice versa) will utilize each factor until its relative cost per unit of production is just equal to the rate at which the firm can, at the margin, substitute one of the factors for another. This suggests that an increase in the price of one productive input may lead to greater use of the substitutable input. Market wages, then, would also reflect the marginal firm's technical ability to substitute factors.

2. Explanations of contingent employment.

The conventional supply-side explanation of contingent employment is that the workforce has changed, increasing the supply of workers--especially women--seeking flexible work arrangements in order to accommodate child rearing and other domestic obligations.(43) These workers may be secondary earners in a dual or multiple-earner household, or sole earners for whom a contingent job will make it easier to juggle market and family obligations.(44) Contingent jobs may also give workers a chance to attend school, acquire training, or sample different jobs.(45) Some researchers emphasize that temporary help agencies provide just such training and job opportunities.(46)

Moreover, they argue, contingent workers may enjoy a higher wage in these jobs than they would otherwise, given their particular skills, training, and experience.(47) Indeed, the parties may have a tacit understanding that the employer will train unskilled workers in exchange for a "subminimum training wage."(48) A similar explanation is offered in connection with minority workers, also overrepresented in the contingent workforce.(49) These workers may on average have lower levels of education and training, and thus choose contingent jobs as a way to acquire needed human capital, or explore career options.(50) Finally, several commentators also laud contingent jobs as "safety-nets" that allow workers to supplement income in retirement (when employers may resist recruiting them for permanent jobs), or to buffer themselves from unemployment during the transition between layoff and a new stable job.

These human capital and worker choice variables, then, combine with the employer's production needs and preferences--demand-side factors--to create a market for contingent jobs. Demand has shifted in a time of globalization of the economy and technological change.(51) Employers have turned to contingent staffing practices in order to increase their so-called "flexibility," i.e., their ability to anticipate and respond to changing production needs.(52) Flexibility at the level of the firm may be either "functional" or "numerical." An employer seeking functional flexibility may undertake to re-design production tasks and training programs to enable workers to move across tasks and job categories as the needs of production may dictate.(53) While firms may choose to hire and train some permanent workers to be functionally flexible, some types of expertise will be needed only for a short time, making it efficient to outsource these tasks. This practice of achieving functional flexibility through outsourcing is especially salient in the manufacturing sector, though it may also occur in the expanding service economy.(54) Functional flexibility might, to some extent, entail the direct use of contract and temporary workers who have skills that are not firm-specific.(55) Construction workers, actors, and high-tech professionals, for example, are likely to market their skills to multiple ventures, without an expectation of staying in one place. Another functional strategy a firm may adopt is to hire subcontractor firms or other agents (who may in turn employ temporary employees, leased workers, or independent contractors) to handle part or all of certain areas of management or production; in this way, the firm can better concentrate on its areas of "distinctive competence."(56)

As it happens, contingent workers are concentrated in the service and construction industries, and correspondingly within occupations that prevail in these industries.(57) Contingent workers are also overrepresented in the occupational category of "professional specialty."(58) Although these observations may be consistent with the "functional" explanation,(59) they may be consistent with other, coextensive explanations as well. For example, the largest group of professional contingent workers is teachers, although the "functional flexibility" account seems inapt in describing the experience of contingent teachers.(60)

Employers may simply seek "numerical" flexibility, i.e., to minimize the cost of adjusting the size of their labor force to meet the needs of production. This might be achieved by establishing relatively weak affiliations with some subset of employees, such that episodes of employment can be initiated and terminated, or hours of work varied, without costly violations of legal rules or customary norms of the workplace.(61) For example, employers may wish to avoid the direct costs of wrongful termination lawsuits and unemployment insurance liability.(62) Similarly, employers may rationally seek to avoid the indirect costs of legal protections, such as the need to screen potential candidates more rigorously and document personnel decisions more carefully.(63) Alternatively, firms may use contingent arrangements as a way to reduce downstream risk by allowing them to observe workers for a period of time before committing to a permanent offer.(64)

A related claim is that employers may prefer workers with part-time, temporary or weak attachments to the firm over full-time, permanent employees because it allows them to save on wage costs and avoid subsidizing costly employee benefits.(65) In this view, even if there is a class of workers who hold contingent jobs despite a preference for traditional jobs, we owe their existence to regulations mandating costly welfare protections for permanent workers. Employers may try to avoid these regulatory "taxes" by hiring workers for whom benefits are not mandatory. Some workers may prefer wage over benefits-based compensation, but in other cases, contingent hiring (or reclassification of workers as temporary workers and/or independent contractors) may occur even though both employer and employee would be better off if the employee increased work hours or attachment.

In the orthodox neoclassical view, then, neither the existence nor the growth of contingent employment is a "problem": they are merely features of a variegated market in which a diverse array of workers with different skills, abilities, and preferences are hired by firms that have corresponding needs and preferences. Labor market arrangements are a function of both workers' human capital and tastes, and employers' production demands and strategic decisions.

3. Policy prescriptions.

Orthodox neoclassicists commonly contend that the group of contingent workers whose status is a "problem" (because they seek but cannot obtain regular employment) may be simply too small to warrant policy intervention. They argue, in turn, that since the vast majority of contingent arrangements accommodate the mutual preferences of workers and employers, further market intervention is ill-advised. Regulations designed, for example, to extend benefits to contingent workers will simply make it more costly for employers to hire them and, though contingent employment may be reduced, it will not necessarily be replaced with permanent employment.(66) Or employers may simply decline to offer private benefits if forced to make them available across-the-board.(67) Further still, some argue that regulations should reward, rather than punish temporary help agencies for the skills training and job placement function they perform.(68)

Other critics of regulations designed to reduce the harms of contingent employment adopt a more explicitly deregulatory posture: To the extent that "regulatory taxes," as described earlier, create inefficiencies in the economy, the appropriate policy solution is to eliminate them. They argue that the elimination of mandated benefits and job security would create efficiencies in the form of higher wage compensation for those who desire it and more latitude for employers to build productivity-enhancing incentives, such as pay-for-productivity schemes, into compensation arrangements.(69) The resulting increase in productivity might, in mm, increase employers' willingness to commit to employees on a permanent basis.(70)

4. Critique of orthodox neoclassical account.

A range of familiar critiques challenge the orthodox neoclassical prediction that labor markets clear and that equilibrium wages and employment will be Pareto efficient.(71) For example, labor markets may fail because workers have inferior information about the features of a particular job. Also, there may be limits on workers' abilities to foresee the effects of their decisions (sometimes termed "bounded rationality"). They may underestimate the possibility that their initial labor market investment decisions will take them down bad labor market "paths," leading to future disadvantage. Or they may "choose for today," preferring high wages now and failing to see that down the road, they will require health or pension benefits. Other critics contend that even where there is full information and rational behavior, public goods problems may plague contracting, especially among nonunion workers. Thus, all workers may desire a particular employment condition, yet no individual worker has the incentive to incur the initial costs of demanding it. Similarly, signaling problems may arise. For example, even if all employees desire job security, no individual employee will demand it because of the risk that she will send a bad signal to the employer that she is the type of employee who needs it. Or employers may decline to offer job security or benefits because doing so may attract the "wrong" kind of worker, i.e., those who are most likely to shirk or fall ill. As such, the suboptimal status quo persists. It is also commonplace to identify market power--whether oligopolies or unions--as creating a socially inefficient undersupply of good jobs. All of these arguments go to the reasons why the labor market may fail to produce efficient bargaining outcomes, contrary to orthodox neoclassical assumptions.

My analysis takes on a narrower question. In simple aggregate comparisons using the BLS definition of contingent employment, contingent workers earn less than noncontingent workers.(72) They are also far less likely than noncontingent workers to have health or pension benefits.(73) The discussion to follow critiques the orthodox neoclassical predictions that contingent jobs simply match workers having particular human capital, endowments, and preferences with employers who have compatible demands, and that contingent jobs pay less because they require less skill, or attract less productive workers.

Involuntariness. An important aspect of the orthodox neoclassical account is that supply-side (worker preferences) as well as demand-side factors drive contingent employment. A problem for these analysts, then, is what to make of "involuntary" contingent workers, i.e., those who would prefer permanent employment but cannot find it. A very high percentage of contingent workers--56%--describe themselves as "involuntary" contingent,(74) though self-reporting of this sort may lead to inflated estimates for reasons discussed later in Part II.B. Yet there is also a large group whose specific reasons for working in a contingent job imply that they sought but were unable to obtain suitable employment: "employer laid off, but rehired"; "could only find this type of employment"; "job may lead to a permanent one."(75) If markets function as they should in the orthodox neoclassical account, workers should be matched to jobs that fit with their preferences and abilities. Orthodox neoclassicists' response is that those workers who report that they have tried but failed to find a permanent job may overestimate their own quality and/or reject jobs for which they are suited.

Contingent workers are "paid what they're worth." The relative youth of contingent workers makes it plausible that they seek contingent employment as a way to sample jobs and obtain training, perhaps with the hope that contingent employment will serve as a stepping stone to a permanent job. Temporary jobs may permit workers to learn about their labor market options and preferences relatively cheaply. Yet early career "churning" may represent different things for different workers. For high school graduates, it is indeed likely to be a way to sample potential occupations.(76) A disproportionate percentage of contingent workers, however, are high school dropouts.(77) Among drop-outs, churning is likely to reflect unproductive time in dead-end jobs or a desire to mix leisure with work in order to mimic the patterns of college friends in the same cohort.(78)

Contingent jobs are also hypothesized to provide on-the-job training. Under such a view, the lower wage earned by low-skill temporary workers is like a placement fee for eventual permanent employment. Thus, for example, the authors of one recent article speculated that temporary help firms use tenure and client reports to determine the wages of their lower skilled workers, thus creating an "enterprise internal labor market."(79) They based this claim on their empirical finding that wages of low-skill temporary workers are determined primarily by their experience in the temporary help industry.(80) The fact that experience predicts wages, however, does not uniquely support the "training wage" hypothesis. Wages could, for example, rise with experience yet still never compensate a worker for the full value of his marginal product. As for the "stepping stone" hypothesis, a different analysis, using data from the Current Population Survey between 1983 and 1993, reported that more than half of those employed as temps during that time had permanent jobs one year later.(81) However, the same study found temporary workers on average more than twice as likely as permanent workers to be unemployed one year later.(82) Because no long-term data are available to track the fates of workers identified in the BLS data on contingent employment, it is impossible to know what proportion of those workers will ultimately move into permanent jobs. A BLS economist, however, suggests that the expectations of temporary and on-call workers of moving into permanent jobs probably exceed reality.(83) In sum, the empirical literature on long-term outcomes for contingent workers is thin and contradictory. In light of this, it seems premature to claim convincing support for the training wage and stepping stone hypotheses.

More to the point, perhaps, the training wage and stepping stone hypotheses are based on the assumption that contingent workers have lower human capital than other workers. Unfortunately, there have been relatively few empirical efforts to measure the wage gap while holding human capital characteristics constant. The aforementioned Israeli study concluded that in low-skill occupations, temporary workers had lower returns to human capital than their permanent counterparts.(84) An American study analyzing data from the 1983-1993 Current Population Survey reached similar results. Controlling for age, race, sex, education, and geographic region, part-time status, union status, and occupation reduced but did not eliminate the wage gap between temporary and permanent workers in blue collar and pink collar occupations, though it eliminated it for white collar workers.(85) In this latter study, the researchers examined benefits as well as wages and found that controlling for the characteristics mentioned above did not significantly reduce the dramatic gap in private health insurance coverage.(86)

Why might some workers be willing to accept temporary jobs when the wages are lower than those of permanent workers in the same occupation who have comparable human capital? One possibility is that they do not, in fact, possess comparable human capital. There may be forms of human capital that simply cannot be captured by even the most sensitive analysis. Differences between workers in terms of quality of education, motivation, ability to get along with others, manual dexterity, reliability, and so forth, may explain the residual gap.(87) Still, it remains puzzling why, if these forms of productivity are so observable to employers that they can put a price on them, econometricians remain unable to observe them.

Another (unsettling) hypothesis is that the wage gap is the result of discrimination against the Hispanics, African-Americans, and women who are overrepresented in contingent jobs. Controlling for occupation or industry eliminates women's disproportionate representation in contingent jobs. In other words, women tend to be concentrated in occupations and industries that have more contingent jobs.(88) This latter observation, however, merely restates rather than explains the problem, or at least raises a new question. Why do women and contingent jobs tend to converge in certain occupations and industries? This segregation may itself be the result of discrimination. As for race, blacks too tend to be concentrated in industries that have high rates of contingency.(89) In contrast with women, however, blacks additionally have markedly higher rates of contingency than whites even in the occupations and industries that themselves contain the most contingent workers.(90) Thus, blacks are not only more likely to work in high-contingency industries, but they are also more likely than whites within those industries to hold contingent jobs.

In sum, orthodox neoclassicists can claim partial support for the hypothesis that contingency status is the consequence of human capital differences and revealed worker preferences. For some workers, contingent jobs undoubtedly serve as a source of training and a stepping stone to permanent employment. By most empirical measures, however, there is wage gap between contingent and non-contingent workers that persists even after accounting for conventional measures of human capital. Certainly, this persistent gap may be merely due to omitted variable bias, i.e., econometricians may have overlooked or cannot observe certain human capital variables that correlate with contingent status and explain the wage gap. But for those who hypothesize that contingency is a labor market problem, this explanation fails to meet the burden of persuasion. In other words, the evidence does not rule out the competing hypothesis that some workers are underemployed--unable to obtain jobs on the same terms as others with comparable preferences and human capital attributes. Part II.C discusses what policy interventions, if any, are appropriate if persistent underemployment exists. First, though, I take up radical theories of contingency.

B. Strong Segmentationist Accounts

The following passage from a recent law review article is typical of the strong segmentationist view of contingent employment:
   The economic realities of the market are more and more becoming a tale of
   two cities. Employers are increasingly employing higher wage, higher
   skilled workers while using the labor of lower skilled, lower wage workers
   through contracting, leasing, or temporary agency arrangements. As a
   result, low-wage workers are kept out of internal labor markets where they
   would gain benefits, such as health insurance, that they are unable to
   obtain in the external labor market. Furthermore, these arrangements allow
   employers to distance themselves from the exploitation of low-wage workers
   while still benefiting from their exploitation. Thus the use of contractors
   is creating two labor markets--one in which workers are motivated by the
   prospects of advancement, participation and job security, and the other in
   which workers are motivated by insecurity and fear. Disturbingly, this
   segmentation is often along racial, ethnic and gender lines.(91)


The strong segmentationist school emphasizes a barrier between "good" and "bad" jobs, caused at least partly by structural constraints imposed by society and employers. These assumptions invigorate their assertion that the negative features of contingent employment, both economic and psychological, are "harms" which justify active regulatory intervention.

1. Overview of the model.

The rhetoric of segmentation draws on the labor market segmentation paradigm, a descriptive account of labor market institutions to which both strong segmentationist and New Keynesian accounts are heavily indebted. This paradigm forms the backbone of their claims that "bad" jobs may persist in ways that orthodox neoclassical models reject. Peter Doeringer and Michael Piore's Internal Labor Markets and Manpower Analysis, first published in the early 1970s, is perhaps the most influential work on labor market segmentation.(92)

The core premise of labor market segmentation theory is that the economy has multiple, distinct labor markets in which working conditions, wage bargains, and job stability differ systematically.(93) In the simplest version, there are two labor markets: primary and secondary.(94) The chief distinction between them is that primary sector workers have greater access to "internal labor markets." Stated simply, an internal labor market is a set of bureaucratic rules and customs that allocate job security and wages. Ports of entry into these labor markets are limited, and workers within internal labor markets are buffered from ordinary competitive forces. The paradigmatic internal labor market worker has high wages resistant to demand shifts, due process in the event of discipline or dismissal, "career ladders" by which she can advance to higher-echelon positions, substantial returns to her investment in education and training, and most salient of all, job security. In secondary markets, by contrast, access to internal labor markets is severely limited. Jobs in secondary markets tend to require fewer skills, have poorer working conditions, and entail closer performance monitoring. Workers have fewer protections against arbitrary treatment by the employer, fewer opportunities for advancement, higher turnover, and lower job security. According to this account, the employment levels and wages of workers in the secondary labor market vary with supply and demand factors in a manner more closely matching the way neoclassical models predict competitive labor markets function.

The initial decision to enter into the secondary labor market may well be influenced by a worker's preference for transitory employment in a noncommittal arrangement because of other obligations or desires, such as time for family or other interests, a counter-cultural desire to occupy fringe labor markets, or temporary immigration status.(95) Early accounts of labor market segmentation, however, were concerned in significant part with less clearly exogenous labor market factors. Doeringer and Piore's paradigm strongly emphasized the structural and historical determinism of labor market outcomes.(96) They posited "mobility chains" determined by social and historical contingency: Rather than resulting solely from education, experience, and preferences, the segment into which a worker initially falls also depends on social class, family background, neighborhood, race, gender, etc. The point here is not only that firms may discriminate against workers on the basis of, say, race at the point of entry ("we don't hire minorities") or indirectly through actions that would be recognized under adverse impact theories of discrimination (ability testing where the testing instrument is biased in favor of whites). Doeringer and Piore argue that still more nuanced forms of exclusion may occur through ordinary recruitment channels and thus be overlooked by ordinary prohibitions against discrimination.(97) A firm's reliance on information and referrals from the friends, civic associations, churches, school counselors, and other community contacts of incumbent workers will tend to perpetuate racial employment patterns, both in terms of who gets hired, and where those who do get hired are placed within the firm (i.e., high-skill versus low-skill areas of production). Job seekers may also self-select: Those who do not have an "in" with a firm or area of production, or do not see a critical mass of workers in particular jobs who share their characteristics, may decide not to apply. Members of minority groups may be systematically excluded in this fashion. Initial placement, in turn, influences the long run career path of the worker.

While some of these observations (e.g., that a worker's initial port of entry can influence her subsequent path in the labor market) are consistent with human capital theories of the orthodox neoclassical school, the key distinction of Manpower Analysis is epistemological. It emphasizes how significantly behavioral traits and preferences are determined by learned experience. To the extent behaviors within the primary and secondary sectors give rise to different sets of rewards and punishments, workers learn different kinds of behaviors consistent with local reinforcement schema; they also develop behavioral patterns through imitation and conformity with group norms.(98) These patterns of learning, in mm, deeply affect labor market outcomes. For example, skilled blue collar workers (Doeringer and Piore's "lower tier primary sector" workers) receive on-the-job training rather than formal education. They thus become experts in the stable, routine skills within their sphere of the production process, which permits them to function without much direct supervision. Also important, however, blue collar workers develop habits that conform with the customs and norms of the blue collar shop floor culture.(99) This entrenchment, argue the authors, then limits their access to the upper tier of the primary sector, where workers trained to engage in more abstract kinds of problem-solving develop different organizational cultural traits.

Extending this further, in secondary markets, where jobs are unstable and undesirable, experiences at work may reinforce the attitudes that confine secondary workers to secondary markets. Where supervisors are harsh or abusive, for example, workers might learn adaptive behaviors that help them resist abuse in their current jobs but would be considered insubordinate by primary sector standards.(100) Doeringer and Piore further argue that adaptive learning shapes patterns of affiliation and mobility in society more broadly: If workers in secondary jobs learn few skills, are treated poorly, and know that retention prospects are low, they may learn adaptively to place little value on job security. In this view, workers in unstable secondary markets learn to rely more heavily on friendships and attachments outside the workplace, thus devolving still further away from the possibility of cultivating networks of attachments that could facilitate internal promotion.(101) Some workers may find themselves trapped in secondary markets despite their ability and desire for primary employment. Their communities, families, or friends from school may encourage them to enter into the labor market through secondary channels from which it is later difficult to move, or firms may tend to select them for secondary rather than primary jobs because they perceive them as better suited to secondary employment. Thus, some workers' secondary status is "involuntary" because they seek (or initially sought) alternatives they are capable of performing, but cannot obtain them.

In sum, the segmentationist paradigm differs in two important respects from the orthodox neoclassical account.(102) First, segmentationists posit that there are workers qualified and able to work in "good" jobs, but unable to gain access. In contrast, the orthodox model argues that the distributional characteristics of labor markets simply reflect the "quality" or preferences of the workers, but that all markets eventually clear.(103) Second, the segmentationist concepts of constrained choices and path-dependent preferences generally runs counter to the orthodox neoclassical conception that "tastes" are exogenous.(104) Such concepts would imply that workers' early educational and labor market experiences, along with their status within labor market institutions, profoundly affects their personality, goals, and sense of their own potential.

As mentioned at the outset of this section, the segmentationist account strongly influenced both strong segmentationist and New Keynesian theories. In the remainder of this Part, I will describe the radical variant of the segmentationist account embraced by strong segmentationists. When I discuss New Keynesian theories in Part II.C, however, I will return to many of the themes from the foregoing paragraphs.

2. Explanations of contingent employment.

One can trace distinct themes from the labor market segmentation model in the strong segmentationist account of contingent work. Numerous commentators explicitly ascribe "secondary labor market" status to contingent workers.(105) Contingent workers are also variously described as "ring," "periphery," or "externalized" workers,(106) though contemporary models often contemplate a more nuanced or slightly different relationship between primary and secondary workers than the simple dual model I have described. Strong segmentationists are especially influenced by the more radical strain of labor market segmentation theory, which overlapped with Doeringer and Piore's work but strongly emphasized class-driven employer control mechanisms. Rather than explaining segmentation on the basis of the actual preferences of certain workers for unstable employment, the entrapment in secondary markets of others who resemble secondary workers and whom employers therefore falsely label as suitable and desirous of secondary jobs, and the path-dependent evolution of entrenched segments due to employer expectations and workers' adaptive behaviors, radicals explained it on the basis of the divide-and-conquer imperatives of capitalists.

In the radical view, the profit-seeking capitalist seeks to avoid the spread of solidarity that might lead workers to demand a larger share of production rents (in the form of higher wages, better working conditions, etc.).(107) The principal way to do so is to prevent workers in less desirable jobs from identifying with workers in more desirable jobs and developing a class consciousness about the "oppressiveness" of their working environment.(108) Methods for preventing the emergence of class consciousness might include stratifying jobs by educational requirements and specialization, assigning workers least likely to identify from the outset with privileged white collar workers (such as minorities) to secondary jobs, and creating organizational hierarchies designed both to motivate and to balkanize workers of differing status.(109) The radical view therefore finds the basis for segmentation in the decisions of capitalists to exploit and cultivate class, gender, and racial divisions, rather than in the interaction between the supply and preferences of workers and the historical evolution of labor market paths and institutions.(110) Two key factors, then, characterize the radical or "labor process" theories. First, the nature of the workplace (demand factors), rather than solely the characteristics and preferences of workers (supply factors), cause labor market segmentation. Second, within the bounds of existing technical and historical constraints, the capitalist uses his considerable leeway over the organization of production as a way to control labor and extract labor power.(111)

Having described radical theorists' distinct explanation of labor market segmentation, I shall illustrate how strong segmentationists embrace this radical view in the debate about contingency. They do so by making two linked claims. First, they argue that much of contingent employment is involuntary. Second, they argue that the involuntariness and other features of contingent employment foster "alienation," which, in turn, further perpetuates labor market disadvantage. Let me discuss each in turn.

Strong segmentationists stress the degree to which the rise of contingency stems not from the choices of workers, but from conscious decisions by employers to create a division between career jobs and "dead-end" jobs. While acknowledging the view that intensification of domestic and international competition, technological change, and escalating payroll costs may be important factors in the drive toward contingent staffing, strong segmentationists particularly emphasize the role of "externalization" of jobs as a way for employers to exert control over labor.(112) They also reject interpreting the influx of married women and young workers into the workforce over the past twenty-five years as evidence that workers increasingly seek contingent jobs.(113)

Strong segmentationists' prime target for advancing this first claim is mainstream labor economists' (e.g., the Bureau of Labor Statistics (BLS)) definition of "voluntary" labor market status. A standard inquiry in survey instruments used to measure the prevalence of part-time and insecure employment is the individual's reason for working in his current employment. The conventional indicator of "involuntariness" is when the worker gives an "economic" reason for doing so, e.g., "could only find this kind of employment," or "this job may lead to a permanent one." In this view, only constraints in available job choices--"hard constraints"--lead to classification as involuntary. Those who choose contingent jobs because of family or other personal obligations are typically excluded from the category of involuntary.(114) According to this measure, only 34.7% of (BLS-defined) contingent workers are involuntary.(115) Yet the percentage of these workers who say that they would rather be working in a secure job is considerably higher--55.8%.(116)

What explains the discrepancy? While not denying the relevance of hard constraints, radical critics argue that statisticians are wrong not to count as involuntary at least some workers who give "personal" reasons for working in a flexible job. In a "constrained choices" view, cultural stereotypes and community norms that reinforce, for example, traditional family and economic roles of men and women have a coercive influence on women's choices.(117) Thus strong segmentationists view at least some of the reasons labeled as personal--such as a woman's decision to stay home to raise children--as additional constraints on the manifestation of true preferences ("soft constraints").(118) Such arguments are directed at those who might infer that some groups in society have "bad jobs" because they are uncommitted, unreliable, unmeritorious, or part of a "culture" lacking in ambition. These arguments shift the emphasis to non-individuated social and geographic influences such as class, race, gender, and community.(119) Importantly, though, the arguments are framed not only in terms of society-wide reproduction of hegemonic social attitudes, but also in terms of employer control. Thus an employee's choices may be constrained precisely because an employer decides to limit the set of jobs available to particular groups of workers. Capitalizing on gendered and racialized understandings of appropriate work, the employer is able to enlist these workers for routinized, limited opportunity, low-discretion jobs.(120)

The proposition that many contingent workers' choices are constrained in a fashion that reflects and reproduces forms of social inequality anticipates the second strong segmentationist claim: that contingent employment is alienating. Strong segmentationists argue that contingent jobs are "bad jobs" in part because secondary status may lead to other, sometimes latent, emotional costs: loss of self-esteem, loneliness, powerlessness. Some label these three phenomena collectively as "alienation."(121) Thus, for example, in their study of temporary academic employees, Katovich & Hardesty write of a worker's alienation from herself. They describe the temporary worker as vulnerable to the kind of anomie Durkheim described, because the ever-present "dangling possibilities" of becoming permanent create a world that is "simultaneously hopeful and grim."(122) Other burdens are loneliness and separation from social life, that is, the inability to form lasting or genuine friendships, invisibility in terms of one's views not being taken seriously, and sometimes physical separation from others in the workplace.(123) Finally, strong segmentationists argue that contingent employment may lead to class division and disempowerment in society more generally.(124) For example, employers may monitor contingent employees' behavior in a fashion that represents the exercise of control over workers by employer-capitalists.(125)

This second strong segmentationist argument regarding alienation links neatly to the epistemological strain in classic treatments of labor market segmentation theory; that is, in a path-dependent fashion, contingent workers' experiences in unstable jobs will affect their knowledge and assessment of their opportunities.(126) It is argued that contingent workers' identity is shaped and their ambitions undermined by their low status assignments, the demoralizing effect of the threat of unemployment, their detachment from other workers, and their powerlessness.(127) In other words, workers will internalize the idea that, whether by desert or fate, they are not valuable workers, are not part of workplace life, or are not pegged for the career track. They thus set lower expectations for themselves, and are less motivated than workers in the "core" to establish career-enhancing networks within the firm or to acquire human capital.(128)

3. Reform proposals.

In light of the harm they perceive from the advent of contingent employment, segmentationists advocate reforms that fall generally into two categories: (1) extending greater regulatory protection to contingent workers, such as reducing the statutory thresholds for employment-based social benefits; and (2) reforming collective bargaining laws. Both aim to address the perceived harms of secondary labor markets by, in essence, converting contingent jobs into "good" jobs, or at least reducing the gap between contingent workers and their counterparts in the primary sector. Many commentators advocate both types of reform.(129)

Minimum standards and other substantive regulations. Many reformers advocate changing state and federal regulations to extend employment benefits to contingent workers.(130) These efforts have garnered some interest on the part of legislators, but their discussion and research has yet to result in concrete law reform.(131) Reformers advocate enacting a system of health coverage on a universal or prorated basis,(132) pension portability,(133) wage floors,(134) and a more liberal application of wrongful termination rules and severance protections such as unemployment insurance.(135) Some would enact laws directly regulating the temporary help industry and employee leasing companies, requiring, for example, that they demonstrate sufficient net worth to cover obligations to employees such as payroll, tax withholding, and payment of medical benefits.(136) Others seek to prohibit employers from deliberately reclassifying employees as independent contractors even though many such "contractors" remain dependent on a single contract of employment.(137) Although recent litigation has successfully challenged such practices,(138) reliance on piecemeal enforcement may be inadequate, particularly where the workers affected are unsophisticated litigants.

Collective bargaining. Extending collective bargaining to contingent workers is the second type of reform proposal. In contrast to the regulatory interventions described in the previous section, these proposals are designed to help contingent workers to collectively negotiate tailored wage, benefit, and job security provisions with the employer.(139) Critics argue that traditional organizing techniques fail to reach contingent workers because, inter alia, such techniques are based upon assumptions of worksite attachment, fixed bargaining units, and the gradual on-site recruitment of workers into the union.(140) This has led reformers to urge Congress to modify the National Labor Relations Act. There has been little serious effort to push for protective policies that might stem the decline of the post-war model of stable oligopoly primary sector firms and strong plant-based bureaucratic unions.(141) Instead, advocates suggest reforms and methods that would facilitate new forms of bargaining adapted to the contingent workforce.

There have been several proposals to revitalize "occupational unionism." Used historically by the tradespeople who moved from one jobsite to another, occupational or "craft" unions organized occupations rather than industries, emphasizing steady access to work as it became available in the market, rather than rights to individual jobs.(142) The occupational model was recently repopularized in the policy context of contingent employment, most notably by Dorothy Sue Cobble's historical study of waitress unions.(143) The striking aspect of Cobble's study is that it challenges the received wisdom that craft unions are viable only among elite tradespeople for whom the lynchpin of union solidarity is skill-based "craft pride."(144) Cobble's waitresses resembled many existing service industry workers whose jobs may be perceived as inadequately high-skilled to form the basis for union solidarity under the traditional craft paradigm. Yet between the mm of the century and the late 1960s, female-dominated groups of hotel and restaurant workers successfully organized along occupational (and skill) lines. The key features of the waitress unions were union control over the training and supply of labor, provision of portable benefits managed by the union, and peer-driven discipline and work performance standards.(145) Job assignments were determined by a union-managed hiring hall (according to factors such as how long it had been since a worker's last job assignment), rather than rigid systems of seniority, thus achieving a more egalitarian distribution of available work opportunities than industrial unions.

Post-war restrictions on union activities such as the closed shop and secondary boycotts contributed to the decline of waitress unions. Cobble envisions adapting the craft model to the needs of non-traditional workers, with a keen emphasis on how these unions might help workers augment human capital and career advancement. This would occur through a combination of union involvement in training and the promotion of worker- or state-run employment bureaus which would match workers to job assignments. Necessary legal changes would include expanding the definition of "employee" under the NLRA, giving unions greater flexibility in determining bargaining units (so that they might include more part-time and temporary workers), allowing unions to participate in the hiring, discipline, and training of members, and removing restrictions on the use of pre-hire agreements and secondary boycotts, currently available only to limited classes of workers such as the building trades under the NLRA.(146) Extending these exemptions to other sectors would make multi-employer hiring-hall style unions possible for a broader cross-section of the workforce.

A related proposal is to encourage geographically-based occupational groupings, as were formed in the Justice for Janitors campaigns in various cities.(147) According to this model, the key to improving the lot of contingent workers is to identify non-work social organizations such as community, ethnic, racial, and religious groups and to turn these collectives, which may otherwise entrench labor market disadvantage, into channels for union solidarity and occupational mobility.(148) Taking Wial's proposal as an example, before embarking on a unionization drive, union organizers would locate neighborhoods, clubs, churches, or other gathering places containing cohorts of workers with shared identity or occupational interests.(149) Union organizers would then facilitate the creation of "pre-collective bargaining associations" based on these groups, in which workers might identify common interests and learn how to organize. These methods might reach workers who often remain unorganized. Also, importantly, they might help target and redirect groups of workers who tend to follow dead-end job mobility paths into more promising ports of entry into the labor market.

Cobble and Wial share with other advocates the conviction that, in addition to lobbying for or managing such things as (portable) employee benefits programs--which might also be achieved through non-union linked regulatory reform--unions could play a strong role in hiring, training, and job referrals.(150) Unions might also work in partnership with the public sector or private groups.(151)

4. Critique of strong segmentationist arguments.

To summarize sections 1-3 of this Part, strong segmentationists argue that while "hard" mobility constraints (lack of suitable jobs) explain some of the involuntariness of contingent employment, "softer" systemic forms of coercion based on employer control strategies and entrenched social norms are also at play. Capitalists enrich themselves through the use of devices--like segmented labor markets and contingent arrangements--that shift productive surplus away from labor. Whether or not this is efficient, strong segmentationists argue, it leads to unjust distributional outcomes. Compounding these "hard" and "soft" constraints is the phenomenon by which workers' entry into involuntary contingent jobs may lead to a vicious circle of further harms: unhappiness, loneliness, and boredom, as well as a loss of the self-esteem and motivation necessary to acquire and maintain useful human capital and achieve one's potential. The possibility that worker preferences are path-dependent, segmentationists assert, offers additional justification for labor market interventions designed to ameliorate inequalities in employment opportunities and outcomes.

In order to determine the plausibility of this account, one must ask first whether the underlying assertions about worker coercion and alienation are themselves convincing. One must then ask whether the above reform proposals--wider availability of mandated benefits and protections and expanded forms of collective bargaining--are a good "fit" for this theoretical account. I find the segmentationist response to both of these questions only partly satisfactory.

Involuntariness. It seems wholly plausible that some workers' choices are limited by "hard" constraints. The actual jobs available to people who wish for flexible schedules may not be fully suited to their combination of preferences and capabilities,(152) or workers may be unable to leave communities that lack good jobs because of family ties, limited housing options in other communities, and so on. The existence of "soft" constraints also seems plausible. Many who work in contingent jobs for so-called "personal" reasons are surely pressured to do so because traditional understandings of the appropriate division of labor operate within family, community, and workplace.(153) Choices may also be distorted by legal institutions: Tax laws, family laws, and the provision of social services (such as child care) may create incentives that reinforce these traditional understandings or reduce worker mobility.(154) Women, minorities, new immigrants, and other groups historically less successful in labor markets likely face these kinds of constraints disproportionately. The existence of such "soft" constraints means that conventional measures of involuntariness (which, again, include only those workers who have attempted yet failed to obtain permanent employment) may understate the problem.

As intuitive as the notion of soft constraints is, however, the existence, extent, and consequences of such limitations on freedom are difficult to isolate in practice. Two major challenges arise. First, while strong segmentationists may be legitimately skeptical of the probativeness of conventional measures of involuntariness, relying on other techniques to obtain evidence of employer- or society-driven coercion may lead to equally troubling inaccuracies. Take, for example, appeals to self-reported preferences. Contingent workers who say they would prefer a permanent job may conflate lack of satisfaction with involuntariness. To be sure, many contingent workers may be "stuck" in unsatisfying career paths that are difficult to leave. They may prefer more interesting tasks or better working conditions. Yet this does not mean they are involuntarily employed in a sense that is meaningful from a reform standpoint. More importantly, measures based on self-reporting fail to take into account whether workers' aspirations are commensurate with their own human capital, endowments, and capabilities. A worker who says she would prefer a permanent job may hold an insecure position because the employer has accurately judged that she has a lower level of education or talent than others the employer is willing to hire permanently. To put this point differently, policies to create "better" jobs are misguided if the workers who want them are not qualified.

Consider strong segmentationist reforms designed to overcome income stream interruption and lack of benefits by prorating unemployment and fringe benefits. Forcing employers to offer benefits to workers who have part-time schedules or less secure promises of future employment may, in the end, be counter-productive. Employers' willingness to make new hires hinges, at least in part, on the compensation package they anticipate having to pay workers. If prospective employers of contingent workers know that they will be required to offer those workers additional benefits, then they might offer commensurably lower wages to compensate for the anticipated cost of non-pecuniary compensation. However, to the extent that some contingent work is truly voluntary (i.e., some workers prefer higher wages and flexibility over lower wages and better non-wage protections) this scenario would make certain workers worse off. Alternatively, the employer may offer the mandated benefits without reducing wages but choose to employ fewer workers, thereby pushing some workers into involuntary unemployment. In either case, a plausible result of such mandatory regulations would be to replace one type of unchosen harm with another, often to the detriment of both workers and employers.

This is not to say that lack of adequate benefits coverage for contingent workers is a trivial problem. Nor do I mean to suggest that the state should do nothing towards a solution, nor even that it would be impossible to defend the potential trade-offs between mandated, cross-the-board employee welfare protections and depressed wages or increased unemployment. My point is that many strong segmentationist reform advocates fail to explicitly defend these trade-offs and as such, they make a less convincing case for themselves.

A second burden for strong segmentationist theories of "involuntariness" is to distinguish between labor market and "pre-labor market" problems, that is, between those that result from the functioning of the labor market, and those due to other societal variables. Strong segmentationists argue that even the liberal survey instrument of self-reporting understates labor market coercion because workers themselves may underestimate the degree to which their choices are constrained by employers or society. Everyone is constrained somehow by family, place of birth, physical and mental endowments, appearance, wealth, and so on. But historical and social constraints on peoples' capacities to lead ideal lives, real and troubling as they may be, are not always labor market problems. Thus, if a worker's choices are constrained because of a boss who exerts oppressive control in order to maximize profits, then the solution may be labor market regulations that require safety precautions, unemployment insurance funded by taxes on employers, or payment of a minimum wage. If, on the other hand, women disproportionately choose part-time work because of constraining social attitudes about the role of women in the family, policies to subsidize the replacement of part-time jobs with full-time, secure alternatives may misfire. Similarly, if because of poverty a technically gifted worker never had the chance to go to college and earn an engineering degree, the problem will not be solved by expanded collective bargaining or mandated benefits. In the latter two examples, reform of "pre-market" institutions would be a more vital and immediate imperative than labor market reform per se.

I argued in Part I that one way to approach the matter would be to focus only on those workers who both would prefer a different job and who are capable of performing that job--in other words, those who are underemployed. If mandating cross-the-board benefits were to create better jobs, and workers currently underutilized were both qualified and willing to accept these superior jobs, then firms may receive sufficient returns to the increased compensation to render the change efficient.(155)

Recall, however, that a crucial assumption of strong segmentationists is that workers may get trapped in bad career paths as a function of their port of entry into the labor market. Informal networks based on neighborhoods of origin, friendships, and other social networks may, in mm, be important determinants of these ports of entry, further entrenching labor market "ghettoes" created along ethnic, class, and gender lines.(156) Being pushed from traditional to temporary or part-time employment may lead to similar pathologies.(157) Underemployment, whether among "downsized" workers or new entrants into the workforce, is surely more worrisome if it triggers a harmful pattern of continued underemployment or unemployment, either by directly stigmatizing workers in the eyes of future employers, or by leading workers to lower their expectations of themselves and give up searching for better work. Either of these "secondary effects" of underemployment would in mm hurt workers' chances of finding a good job in the future. Disturbingly, there is substantial evidence that these self-reinforcing phenomena occur.(158) Despite my earlier criticisms, I find this aspect of the strong segmentationist story compelling. It contemplates workers who, in a deeper sense than implicated by the discrete phenomenon of the skills-job mismatch, may underutilize their human and labor market potential. In essence, it contemplates the paths never taken. An important question, then, is whether strong segmentationist proposals are responsive to this concern.

Advocates of proposals to organize contingent workers envision unions as able to create a fairer distribution of labor market opportunity. Unions may negotiate better pay and benefits for their members and expand the base of inclusion. On average, jobs should improve, and extremes in the distribution of opportunities should flatten as more workers obtain a "share" of available jobs through hiring halls. Isolated unionizing drives, however, even city-by-city (as opposed to plant-by-plant), may be inadequate to the task. Some advocates propose schemes that would incorporate outreach and manpower programs into unions to avoid the informational barriers and the training-jobs mismatch that inhibit labor market mobility.(159) Of course, the efficacy of existing federal and state training programs, particularly those that pertain to youth, has been controversial.(160) The long-term success of innovative hybrid initiatives may depend on the outcome of ongoing local experimentation. In any event, my inquiry here is different. Assuming the strong segmentationists are correct that ameliorating both hard and soft constraints on choice is crucial to achieving labor market equity, these proposals are a promising "fit."

Alienation. The second strong segmentationist criticism of contingent employment is that it creates or magnifies "alienation." As discussed earlier, the term "alienation" is used by strong segmentationists to mean a range of things: personal anomie and indignity, loneliness or separation from others, and powerlessness in society more broadly.(161) However, it poses several ambiguities. First, the kinds of alienation suffered by individual contingent workers may be idiosyncratic. What is alienating for one worker may be liberating for another. Freedom from the responsibilities of integrating into the political and social life of the workplace, for example, may allow a worker to channel his time into other preferred activities.(162) Moreover, the opportunity to work despite disabling conditions, to explore different jobs as a way to find oneself, to avoid forming emotional attachments, or to get away from the drudgery of housework, all may be seen as identity-enhancing aspects of contingent employment.(163) Second, causality may be difficult to discern: Is alienation due to contingency per se, or other factors correlated with contingency? For example, perhaps members of the demographic groups concentrated in contingent jobs are more likely to experience alienation than others, whatever job they hold. Or perhaps occupations having higher rates of contingency (such as lower skilled service and clerical jobs) would be "alienating" even if permanent. Recent experiments attempting to measure the relationship between temporary employment and job satisfaction, salutary as they may be, suffer shortcomings stemming from these problems of causality.(164) Third, different dimensions of alienation--boredom, exclusion, lack of skill development, lack of confidence, and lack of autonomy--cut across very different dimensions of the human experience and may require vastly different responses.

I mentioned above that strong segmentationists must defend the complex trade-offs that may result from trying to create good jobs through labor market intervention. A common defense of collective bargaining is that, while it may not eliminate these trade-offs, unions offer countervailing social benefits for members by substituting for traditional jobs as a locus for work-related attachments, job satisfaction, and so on.(165) It is difficult to know, however, whether contingent workers will indeed overcome feelings of psychological detachment and "outsiderness" through the labor market institution of unions. Studies like Cobble's history of waitress unions suggest that this type of union may well create beneficial forms of community through union membership. However, it is also possible that these unions would depend on forms of social conformity that themselves undermine worker autonomy.(166) Perhaps resolution of these problems ought to be left to nonlabor-market-based institutions such as identity group organizations, religious affiliations, hobby clubs, and so on. But these also may be unavailing if workers with, for example, multiple part-time jobs have no time to join. Thus, despite compelling accounts of the degrading and undesirable features of many contingent jobs, and despite quite plausible speculation about its consequent social costs, further efforts to define alienation as it affects contingent (or underemployed) workers, determine its antecedents, and assess its pervasiveness are necessary to explore more fully the limits of and possibilities for responsive reform.(167)

In sum, the strong segmentationist account of contingent work resonates with fears held by many about a rising underclass of workers trapped in dead-end jobs. Multiple constraints on choices--both "hard" constraints on the availability of jobs and "soft" constraints arising from certain forms of social coercion--may operate to create an inequitable distribution of labor market opportunity. Subsequent skill erosion, loss of confidence, and loss of motivation may create a self-fulfilling prophecy of involuntary contingency, further compounding the harms. However, even accepting, as I do, that the self-fulfilling prophecy phenomenon occurs, the strong segmentationists' theory of what causes this ultimate outcome--forms of oppression by capitalists and society which constrain workers' choices and produce alienation--seems at odds with evidence that many contingent workers have jobs that satisfy and challenge them fully. Concepts of involuntariness and alienation need sharper definition and refinement for purposes of devising sound policy responses and fully appreciating the trade-offs that may result.(168) Until these challenges are surmounted it is prudent to explore further for a more complete positive account of contingent employment. It is to this inquiry that I now turn.

C. New Keynesian Accounts

The commonsense intuitions that flow from both the orthodox neoclassical and strong segmentationist models perhaps represent their best and worst respective traits. On the one hand, each account provides an elegant paradigm. On the other hand, however, each claims too much about its explanatory power, casting doubt on the efficacy of its particular regulatory agenda. Moreover, the models analyzed above share little common ground with one another, making it seem difficult to adopt attractive features of one account without rejecting the other out of hand. In this section, I turn to a third, "New Keynesian" paradigm, that manages to carve out a middle ground between the polar approaches described above.

1. Overview of models.

Many labor economists otherwise sympathetic to neoclassical approaches have long been troubled by the empirical observation that unemployment and entrenched labor market segmentation persist contrary to the predictions of market-clearing models. Perhaps inspired by this puzzle, a number of New Keynesian theoretical accounts have emerged in recent years.(169)

While relaxing a number of market-clearing assumptions, New Keynesian models retain several central tenets of neoclassicism. Specifically, they accept the following: (i) firms make hiring decisions so as to maximize profits; (ii) workers make labor market decisions so as to maximize utility; and (iii) owners' and workers' expectations are rational. Nevertheless, New Keynesian models resist "assuming away" the phenomenon of persistent unemployment. To the contrary, they offer numerous reasons why one might not expect wages and prices to adjust frictionlessly to eliminate a transitory surplus in the supply of labor. In the face of such wage "stickiness," some portion of the workforce is simply unable to find work on the same terms as others, despite ostensibly comparable productivity and tastes. "Underemployment," as I am using the term, is a variant of this concept: The underemployed worker is employed, rather than unemployed, but in a job that undervalues her motivation, talents, training, and experience relative to other workers.

Despite their primary focus on the microfoundations of unemployment, New Keynesian models are heavily indebted to the core labor market segmentation theory described in Part II.B.1. Like the strong segmentationists, New Keynesians contemplate that some workers will be forced, despite their true preferences and talents, into secondary labor markets. They also contemplate that path-dependency may shape long-term outcomes. In this sense they stand apart from orthodox neoclassicists. Yet the notion of "involuntariness" embraced by New Keynesians is different from that embraced by strong segmentationists in two important respects. First, fewer things are perceived as involuntary: New Keynesians share with neoclassical orthodoxy a greater faith in the validity (at least in the short run) of workers' facial preferences and talents in determining individual labor market outcomes. Second, they explain any existing involuntariness within labor markets as caused by factors that may have nothing to do with employer control or assertion of power over workers. Instead, New Keynesians hypothesize that involuntariness depends primarily on (potentially non-correctable) failures in otherwise efficient markets.

Strikingly, although many observers in contemporary debates about contingency recognize the connection between unemployment and contingent employment, very few have marshaled analytic arguments from the New Keynesian tradition to their aid. Conversely, in the explosive current literature on New Keynesian models of unemployment and labor market segmentation, all but a handful of investigators either ignore or relegate to footnotes any explicit application of their models to contemporaneous policy debates about contingent employment. A central contribution of my analysis is to suggest that despite this oversight, New Keynesian labor market models may be more promising as explanations of underemployment than unemployment, and further, may fill important gaps in existing popular accounts--both orthodox neoclassical and strong segmentationist--of the causes and consequences of contingent employment.

2. Explanations of underemployment.

Although several hypotheses fall within the New Keynesian framework, I shall focus on two: efficiency wage and insider-outsider. These two accounts offer the best insight into the phenomenon of contingent employment.

Efficiency wages. Efficiency wage models fall into several categories: adverse selection or "signaling" models; moral hazard or "shirking" models (and closely related turnover models); and "gift exchange" models. Those I discuss all assume that employers can use wages instrumentally to overcome problems caused by information asymmetries. For instance, employers may lack reliable information about workers' potential productivity, or may be unable to monitor the effort workers expend from day to day. Because of these information gaps, employers devise rational strategies to screen workers, to elicit effort, or both. One such strategy is to set wages above the market clearing level.(170)

"Signaling" accounts center on information asymmetry in the hiring market. Rational employers who lack information about the prospective productivity of competing job applicants may use easily identified proxies to sort workers according to predicted quality.(171) Such proxies or "signals" include educational characteristics, physical appearance, and job history. The employer estimates a candidate's potential productivity by looking at the productivity of workers with the same characteristics.(172) According to Michael Spence's classic account, workers' knowledge of employer sorting practices may in turn lead to feedback effects: Workers may act strategically to improve the quality of the signal they send, or may exert little effort at a job search in areas where they know the signal they send will be disfavored.(173)

Another type of signaling may occur when an employer uses the wage at which an employee is willing to work to infer her quality.(174) According to this account, a worker's very act of agreeing to work for a wage X implies that the monetary value of her marginal productivity must be equal to or less than X. While a rational employer may set a high wage in order to attract superior workers, she will also attract some low-productivity workers who will benefit from the employer's inability to accurately assess their productive capacity. The consequence is that high- and low-productivity workers will compete for the same high wage positions, and at least some high-productivity workers will be crowded out by low-productivity workers. In equilibrium, some workers will be unemployed despite preferences and abilities identical to those of workers who are hired. Over time, the gulf between the workers crowded out and those employed in high-wage jobs may increase, as unemployed workers' skills and motivation atrophy. Thus an initially arbitrary dichotomy between equally qualified workers may become a self-fulfilling prophecy. One could also tell a story in which the initial dichotomy was not arbitrary: If employers use the rational but discriminatory sorting practices described by Spence and others to choose among those competing for high wage jobs, the workers who get crowded out may tend to be women and minorities.

How, then, might the signaling hypothesis fit with what we observe about contingent employment? Some contingent workers may be the rationed-out workers who queue with aspirations of eventually landing a high-wage job.(175) At least initially, these workers need not have inferior human capital traits compared with their permanent counterparts, a prediction that coincides with empirical observations.(176) However, to the extent that underemployment and job churning may lead to path-dependent, productivity-reducing consequences (such as skill atrophy, loss of motivation, and so on), they may eventually lead to an actual productivity differential between workers who were initially indistinguishable.(177) Thus, the first problem is that workers pushed into contingent jobs will on average experience skill erosion. Second, employers may interpret a worker's contingent status as a signal that she is a lower quality worker and is willing to accept an inferior job in terms of pay, benefits, and security. A rational employer may, all things being equal, prefer a worker who has not previously been terminated or moved through multiple jobs to one who has, regardless of the underlying explanation.(178) A criticism of the signaling model of efficiency wages is that it depends on the proposition that a firm will be unable to measure workers' efforts or abilities even after they are hired, thus making it impossible to weed out unproductive workers.(179) One answer to this criticism is that in difficult-to-monitor jobs, such as many service occupations, it may be as difficult to detect a worker's quality after hiring as it was before.(180)

Efficiency wage accounts, based on the problem of moral hazard or "shirking," tell a different story. In the shirking model, employers use high wages to elicit effort (rather than information) from workers. One way to elicit effort is through closer monitoring, which will increase the likelihood of catching a worker who shirks. This, however, may be costly for the employer (it may, for example, require hiring more supervisors). Another way to elicit effort is to increase the penalty associated with shirking.(181) Under this method, workers know that if they are caught shirking they may be dismissed. Accordingly, the employer increases wages above the competitive level, thereby raising the cost to workers of being dismissed. In the face of this threat, workers increase their efforts, which in turn increases their productivity.(182) So long as the increased wage bill does not exceed the sum of monitoring costs avoided and increased revenues from enhanced productivity, this will be an efficient strategy. Moreover, even if other employers follow suit, so that eventually everyone pays wages above workers' marginal products, the resulting unemployment will once again result in job rationing, thereby maintaining the incentives for the worker not to shirk. As with the signaling model, then, the shirking model predicts a persistent pool of unemployed workers whose abilities and motivations are identical to those of workers employed at high wages.(183)

Applying the "shirking" model to the phenomenon of contingent employment, one could envisage a situation where employers voluntarily set the wage for primary jobs above the market-clearing level, and offer other attractive features such as fringe benefits and a reward of loyalty with rising wages and longer-term career prospects.(184) Workers pushed out of primary markets may choose not to work at all, or they may accept jobs with lower compensation, insecurity, reduced hours, and other paradigmatic "secondary" characteristics identified with contingency, while queuing for rationed primary jobs. Thus, the efficiency wage mechanism creates underemployment--an equilibrium surplus of workers who are both capable and desirous of high wage "core" jobs yet are unable to obtain them. The existence of these queued contingent workers may induce effort on the part of "core" employees who fear slipping out of the primary workforce and into the queue. Moreover, an employer may reduce the wage premium required to induce work effort among core workers by bringing contingent workers in-house. Creating a last-hired first-fired "buffer" class will reduce layoff probability for core workers, which accordingly reduces the wages that must be paid to induce them to work hard.(185) An alternative account of the role of in-house contingent workers is that to the extent some of these workers perform the same tasks as permanent employees (like word processing), their presence will increase effort by serving as a close and constant reminder to permanent workers that sub-contracting or temporary employment is always a future possibility.(186)

"Gift exchange" models posit that workers, when paid a higher wage, will respond by working harder because they perceive that the employer is treating them generously.(187) This idea is based on the sociological observation that employees and their employer will develop norms of contact that value the element of generosity and non-commodification implied in a gift-giving relationship. The workers give the employer the gift of work effort beyond the minimum required, and the employer in return gives wages above the market-clearing rate. Workers also maintain norms of fairness and equal treatment among themselves. Individuals may derive utility from pulling more than their "share" of the load, and not having to see less productive coworkers suffer harsher discipline or lower pay.(188) Why doesn't a firm just fire the least productive workers in the group and thus establish a new minimum? The workers left behind might view the layoff of their peers as a repudiation of the gift exchange relationship and reduce their own productivity below the new standard.

Using a gift exchange account to explain contingent employment, one might hypothesize that employers "externalize" selected jobs within the firm (e.g., word processors) in order to shift the inferior lower-paying positions, and risk of layoffs, to contingent workers. By doing so, firms may accomplish two things. First, buffering the core workers from job insecurity may be seen as an act of generosity deserving reward in the form of harder work. Second, giving the lower paying jobs to outsiders may eliminate the potential morale problems of adopting a two-tier wage structure within the firm.(189) These two mechanisms for enhancing morale (or stemming loss of morale) may increase the effort of core workers who will be correspondingly buffered from risk.(190)

The shirking and gift exchange accounts raise several puzzling theoretical questions. Why, for example, don't workers forced into the queue simply bid down the high wages? One response is that while queued workers may offer to work for a lower wage, an employer will have no incentive to accept lower bids; doing so lowers the implicit "penalty" to current employees for getting caught shirking, which then makes it more difficult to induce optimal effort by employees.(191) Another response is that workers will adopt a norm against underbidding. As players in an infinitely repeated game, workers will refuse to "defect" (underbid) in order to avoid the future prospect of a new equilibrium wage at the (low) reservation price.(192) A second, related, puzzle is how firms that choose the high wage strategy survive. Won't competitor firms that pay low wages enter the market and prevail, given how heavy a toll even a marginal wage increase can exact on a firm's profits? Again, however, to the extent that wage premia contribute directly to productivity by increasing workers' effort in the absence of costly increased monitoring, the high wage firm should be able to compete against other firms that adopt a different compensation-monitoring strategy.(193) A third criticism is that efficiency wages theorists fail to explain why employers would use efficiency wages in order to prevent shirking rather than pursuing the costless strategy of requiring workers to post bonds, i.e., pay the equivalent of "entrance fees" that are subject to forfeiture upon being caught shirking.(194) The absence of this practice, the critics argue, suggests a flaw in the logic of the efficiency wages account. Several responses may be offered to this criticism. In addition to the possibility that forfeiture of bonds may be an unenforceable form of liquidated damages, unemployed or underemployed workers may lack the wealth to "purchase" good jobs. Alternatively, it is possible that those who can afford the fees decline to pay them for fear that the employer will renege on its promise and terminate them in bad faith, leading to forfeiture of the bond.(195)

Insider-Outsider Model. Like efficiency wage models, the "insider-outsider" (or "influence costs") account was also devised to explain the persistence of involuntary unemployment.(196) The insider-outsider account is generally seen to complement--rather than compete with--efficiency wage models. Insiders are experienced employees already established within a firm. They may be entrenched in the firm in a variety of ways. For example, they may have acquired job security in the form of explicit or implicit contractual promises, or regulations restricting termination. They also typically are in positions to exert influence within the firm through personal friendships with other employees and their firm-specific expertise. In contrast, outsiders exist in a competitive "external" labor market. They may be employed or unemployed, but if employed, they have little or no expectation of long-term attachment.

According to this theory, involuntary outsider status persists because it is too costly for employers to hire outside workers to replace insiders, even when the wage outsiders would be willing to accept (their "reservation wage") is lower than the reservation wage of insiders. This is because inside workers behave strategically to make turnover costly for the employer, or in other words, engage in "rent-seeking" behavior. They harass and refuse to cooperate with newcomers to the firm by, for example, refusing to share firm-specific knowledge.(197) This reduces the newcomers' marginal productivity to a level below what it would be in a "cooperative" environment, thus reducing outsiders' chances of being accepted as permanent employees. It also discourages outsiders from seeking insider status, that is, competing for jobs within the firm by offering to work at wages below the insider wage. In this way, insiders are "buffered" from market competition.(198)

An important theme in this account, as with efficiency wage models, is that the determination of whether a worker is an insider or an outsider is not based purely on the natural endowments or preferences of the workers. The costs of turnover result from legislation, technology, and the social relations between workers which permit rent-seeking through non-cooperation and harassment. These factors constrain the opportunities of outsiders and thus deny them jobs, work hours, security, or wages equivalent to the value of their marginal product.

Many workers described as "contingent" in popular debates could fit the description of outsiders.(199) The contingent workers whom reformers seek to aid may be the very workers insider-outsider models predict will be harassed by insider incumbents if they attempt to underbid insider wages. It is worth emphasizing again that, if this account is correct, it would predict that rational market behavior on the part of workers and employers will lead to long-run harmful effects on those workers pegged as the "outsiders within."

Having described two New Keynesian theories and their potential application to the problem of contingent employment, however, we are left with two critical questions. First, is there empirical support for them? Second, if so, what are the policy implications?

3. Assessment and policy implications.

Evidence. There is a large empirical literature spanning twenty years examining the comparative viability of orthodox versus non-market clearing models of labor markets. Evidence of interindustry and interemployer wage differentials that cannot fully be explained by human capital factors, different wage determination mechanisms in different sectors of the labor market, and queues of workers who are qualified and motivated to work in primary jobs offer support for theories of equilibrium underemployment.(200) More recently, however, researchers have turned their attention to more particularized tests of distinct non-market clearing models.

An obvious question for both the shirking and gift exchange hypotheses is whether efficiency wages do indeed increase effort among workers. Analysts have struggled with this question. Merely showing a positive correlation between wages and productivity is inadequate, for that is equally consistent with the predictions of orthodox competitive accounts of labor markets. Sensitive to this problem, several researchers have reported a positive influence of wages on productivity even after taking into account the contribution of human capital.(201) Several studies have concluded that the increase in productivity actually offsets the higher wage bill, thus offering evidence of the net efficiency of a high wage strategy.(202) Residual concerns about the adequacy of such measures have spurred a further line of studies examining other behaviors hypothesized to occur when employers pay efficiency wages. Thus, for example, both gift exchange and shirking models assume that employers who use efficiency wages do so to motivate workers whose performance is difficult to monitor. As such, one would expect employers who pay higher wages to spend less time on monitoring-related activities. Investigations of this relationship have led to mixed findings, although studies failing to find a trade-off have been criticized for omitted variable bias and the use of noisy measures of supervisory intensity.(203)

Although these studies add to the weight of evidence in support of efficiency wage accounts, one problem is that they draw few explicit links to debates about contingent employment. There are, however, a few researchers carving out a research agenda in this area. In one study, researchers linked the efficiency wage hypothesis to their observation that high-wage firms are more likely than low-wage firms to contract out low-skill, competitively compensated jobs (like janitorial jobs) to outside contractors.(204) They speculate that firms pay high wages only to core staff for whom monitoring is difficult; where the marginal return to paying efficiency wages is lower--e.g., in the case of janitors--firms will use outside contractors.(205) The use of outside contractors, in these researchers' estimations, might also reduce morale problems arising from a two-tier compensation scheme among permanent workers. Consistent with this hypothesis, one study reports that when three electronics plants hired temporary workers, morale problems arose due to core workers' complaints that the temporary workers were mistreated.(206)

Similarly, a few researchers have tried to measure insider-outsider harassment of temporary workers by union incumbents.(207) In one study of manufacturing firms in Spain, where labor markets are heavily regulated, researchers observed harassment of temporary workers by union incumbents, noting that this is consistent with "insider-outsider" theories.(208) Nonetheless, the authors did not observe the effect in other European countries they examined. In particular, the effects were not present in England, which like the United States, has a highly deregulated economy with fewer state protections of workers.

Though providing interesting insights, these studies are incomplete. The speculation, for example, that a firm will hire outside contractors in order to avoid the potential morale problems commensurate with dual wage structures, though consistent with a gift exchange hypothesis, has not been subject to rigorous comparison with alternative New Keynesian hypotheses. This leads me to puzzle over at least three questions that might be tested through more systematic empirical enquiry.

First, analysts must examine whether and which categories of contingent workers are underemployed, in the sense of having human capital and motivation equivalent to other workers employed in non-contingent jobs. A second step is to test different New Keynesian wage-setting theories within firms and industries that make use of contingent workers. Augmenting conventional econometric methods with ethnographic and psychological methods will be especially helpful. If workers respond to wages above market-clearing levels, which cognitive phenomena (fear of dismissal or a norm of reciprocity with the employer) induce their additional effort? How does the presence or absence of contingent colleagues affect their motives? A third key area for further empirical inquiry is whether and to what degree hysteresis affects underemployed contingent workers. Research to this point has focused primarily on unemployed workers, leaving a gap in debates about the consequences contingent employment.(209)

Policy. Assuming for the remainder of this section that non-market clearing models offer a persuasive explanation of contingent employment, what does this mean for labor policy? Collectively, the New Keynesian models suggest that contingent jobs are a symptom of deeper efficiency and distributional problems. The efficiency wage, shirking, and insider-outsider models all predict the existence of persistent underemployment, a phenomenon in which identically talented and motivated workers may have different quality jobs. These insights may justify law reform.

Pursuing such reform, however, poses numerous challenges. The first challenge is that the predicted effect of a given intervention depends on which New Keynesian account one adopts. Confounding this problem is the likelihood that each of these wage-setting stories is true in some settings, some of the time. Second, even assuming one could accurately predict the effects of a given intervention, one must also assess whether the intervention is justified in light of the trade-offs that may result.

Analyzing proposals to liberalize the availability of unemployment insurance serves as an example of the first type of challenge.(210) Assume first that the "signaling" account of efficiency wages explains underemployment. Losing one's job in the "core" would mean facing either unemployment, or employment in an inferior position and likely turnover through a succession of jobs. The level and availability of unemployment insurance would affect conditions in both sectors. Wider availability of unemployment benefits to workers with a history of "churning" ought to reduce their efforts at job search. With fewer lower-productivity workers in the pool of applicants, employers may not need to set wages as high in order to attract relatively high-productivity applicants.(211) The wage savings to the employer would mean it could create more core jobs, thus reducing unemployment and/or underemployment.

By contrast, consider the predicted effect of the same intervention under the shirking model. Here, expanded unemployment insurance would again make external jobs more desirable, but this time, according to the assumptions of the model, the effect would be to reduce the cost to incumbents of being dismissed. Employers would need to increase wages further still in order to motivate their incumbent employees. With higher expenditures on effort-inducing wages (as well as on unemployment insurance taxes), the employer would be forced to lay off some workers, and unemployment/underemployment would rise.(212)

Finally, in gift exchange models, an increase in the quality of external conditions would raise the price required to make the inside wage appear like a gift, which would also result in increased unemployment.

The problem of assessing the effect of interventions is further complicated by the presence of dynamic effects. Increasing the quality or duration of unemployment benefits (as opposed to liberalizing the threshold for eligibility) may increase the attractiveness to core workers of queuing in the pool of unemployed workers rather than taking an interim contingent job. Yet it might also induce unemployed contingent workers to join the queue for primary jobs, rather than seeking re-employment in the contingent sector. These effects might, in turn, both reduce the ability of firms to motivate workers and increase the rate of unemployment.(213) It is not clear whether the incentive effects of increased unemployment or the disincentive effects of more attractive unemployment terms would dominate. Granted, these cautionary lessons about dynamic labor market forces are not unique to an assessment of efficiency wage models, but they nonetheless illustrate some of the difficulties faced by empirical researchers.

Even assuming one can identify the applicable New Keynesian model within a particular setting, a second policy challenge is to weigh the potential trade-offs associated with a given intervention. Many of the New Keynesian accounts ultimately turn on the existence of information asymmetries. Because it is generally impossible to "undo" such information problems, the best solution is typically "second best." This means that a prudent legal rule may prospectively trade off certain types of inefficiencies in order to create optimal incentives. For instance, the shirking model creates one type of social cost (persistent unemployment), but also gives workers in the primary sector an incentive to exert effort and a reward for doing so--neither of which would have existed otherwise. One must be aware that some legal interventions will tend to reduce these incentives and rewards.

To illustrate the second challenge, consider reform proposals popularized by the new segmentationists to enhance the collective bargaining power of contingent workers. Under a shirking account of efficiency wages, unionization of contingent workers would make the prospect of falling from a secure job into contingency less frightening, forcing employers to pay higher wage premia to motivate their difficult-to-monitor core workers. The result might be a smaller, more elite class of permanent workers, and more unemployment and underemployment. A reduction in core employment might be quite defensible, however, if the increase in quality of contingent jobs due to unionization were sufficient to dampen the hardship of loss of permanent employment for those at the margin, and if barriers to entry for the new union jobs were low enough that displaced workers could indeed obtain them. In terms of insider-outsider models, proposals to extend union membership to casual, temporary and part-time workers would be a step towards increasing outsider power. This might simply heighten conflict. Alternatively, it might stem rent-seeking behavior among insiders and create a more egalitarian distribution of the benefits of union activity. The potential redistributive benefit of more widespread unionization may make this strategy preferable to increasing the concentrated monopoly power of insider union incumbents.

Another potential benefit of the new forms of unionization is that they may help workers obtain training. But a worker who is truly underemployed does not need skills; she needs a better job. Jobs linked to occupational or geographic unions, of course, may be better jobs if unions negotiate better wages, portable benefits, and other material gains for their members. Better wages and benefits, however, may be insufficient: A defining feature of paradigmatic "good" jobs is that they provide on-the-job training and opportunities for advancement through internal labor markets.(214) Thus the incorporation of progressive skill development and job-matching programs into unions is crucial. These programs may allow the union to cultivate career ladders that substitute for single-firm internal labor markets. Assessing the relative merits of competing reforms, too, is a related component of a prudent policy choice. The potential for unions to provide substitute "career ladders" in the absence of traditional internal labor markets may give this type of reform an advantage over proposals to expand the availability of unemployment insurance and other substantive benefits.

As I argued in Part II.B, collective bargaining might also stem the negative feedback effects--skill atrophy and loss of motivation--of contingent status. Such reforms may create economy-wide efficiencies as well as distributional gains. However, accepting the notion that a worker will settle into habits, skills, and aspirations adapted to the job she currently holds, it follows that an underemployed worker will, over time, cease to be underemployed. The worker's human capital and motivations will eventually match her labor market attainment, even though that level of attainment falls short of what it might have been if earlier events had been different. On the other hand, even if eliminating underemployment by reducing the skills-job mismatch and increasing worker mobility cannot reverse the career trajectories of lives already lived, it may affect the paths of future generations.

CONCLUSION

Extending New Keynesian theories to the policy debate about contingent employment is compelling in its hybrid approach. Perhaps predictably, its policy implications are complex. Cross-the-board reform proposals may fail to respond to the heterogeneous group of workers who hold contingent jobs. Understanding the dynamic relationship between regulatory reforms and unemployment and underemployment equilibria is still another challenge. Despite these hurdles, the application of New Keynesian models to the phenomenon of contingent employment is one of the most promising--and underexplored--avenues for experimentation.

(1.) I borrow a turn of phrase from Kenneth L. Karst, The Coming Crisis of Work in Constitutional Perspective, 82 CORNEL L. REV. 523 (1997).

(2.) Compare William Bridges, The End of the Job, FORTUNE, Sept. 19, 1994, at 62, Lance Morrow, The Temping of America, TIME, Mar. 29, 1993, at 40, and Janice Castro, Disposable Workers, TIME, Mar. 29, 1993, at 42, with Linda McDonnell, Temporary Agency Fills Need for the Disabled, PHIL. INQUIRER, Dec. 12, 1992, at D7, Ellen Paris, A Business Even the Patients Like, FORBES, May 14, 1990, at 104, and Robert Rose, A Foot in the Door, WALL ST. J., Feb. 27, 1995. Examples of recent journal symposia addressing the subject of contingent employment are in 52 WASH. & LEE L. REV. (1995) and 17 J. LAB. RES. (1996). Government interest is apparent from the 1993 appointment of a commission on the future of labor issues, including contingent employment. See COMM'N ON THE FUTURE OF WORKER-MANAGEMENT RELATIONS, U.S. DEP'TS OF LAB. AND COMM., FACT FINDING REPORT 93-103 (1994) [hereinafter DUNLOP COMMISSION REPORT]. The Senate and the House have also held Committee Hearings on the topic. See Conference on the Growing Contingent Work Force: Flexibility at the Price of Fairness?: Hearings on S. 472 Before the Subcomm. on Labor of the Senate Comm. on Labor and Human Resources, 103d Cong. (1994); Toward a Disposable Workforce: The Increasing Use of "Contingent" Labor: Hearings on S. 620 Before the Subcomm. on Labor of the Senate Comm. on Labor and Human Resources, 103d Cong. (1993); Rising Use of Part-Time and Temporary Workers: Who Benefits and Who Loses?: Hearing Before a Subcomm. of the House Comm. on Gov't Operations, 100th Cong. (1988).

(3.) See Mark Berger, Unjust Dismissal and the Contingent Worker: Restructuring Doctrine for the Restructured Worker, 16 YALE L. & POL'Y REV. 1, 7-8 (1997) (detailing ways in which the judiciary has treated contingent workers as having weaker entitlements than regular workers to protection under unjust dismissal laws). For a treatment linking growing income inequality to the rise of contingent employment, see BENNETT HARRISON & BARRY BLUESTONE, THE GREAT U-TURN: CORPORATE RESTRUCTURING AND THE POLARIZING OF AMERICA 74-75 (1988).

(4.) The test for determining who is a covered "employee" varies from statute to statute, but most draw on the common law for assistance, often emphasizing who exercises "control" over the worker. Although workers employed by, for example, a temporary agency or leasing service are usually considered employees of the agency or service, and accordingly entitled to benefits afforded "employees," it may be difficult to show that a temporary worker who moves among several agencies (a fairly common practice) is under the control of anyone. For articles reviewing these various statutory tests and discussing how they may tend to exclude contingent employees, see Jennifer Middleton, Contingent Workers in a Changing Economy: Endure, Adapt, or Organize? 22 N.Y.U. REV. OF L. & SOC. CHANGE 557, 576-83 (1996); Mary O'Connell, Contingent Lives: The Economic Insecurity of Contingent Workers, 52 WASH. & LEE L. REV. 889 (1995); Development in the Law: Employment Discrimination, 109 HARV. L. REV. 1568, 1655-62 (1996) [hereinafter Employment Discrimination].

A related problem is that nontraditional employees may be excluded when determining whether a firm employs sufficient workers to reach the statutory trigger point for status as an "employer," and thus firms making heavy use of these kinds of workers may tend to escape liability. For example, the Worker Adjustment and Retraining Notification (WARN) Act defines "employer" as a business employing 100 or more employees, excluding part-timers, or 100 or more employees who in the aggregate work at least 4000 hours per week exclusive of overtime. 29 U.S.C. [sections] 2101(a). For a more extended discussion of this issue, see Catherine Connolly, The Failure of WARN to Warn: An Analysis of Work-Status Exceptions in the Worker Adjustment and Retraining Notification Act (WARN), 29 LAND & WATER L. REV. 557 (1994).

(5.) Detailed analyses of the ways in which these and other eligibility requirements may serve to disqualify contingent workers from unemployment insurance can be found in YOUNG-HEE YOON, ROBERTA SPALTER-ROTH & MARC BALDWIN, UNEMPLOYMENT INSURANCE: BARRIERS TO ACCESS FOR WOMEN AND PART-TIME WORKERS 33-40 (National Comm'n for Employment Policy Research Report No. 95-06, 1995) (dealing specifically with part-time workers); Martin Malin, Unemployment Compensation in a Time of Increasing Work-Family Conflicts, 29 U. MICH. J.L. REFORM 131, 136-52 (1996); Stephen Bingham, Replace Welfare for Contingent Workers with Unemployment Compensation, 22 FORDHAM URB. L.J. 937, 944-47 (1995); Deborah Maranville, Changing Economy, Changing Lives: Unemployment Insurance and the Contingent Workforce, 4 B.U. PUB. INT. L.J. 291 (1995).

(6.) Commentators elsewhere have discussed adverse selection problems and other market failures that may interfere with individual workers' ability to contract for benefits. See, e.g., David Charny, The Employee Welfare State in Transition, 74 TEX. L. REV. 1601, 1608 (1996). Others have discussed these problems in the context of job security. See, e.g., Pauline T. Kim, Bargaining with Imperfect Information: A Study of Worker Perceptions of Legal Protection in an At-Will World, 83 CORNELL L. REV. 105 (1997) (finding empirical evidence that employees overestimate the degree of job security); Walter Kamiat, Labor and Lemons: Efficient Norms in the Internal Labor Market and the Possible Failures of Individual Contracting, 144 U. PA. L. REV. 1953 (1996); Mark Kelman, Could Lawyers Stop Recessions? Speculations on Law and Macroeconomics, 45 STAN. L. REV. 1215, 1252 n. 131 (1993) [hereinafter Kelman, Law and Macroeconomics]; David I. Levine, Just-Cause Employment Policies in the Presence of Worker Adverse Selection, 9 J. LAB. ECON. 294 (1991). But see J. Hoult Verkerke, Employment Contract Law, in THE NEW PALGRAVE DICTIONARY OF ECONOMICS AND LAW (1998) (questioning the empirical relevance of the signaling hypothesis); J. Hoult Verkerke, An Empirical Perspective on Indefinite Term Employment Contracts: Resolving the Just Cause Debate, 1995 WIS. L. REV. 837, 902-05 (arguing based on survey data that the just cause default reflects employer and employee preferences, and finding no empirical evidence of the signaling hypothesis).

(7.) RICHARD S. BELOUS, THE CONTINGENT ECONOMY: THE GROWTH OF THE TEMPORARY, PART-TIME AND SUBCONTRACTED WORKFORCE (1989).

(8.) Id. at 3. Belous's description of contingent workers, whom he contrasts with "core" workers who have the opposite characteristics, is as follows:
   [C]ontingent workers have a weak affiliation with a specific employer and
   do not have a significant stake in a company.... They do not show long-term
   attachment to a company, and they often do not have a real measure of job
   stability. Employers generally do not make implicit contracts with
   contingent workers.


Id. at 6.

(9.) See id. at 16-17. Estimates varied depending on whether a conservative or liberal estimate was used. The conservative estimate adjusted for possible double or triple counting due to overlaps between the categories. Belous also determined that between 1980 and 1988, using the conservative estimate, the contingent workforce grew at a rate 40% faster than employment generally. Using the liberal estimate, the contingent workforce grew 100% faster, i.e., twice as fast. Id. at 17. Belous later updated his figures, showing slower relative growth of contingent employment after 1988. See Richard Belous, The Rise of the Contingent Workforce: The Key Challenges and Opportunities, 52 WASH. & LEE L. REV. 863, 868 (1995).

(10.) Part-time workers (those who work fewer than 35 hours per week) are a large segment of the workforce as a whole: From 1980 to 1993, approximately 18 to 20% of total employment was part-time. See BUREAU OF LABOR STATISTICS, EMPLOYMENT & EARNINGS (1980-1993).

(11.) See Chris Tilly, Dualism in Part-Time Employment, 31 INDUS. REL. 330, 336 (1992) (distinguishing between "secondary" and "retention" part-time jobs, with the latter being relatively secure because employers create such jobs in order to retain certain workers who seek flexibility); see also Stewart J. Schwab, The Diversity of Contingent Workers and the Need for Nuanced Policy, 52 WASH. & LEE L. REV. 915, 918 (1995) [hereinafter Schwab, Nuanced Policy] (making a similar observation). That said, part-time workers are still more likely than workers as a whole to possess Belous's indicators of contingency. For example, the job tenure of part-time workers is shorter than among workers generally. See, e.g., Rebecca Blank, Understanding Part-Time Work, in RESEARCH IN LABOR ECONOMICS 137, 142 (Laurie Bassi & David Crawford eds., 1990) (finding that "the propensity to leave part-time work is high"); see also JAMES REBITZER & LOWELL TAYLOR, WORK INCENTIVES AND THE DEMAND FOR PRIMARY AND CONTINGENT LABOR 12-13 (National Bureau of Econ. Research Working Paper No. 3647, 1991) [hereinafter, REBITZER & TAYLOR, WORK INCENTIVES] (demonstrating that part-time workers have a shorter median job tenure and weaker workforce attachment than full-time workers).

(12.) See BUREAU OF LABOR STATISTICS, EMPLOYMENT & EARNINGS 190 (1996). The Department of Labor's Current Population Survey counts as "involuntary" part-time any individual who works part-time for what the Survey labels "economic" reasons, i.e., because of slack work or business conditions, because the work is seasonal, because the job started or ended during the survey week, or because she could only find part-time work. In 1996, approximately 19% of part-time workers, or 3.5% of Americans at work, were working part-time involuntarily according to this definition. I will discuss critiques of this definition of involuntariness in greater detail below.

(13.) See generally, JEREMY RIFKIN, THE END OF WORK: THE DECLINE OF THE GLOBAL LABOR FORCE AND THE DAWN OF THE POST-MARKET ERA (1995).

(14.) Id. at 191.

(15.) Id. at 194.

(16.) The relevant questions of the special survey were as follows:
   (1) Some people are in temporary jobs that last only for a limited time or
   until the completion or a project. Is your job temporary?

   (2) Provided the economy does not change and your job performance is
   adequate, can you continue to work for your current employer as long as you
   wish?


Respondents who answered yes to the first question and no to the second, were asked further questions designed to weed out individuals who, for personal reasons, were holding a "temporary" job that offered the option of permanent employment. Thus, a worker would be excluded if she indicated that, but for personal reasons such as voluntarily retiring or returning to school full-time within the year, she could continue working in that job. BUREAU OF LABOR STATISTICS, CONTINGENT AND ALTERNATIVE EMPLOYMENT ARRANGEMENTS 4 (1995) [hereinafter SURVEY 95].

(17.) See BUREAU OF LABOR STATISTICS, CONTINGENT AND ALTERNATIVE EMPLOYMENT ARRANGEMENTS, Tbl. 1 (1997) [hereinafter SURVEY 97]. This amounts to somewhere between 2.4 and 5.6 million American workers. The special survey permitted the BLS to devise three separate estimates of the prevalence of contingent work. In all subsequent references in this article to the BLS/CPS estimate, I will adopt the most liberal estimate, Estimate 3 (4.4%), which defines a contingent worker as any wage and salary worker who does not expect her job to last (except if she plans to leave for personal reasons), or any worker who has been self-employed or an independent contractor for one year or less and does not expect her current employment to last for more than one year (again, except if she plans to leave for personal reasons). The estimate includes employees of temporary help agencies who have been and expect to be employed for a given client for one year or less, even if they remain working for the same temporary help service for more than one year.

(18.) See U.S. DEP'T OF LABOR, REPORT ON THE AMERICAN WORKFORCE 25 (1995) [hereinafter RAW]. Only 10% of part-time workers are contingent using the BLS definition. See id. at 26.

(19.) See id. at 37. The 10% contingency rate in part-timers is more than double the estimated rate--4.4%--for all workers under the BLS definition. See id. at 25-26. Also, the biggest subgroup within BLS's definition of contingency is part-time workers: Forty-three percent of BLS contingent workers are part-time, depending on the technique used to count contingent workers. See id. at 28. A dramatic 66.5% of temporary help workers are contingent, and sizeable percentages of on-call workers/day laborers (21.5) and workers in contract firms (19.8) are also contingent. See id. at 43.

(20.) See id. at 43-44 (noting that only 3.8% of independent contractors are contingent). The survey did not, unfortunately, discern whom among the independent contractors were former employees whose status had been "converted" by the employer to "independent contractor" in order to avoid legal requirements; depending on their interpretation of the words "employed" and "self-employed," two employees in the same job might give different answers to the question of whether they worked for an employer or were self-employed. See SURVEY 95, supra note 16, at 5.

(21.) Although the Current Population Survey shows a decline in the prevalence of contingent employment in recent years (from between 2.2 and 4.9% in 1995, to between 1.9 and 4.4% in 1997), these trend data must be viewed with caution because of the short time span over which they were collected. There have been only two supplements to the CPS measuring the prevalence of contingent work, the first in February 1995, and the second in February 1997. See SURVEY 95, supra note 16; SURVEY 97, supra note 17.

(22.) Job security in the United States, however, may well lag behind that in other nations. See, e.g., Katharine G. Abraham & Susan N. Houseman, Job Security and Work Force Adjustment: How Different Are U.S. and Japanese Practices?, in EMPLOYMENT SECURITY AND LABOR MARKET BEHAVIOR: INTERDISCIPLINARY APPROACHES AND INTERNATIONAL EVIDENCE 180 (Christoph Beuchtemann ed., 1993) (noting that the United States provides workers with less job security than most European countries and Japan); William B. Gould IV, Employment Protection and Job Security Regulation in the United States and Japan: A Comparative View, in EMPLOYMENT SECURITY, supra, at 165 (arguing that "the law relating to the protection of workers in connection with collective dismissal remains antiquated by European and Japanese standards").

(23.) See, e.g., Katharine G. Abraham, Restructuring the Employment Relationship: The Growth of Market-Mediated Work Arrangements, in NEW DEVELOPMENTS IN THE LABOR MARKET: TOWARD A NEW INSTITUTIONAL PARADIGM 86-90 (Katharine G. Abraham & Robert McKersie eds., 1990) [hereinafter Abraham, Restructuring]; KATHARINE G. ABRAHAM, FLEXIBLE STAFFING ARRANGEMENTS AND EMPLOYERS' SHORT-TERM ADJUSTMENT STRATEGIES (National Bureau of Econ. Research Working Paper No. 2617, 1988); Angela Clinton, Flexible Labor: Restructuring the American Work Force, 120 MONTHLY LAB. REV., Aug. 1997, at 3; Katharine G. Abraham & Susan K. Taylor, Firms' Use of Outside Contractors: Theory and Evidence, 14 J. LAB. ECON. 394 (1996) [hereinafter Abraham & Taylor, Outside Contractors].

(24.) The number of people employed in the personnel supply industry (which includes those employed by temporary help services, employment agencies, and employee leasing services) tripled between 1980 and 1990, and grew by an additional 60% between 1990 and 1995. See BUREAU OF THE CENSUS, U.S. DEP'T OF COMM., STATISTICAL ABSTRACT OF THE UNITED STATES 421 (1996) [hereinafter STATISTICAL ABSTRACT] (tabulation by author). For a more complete discussion of the various ways in which one might attempt to measure the growth of temporary employment, see Lewis M. Segal & Daniel G. Sullivan, The Growth of Temporary Services Work, 11 J. ECON. PERSP. 117, 117-18 (Spring 1997) [hereinafter Segal & Sullivan, Growth of Temporary Services] and Karylee Laird & Nicolas Williams, Employment Growth in the Temporary Help Supply Industry, 17 J. LAB. RES. 663, 673-74 (1996).

(25.) See note 20 supra and accompanying text.

(26.) Temporary help service workers are only about 2% of the workforce. See STATISTICAL ABSTRACT, supra note 24, at 421.

(27.) See Anne E. Polivka, Are Temporary Help Agency Workers Substitutes for Direct Hire Temps? Searching for an Alternative Explanation of Growth in the Temporary Help Industry 2-3 (Dec. 1996) (unpublished manuscript on file with the Stanford Law Review).

(28.) For helpful reviews of empirical evaluations of the question, see RAW, supra note 18, at 11-23 (1995); Stefanie R. Schmidt & Shirley V. Svorny, Recent Trends in Job Security and Stability, 19 J. LABOR RES. 647 (1998); Thomas Nardone, Jonathan Veum & Julie Yates, Measuring Job Security, 120 MONTHLY LAB. REV., June 1997, at 26; Dave E. Marcotte, Declining Job Stability: What We Know and What it Means, 14 J. POL'Y ANALYSIS & MGMT. 590 (1995).

(29.) See Marcotte, supra note 28, at 593. See, e.g., BUREAU OF LABOR STATISTICS, EMPLOYEE TENURE IN THE MID-1990S tbl. 1 (1997) (using median job tenure as a proxy for stability); STEPHEN ROSE, DECLINING JOB SECURITY AND THE PROFESSIONALIZATION OF OPPORTUNITY (National Commission for Employment Policy Research Report No. 95-04, 1995) (using turnover as a proxy for stability); HENRY S. FARBER, ARE LIFETIME JOBS DISAPPEARING? JOB DURATION IN THE UNITED STATES: 1973-1993 (National Bureau of Econ. Research Working Paper No. 5014, 1995) (using median job tenure as a proxy for stability); see also Kenneth Swinnerton & Howard Wial, Is Job Stability Declining in the U.S. Economy?, 48 INDUS. & LAB. REL. REV. 293 (1995); FRANCIS X. DIEBOLD, DAVID NEUMARK & DANIEL POLSKY, JOB STABILITY IN THE UNITED STATES (National Bureau of Econ. Research Working Paper No. 4859, 1994); Francis X. Diebold, David Neumark & Daniel Polsky, Is Job Stability Declining In The U.S. Economy?, 49 INDUS. & LAB. REL. REV. 348 (1996) (using historical retention rates as a proxy for stability).

(30.) SURVEY 95, supra note 16, at 4.

(31.) The term "implicit contract" is not defined by the BLS; the BLS may, however, mean to refer to a broader theoretical concept postulated by labor economists. According to implicit contracting theory, an employee and employer make bilateral investments in one another at the outset of the employment relation: both invest in the employee's development of firm- and industry-specific capital. The employee receives a promise that her wages will rise with job tenure, akin to a set of installment payments as consideration for her investments in firm-specific human capital. Employers, in turn, will resist lowering the wage of longer-term (higher skilled) employees for fear that they will quit and expensive investments will be required to hire and screen new workers. Unless the employer decides to invest in the long-term capital of a worker within the firm, the worker is unlikely to be assigned to the sorts of tasks that would lead to firm-specific investments and corollary buffering from layoff due to short-term demand shocks in the future. I discuss implicit contracting in more detail at note 103 infra.

(32.) It is simply not clear how existing popular perceptions of the trend towards "downsizing" in the United States cut. Workers may well be unduly nervous about their employment, particularly if there is an increasing tendency on the part of employers to expressly disclaim for-cause arrangements when workers commence employment. Conversely, workers may tend to assume, falsely, that their own employment is secure even though others around them will lose their jobs. A recent noteworthy study, however, may lend support to the reliability of expectations-based data. See JEFF DOMINITZ & CHARLES F. MANSKI, PERCEPTIONS OF ECONOMIC INSECURITY: EVIDENCE FROM THE SURVEY OF ECONOMIC EXPECTATIONS 26-30 (University of Wisconsin-Madison Institute for Research on Poverty Discussion Paper No. 1105-96, 1996) (finding a high degree of correspondence between survey respondents' expectation of probability of job loss in the subsequent year, and actual percentage of respondents who, when polled one year later, reported a job loss some time in the previous 12 months). Although there are differences between the Survey of Economic Expectations and the CPS Supplement on contingent employment, as well as some limitations of the SEE comparison, see id. at 27 n.5, the data are certainly suggestive.

(33.) See Daniel C. Feldman, The Nature, Antecedents and Consequences of Underemployment, 22 J. MGMT. 385 (1996); Gerald Glyde, Underemployment: Definition and Causes, 11 J. ECON. ISSUES 245 (1977).

(34.) Others, on whose treatments I have drawn, operationalize it somewhat differently. For example, Feldman, supra note 33, at 388, sets out a five-part totality-of-the-circumstances test which includes situations where a worker earns 20% less than in her previous job, as well as situations in which there is a job-human capital mismatch. Feldman also would include all part-time and temporary workers, who, based on self-reported preferences, say they would rather not be working part-time or temporarily. See id. I think the latter

criterion of involuntariness is overinclusive. Similarly, Glyde, supra note 33, at 246-47, does not include the preference for superior employment among his criteria.

(35.) Human capital is essentially the investment in human resources--schooling, on-the-job experience, skills training--expected to improve productivity. For the classic definition of human capital, see GARY S. BECKER, HUMAN CAPITAL: A THEORETICAL AND EMPIRICAL ANALYSIS, WITH SPECIAL REFERENCE TO EDUCATION 98-101 (1st ed. 1964). Note that human capital is correlated with, but not the same as, the worker's natural endowments and talents.

(36.) It may turn out that a person in fact has identical talents, education, experience, and desires to work as those who were hired, but he cannot get hired himself because the features of his credential (or his own characteristics) falsely signal to employers that he is less qualified. This person would be underemployed by my definition. The phenomenon whereby false signaling may lead to racially discriminatory hiring practices at elite law firms is discussed in David B. Wilkins & G. Mitu Gulati, Why Are There So Few Black Lawyers in Corporate Law Firms? An Institutional Analysis, 84 CAL. L. REV. 493, 520-23, 554-64 (1996), and David Charny & G. Mitu Gulati, Efficiency Wages, Tournaments, and Discrimination: A Theory of Employment Discrimination for "High-Level" Jobs, 33 HARV. C.R.-C.L.L. REV. 57 (1998). I will return to the problem of signaling imperfections in Part II.C. infra.

(37.) For classic treatments of this argument, see generally Herbert Gintis, Consumer Behavior and the Concept of Sovereignty: Explanations of Social Decay, 62 AM. ECON. REV. 267 (1972) and Mark Kelman, Choice and Utility, 1979 WIS. L. REV. 769. See also JON ELSTER, SOUR GRAPES 109-40 (1983) (discussing adaptive preferences); Cass R. Sunstein, Preferences and Politics, 20 PHIL. & PUB. AFF. 3, 19-24 (1991) (noting that preferences adapt to undue limitations in available opportunities or to unjust background conditions); Jon Elster, Self-Realization in Work and Politics: The Marxist Conception of the Good Life, 3 SOC. PHIL. & POL'Y 97, 123 (1986) (arguing that the absence of self-realization may be caused by an absence of opportunities).

(38.) As I will discuss in Part II.C infra, these path-dependent externalities may be triggered not only when there is inequality of treatment between comparably skilled and motivated workers because some workers receive less than they are "worth" (i.e., less than their marginal product), but also when there is inequality because some workers receive more than the value of their marginal product.

(39.) The process by which markets clear is known as "tatonnement" (price adjustment). Any imbalance between quantifies supplied and demanded will result in a change in price that counteracts the imbalance. Should the supply of laborers at the prevailing wage increase for some reason, wages will fall. This will cause some workers to leave the labor force voluntarily, but other workers will enter into the labor force because of the increased ability of employers to hire new workers at the lower wage, thus returning the market to equilibrium. Unless government regulation dampens this process of "tatonnement," non-working individuals must be so voluntarily: they either prefer some non-employment activity (e.g., leisure, acquiring human capital, working outside of the labor market) to work at the prevailing wage, or they have elected to take advantage of the relatively low opportunity cost of engaging in job search or investment in human capital.

(40.) I have included human capital as a "supply-side" phenomenon, although one could certainly postulate that much of the value of a worker's human capital comes from the intensity of the demand for those skills on the other side of the market.

(41.) It is generally presumed that training, education, and experience tend to increase the marginal productivity of workers, thereby raising their equilibrium wages.

(42.) Note that it is not simply the wage that embodies these features; benefits and job stability are also part of the function.

(43.) See, e.g., Laird & Williams, supra note 24, at 667, 667-78 (providing a supply-side explanation of increasing temporary employment). See also SURVEY 97, supra note 17, tbl.2 (reporting that 51% of contingent workers in the 1997 survey were women, compared to 46% of noncontingent workers).

Note that in this section, where I discuss empirical research, I focus mainly on the BLS definition of contingent employment, but also on some studies of temporary agency workers. I do this for greater clarity of illustration, rather than trying to juggle parallel empirical literatures on the various kinds of workers, such as part-time workers, included within the Belous definition.

(44.) See, e.g., STANLEY NOLLEN & HELEN AXEL, MANAGING CONTINGENT WORKERS: HOW TO REAP THE BENEFITS AND REDUCE THE RISKS 164 (1996) (reporting that contingent workers are more likely than traditional workers to be single parents).

(45.) Contingent workers are dramatically more likely than noncontingent workers to be young and enrolled in school. In February 1997, contingent workers were more than twice as likely to be between the ages of 16 and 24 (30.3%) than noncontingent workers (13.4%). See SURVEY 97, supra note 17, tbl.2. Indeed, the CPS estimations may underestimate the proportion of workers in this age bracket who are effectively contingent: Even those young workers classified as noncontingent because of their stated expectations seem highly likely to switch jobs in the short-term, if not within the year then soon after that. In addition, contingent workers are about three to four times as likely as noncontingent workers to be enrolled in school. See RAW, supra note 18, at 31 tbls.1-9 (17.7% of contingent workers were enrolled in school during the 1995 survey, versus 5.3% of noncontingent workers).

(46.) According to the 1994 National Association of Temporary and Staffing Services (NATSS) survey, 29% of temporary workers receive more than 20 hours of general training from their temporary help company. See Segal & Sullivan, Growth of Temporary Services, supra note 24, at 131. According to the 1995 NATSS survey, 37% of workers found permanent employment during their stints as temporary employees, though not all found the job as a direct result of the temporary assignment. See Edward Lenz, Flexible Employment: Positive Work Strategies for the 21st Century, 17 J. LAB. RES. 555, 558 (1996) [hereinafter Lenz, Positive Work Strategies].

(47.) See, e.g., Maria O'Brien Hylton, The Case Against Regulating the Market for Contingent Employment, 52 WASH. & LEE L. REV. 849, 856 (1995) [hereinafter Hylton, Against Regulating] (asserting that if contingent workers earn less than others, it is because they are less productive); Dwight Lee, Why Is Flexible Employment Increasing?, 17 J. LAB. RES. 543, 545 (1996) (noting that there are plenty of permanent jobs available for "those who want them and have the required skills and work habits").

(48.) See Hylton, Against Regulating, supra note 47, at 856-57; see also Segal & Sullivan, Growth of Temporary Services, supra note 24, at 131 (adding that, in order to prevent workers from leaving once they are trained, the temp agency tends to spread training between assignments, thus inhibiting flight).

(49.) See SURVEY 97, supra note 17, tbl.2 (reporting that 12% of contingent workers are Hispanic, compared with 9% of noncontingent workers, and about 11% of contingent workers are black, compared with 10.6% of noncontingent workers) If we look at the narrower estimates of contingency--Estimates 1 and 2--blacks are overrepresented at 13%. Other ethnic groups are not isolated in the data. See id.

(50.) See Lenz, Positive Work Strategies, supra note 46, at 560. Lenz also quotes a statement by Richard Belous in a 1993 Washington Post article that temporary help firms "have done more to train inner-city residents than all the government training programs combined." Id. at 558.

(51.) These arguments are made in detail elsewhere. See NEW YORK TIMES, THE DOWNSIZING OF AMERICA 14-20 (1996); NOLLEN & AXEL, supra note 44, at 15-23; Thierry Noyelle, Toward a New Labor Market Segmentation, in SKILLS, WAGES AND PRODUCTIVITY IN THE SERVICE SECTOR 212, 213-14 (Thierry Noyelle ed., 1990) [hereinafter Noyelle, Labor Market Segmentation].

(52.) Note that this usage differs from the orthodox usage. In orthodox terms, in a flexible market, wages adjust quickly following changes in prices or unemployment levels. Another, closely related use of the term flexibility refers to the regulatory climate of a labor market. A less regulated labor market may be said to be more "flexible" on the assumption that the elimination of protective labor legislation will reduce unemployment because it will reduce the "taxes" associated with hiring and retaining additional employees. Although the distinction between the orthodox and newer understandings may be seen as a distinction between macro and firm-level perspectives, it is misleading to generalize this observation too broadly. Firm-level flexibility contributes to economy-wide flexibility because it influences the ease with which labor costs will shift in response to demand fluctuations. Authorities discussing the meaning of "labor market flexibility" include ROBERT BOYER, THE SEARCH FOR LABOUR MARKET FLEXIBILITY (1988); Melvin M. Brodsky, Labor Market Flexibility: A Changing International Perspective, 117 MONTHLY LAB. REV., Nov. 1994, at 53 and Ricardo Lagos, Labor Market Flexibility: What Does It Really Mean?, 54 CEPAL REV. 81 (1994).

(53.) See THOMAS A. KOCHAN, HARRY C. KATZ & ROBERT B. MCKERSIE, THE TRANSFORMATION OF AMERICAN INDUSTRIAL RELATIONS 120, 188 (1986); Mark Barenberg, Democracy and Domination in the Law of Workplace Cooperation: From Bureaucratic to Flexible Production, 94 COLUM. L. REV. 753, 884 (1994); Michael Piore, Perspectives on Labor Market Flexibility, 25 INDUS. REL. 146, 159-60 (1986); Charles Sabel, Moebius-Strip Organizations and Open Labor Markets: Some Consequences of the Reintegration of Conception and Execution in a Volatile Economy, in SOCIAL THEORY FOR A CHANGING SOCIETY 23, 27-28 (Pierre Bourdieu & James Coleman eds., 1991) [hereinafter, Sabel, Moebius-Strip Organizations].

(54.) While the precise criteria for inclusion in this "functionally flexible" model have yet to be fully refined, numerous examples of "flexible production complexes" have been provided in the literature. See, e.g., Michael Storper & Allen J. Scott, Work Organisation and Local Labour Markets in an Era of Flexible Production, 129 INT'L LAB. REV. 573, 584-88 (1990) (citing examples from craft-based manufacturing that include the New York fashion clothing industry, the film industry in Los Angeles, the textile industry in the so-called "Third Italy," the printed circuits industry in southern California, and the non-craft based example of high-technology manufacturing in major centers near Boston, San Francisco, and Los Angeles). On services, see id. at 587-88 (citing business and financial services--commodities futures agents in Chicago and advertising executives in New York--as employing functionally flexible techniques). Note that not all flexible firms fit the functional model--e.g., in the fashion and computer industries, both functional and numerical techniques are employed for achieving flexibility, provoking one commentator to describe some such operations as "low-skill" "fly-by-night" "sweatshops." Susan Christopherson, Emerging Patterns of Work, in SKILLS, WAGES AND PRODUCTIVITY IN THE SERVICE SECTOR 11, 24 (Thierry Noyelle ed., 1990).

(55.) See Alison Davis-Blake & Brian Uzzi, Determinants of Employment Externalization: A Study of Temporary Workers and Independent Contractors, 38 ADMIN. SCI. Q. 195, 217 (1993) (observing that jobs requiring firm-specific skills are unlikely to be filled by temporary workers); Rosabeth Moss Kanter, Nice Work If You Can Get It: The Software Industry as a Model for Tomorrow's Jobs, AM. PROSPECT, Fall 1995, at 52 (describing the rise of short-term, project-based employment in the computer software industry).

(56.) See Noyelle, Labor Market Segmentation, supra note 51, at 216 (observing the trend toward specialization and disintegration of firm operating units); Jeffrey Pfeffer & James Baron, Taking the Workers Back Out: Recent Trends in the Structuring of Employment, 10 RES. ORG. BEHAV. 257, 273 (1988); Stephen Blakely, Executive `Temps' Cut Costs, 85 NATION'S BUS. 61 (1997) (describing the use of temporary managers); Margaret Kirk, The Temps in the Gray Flannel Suits, N.Y. TIMES, Dec. 17, 1995, at F13 (describing the rise of temporary workers among executives and professionals); Pamela Burdman, The New Executive Temp, S.F. CHRON., Aug. 10, 1992, at C1 (same). Note, however, that although firms' (especially small firms') growing reliance during the eighties and nineties on outside contractors for a number of specialized services, such as accounting, was driven by a demand for the comparative advantages and economies of scale offered by such providers, see Abraham & Taylor, Outside Contractors, supra note 23, at 412, 416, this does not necessarily imply that the workers employed by these services are all, or even mainly, contingent workers.

(57.) See SURVEY 97, supra note 17, tbl.4. Fifty-six percent of contingent workers were in service industries, compared with 36% of non-contingent workers; 10% of contingent workers were in the construction industry, as compared to 6% of noncontingent workers. See id. Not surprisingly, more contingent than non-contingent workers--15% versus 13%--were in service occupations (e.g., waitresses, cooks, bartenders, child care workers, janitors). See id. Contingent workers are also more likely to be in administrative support, including clerical, jobs--20% versus 14%. See id.

(58.) See id. Twenty-one percent of contingent workers fell within the "professional specialty" category, compared to 15% of noncontingent workers. Id. "Professional specialists" include teachers, editors, reporters, photographers, actors and directors, and athletes. On the professionalization and paraprofessionalization of contingent employment, see THIERRY J. NOYELLE, BEYOND INDUSTRIAL DUALISM: MARKET AND JOB SEGMENTATION IN THE NEW ECONOMY 116-17 (1987) [hereinafter NOYELLE, BEYOND INDUSTRIAL DUALISM]. On different kinds of temporary professionals, see generally Vincent R. Johnson & Virginia Coyle, On the Transformation of the Legal Profession: The Advent of Temporary Lawyering, 66 NOTRE DAME L. REV. 359 (1990); Celeste Zike-Harris, Statutory Regulations of Nurse Staffing Agencies: Can Quality Health Care Be Assured Without It?, 34 ST. LOUIS U. L.J. 1129, 1129-30 (1990); Timothy Egan, A Temporary Force to Be Reckoned With: Wanted: High-Tech `Hired Guns' and Agencies to Provide Them, N.Y. TIMES, May 20, 1996, at D1; Kanter, Nice Work If You Can Get It, supra note 55, at 54-55. Note, however, that while contingent workers may be overrepresented within certain professional categories, the same is not true for managerial occupations. Contingent workers are only half as likely to hold executive and managerial jobs as non-contingent workers. See SURVEY 97, supra note 17, tbl.4 (noting that 7% of contingent workers hold executive, administrative, or managerial occupations compared with 15% of non-contingent workers).

(59.) Note that contingent workers in the construction industry or in some professions may earn wages that are actually higher than those for non-contingent workers in the same occupation. See Steven Hipple & Jay Stewart, Earnings and Benefits of Contingent and Noncontingent Workers, 119 MONTHLY LAB. REV., Oct. 1996, at 22, 25 tbl.2 (noting that contingent construction workers have a median weekly income 36 dollars higher than their noncontingent counterparts); Stanley D. Nollen, Negative Aspects of Temporary Employment, 17 J. LAB. RES. 567, 570 (1996) (noting that contingent workers in "professional specialties" have a median hourly income 37% higher than their noncontingent counterparts).

(60.) See Anne E. Polivka, A Profile of Contingent Workers, 119 MONTHLY LAB. REV., Oct. 1996, at 10, 12 [hereinafter Polivka, Profile] (noting that in the 1995 Survey, teachers accounted for more than 10% of all contingent workers). For critical commentary on the prevalence of contingency among teachers, especially women, see EMILY ABEL, TERMINAL DEGREES: THE JOB CRISIS IN HIGHER EDUCATION (1984); Debra Philo Fourlas, Longterm Substitutes: They're Teachers, Too, 95 DICK. L. REV. 577 (1991).

(61.) See Garth Mangum, Donald Mayall & Kristin Nelson, The Temporary Help Industry: A Response to the Dual Internal Labor Market, 38 INDUS. & LAB. REL. REV. 599, 603-05 (1985) (discussing techniques for avoiding worker attachment).

(62.) See Berger, supra note 3, at 35-36 (arguing that employers may hire contingent workers because they are usually barred from pursuing wrongful termination claims). As for unemployment insurance, the employer must pay a payroll tax only for "covered employees," a category which tends to exclude contingent employees at a disproportional rate (for a discussion of why, see note 5 supra and accompanying text). Similarly, the employer's "experience rating" (the cost of premiums based on past record of layoffs) is based solely on layoffs of covered employees. For a contrary argument, see Lars Osberg, Concepts of Unemployment and the Structure of Employment, 158 ECON. APPLIQUEE 157, 165 (1995) (arguing that as more firms switch to a strategy of "casualizing" workers, total unemployment insurance claims will increase, thus driving up the price of unemployment insurance premiums).

(63.) Also at stake may be the costs of retaining lower-productivity workers or declining to hire potentially high-productivity workers due to fear of lawsuits, and decisions against speculating in risky ventures in ease they culminate in the need for layoffs. See generally JAMES DERTOUZOS & LYNN KAROLY, LABOR MARKET RESPONSES TO EMPLOYER LIABILITY 37-38 (1992); Ian Ayres & Peter Siegelman, The Q-Word as Red Herring: Why Disparate Impact Liability Does not Induce Hiring Quotas, 74 TEX. L. REV. 1487, .1488-89 (1996); Douglass Williams & Richard Sander, The Prospects for "Putting America to Work" in the Inner City, 81 GEO. L.J. 2003, 2054-57 (1993); John Donohue & Peter Siegelman, The Changing Nature of Employment Discrimination Litigation, 43 STAN. L. REV. 983, 1024 (1991).

(64.) See Vicki Smith, Institutionalizing Flexibility in a Service Firm: Multiple Contingencies and Hidden Hierarchies, 21 WORK & OCCUP. 284, 298-99 (1994) (case study of firm that reported the practice of hiring temporary workers for purpose of screening them for permanent employment); Segal & Sullivan, Growth of Temporary Services, supra note 24, at 128-29 (discussing European evidence consistent with this hypothesis with respect to temporary workers).

(65.) See Abraham, Restructuring, supra note 23, at 100-05 (concluding based on analysis of mid-1980s Current Population Survey data that wage savings are likely to be a key factor influencing increasing in firm reliance on outside contract service providers for low-skill work, though the same is not true with respect to the use of agency temporaries). However, employers cite cost reduction as their reason for using contingent workers less frequently than some may predict. One study found that less than one-quarter of firms in national survey cited cost control as a reason for using temporary, on-call, or short-term hires; slightly more (26.9%) said they use independent contractors for this reason. By contrast, each of the following were cited by more than half of respondents as reasons for using temporary, on-call, or short-term hires: to carry out special projects or handle a surge in workload, to meet seasonal peaks in demand, and to fill in for temporary employee absences. See U.S. DEP'T OF COMMERCE, CONTINGENT STAFFING ARRANGEMENTS IN SMALL AND LARGE FIRMS, at tbls.3.7, 4.7, 5.7 & 6.7 (1990). Moreover, as an empirical matter, it may not prove to be cost-effective to hire contingent workers. See NOLLEN & AXEL, supra note 44, at 61-68; Abraham, Restructuring, supra note 23, at 104-05 (noting that wage and benefits savings from hiring contingent workers may be outweighed by increased training and other fixed costs). At some level, of course, it always comes down to a balancing of the aggregate costs of operating a business enterprise and the particular trade-offs, which may include occasionally choosing more facially costly but easily terminable short-term employees. In other words, employer's self-perceptions may point to considerations other than cost-cutting, but the decision to use contingent workers would seem always to turn ultimately on aggregate efficiency.

(66.) See Hylton, Against Regulating, supra note 47. at 858; Edward Lenz, "Contingent Work"--Dispelling the Myth, 52 WASH. & LEE L. REV. 755,769 (1995) [hereinafter Lenz, Dispelling the Myth].

(67.) See Hylton, Against Regulating, supra note 47, at 857.

(68.) See Lenz, Positive Work Strategies, supra note 46, at 563-64 (calling for incentives such as tax credits for temporary help agencies).

(69.) See, e.g., Lee, supra note 47, at 545-47 (describing contingent jobs as "good responses to bad policies").

(70.) See Lowell Gallaway, Public Policy and Part-Time Employment, 16 J. LAB. RES. 305, 312 (1995) ("[T]he statistical evidence indicates that an extra 1000 pages in the Federal Register reduces the percent of involuntary part-time employment by 0.03 percentage points, or by approximately 35,000 jobs.").

(71.) Many commentators discuss the individual arguments mentioned only briefly in this paragraph. For general overviews of these critiques, see RICHARD EDWARDS, RIGHTS AT WORK: EMPLOYMENT RELATIONS IN THE POST-UNION ERA 42-76 (1993); PAUL WEILER, GOVERNING THE WORKPLACE: THE FUTURE OF LABOR AND EMPLOYMENT LAW 74-78 (1990).

(72.) See SURVEY 97, supra note 17, tbl. 13 (noting BLS-defined contingent workers earn $417 per month in nominal wages, compared with $510 for noncontingent workers). This is true even controlling for part-time status: Full time contingent workers still earn only about 80% of the earnings of full-time noncontingent workers, and a similar pattern appears if the sample is limited to part-time workers. See Hippie & Stewart, supra note 59, at 22-23. As mentioned earlier in this section, I am focusing mainly on contingent workers as defined by the BLS. If I were to focus instead on, e.g., part-time workers, I would reach similar conclusions. There is ample evidence, for example, that part-time workers receive lower pay and benefits than full-time workers, even controlling for human capital characteristics. See Blank, supra note 11, at 146-47 (reviewing studies that have used multivariate regressions to compare the wages of part-time and full-time workers); Sarah Snider, Characteristics of the Part-Time Workforce and Part-Time Employee Participation in Health and Pension Benefits, 16 J. LAB. RES. 239 (1995).

(73.) See SURVEY 97, supra note 17, tbl.9. Of contingent workers 66.1% had health insurance from any source, including a family member, as compared to 82.2% of noncontingent workers. Contingent workers were also much less likely than noncontingent workers to have an employer provided pension: Only 15% of contingent workers, as compared with 44% noncontingent workers, had employer-provided pension coverage. See id.

While one might assume that wages and fringe benefits would be inversely related, this prediction is not always borne out. At least in the case of part-time work, jobs that pay low wages also tend to have low fringe benefits. See Blank, supra note 11, at 148; see also SURVEY 97, supra note 17, at 3 (noting that median weekly earnings of full-time contingent workers are 82% of the earnings of non-contingent workers).

(74.) See SURVEY 97, supra note 17, at 3.

(75.) See Anne Polivka, Into Contingent and Alternative Employment:By Choice?, 119 MONTHLY LAB. REV., Oct. 1996, at 55, 65 [hereinafter Polivka, By Choice?] (reporting that those giving economic reasons for contingent employment collectively amount to 34.7%).

(76.) See RAW, supra note 18, at 19-21.

(77.) See SURVEY 97, supra note 17, tbl.3 (10.4% of contingent workers lack high school diplomas versus 8.5% of others).

(78.) See RAW, supra note 18, at 20. Churning may occur for still other reasons. See J. Hoult Verkerke, Legal Regulation of Employment Reference Practices, 65 U. CHI. L. REV. 115, 140-47 (1998) (observing that churning may occur where a series of employers, lacking information about an employee's unsuitability, hire and discharge the employee).

(79.) See Yinon Cohen & Yitchak Haberfeld, Temporary Help Service Workers: Employment Characteristics and Wage Determination, 32 INDUS. REL. 272, 286 (1993).

(80.) Others have made similar empirical observations. See, e.g., Lenz, Dispelling the Myth, supra note 66, at 764 (citing an unpublished Industry Training Survey conducted for NATSS showing that temporary workers who work on multiple assignments receive higher wages than on their previous assignment).

(81.) See Segal & Sullivan, Growth of Temporary Services, supra note 24, at 122. The status of workers identified by the survey as employed in personnel supply services in one survey year were examined one year later to determine what proportion had made a transition into permanent employment. These data were then compared with data for all workers.

(82.) Id. at 123 (indicating that 6.5% of temporary workers versus 2.6% of permanent workers were unemployed one year later). Temporary workers were also more than twice as likely than other workers to be out of the workforce one year later (14.0% versus 6.7%). Id. The data are even more stark for blue collar workers, with 10.7% of temporary workers unemployed one year later, as compared to 3.4% of permanent workers. See id.

(83.) See Polivka, By Choice?, supra note 75, at 73 (comparing the percentage of workers currently employed in temporary and on-call jobs who say they expect that their job to lead to permanent employment in that firm with the smaller percentage of traditional workers who actually started working for their previous employer through a temporary position). Of course these data are suggestive but not decisive: workers may hold temporary jobs with one employer but then accept permanent positions elsewhere.

(84.) The same was not true in high-skill occupations, where returns to human capital were equivalent for temporary and permanent workers. See Cohen & Haberfeld, supra note 79, at 280, 283. The study compared two samples of Israeli workers--temporary help services workers (THS) and non-THS--in four occupations. The high skill occupations examined were bookkeeper and typist, and the low skill occupations were clerk and key-punch operator. The independent variables regressed against wages included age, ethnicity (Israeli versus "eastern born" Jews), education, gender, marital status, and industry. However, because the authors were not able to measure and compare non-wage compensation in the form of fringe benefits, these data have limitations that may be quite significant. See id. at 274 n.6.

(85.) Segal & Sullivan, Growth of Temporary Services, supra note 24, at 125. The gap between temporary and permanent workers' wages in blue collar occupations was reduced from 29.4% to 15.6%; for pink collar workers it went from 17.7% to 12%. When the regression was estimated using an individual fixed effect, a method which attempts to include unobserved characteristics like quality of education and number of years of job experience, the gap was reduced further to 3% for blue collar, 5.7% for pink collar, and 2% for white collar but still it was not eliminated.

(86.) See id. 126 (Unfortunately, the researchers did not specify the precise results of the analysis.). The relative youth of contingent workers may also help explain the pension gap. Young workers--even those who are non-contingent--are overwhelmingly less likely than older workers to plan for retirement even when they are eligible. Thus the overrepresentation of youth among contingent workers may help explain the striking gap in pension coverage between contingent and non-contingent workers. This hypothesis finds support in an analysis by Hippie & Stewart, supra note 59, at 29, which examines pension coverage in both contingent and noncontingent workers. Of noncontingent workers who are eligible to participate in a pension plan, only 26% between the ages of 16 and 19, and 61% aged 20 to 24 have pension coverage, compared to 88% of noncontingent workers as a whole. The figures are even starker for contingent workers. Although younger contingent workers are less than half as likely as all contingent workers to be eligible for pensions through their employer, they are also much less likely to participate even when they are eligible: Only 4% of contingent workers between 16 and 19 years of age who are eligible to participate in employer provided plans, and 48% of those between 20 and 24, actually have pension coverage, while 69% of all eligible contingent workers have coverage.

(87.) See Mark Montgomery & James Cosgrove, Are Part-Time Women Paid Less? A Model With Firm-Specific Effects, 33 ECON. INQUIRY 119, 131 (1995) (unable to explain fully the wage gap between full-time and part-time workers using regression analysis, arguing that a woman's decision to work part-time rather than full-time might be correlated with lower reliability and enthusiasm).

(88.) See Polivka, Profile, supra note 60, at 14-16 (analyzing 1995 survey data); see also Segal & Sullivan, Growth of Temporary Services, supra note 24, at 121 (noting that, at least in the temporary workforce, the percentage of men has increased rapidly in recent years, going from 24% in 1988 to 39% in 1993).

(89.) The source on which I rely does not supply industry and occupation figures for other races.

(90.) See Polivka, Profile, supra note 60, at 11, 14 & 15.

(91.) Jonathan Hiatt, Policy Issues Concerning the Contingent Workforce, 52 WASH. & LEE L. REV. 739, 744 (1995). Jonathan Hiatt is general counsel for the AFL-CIO.

(92.) See PETER B. DOERINGER & MICHAEL J. PIORE, INTERNAL LABOR MARKETS AND MANPOWER ANALYSIS (2d. ed. 1985) [hereinafter DOERINGER & PIORE, MANPOWER ANALYSIS]. I shall focus heavily on this treatment, and on other work of Michael Piore, though others were also very influential in the development of this literature. See DAVID GORDON, RICHARD EDWARDS & MICHAEL REICH, SEGMENTED WORK, DIVIDED WORKERS: THE HISTORICAL TRANSFORMATION OF LABOR IN THE UNITED STATES (1982); RICHARD EDWARDS, CONTESTED TERRAIN: THE TRANSFORMATION OF THE WORKPLACE IN THE TWENTIETH CENTURY (1979) [hereinafter EDWARDS, CONTESTED TERRAIN]; Richard Edwards, The Social Relations of Production in the Firm and Labor Market Structure, in LABOR MARKET SEGMENTATION 3 (Richard Edwards, Michael Reich & David M. Gordon eds., 1975) [hereinafter Edwards, Social Relations of Production]; Michael Reich, David M. Gordon & Richard C. Edwards, A Theory of Labor Market Segmentation, 63 AM. ECON. REV. 359 (1973). As we shall see, the strong segmentationists tend to put a more radical spin on the basic segmentationist idea, more consistent with the ideas of Gordon, Edwards, and others, than do the New Keynesians discussed in Part II.C.

(93.) Economists trace this idea to "neo-institutionalist" scholarship of the 1940s and 1950s. See, e.g., Clark Kerr, The Balkanization of Labor Markets, in LABOR MOBILITY AND ECONOMIC OPPORTUNITY 92 (1954). The progenitors of neo-institutionalism, in turn, were the institutional labor scholars of Wisconsin--most famously, John Commons, Richard Lester, and Selig Perlman--whose crowning, triumphant glory was the New Deal. Their interdisciplinary style, strong policy orientation, and conviction that technology, social institutions, and customary relations were integral to the formation of preferences became earmarks of the so-called institutionalist school, though lack of any unified modern theory of institutionalism has led one commentator to describe it as a "friendly alliance between otherwise disparate individuals united primarily in opposition to mainstream economics." PHILIP KLEIN, BEYOND DISSENT: ESSAYS IN INSTITUTIONAL ECONOMICS 115 (1994).

A more apt characterization seems not so much "friendly alliance" as contest over ownership. Many lay claim to (or are claimed to possess) the mantle of institutionalism, from Austrian school theorists such as Hayek, RICHARD LANGLOIS, ECONOMICS AS PROCESS: ESSAYS IN THE NEW INSTITUTIONAL ECONOMICS (1986); Malcolm Rutherford, What is Wrong with the New Institutional Economics (and What is Still Wrong with the Old)?, 1 REV. POL. ECON. 299 (1989), to neo-classical law-and-economics scholars like Posner and game theorists in the comparative institutional vein, Rutherford, supra, to adherents of the idiosyncratic exchange and implicit contracting theories of Williamson and others, Clark Kerr, The Neoclassical Revisionists in Labor Economics (1940-1960)--R.I.P., in HOW LABOR MARKETS WORK 1 (Bruce Kaufman ed., 1988), to what this article calls the New Keynesians, Sanford Jacoby, The New Institutionalism: What Can It Learn From the Old?, 29 INDUS. REL. 316 (1990), to the `moderate radical' sociologically-influenced theories associated with Akerlof and Piore, Jacoby, supra; Yngve Ramstad, Institutional Economics and the Dual Labor Market Theory, in INSTITUTIONAL ECONOMICS: THEORY, METHOD, POLICY 173 (Marc Tool ed., 1993).

(94.) Primary markets were further sub-divided into upper-tier and lower-tier markets. The upper tier was occupied by well educated white collar professional and managerial workers. The lower-tier of the primary sector was occupied by unionized laborers and skilled trades or craftspeople. This two-segment model has become less important, abandoned in favor of the more generalized notion that labor markets are divided segmentally as opposed to continuously. See SUZANNE BERGER & MICHAEL PIORE, DUALISM AND DISCONTINUITY IN INDUSTRIAL SOCIETIES 2 (1980).

(95.) Even where the initial decision is fully voluntary (even as segmentationists see it) it nonetheless may be so imprudent as to be a source of concern for policy-makers. Workers who enter these markets may, for example, underestimate the degree to which they compromise their long-term ability to acquire valuable firm-specific human capital. For more on this point, see notes 98-101 infra and accompanying text.

(96.) See, e.g., Michael Piore, Fragments oft "Sociological" Theory of Wages, 63 AM. ECON. REV. 377, 382 (1973) [hereinafter Piore, "Sociological" Theory] (noting that "even the most casual inspection suggests that many existing norms reflect some earlier market relationship").

(97.) See DOERINGER & PIORE, MANPOWER ANALYSIS, supra note 92, at 139-40.

(98.) See Michael Piore, Notes for a Theory of Labor Market Stratification, in LABOR MARKET SEGMENTATION 125, 131-32 (Richard Edwards, Michael Reich & David M. Gordon eds., 1975) [hereinafter Piore, Notes for a Theory].

(99.) Id. at 133.

(100.) See DOERINGER & PIORE, MANPOWER ANALYSIS, supra note 92, at 170.

(101.) See id. at 171.

(102.) Good reviews of the empirical controversies surrounding the veracity of labor market segmentation models appear in MALCOLM SAWYER, THE CHALLENGE OF RADICAL POLITICAL ECONOMY: AN INTRODUCTION TO THE ALTERNATIVES TO NEO-CLASSICAL ECONOMICS 92-96 (1989); William Dickens & Kevin Lang, Labor Market Segmentation Theory: Reconsidering the Evidence, in LABOR ECONOMICS: PROBLEMS IN ANALYZING LABOR MARKETS 141 (William Darity, Jr. ed., 1993); William Dickens & Kevin Lang, A Test of Dual Labor Market Theory, 75 AM. ECON. REV. 792 (1985); Randy Hodson & Robert Kaufman, Economic Dualism: A Critical Review, 47 AM. SOCIOL. REV. 727 (1982); James Rebitzer, Radical Political Economy and the Economics of Labor Markets, 31 J. ECON. LIT. 1394, 1417-21 (1993).

(103.) Orthodox neoclassical theorists have argued, for example, that it is impossible to rule out the possibility that "job clusters" are post-hoc descriptions of an equilibrium already sorted according to productivity and human capital factors, i.e., just as a competitive market model would predict. Critiques to this effect include Glen Cain, The Challenge of Segmented Labor Market Theories to Orthodox Theory: A Survey, 14 J. ECON. LIT. 1215, 1231-32 (1976) (listing studies which find no support for this dual-market theory).

Moreover, orthodox neoclassicists recognize the existence of internal labor markets, but tell quite a different story about them, focusing on the presence of implicit contracting (IC). In the implicit contracting model, there are numerous points during the life-cycle of the employee in which wages will diverge from the employee's marginal product and remain that way for longer periods than one would predict if one assumes that wages and prices will adjust in response to supply and demand shifts. Nonetheless, the IC model assumes that on average the wage will equal the worker's lifetime marginal product, and in the event it does not, this is the result of shocks to the economy that were impossible for the parties to predict when they entered into the contract. Ex ante, the parties made rational, voluntary allocational decisions that would permit changes in the relationship between the marginal product and wage levels over the lifetime of the employment relationship, but which, on average, would be market-clearing. IC models would predict the existence of contingent employment arrangements. See Michael L. Wachter & Randall D. Wright, The Economics of Internal Labor Markets, 29 INDUS. REL. 240, 243-44 (1990); see also Sherwin Rosen, Implicit Contracts: A Survey, 23 J. ECON. LIT. 1144, 1152 (1985); OLIVER WILLIAMSON, THE ECONOMIC INSTITUTIONS OF CAPITALISM 245 (1985) (whose "asset specificity" model specifically refers to the "spot market" that will develop for individuals who are unskilled or whose skills are nonspecific). Williamson in turn cites ARTHUR OKUN, PRICES AND QUANTITIES: A MACROECONOMIC ANALYSIS 63 (1981). For an alternative explanation of short-term employment arrangements emphasizing the influence of productivity trade-offs and Coasean "transaction costs" within firms, see Harold Demsetz, The Theory of the Firm Revisited, 4 J.L. ECON. & ORG., Spring 1988, at 141, 160.

(104.) This invocation of path-dependency should be understood as distinct from that invoked in chaos theory. For a discussion of this distinction and applications of path-dependency in the legal literature outside the employment field, see Mark Roe, Chaos and Evolution in Law and Economics, 109 HARV. L. REV. 641 (1996); see also Marcel Kahan & Michael Klausner, Path-Dependence in Corporate Contracting: Increasing Returns, Herd Behavior and Cognitive Biases, 74 WASH. U. L.Q. 347 (1996).

(105.) See, e.g., Dorothy Sue Cobble, Making Postindustrial Unionism Possible, in RESTORING THE PROMISE OF AMERICAN LABOR LAW 285, 293 (Sheldon Friedman, Richard W. Hurd, Rudolph H. Oswald & Ronald L. Seeber eds., 1994) [hereinafter Cobble, Postindustrial Unionism]; Joanne Conaghan, The Invisibility of Women in Labour Law: Gender-Neutrality in Model-Building, 14 INT'L J. SOC. L. 377, 380-81 (1986); George Gonos, The Contest over "Employer" Status in the Postwar United States: The Case of Temporary Help Firms, 31 L. & SOC'Y REV. 81, 86-87 (1997); Hiatt, supra note 91, at 744; Karl E. Klare, Toward New Strategies for Low-Wage Workers, 4 B.U. PUB. INT. L.J. 245, 264 (1995); Eileen Silverstein & Peter Goselin, Intentionally Impermanent Employment and the Paradox of Productivity, 26 STETSON L. REV. 1, 2 (1996).

(106.) For examples of core/periphery rhetoric in the literature on contingent employment, see Mangum et al., supra note 61, at 600, 604-05; BELOUS, supra note 7, at 5-6. This terminology owes a debt to developments in industrial sociology that paralleled the rise of labor market segmentation theory in economics, particularly the publication of Robert Averitt's book, The Dual Economy, which was part of a longer historical line of scholarship in sociology dealing with the stratification of the economy. Averitt described a dual or segmented economy, in which there were "center" and "periphery" firms (determined according to their relative market power), as opposed to primary and secondary labor markets (avenues for hiring and promotion). Doeringer and Piore explicitly linked their own ideas to those of Averitt. See DOERINGER & PIORE, MANPOWER ANALYSIS, supra note 92, at 163 (commenting that secondary workers may tend to work in peripheral firms). Thus labor market segmentation and stratification theories are conceptually distinct, though correlated. For a discussion of the convergence of economic segmentation and sociological stratification theories, see generally Ida Harper Simpson, The Sociology of Work: Where Have the Workers Gone?, 67 Soc. FORCES 563 (1989); Kevin Lang & William T. Dickens, Neoclassical and Sociological Perspectives on Segmented Labor Markets, in INDUSTRIES, FIRMS, AND JOBS: SOCIOLOGICAL AND ECONOMIC APPROACHES 65 (George Farkas & Paula England eds., 1988).

(107.) Commentators sympathetic to this view include DOERINGER & PIORE, MANPOWER ANALYSIS, supra note 92. For good summaries of these theories, see also DAVID M. GORDON, THEORIES OF POVERTY AND UNDEREMPLOYMENT: ORTHODOX, RADICAL, AND DUAL LABOR MARKET PERSPECTIVES 53-81 (1972); Howard M. Wachtel, Class Consciousness and Stratification in the Labor Process, in LABOR MARKET SEGMENTATION 95 (Richard C. Edwards, Michael Reich & David M. Gordon eds., 1973).

(108.) See GORDON, supra note 107, at 73.

(109.) See id. at 73-77.

(110.) See EDWARDS, CONTESTED TERRAIN, supra note 92, at 179 ("[T]he system of control ... creates the context within which experience, training, schooling, skills, and other attributes assume their importance."). I should note here that Piore, at least, expressly disavowed what he called the more "conspiratorial" radical ideology. See BERGER & PIORE, supra note 94, at 49-50 (describing secondary markets as "found rather than made," in the sense that certain groups--such as women, youths, and immigrant laborers--because of their immediate social goals or roles, filled the secondary niche that was itself created by flux and uncertainty in the economy. Similarly, the authors argued that the labor legislation created in the 1930s, although it left room for a pre-existing secondary sector that would fall outside the ambit of the legislation, was not designed to create a subordinate class.). But see EDWARDS, CONTESTED TERRAIN, supra note 92, at 178 (implying, critically, that Piore's work treated the relationship between workers' pre-existing skills, training, experience, schooling, etc. and the organization of the labor process as ad hoc, thus offering no explanation other than the characteristics of labor supply for the emergence of the secondary sector).

(111.) See EDWARDS, CONTESTED TERRAIN, supra note 92, at 178-80.

(112.) See, e.g, Lonnie Golden, The Expansion of Temporary Help Employment in the U.S., 1982-92: A Test of Alternative Economic Explanations, 28 APPLIED ECON. 1127, 1131 (1996) ("Employers may pursue more flexibility in labour input as a way to restore management prerogative or to increase their control over work schedules or work assignments in the production process."); Heidi Gottfried, In the Margins: Flexibility as a Mode of Regulation in the Temporary Help Service Industry, 6 WORK, EMPLOYMENT & SOC'Y 443,456 (1992) ("[F]lexibility appears to be a new post-Fordist strategy for increasing capital accumulation and worker regulation.").

(113.) See ROBERT E. PARKER, FLESH PEDDLERS AND WARM BODIES: THE TEMPORARY HELP INDUSTRY AND ITS WORKERS 138 (1993) ("The notion that recent growth has been due to voluntary participation, mainly by women who want a career and a domestic life, is seriously flawed ... Many are victims of the wide-scale corporate restructuring (or downsizing) that has been occurring since the early 1970s."); see also Eileen Appelbaum, Restructuring Work: Temporary, Part-Time, and At-Home Employment, in 2 COMPUTER CHIPS AND PAPER CLIPS: TECHNOLOGY AND WOMEN'S EMPLOYMENT 268, 305-06 (Heidi I. Hartmann, Robert E. Kraut & Louise A. Tilley eds., 1987) (arguing "it is not certain" that the temporary work increase reflects the preferences of women); Lonnie Golden & Eileen Appelbaum, What Was Driving the 1982-88 Boom in Temporary Employment? Preference of Workers or Decisions and Power of Employers?, 51 AM. J. ECON. & SOC. 473 (1992) (arguing that the volatility in labor demand, intensified price competition, and the absence of employee benefit contributions predominated over changes in the demographic composition of the workforce in fueling the increase in contingent workers).

(114.) According to one study, 44% of contingent workers give "personal" reasons for working in contingent jobs. The most common is the wish to accommodate schooling or training (14.7%), followed by "flexibility of work schedule" (9.2%), "other family or personal obligations" (3.5%), and "child care problems" (0.6%). Another 16% give reasons that fall into the catch-all category of "other personal reasons." See Polivka, supra note 75, at 65.

(115.) See id.

(116.) See RAW, supra note 18, at 31.

(117.) See, e.g., DEBORAH L. RHODE, SPEAKING OF SEX: THE DENIAL OF GENDER EQUALITY 153 (1997) ("[W]omen who make families a priority ... hear that they are not sufficiently `committed' to their careers. `Choice' on such terms is not the answer; it's part of the problem."); see also Kathryn Abrams, Choice, Dependence, and the Reinvigoration of the Traditional Family, 73 IND. L.J. 517, 526, 531 (1998) (asserting that working women are beset by pressures from perturbed spouses, conservative pundits, and the mainstream media); Nancy E. Dowd, Work and Family: The Gender Paradox and the Limitations of Discrimination Analysis in Restructuring the Workplace, 24 HARV. C.R.-C.L.L. REV. 79, 90 (1989) (arguing that work-family conflict arises in part from views of appropriate gender roles); Lucinda M. Finley, Choice and Freedom: Elusive Issues in the Search for Gender Justice, 96 YALE L.J. 914, 936-37 (1987) (reviewing DAVID L. KIRP, MARK G. YUDOF & MARLENE STRONG FRANKS, GENDER JUSTICE (1986), and arguing that notions about proper gender roles have strongly influenced women to structure their work lives around family); Frances E. Olsen, The Family and the Market, A Study of Ideology and Legal Reform, 96 HARV. L. REV. 1497, 1547 (1983) (arguing that family obligations and childhood experiences influence women's behavior in the labor market). Similar arguments could be made about the influence of ethnic and racial community membership on labor market choices.

(118.) A good deal more has been written on the social construction of and constraints upon choices than I can adequately address in a footnote, but the ideas cross disciplines. Some examples from other disciplines are L. Richard Della Fave, The Meek Shall Not Inherit the Earth: Self-Evaluation and the Legitimacy of Stratification, 45 AM. SOC. REV. 955 (1980) (sociological treatment theorizing that those who do poorly in the distribution of social rewards may grant legitimacy to the very social structures and hegemonic norms that deny them status, adopting those norms as part of their self identity); Herbert Gintis, Education, Technology, and the Characteristics of Worker Productivity, AM. ECON. REV., May 1991, at 266 (1972) (criticizing neoclassical economic claim that worker preferences are exogenously determined, and arguing that social institutions such as schools affect the tastes and personalities of future workers); George Sher, Our Preferences, Ourselves, 12 PHIL. & PUB. AFF. 34 (1983) (arguing that feminists who favor state intervention to override women's traditional preferences on the basis that they are induced by sexual stereotypes and conditioning have failed to provide sound philosophical justifications to support their claim).

Note that there are other, distinct understandings of the concept of "constrained choices." For example, this term may refer to less systematic constraints at the level of individual bargaining, such as contracts formed under duress. See MICHAEL J. TREBILCOCK, THE LIMITS OF FREEDOM OF CONTRACT 78-101 (1993). That preferences are irrational because of cognitive bias is yet another distinct claim, though one not always cleanly distinguished from other claims about socially and historically constrained choices. For a critique of this sort, see Mark Kelman, Book Review, 35 J. ECON. LIT. 2049, 2050 (1997) (reviewing CASS R. SUNSTEIN, FREE MARKETS AND SOCIAL JUSTICE (1997)).

(119.) See E. Michael Foster, Labor Economics and Public Policy: Dominance of Constraints or Preferences?, in LABOR ECONOMICS: PROBLEMS IN ANALYZING LABOR MARKETS 269, 270-71 (William Darity, Jr. ed., 1993). This concept provides the foundation for the left's attack upon orthodox neoclassical reliance on "revealed preferences" (i.e., that the jobs workers occupy are the truest surrogates for their actual preferences).

(120.) See, e.g., June Lapidus, Family Structure, Flexible Employment, and Labor Market Segmentation: Evidence From a Study of the Temporary Help Industry, 3 INT'L CONTRIBUTIONS TO LAB. STUD. 91, 99 (1993) ("Employment in the temporary help industry is two-thirds female not because particular groups of women workers choose these jobs as compatible with family responsibilities. Rather, they are female precisely because these jobs were created with women workers as the prototype."); Gottfried, supra note 112, at 448-49 ("Flexibility is a gendered mode of regulation in both targeting women for recruitment as a flexible labor force and encoding temporary clerical jobs as women's work. Temporary help service (THS) worker firms principally hire women for temporary clerical jobs based on the assumption that women constitute a flexible labor force due to their assumed responsibility for reproductive household labor.").

(121.) See, e.g., Jackie Krasas Rogers, Just a Temp: Experience and Structure of Alienation in Temporary Clerical Employment, 22 WORK & OCCUPATIONS 137, 142 (1995) (separating forms of alienation into these three classical Marxian categories). 1 make no attempt to canvass the extensive literature on the meaning of the controversial concept of alienation. I limit my discussion to a brief treatment of writings on contingent employment, with apologies for my lack of adherence to or development of any precise taxonomy of forms of alienation in the Marxian understanding. I leave that considerable task to others.

(122.) Michael A. Katovich & Monica J. Hardesty, The Temporary, 7 STUD. SYMBOLIC INTERACTION 333, 336 (1986). The authors also quote one informant as saying, "just the word `temporary,' it's like being terminated everyday." Id. at 341.

(123.) See id. at 337-38, 340-41 (describing the temporary worker as neither an insider nor an outsider, but instead a "refugee"); PARKER, supra note 113, at 107-08; Krasas Rogers, supra note 121, at 149-50.

(124.) See, e.g., Jackie Krasas Rogers & Kevin D. Henson, "Hey, Why Don't You Wear a Shorter Skirt?" Structural Vulnerability and the Organization of Sexual Harassment in Temporary Clerical Employment, 11 GENDER & SOC'Y 215, 234 (1997) ("Temporary employment and its intensification of unequal power relations should be framed as a social problem organized by gender and racial-ethnic inequality rather than a natural evolution of work in U.S. society. As social problems, they must be explored as structural rather than individual phenomena."); CYNTHIA NEGREY, GENDER, TIME, AND REDUCED WORK 27-28 (1993) ("[R]educed work exists within a larger set of social relations characterized by gender inequality and labor market segmentation.").

(125.) See, e.g., Katovich & Hardesty, supra note 122, at 344; Krasas Rogers, supra note 121, at 152-56 (describing the pressures temporary workers feel to look friendly and busy, or dress differently for different job assignments so as to "look the part"); KEVIN D. HENSON, JUST A TEMP 113-43, 145-50 (1996) (same); see also Gottfried, supra note 112 (describing the flexible frontier of control exercised by temporary help workers).

(126.) Note that even if secondary labor markets "clear" and workers who lose their jobs will be able to find others, the threat of job loss may be very anxiety-inducing. The trauma and disruption of job loss and subsequent search may be quite significant. See generally Arthur H. Goldsmith, Jonathan R. Veum & William Darity, Jr., The Psychological Impact of Unemployment and Joblessness, 25 J. SOCIO-ECONOMICS 333 (1996); Knut Gerlach & Gesine Stephan, A Paper on Unhappiness and Unemployment in Germany, 52 ECON. LETT. 325 (1996).

(127.) See Marion Crain, Confronting the Structural Character of Working Women's Subordination: Collective Action vs. Individual Rights Strategies, 3 KANS. J.L. & PUB. POL'Y 26, 31 (1994) (arguing that women's economic subordination--caused in part because of relegation to contingent, dead-end, and low-wage jobs--has led to feelings of hopelessness stemming from blocked opportunities and low self-esteem); HENSON, supra note 125, at 56 (observing a "shifting of anger boundaries" as temporary workers learn to tolerate abusive working conditions); SONIA OSPINA, ILLUSIONS OF OPPORTUNITY: EMPLOYEE EXPECTATIONS AND WORKPLACE INEQUALITY 190-94 (1996) (noting that provisional [probationary] clerical workers in the civil service whose jobs were of relatively low status developed work identities that reified their own subordination and led to low self evaluations); RIFKIN, supra note 13, at 197 ("[M]illions of workers ... experience their own individual deaths, daily, at the hands of profit-driven employers and a disinterested government. They are the ones who are waiting for pink slips, being forced to work part-time at reduced pay, or being pushed onto the welfare rolls. With each new indignity, their confidence and self-esteem suffer another blow. They become expendable, then irrelevant, and finally invisible....").

(128.) Legal scholars in the area of employment discrimination have also raised structuralist arguments, suggesting that workers may internalize their own experience of harassment or discrimination and thus adjust their own behavior in a manner that reinforces their own subordinated status. See, e.g., Kathryn Abrams, Gender Discrimination and the Transformation of Workplace Norms, 42 VAND. L. REV. 1183, 1205 (1989); Martha Chamallas, Structuralist and Cultural Domination Theories Meet Title VII: Some Contemporary Influences, 92 MICH. L. REV. 2370, 2378-79 (1994); Charny & Gulati, supra note 36, at 64-65; Marion Crain, Women, Labor Unions and Hostile Work Environment Sexual Harassment: The Untold Story, 4 TEX. J. WOMEN & L. 1, 28-29 (1995); Vicki Schultz, Telling Stories about Women and Work: Judicial Interpretation of Sex Segregation in the Workplace in Title VII Cases Raising the Lack of Interest Argument, 103 HARV. L. REV. 1749, 1825-27 (1990). But see Regina Austin, Employer Abuse, Worker Resistance and the Tort of Intentional Infliction of Emotional Distress, 41 STAN. L. REV. 1, 44 (1988) (arguing that subordinated groups affirmatively attack the structural constraints imposed upon them by engaging in oppositional behavior in the workplace).

(129.) For overviews, see Francoise J. Carre, Virginia duRivage & Chris Tilly, Representing the Part-Time and Contingent Workforce: Challenges for Unions and Public Policy, in RESTORING THE PROMISE, supra note 105, at 314; Virginia L. duRivage, New Policies for the Part-Time and Contingent Workforce, in NEW POLICIES FOR THE PART-TIME AND CONTINGENT WORKFORCE 89 (Virginia L. duRivage ed., 1992); Middleton, supra note 4.

An important caveat is that although the proposals discussed are consistent with what strong segmentationists generally advocate, not all of the law reformers whose ideas I discuss in this section explicitly adopt a theory of the labor market (strong segmentationist or otherwise). Many law reform writers--misguidedly, in my view--focus only briefly on underlying explanations for the labor market problems they seek to fix. At least part of my aim in this article is to demonstrate that in a world of competing ideas about whether and how to regulate labor markets, it is important to demonstrate that one's proposals are grounded in a persuasive theory of labor market behavior.

(130.) See, e.g., POLLY CALLAGHAN & HEIDI HARTMANN, CONTINGENT WORK: A CHART BOOK ON PART-TIME AND TEMPORARY EMPLOYMENT 31 (1991); O'Connell, supra note 4, at 903-13.

(131.) See note 2 supra; Middleton, supra note 4, at 583-86 (reviewing state and federal initiatives). Some reform bills have been introduced into Congress. Senator Metzenbaum's unsuccessful Contingent Workforce Equity Act would have extended the application of federal labor laws to part-time and temporary workers, and liberalized the coverage "triggers" in many statutes. See 140 CONG. REC. S 14247 (daily ed. Oct. 5, 1994) (statement of Sen. Metzenbaum). A version of this bill was later introduced in the House by Representative Velazquez. See H.R. 3657, 104th Cong. (1996). Representative Pat Schroeder's unsuccessful Part-time and Temporary Workers' Act would have pro-rated health and pension benefits to part-time and temporary workers. See 135 CONG. REC. E2013 (daily ed. June 7, 1989) (statement of Rep. Schroeder). Representative Schroeder reintroduced the bill in the 104th Congress. See H.R. 3682, 104th Cong. (1996). Other examples include the Economic Equity Act of 1996, sponsored by Representative Morella, H.R. 3857, 104th Cong. (1996), which would, inter alia, amend the Internal Revenue Code to expand approval of state unemployment compensation laws to provide eligibility for more part-time and temporary workers; the Fairness in the Workplace Commission Act, sponsored by Representative Kennedy, H.R. 2997, 105th Cong. (1997), which would create a commission to study wages inequity between part-time and temporary employees and their full-time counterparts; and the National Commission on Fairness in the Workplace Act, S. 1453, 105th Cong. (1997).

(132.) See HARRISON & BLUESTONE, supra note 3, at 189-90; duRivage, supra note 129, at 94-99.

(133.) See TERESA GHILARDUCCI, GARTH MANGUM, JEFFREY S. PETERSEN & PETER PHILIPS, PORTABLE PENSIONS FOR CASUAL LABOR MARKETS: LESSONS FROM THE OPERATING ENGINEERS CENTRAL PENSION FUND 167-85 (1995).

(134.) See JOEL HANDLER & YEHESKEL HASENFELD, WE THE POOR PEOPLE 106-13 (1997).

(135.) See, e.g., Berger, supra note 3, at 46-47 (advocating expanded protection of contingent workers by wrongful termination rules); Bingham, supra note 5, at 950-54 (advocating expanded unemployment compensation); HANDLER & HASENFELD, supra note 134, at 136-39 (same); Maranville, supra note 5, at 299 (same).

(136.) See CALLAGHAN & HARTMANN, supra note 130, at 32; Carre et al., supra note 129, at 322; Jonathan P. Hiatt & Lynn Rhinehart, The Growing Contingent Work Force: A Challenge for the Future, 10 LAB. LAW. 143, 158 (1994). Some states have enacted just such legislation. See Middleton, supra note 4, at 585-86 (reviewing state legislation that, for example, requires that either agency, or client employer, or both, pay workers' compensation, requires agencies to comply with minimum standards of net worth to ensure compliance, and mandates disclosure of benefits plans).

(137.) I have been unable to find an empirical study measuring the prevalence of this practice, but the Dunlop Commission heard testimony reporting a number of instances where it has occurred. See DUNLOP COMMISSION REPORT, supra note 2, at 93-94.

(138.) See Vizcaino v. Microsoft Corp., 97 F.3d 1187 (9th Cir. 1996), reh'g en banc, 120 F.3d 1006 (1997), and cert. denied, 118 S. Ct. 899 (1998) (holding in-house workers who the employer had reclassified as "independent contractors" to be "employees" for purposes of the common law test used for eligibility for ERISA benefits, notwithstanding that the employer had required them to sign agreements acknowledging their independent contractor status and waiving their claims to company benefits.).

(139.) Others have focused on how loopholes in the NLRA leave room for "flexible" practices that evade protections given to existing union members. See Craig Becker, Labor Law Outside the Employment Relation, 74 TEX. L. REV. 1527, 1537-53 (1996) (arguing that the definitions of "employee," "employer," and "secondary employer" under the NLRA create incentives for the strategic use of subcontracting arrangements to avoid duties under the NLRA. "Client" firms may exploit their non-employer status under the NLRA to undermine duty to bargain and successorship rules, and remain immunized from secondary boycotts).

(140.) See Howard Wial, New Bargaining Structures for New Forms of Business Organization, in RESTORING THE PROMISE, supra note 105, at 303,309 (noting that bargaining unit determination rules, where the NLRB determines bargaining units according to a "community of interest" test, identify single firms as the relevant community, unless there is precedent recognizing a broader community, and arguing that multi-employer communities of service workers fail to meet either criterion); Klare, supra note 105, at 262, 270-71 (describing the difficulty of union organizing in the building services industry).

(141.) But see MICHAEL LIND, THE NEXT AMERICAN NATION: THE NEW NATIONALISM ANt) THE FOURTH AMERICAN REVOLUTION 319, 322-26 (1995) (advocating the imposition of tariffs on U.S. manufacturing abroad, restricted immigration, laws making it illegal or difficult to hire temporary workers, and direct government-linked, rather than employer-linked, provision of benefits and employee protections, in the hopes of creating "middle class service careers").

(142.) For earlier descriptions and "re-inventions" of the craft administration model, though directed less specifically toward contingent work, see generally MICHAEL J. PIORE & CHARLES F. SABEL, THE SECOND INDUSTRiAL DIVIDE: PROBLEMS FOR PROSPERITY (1984); CHARLES F. SABEL, WORK AND POLITICS 194-231 (1982). For a classic treatment of the craft administration model, see Arthur L. Stinchcombe, Bureaucratic and Craft Administration of Production: A Comparative Study, 4 ADMIN. SCI. Q. 168 (1959).

(143.) See Cobble, Postindustrial Unionism, supra note 105; Dorothy Sue Cobble, Organizing the Postindustrial Work Force: Lessons from the History of Waitress Unionism, 44 INDUS. & LAB. REL. REV. 419 (1991) (analyzing waitress unions that existed between the turn of the century and the 1960s).

(144.) Note that skepticism of this nature persists among some who advocate collective bargaining among service workers. See Howard Wial, The Emerging Organizational Structure of Unionism in Low-Wage Services, 45 RUTGERS L. REV. 671, 679, 686 (1993) [hereinafter Wial, Emerging Organizational Structure].

(145.) See Cobble, Postindustrial Unionism, supra note 105, at 292-94.

(146.) See id. at 294-302 (discussing these and other reforms).

(147.) See Wial, Emerging Organizational Structure, supra note 144, at 690-92. On the Justice for Janitors Campaign, see also Richard Hurd & William Rouse, Progressive Union Organizing: The SEIU Justice for Janitors Campaign, 21 REV. RADICAL POL. ECON. 70 (1989); Klare, supra note 105, at 270-73.

(148.) See, e.g., Howard Wial, Getting a Good Job: Mobility in a Segmented Labor Market, 30 INDUS. REL. 396, 414-15 (1991) [hereinafter Wial, Getting a Good Job] (arguing that collective bargaining is a way to convert secondary jobs into primary jobs, and to disrupt norms that inhibit mobility in disadvantaged communities).

(149.) Wial, Emerging Organizational Structure, supra note 144, at 696. The idea of harnessing the pre-existing solidaristic impulse of identity- and community-group associations shares common ground with CHARLES C. HECKSCHER, THE NEW UNIONISM: EMPLOYEE INVOLVEMENT IN THE CHANGING CORPORATION 177-91 (1988).

(150.) See, e.g., Alan Hyde, Employee Organization in High-Velocity Labor Markets, N.Y.U 50TH ANNUAL CONFERENCE ON LAB. PROC. (forthcoming 1998); NOYELLE, BEYOND INDUSTRiAL DUALISM, supra note 58, at 124; Silverstein & Goselin, supra note 105 (proposing that section 8(f) of the NLRA extend beyond workers in the construction and building trades, so that unions and professional associations could establish employment offices that would train workers and compete with employment agencies to secure them work assignments, yet arguing against the need to exempt these associations from the prohibition on secondary boycotts).

(151.) See Thomas R. Bailey & Annette D. Bernhardt, In Search of the High Road in a Low-Wage Industry, 25 POL. & SOC'Y 179, 196-97 (1997); CHARLES B. CRAVER, CAN UNIONS SURVIVE? THE REJUVENATION OF THE AMERiCAN LABOR MOVEMENT 72 (1993); PETER B. DOERINGER, TURBULENCE IN THE AMERiCAN WORKPLACE 211-12 (1991); HANDLER & HASENFELD, supra note 134, at 146. On the potential for government-sponsored community job creation programs to ameliorate hardships borne by contingent workers, see Rachel Geman, Safeguarding Employee Rights in that Post-Union World: A New Conception of Employee Communities, 30 COLUM. J.L. & SOC. PROB. 369, 403-04 (1997).

Still others push toward related models of worker-owned cooperatives to achieve many of the same ends for contingent workers. See Middleton, supra note 4, at 611-13 (describing attempts to establish worker-owned cooperatives of clerical workers, orange growers, domestic workers, and landscape workers); Peter Pitegoff, Child Care Enterprise, Community Development, and Work, 81 GEO. L.J. 1897, 193743 (1993) (advocating community-based, worker-owned cooperative child care centers).

(152.) For a fuller discussion of this notion, see EDWARD J. MCCAFFERY, TAXING WOMEN 246-48 (1997) (discussing the problem of "incomplete markets" in which, for example, a married woman seeking a 20-hour job may be forced to make an all-or-nothing choice--zero hours or 40 hours--because of a lack of intermediate options that would permit employment at her efficient wage).

(153.) Maria Hylton criticizes this idea. See Hylton, Against Regulating, supra note 47, at 855 ("Labeling employment decisions influenced by personal choices about transportation, children and so forth `involuntary' is dangerous because the `involuntary' label creates a class of `victims' whose `subordinated' status is a function of purely voluntary decisions.").

(154.) There is a large literature on institutionally-created incentives. Recent overviews of how incentives created by legal institutions may reproduce the traditional household division of labor appear in RHODE, supra note .117, at 141-214. See also Symposium on Institutional Barriers to Women in the Workplace, 6 UCLA WOMEN'S L.J. 279 (1996). On how legal rules, such as child custody laws, might blunt worker mobility, see Kelman, Law and Macroeconomics, supra note 6, at 1276-77.

(155.) An important question, then, is why firms do not simply offer better jobs to workers who are underemployed. Several hypotheses are advanced in Part II.C, infra, on New Keynesian explanations of underemployment.

(156.) For empirical evidence of this proposition, see, for example, ROLAND BENABOU, WORKINGS OF A CITY: LOCATION, EDUCATION, AND PRODUCTION (National Bureau of Econ. Research Technical Working Paper No. 113, 1991) (finding that "equilibrium segregation can cause entire `ghettos' to drop out of the labor force"); Julia R. Henly, Barriers to Finding and Maintaining Jobs: The Perspective of Workers and Employers in the Low-Wage Labor Market, in BEYOND WELFARE: GETTING REAL ABOUT WORK FOR LOW-INCOME WOMEN: CHALLENGES, STRATEGIES AND INNOVATIONS (Joel F. Handler & Lucie White eds., forthcoming 1998) (analyzing surveys of employers and low income women in Los Angeles); Joleen Kirschenman & Katherine M. Neckerman, "We'd Love to Hire You But": The Meaning of Race for Employers, in THE URBAN UNDERCLASS 203-32 (Christopher Jencks & Paul Peterson eds., 1991) (examining low-income blacks in Chicago); Roger Waldinger, Black/Immigrant Competition Reassessed: New Evidence from Los Angeles, 40 SOC. REV. 365 (1997); Wial, Getting a Good Job, supra note 148 (examining low-income white men from ethnic neighborhoods in South and East Boston).

(157.) See Feldman, supra note 33, at 397-98 (discussing the effect of underemployment on career attitudes); Linda J. Khan & Paula C. Morrow, Objective and Subjective Underemployment Relationships to Job Satisfaction, 22 J. BUS. RES. 211 (1991) (discussing objective and subjective measurements of unemployment); Louis Uchitelle & N.R. Kleinfield, The Price of Jobs Lost, in THE DOWNSIZING OF AMERICA 3, 3-36 (New York Times ed., 1996) (describing individual accounts of the trauma of a drop in job status, including the decline from permanent to contingent).

(158.) On the self-reinforcing effects of unemployment (hysteresis), see RONALD SCHETTKAT, LABOR MARKET DYNAMICS OF ECONOMIC RESTRUCTURING 44, 181-82 (1992); Christopher A. Pissarides, Loss of Skill During Unemployment and the Persistence of Employment Shocks, 107 Q.J. ECON. 1371 (1992). Similar observations with respect to non-traditional employment and intermittent labor force attachment can be found in Marianne A. Ferber & Jane Waldfogel, The Long-Term Consequences of Non-Traditional Employment, MONTHLY LAB. REV., May 1998, at 3; Reuben Gronau, Sex-Related Wage Differentials and Women's Interrupted Labor Careers--the Chicken or the Egg, 6 J. LAB. ECON. 277 (1988); Joyce P. Jacobsen & Laurence M. Levin, Effects of Intermittent Labor Force Attachment on Women's Earnings, MONTHLY LAB. REV., Sept. 1995, at 14. Cf. Edward J. McCaffery, Slouching Towards Equality: Gender Discrimination, Market Efficiency, and Social Change, 103 YALE L.J. 595, 615-24 (1993) (arguing that women's tendency to hold inferior jobs is a self-fulfilling prophecy due to a combination of women's rational human capital investment strategies, employer perception that women are less committed workers, and tax incentives); Wilkins & Gulati, supra note 36, at 540-41, 564-84 (describing the concept of "flatlining" in elite law firms, where in response to noisy signals about the abilities of racial minority associates, firms put these workers on a track that involves fewer and poorer opportunities for skill acquisition, and ultimately, lower chances of partnership).

(159.) Where membership within a protected group can be isolated as the primary factor leading to inefficient job placement, then the answer would seem to lie not in the expansion of benefits, organization of contingent workers, or implementation of outreach and training programs. Rather, antidiscrimination measures, including affirmative action, might help some workers overcome the discriminatory stereotypes that undermine their abilities to demonstrate their marginal product (especially in difficult to monitor jobs). See, e.g., Geman, supra note 151, at 379-90 (arguing that the law should promote "identity communities," groups formed in the workplace to address the concerns of minority employees). But for the many contingent workers who are not members of protected groups, their problem is not discrimination as conventionally understood.

(160.) See, e.g., LARRY L. ORR, HOWARD S. BLOOM, STEPHEN H. BELL, FRED DOOLITTLE, WINSTON LEE & GEORGE CAVE, DOES TRAINING FOR THE DISADVANTAGED WORK? EVIDENCE FROM THE NATIONAL JTPA STUDY (1995) (finding that training provided by federal Jobs Training Partnership Act in the form of general education, classroom training in occupational skills, on-the-job training, job-search assistance, and subsidized work experience is cost effective for men and women, but not for youth). But see Rosella Gardecki & David Neumark, Order From Chaos? The Effects of Early Labor Market Experiences on Adult Labor Market Outcomes, 51 INDUS. LAB. REL. REV. 299 (1998) (arguing that evidence that adult labor market outcomes are for the most part unrelated to early labor market experiences fails to provide support for schools-to-work transition programs).

(161.) See notes 121-125 supra and accompanying text.

(162.) See Katovich & Hardesty, supra note 122, at 347.

(163.) See Virginia L. Olesen & Frances Katsuranis, Urban Nomads: Women in Temporary Clerical Services, in WOMEN WORKING: THEORIES AND FACTS IN PERSPECTIVE 316, 332-34 (Ann H. Stromberg & Shirley Harkess eds., 1978).

(164.) See, e.g., Moshe Krausz, Tamar Brandwein & Shaul Fox, Work Attitudes and Emotional Responses of Permanent, Voluntary, and Involuntary Temporary-Help Employees: An Exploratory Study, 44 APPLIED PSYCHOL.: AN INT'L REV. 217 (1995) (reporting that involuntary temporary workers have lower job satisfaction than either permanent workers or voluntary temporary workers). The researchers employed a problematic experimental design, basing their measures of voluntariness on self-reporting. To the extent that self-reported voluntariness is in essence a measure of whether a worker perceives that she is getting what she wants, one would predict a strong correlation between voluntariness and conventional measures of job satisfaction.

Others using measures of involuntariness that include the skills-job mismatch have also reported negative relationships between job satisfaction and both involuntariness and the skills-job mismatch, but with similar shortcomings due to the use of workers self-reported perceptions. See Daniel C. Feldman, Helen I. Doerpinghaus & William H. Turnley, Employee Reactions to Temporary Jobs, 7 J. MANAGERIAL ISSUES 127 (1995) (finding that individuals who hold temporary jobs involuntarily, as measured by self-reporting of involuntariness and failed attempts to find permanent employment, and who reported that they were employed in temporary jobs that could be performed by someone with less education and experience, scored lower on job satisfaction measures than those who reported that their temporary jobs were voluntary and consistent with their prior education and experience).

(165.) For a general statement of this argument, see, for example, Thomas Kohler, Civic Virtue at Work: Unions as Seedbeds of the Civic Virtues, 36 B.C.L. REV. 279, 300 (1995) (asserting that "by providing a forum for serious deliberation and self-directed action, collective bargaining can ... reduce the fragmentation and isolation that dissolves community and undermines the possibility of authentic self-rule").

(166.) See Sabel, Moebius-Strip Organizations, supra note 53, at 46 (arguing that the new kind of labor movement "might be a world in which workplace autonomy was combined with, even dependent on, forms of social conformity persons of my generation once associated with the postwar U.S. suburbs and those in the United States now associate with Japan").

(167.) The work of Feldman, supra note 33, takes a significant step in this direction.

(168.) Of course, reflecting on the possibility that workers' labor market outcomes are influenced substantially by their initial social and labor market conditions may itself effect social change. If heightened awareness eventually helps to change the social attitudes that lead some groups of workers to underestimate and ultimately under-realize their own potential, then it serves a useful purpose, even if it does not point to clear-cut labor market reforms.

(169.) For overviews of New Keynesianism, see Robert J. Gordon, What is New Keynesian Economics?, 28 J. ECON. LIT. 1115 (1990); Kelman, Law and Macroeconomics, supra note 6, at 1247-59; Michael Syron Lawler, Keynes, Cambridge, and the New Keynesian Economics, in LABOR ECONOMICS: PROBLEMS IN ANALYZING LABOR MARKETS 11 (William Darity, Jr. ed., 1993); Daniel J.B. Mitchell, Keynesian, Old Keynesian and New Keynesian Wage Nominalism, 32 INDUS. REL. 1 (1993); Joseph E. Stiglitz, Methodological Issues and the New Keynesian Economics, in MACROECONOMICS 38 (Alessandro Vercelli & Nicola Dimitri eds., 1992). Implicit contracting (IC) models are also commonly identified as New Keynesian in orientation, but as discussed above, I consider them to be closer to orthodox neoclassical accounts. See note 103 supra.

(170.) See GEORGE A. AKERLOF & JANET L. YELLEN, EFFICIENCY WAGE MODELS OF THE LABOR MARKET 2 (1986) (outlining the benefits to employers of paying higher wages); ANDREW WEISS, EFFICIENCY WAGES: MODELS OF UNEMPLOYMENT, LAYOFFS, AND WAGE DISPERSION 55-56 (1990) (discussing the importance of higher wages in fostering employee goodwill).

(171.) See A. MICHAEL SPENCE, MARKET SIGNALING 7-9 (1974) (discussing the nature of employer uncertainty and the possible responses).

(172.) See id. at 8. Inefficient discrimination may occur even when it is common knowledge that group characteristics do not differ and when employers do not prefer candidates from the same group. See Bradford Cornell & Ivo Welch, Culture, Information, and Screening Discrimination, 104 J. POL. ECON. 542, 542 (1996) (arguing that an employer may choose "same-group" workers based on the belief that it will be better able to judge applicants' unknown qualities when they come from the same cultural background than when they come from a different group).

(173.) See SPENCE, supra note 171, at 10-11. Note the parallels between this account and Docringer & Piore's prediction that workers expecting to be rejected because they fall outside certain community chains of mobility may decide not to apply for jobs for which they are qualified, and thus initiate their careers from an inferior port of entry. Set., notes 95-97 supra and accompanying text; see also Charny & Gulati, supra note 36, at 78-83 (asserting that minority workers may adopt "outlier" human capital investment strategies in order to distinguish themselves from others, even if their mean prior achievement level is the same as non-minority competitors. This may do them little good in the end: the "noisier" signal minorities send may be disfavored by rational employers).

(174.) See generally Andrew Weiss, Job Queues and Layoffs in Labor Markets with Flexible Wages, 88 J. POL. ECON. 526 (1980). In the legal literature, see Michael Selmi, Testing for Equality: Merit, Efficiency, and the Affirmative Action Debate, 42 UCLA L. REV. 1251, 1301-08 (1995) (marshaling this type of efficiency wage argument to defend affirmative action policies).

(175.) Although the literature and I refer to workers who are crowded out as "queued," an important caveat is in order. While a queue suggests a particular ordering of workers, in which those who have waited longest have the first chance at the next available high-wage job, no such orderly queue is contemplated in efficiency wage and insider-outsider models. Indeed, workers in the queue are uncertain of whether and when they will be promoted to a job matched with their preferences and abilities.

(176.) Recall the failure of human capital models to explain fully the gap in compensation between contingent and non-contingent workers. See notes 71-90 supra and accompanying text.

(177.) See note 158 supra and accompanying text.

(178.) See J. Hoult Verkerke, Legal Regulation of Employment Reference Practices, 65 U. CHI. L. REV. 115, 146-49 (1998) (arguing that previous dismissals or past job market "churning" may "scar" a worker, even if the reasons are benign, and send a negative signal to an employer).

(179.) See AKERLOF & YELLEN, supra note 170, at 8.

(180.) I confess that the "positive" signaling model doesn't sit as comfortably with my own instincts as "negative" signaling models of the sort described above, in which employers draw negative implications from signals that may have no relationship to actual productivity. See notes 171-172 supra and accompanying text. It seems less intuitively obvious to me that employers will base hiring decisions on a presumed correlation between a high wage demand and high quality.

(181.) For example, assume the expected cost to the worker of shirking is the product of the probability of dismissal and cost of dismissal. Assuming the employer's goal is to increase the cost of shirking, it may do so by increasing either factor.

(182.) For a concise model of this intuition, see generally Robert Solow, Another Possible Source of Wage Stickiness, 1 J. MACROECON. 79 (1979). One would also expect efficiency wages to reduce costly turnover, and this additional cost avoided should also be factored into the cost-benefit analysis.

(183.) This argument is developed in Carl Shapiro & Joseph E. Stiglitz, Equilibrium Unemployment as a Worker Discipline Device, 74 AM. ECON. REV. 433 (1984). Note also that this model does not assume that all workers who are unemployed are workers who were caught shirking and dismissed, or laid off because of "dismissal-threat" employer strategies. At any given time, some unemployment may also be voluntary, or due to monitoring errors or layoffs unrelated to such employer strategies. The rate of unemployment will nonetheless be above the optimal level in equilibrium.

(184.) Rising age-wage profiles and other non-dismissal-based effort incentives need not be incompatible with efficiency wages. See, e.g., Jeremy I. Bulow & Lawrence H. Summers, A Theory of Dual Labor Markets with Application to Industrial Policy, Discrimination, and Keynesian Unemployment, 4 J. LAB. ECON. 376, 387 (1986) (arguing that although employers may use rising age-wage profiles to prevent quitting, efficiency wages may be necessary to solve effort elicitation problems not fully solved by the age-wage mechanism); Stewart J. Schwab, Life-Cycle Justice: Accommodating Just Cause and Employment at Will, 92 MICH. L. REV. 8, 21-22 (1993) (same). See also Joseph A. Ritter & Lowell J. Taylor, Workers as Creditors: Performance Bonds and Efficiency Wages, 84 AM. ECON. REV. 694, 695 (1994) (arguing that efficiency wage mechanisms may co-exist with bonding-based effort elicitation schemes); David B. Wilkins & G. Mitu Gulati, Redecribing the Tournament of Lawyers: Structuring Internal Labor Markets in the Age of Opportunism, 84 VA. L. REV. (forthcoming 1998) (arguing that law firms employ multiple incentive schemes to motivate workers, including, but not limited to, efficiency wages).

(185.) This account assumes that an employer who pays efficiency wages will need to pay core workers higher premia to induce effort if the credible threat of job loss is higher, because with a rising probability of job loss, anticipated tenure--and lifetime earnings--in the firm is lower. See Bulow & Summers, supra note 184, at 384; REBITZER & TAYLOR, supra note 11, at 2-3.

(186.) See Pfeffer & Baron, supra note 56, at 289.

(187.) See George A. Akerlof, Labor Contracts as Partial Gift Exchange, 97 Q.J. ECON. 543, 54344 (1982); see also Mark Barenberg, The Political Economy of the Wagner Act: Power, Symbol, and Workplace Cooperation, 106 HARV. L. REV. 1379, 1485-87 (describing the Wagner Act as premised upon the theoretical assumptions of gift exchange models).

(188.) See GEORGE A. AKERLOF, AN ECONOMIC THEORIST'S BOOK OF TALES 152 (1984).

(189.) Employers may find it proves harmful to primary employee morale to have both high- and low-paying jobs within an establishment, because core employees will be more productive if there is greater equity among them. For discussion of the morale-undermining influence of wage inequity across employees within a firm, see Robert H. Frank, Are Workers Paid Their Marginal Products?, 74 AM. ECON. REV. 549, 549-52 (1984); Richard McAdams, Relative Preferences, 102 YALE L.J. 1, 46 (1992); Segal & Sullivan, Growth of Temporary Services, supra note 24, at 132.

(190.) See Robert Drago & Richard Perlman, Supervision and High Wages as Competing Incentives: A Basis for Labour Market Segmentation Theory, in MICROECONOMIC ISSUES IN LABOUR ECONOMICS: NEW APPROACHES 41, 51-52 (Robert Drago & Richard Perlman eds., 1989) (describing the importance of job security for primary sector workers); Thomas I. Palley, The Fair-Wage-Effort Hypothesis: Implications for the Distribution of Income and Dual Labor Markets, 24 J. ECON. BEHAV. & ORG. 195, 203-05 (1994) (explaining that wage levels correspond to employee effort levels).

(191.) See Bulow & Summers, supra note 184, at 384. For a critical view, skeptical of the claim that employers will resist negotiating wage deals with workers willing to work at their marginal product, see H. Lorne Carmichael, Efficiency Wage Models of Unemployment--One View, 28 ECON. INQUIRY 269 (1990).

(192.) See ROBERT M. SOLOW, THE LABOR MARKET AS A SOCIAL INSTITUTION 37-50 (1990) [hereinafter SOLOW, SOCIAL INSTITUTION].

(193.) See Bulow & Summers, supra note 184, at 388-89.

(194.) See Carmichael, supra note 191, at 282-85; Derek Laing, Firm Specific Human Capital as an Employer Discipline Device, 32 ECON. INQUIRY 128, 128-29 (1994).

(195.) See Laing, supra note 194, at 128. This latter response assumes that rogue employers' reputations for engaging in bad faith practices will not be disseminated widely enough to drive them from the market.

A variation on the bonding critique is that there does exist an alternative practice capable of eliciting effort, but which is consistent with market-clearing assumptions. Instead of fees or efficiency wages, the critics argue, we observe self-enforcing contracts, i.e., life-cycle implicit contracting practices. Thus workers may begin as apprentices earning low wages, and will work harder because of promises of high wages downstream. But as described above, the existence of these practices is not inconsistent with efficiency wages; both may co-exist yet remain distinct. See note 184 supra. For fuller treatments of these and other arguments disputing the bonding critique of efficiency wage models, see William T. Dickens, Lawrence F. Katz, Kevin Lang & Lawrence H. Summers, Employee Crime and the Monitoring Puzzle, 7 J. LABOR ECON. 331, 336-42 (1989); Kevin Lang & Shulamit Kahn, Efficiency Wage Models of Unemployment: A Second View, 28 ECON. INQUIRY 296, 299-300 (1990).

(196.) See generally ASSAR LINDBECK & DENNIS J. SNOWER, THE INSIDER-OUTSIDER THEORY OF EMPLOYMENT AND UNEMPLOYMENT (1988) (comprising numerous articles published by the authors on the subject during the 1980s). I rely heavily on this source in the description to follow.

(197.) If insiders request too high a wage, or work too little, or create too much disruption within the firm (e.g., by excessive striking or harassment), the employer may indeed replace them with outsiders. Thus, although they may engage in rent-seeking behavior, insiders also have the incentive to provide sufficient value to prevent the employer from replacing them.

(198.) Turnover costs may also be exogenous, that is, independent of the workers' decision-making processes. They may arise in the form of severance pay (e.g., as a result of wrongful dismissal litigation), or from interviewing and screening replacements. Such expenses simply may consist of the costs of advertising and conducting interviews, or they may include costs resulting from inefficiencies in screening for high quality replacements.

(199.) See LINDBECK & SNOWER, supra note 196, at 5, 246-48 (fitting the insider-outsider account into the description of primary and secondary labor markets).

(200.) For a detailed review of these areas of research, see generally Dickens & Lang, supra note 102.

(201.) See Jasmin Ansar, Paul Cantor & Roger W. Sparks, Efficiency Wages and the Regulated Firm, 11 J. REGULATORY ECON. 55 (1997) (using company-level data, finding payment of wage premium increases labor productivity and lowers operating costs); Carl M. Campbell III, Do Firms Pay Efficiency Wages? Evidence with Data at the Firm Level, 11 J. LAB. ECON. 442 (1993) (concluding from a survey of companies that those with higher wages benefit from increased worker effort); David I. Levine, Can Wage Increases Pay for Themselves? Tests with a Production Function, 102 ECON. J. 1102 (1992) (using PIMS line-of-business data, finding positive correlation between changes in wages and changes in total factor productivity); Peter Cappelli & Keith Chauvin, An Interplant Test of the Efficiency Wage Hypothesis, 106 Q.J. ECON. 769 (1991) (using data across plants within the same firm, finding high wages correlated with lower shirking as measured by disciplinary dismissals). For more qualified support of this theory, see Tzu-Ling Huang, Arne Hallan, Peter F. Orazem & Elizabeth M. Paterno, Empirical Tests of Efficiency Wage Models, 65 ECONOMICA 125, 135 (1998) (concluding that 88% of the productivity effect associated with industry wages can be tied to observable human capital, while only 12% is attributable to wages).

(202.) See Campbell, supra note 201; Levine, supra note 201; Ansar et al., supra note 201.

(203.) See James B. Rebitzer, Is There a Trade-Off Between Supervision and Wages? An Empirical Test of the Efficiency Wage Theory, 28 J. ECON. BEHAV. & ORG. 107 (1995) (finding a trade-off between wages and supervision among contract maintenance workers in the petrochemical industry, and arguing that others who failed to find such a trade-off did not control for omitted features of human resources policies (such as screening procedures) that also influence both wages and supervisory intensity); Erica L. Groshen & Alan B. Krueger, The Structure of Supervision and Pay in Hospitals, 43 INDUS. & LAB. REL. REV. 134-S (1990) (explaining the trade-off between supervision intensity and wages in hospitals). But see David M. Gordon, Who Bosses Whom?: The Intensity of Supervision and the Discipline of Labor, 80 AM. ECON REV. 28 (1990) (concluding that supervision is needed to promote effort regardless of the wage premia); Douglas Kruse, Supervision, Working Conditions, and the Employer Size-Wage Effect, 31 INDUS. REL. 229 (1992) (finding that employee-reported frequency of supervision has a negative relationship to pay); Jonathan S. Leonard, Carrots and Sticks: Pay, Supervision, and Turnover, 5 J. LAB. ECON. S136 (1987) (finding little evidence to support the hypothesis that wages and amount of supervision and inversely related).

(204.) See Abraham, Restructuring, supra note 23, at 114; Abraham & Taylor, Outside Contractors, supra note 23, at 396.

(205.) See Abraham, Restructuring, supra note 23, at 95-96; Abraham & Taylor, Outside Contractors, supra note 23, at 417-18.

(206.) See John Francis Geary, Employment Flexibility and Human Resource Management: The Case of Three American Electronics Plants, 6 WORK, EMPLOYMENT & SOC'Y 251, 261-64 (1992). Note that this observation is more consistent with a gift exchange phenomenon than with the predictions under the shirking model, see text accompanying note 184 supra, that creating a last-hired, first-fired "buffer" class of part-time and temporary workers will increase work effort among primary workers as they calculate a higher expected life-time earnings.

(207.) Although there has been scant attention to the relationship between insider-outsider theories and current debates about contingent employment, there is empirical support for the predictions of insider-outsider models regarding the relationship between wages and unemployment levels. For a review of the European and American empirical literature, see Marcel H.C. Lever, Insider-Outsider Effects in Wage Formation: An Empirical Survey, 47 BULL. ECON. RES. 257 (1995).

(208.) See Samuel Bentolila & Juan J. Dolado, Labor Flexibility and Wages: Lessons from Spain, 9 ECON. POL'Y 53 (1994) (finding that unionized insiders increased their "bargaining power" by threatening to be uncooperative with the temporary workers, and that firms met their wage demands rather than risk a loss of productivity due to harassment effects). They also speculated, however, that temporary workers might have a countervailing effect on union bargaining power to the extent that their resistance to concerted strike activity would undermine the union's ability to use strikes to extract wage increases. See id.; cf. Geary, supra note 206, at 263 (noting that a source of tension between core and temporary workers was the belief within the union that temps would not support collective action). But see Osberg, supra note 62, at 171 (arguing that the extent of harassment of newcomers is not substantial, in the way Lindbeck & Snower propose: "I would wager that the readers of this article do not harass the casual employees with whom they share their workplace on a day-to-day basis.").

(209.) Exceptions include Feldman, supra note 33; Feldman et al., supra note 164. See also studies discussed in note 158 supra.

(210.) See Anthony Barnes Atkinson, Institutional Features of Unemployment Insurance and the Working of the Labor Market, in ECONOMIC ANALYSIS OF MARKETS AND GAMES 82, 82-106 (Partha Dasgupta, Douglas Gale, Oliver Hart & Eric Maskin eds., 1992) (reviewing the implications of increased unemployment insurance benefits under various efficiency wage models); Moira Wilson, Efficiency Wages and Unemployment, 4 INT'L CONTRIBUTIONS TO LAB. STUD. 1, 8 (1994) (explaining that an increase in unemployment insurance increases the relative value of primary sector employment by reducing the cost of termination).

(211.) See Joseph E. Stiglitz, Theories of Wage Rigidio, in KEYNES' ECONOMIC LEGACY: CONTEMPORARY ECONOMIC THEORIES 153, 188 (James L. Butkiewicz, Kenneth J. Koford & Jeffrey B. Miller eds., 1986).

(212.) See Bulow & Summers, supra note 184, at 408 (noting that this effect is driven entirely by the wage-setting mechanism; it will occur even assuming unemployment benefits have no effect on workers' search decisions or employers' layoff decisions).

(213.) See Wilson, supra note 210, at 9.

(214.) See Michael Wachter, Primary and Secondary Labor Markets: A Critique of the Dual Approach, 3 BROOKINGS PAPERS ON ECON. ACTIVITY 637, 673 (1974) (arguing that a criterion of "good" jobs is well-developed internal labor markets, and that legislating higher wages would be inadequate to create "good" jobs in secondary labor markets).

Gillian Lester, Acting Professor of Law, UCLA School of Law. I am grateful to Ben Aaron, Rick Abel, Ian Ayres, Gary Blasi, Grace Blumberg, David Chamy, John Donohue, Robert Goldstein, Laura Gomez, Mitu Gulati, Joel Handler, Alan Hyde, Jerry Kang, Ken Karst, Mark Kelman, Ken Klee, Bill Klein, Jerry Lopez, Dan Mitchell, Deborah Rhode, Rick Sander, Gary Schwartz, Seana Shiffrin, Bill Simon, David Sklansky, Eric Talley, Rip Verkerke, Eugene Volokh, John Wiley, and Steve Yeazell for reading earlier drafts of this article and offering thoughtful criticism and guidance. I am also grateful to the UCLA Junior Faculty Reading Group for helpful critique. Excellent research and technical assistance were provided by Kyung Sin Park, Aaron O'Donnell, Jon Gura, Amman Khan, Karen Mathews, Ethan Millar, Z. Jane Gao, Jenny Sievers, and UCLA's outstanding librarians and computing staff. Generous funding was provided by the Social Sciences and Humanities Research Council of Canada, the Laidlaw Foundation, the John M. Olin Program at Stanford Law School, the UCLA Institute for Industrial Relations, the Labor Relations Fund, the UCLA Dean's Fund, and the UCLA Senate.
COPYRIGHT 1998 Stanford Law School
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Lester, Gillian
Publication:Stanford Law Review
Date:Nov 1, 1998
Words:37740
Previous Article:The confounding common law originalism in recent Supreme Court statutory interpretation: implications for the legislative history debate and beyond.
Next Article:The essential structure of judgment proofing.


Related Articles
Football: Axe for ref Rob.
MANAGER'S GUIDE TO CONTINGENCY PLANNING FOR DISASTERS--SECOND EDITION.
AMERICAN CAREER SEARCH INC.
Football: FAI plan for split in league.
Football: Dons to send out SOS Cal.
New contingency contracting simulation module available.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters