classifying work in the new economy - wharton faculty · zations with temporary, leased, and...

CLASSIFYING WORK IN THE NEW ECONOMY

PETER CAPPELLIJR KELLER

University of Pennsylvania

Alternatives to the archetypal model of full-time regular employment are now bothprevalent and wide-ranging. Over a fifth of U.S. workers, and even more globally, nowperform economic work under arrangements that differ from full-time regular employ-ment. Yet most of our management and social science notions about economic workare based on the full-time employment model. We know relatively little about theoperation and consequences of alternative arrangements in part because while thesearrangements vary considerably, they are commonly grouped together for researchpurposes using existing classification systems. We outline an inclusive classificationsystem that distinguishes clearly between employment and its alternatives. It alsodistinguishes among the alternatives themselves by grouping work arrangementsinto categories that share common properties and that are distinct from each other inways that matter for practice and for research. The classification system is based ondistinctions about the sources and extent of control over the work process, the con-tractual nature of the work relationship, and the parties involved in the work rela-tionship. Our classification system is both informed by and reflects the legal distinc-tions among these categories. We explore implications of our system for research andtheory development.

Textbook accounts of important workplacemanagement topics, such as work attitudes andbehavior, organizational culture, and outcomeslike turnover and job performance, are based onthe full-time employment model and the uniquerelationship that employers have with employ-ees. Alternatives to traditional employment,such as independent contracting, temporaryhelp, and “vendors on premises,” now accountfor as much as 20 percent of the individualsdoing economic work in the United States and ahigher percentage in other countries. There islittle reason to believe those textbook accountsshould apply to the array of availablealternatives.

Furthermore, these various alternatives areoften quite different from each other. Yet moststudies lump them together in categories like“nonstandard work” (literally everything butfull-time regular employment) and “contingentwork” (arrangements that lack the potential per-manence of employment). Such classificationswere designed to distinguish good jobs frombad jobs, but changes in the workforce haveblurred even that simple distinction: high-wage

IT contractors in nonstandard work have betterjobs than low-wage fast-food workers in full-time employment, for example. More recently,the variations among alternatives have becomemore extensive: contracting relationships canlast longer than regular employment, organiza-tions hire temporary workers directly ratherthan through agencies, and employees can workpermanently offsite (in their homes, for example)while contractors can have offices at the client’slocation. As a consequence, distinctions that ap-peared to make sense for classifying work in thepast, such as how long a job lasts, no longerappear useful, hindering our ability to buildknowledge about these new arrangements.

At the same time, scholars are becoming in-creasing aware that the organization of workand the broader context in which work takesplace have tremendous implications for how wethink about individual, managerial, and organi-zational outcomes (Ashford, George, & Batt, 2007;Grant & Parker, 2009; Humphrey, Nahrgang, &Morgeson, 2007). So the need to understand thecontext associated with these growing arrange-ments increases. Yet existing classificationsalso obscure differences that are central to man-agement concerns. Consider the example oftemp work. Temps working for agencies maywork on a full-time regular basis as employees

For meaningful feedback and suggestions, we thank VickiSmith, our three anonymous reviewers, Matthew Bidwell,Adam Cobb, Katherine Klein, and John Paul MacDuffie.

� Academy of Management Review2013, Vol. 38, No. 4, 575–596.http://dx.doi.org/10.5465/amr.2011.0302

575Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyrightholder’s express written permission. Users may print, download, or email articles for individual use only.

of their agency, even though they hop from cli-ent to client; temps who are hired directly, incontrast, are employees of the organizationwhere they work and may perform the same jobeach time, but their days of work may be veryerratic. Both arrangements are called temporaryhelp, but they have little in common. Includingboth in studies of “temporary help” would leadto unmanageable measurement error, and con-clusions from studies of one type of temp workwill not generalize to the other.

The absence of clear distinctions among thesealternatives has left scholars interested instudying them without an accepted and sharedvocabulary, impeding communication and accu-mulation of knowledge (Lodahl & Gordon, 1972;Pfeffer, 1993; Salancik, Staw, & Pondy, 1980). Per-haps to avoid having to use what Osnowitz(2010: 10) describes as nebulous concepts like“contingent work,” most studies of alternativearrangements focus on a single arrangement,such as independent contracting (Ashford et al.,2007; Connelly & Gallagher, 2004). There is littleto guide us on the external validity of such stud-ies, however—whether their conclusions applyto any other type of arrangement and, if so,which ones. Nor do we have any guide for form-ing hypotheses across these different arrange-ments on basic questions such as what factorsexplain individual work attitudes and perfor-mance. Consider, for example, the challenge ofstudying the job performance of truck driversworking for a typical shipping company wheresome are full-time regular employees, some areindependent contractors who own their trucks,and some are leased employees under contractto an agency (Belzer, 2002). These different ar-rangements certainly call for separate hypothe-ses. Indeed, the influence of those arrangementsmay overwhelm the importance of the commonattributes of job tasks.

What we need to make progress on under-standing these important and growing alterna-tive arrangements is a classification system fororganizing the various arrangements throughwhich individuals can be engaged in economicwork. As we describe below, lessons from thephilosophy of science suggest that a good clas-sification system should organize elements intocategories where their important attributes aresimilar, where these attributes are distinct fromthose of other categories, where those differ-ences are discrete (rather than continuous), and

where the attributes relate to issues of interest(in this case to management research). We usethese criteria as a means to evaluate currentclassification systems and find them to fall shorton a number of dimensions.

We then develop an inclusive classificationsystem of arrangements for economic work—defined as activity undertaken for another partyin exchange for compensation—that meetsthose objectives. Presented in Figure 1, our sys-tem is based on the theoretical construct of con-trol—specifically, how control over the work pro-cess governs the relationship between theworker and the organization that benefits fromthe worker’s efforts. The most important sourceof that control and how it constrains the organi-zation stems from underlying legal frameworksthat determine the nature of the relationshipand, in turn, a great many attributes about howthe organization can “manage” the workers per-forming the tasks. We outline a classificationsystem that begins with the distinction betweenemployment and contractual arrangements,which are underpinned by employment law andcontract law, respectively. The employer haswell-known powers but also obligations to em-ployees associated with employment law, whilethe client and worker in contractual arrange-ments have a market-based arrangement out-lined in the contract and enforced by the courts.The important distinction within each of thesetwo categories has to do with whether the rela-tionships are direct or triangular in the sensethat they involve third parties such as tempagencies.

While developments in the workplace haveblurred the distinctions created in previous at-tempts to classify work arrangements, legalframeworks create homeostatic pressures tosustain the distinctions across these categories.Most important, this system reveals unique hy-potheses linking factors and outcomes in eachcategory, as well as comparisons across ar-rangements that allow effective tests of thoserelationships.

Prior to describing the most commonly usedclassification systems and describing our at-tempt to address their shortcomings, we brieflydetail the growing use of alternative arrange-ments, since the usefulness of our taxonomyrests in part on how prevalent these arrange-ments are in practice.

576 OctoberAcademy of Management Review

HOW EXTENSIVE AREALTERNATIVE ARRANGEMENTS?

Surveys of employers/organizations providemore accurate reports of the incidence of alter-native arrangements than do surveys of individ-uals because employer respondents are morelikely to understand the distinctions betweendifferent arrangements and can report how often

work is actually being performed. Yet such sur-veys are surprisingly rare.

The most detailed and still the most recentgovernment data on these issues comes from the2000/2001 National Employer Survey, whichasked firms several questions about their use ofalternative arrangements for economic work(see Table 1). Altogether, the percentage of indi-viduals working for an organization at its loca-

FIGURE 1Classification of Economic Work Arrangements

2013 577Cappelli and Keller

tion in a typical week but not full-time regularemployees was 14 percent. These included part-time workers, temps hired directly by the em-ployer, temps hired through agencies, indepen-dent contractors, workers employed throughprofessional employee organizations (PEOs),and the employees of vendors working at one’sestablishment. In a “peak” week, the use of tem-porary helped push that figure up to 20 percent.A more recent survey, albeit with a less repre-sentative sample (Dwyer, 2011), found thatnearly 22 percent of the average respondent’sworkforce consisted of nonemployees. Employ-ers in that survey reported difficulty in evenkeeping track of the array of alternative ar-rangements they were using, as well as in think-ing through how best to choose and then man-age those arrangements.

While many of these alternative arrange-ments began in the United States, they are ac-tually more prevalent in other countries. For ex-ample, temporary help alone constitutes14.5 percent of the European workforce, 9.0 per-cent of the workforce in the G7 countries, and12.1 percent of the workforce in the OEDC mem-ber countries. The percentage of temporaryworkers also grew faster in the first half of the2000s in Europe (14 percent), the G7 countries (6.7percent), and the OEDC countries (10.7 percent)than in the United States (4.7 percent). Mean-while, the staffing sector has grown into a$300 billion-plus worldwide industry, with 72,000private employment agencies providing organi-zations with temporary, leased, and temp-to-

perm workers working as many hours as 9 mil-lion full-time employees (CIETT, 2011). Suchagencies alone provide 1.3 percent of all workerhours in the United States, 1.5 percent in Europe,and 1.7 percent in Japan.

The use of alternative arrangements, espe-cially temporary agency work, increasessharply when economies are recovering fromeconomic downturns and tends to remain highwhen growth returns (Luo, Mann, & Holden, 2010;Peck & Theodore, 2007). The percentage of workbeing performed under alternative arrange-ments may therefore continue to increase, un-derscoring the need for a taxonomy that willhelp us to develop theories able to take intoaccount the difference among these alternatives.

AN APPRAISAL OF PRIORCLASSIFICATION SYSTEMS

One approach to investigating questions re-lated to alternative arrangements is to treateach arrangement as unique and then examinehow it operates. Indeed, much of the existingliterature consists of such studies. One downside to this sui generis approach begins with thefact that the number of arrangements keeps ex-panding and is already quite large, so a man-ageable set of findings about alternative workoverall becomes ever more difficult to assemble.A simpler approach is to begin with a classifi-cation system of these alternative arrangementsthat will reveal the most important similaritiesand differences across them in a shorthandmanner. Such a system greatly simplifies pro-grams of empirical research by suggestingwhich questions should be most interesting forwhich arrangement, what lessons from specificarrangements should be applicable where, andhow comparisons across specific arrangementsmight create useful quasi-experiments for futureresearch.

In order to begin this process, we draw on animportant but underused body of literaturespanning philosophy and the social sciencesabout what makes a good classification system.This helps us understand exactly why prior clas-sification systems fall short of accomplishingthese goals, the problems these shortcomingspose for research and theory development, andhow a carefully constructed classification sys-tem can help overcome these problems.

TABLE 1Use of Alternative Work Arrangements by

Organizationsa

Variable N Mean

Part-time employees 2,943 6.67%Direct hire temporary workers 2,944 2.77%Agency temporary workers 2,949 2.14%

(8.75% in peakweek of year)

Independent contractors 2,793 1.49%Professional employee

organization (PEO) workers2,851 0.87%

Onsite vendor workers 2,818 0.86%Average week total 2,911 14.20%Peak week total 2,793 20.81%

a Source: Authors’ calculations from the 2000/2001 Na-tional Employer Survey.


Evaluating Classification Systems

While there is ongoing debate as to the properapproach to take in constructing a classificationsystem,1 there is agreement on what makes for agood classification after the fact. Bird and Tobin(2010) offer a list of criteria asserted by differentauthors over time, several of which are violatedby what are arguably the three most influentialclassifications developed to date—those thatclassify arrangements as (1) regular or contin-gent, (2) standard or nonstandard, and (3) inter-nalized or externalized.

The first and most fundamental criterion isthat there should be much more similaritywithin the categories than between them(Quine, 1969).2 This occurs when the members ofa category share more properties, allowing forinduction within the category. This is central toestablishing external validity, ensuring thatwhat we what we say about one member of acategory is broadly true of the other members.Second, effective classification systems requirewell-defined boundaries, so the properties usedto organize individual units into categoriesshould represent clear distinctions from eachother, as opposed to having a smooth transitionbetween them. In fact, there is no need to create

a classification system where differences arebased on continuous properties, because we al-ready know how each individual unit corre-sponds to another. Taken together, these criteriaalso suggest that a good classification systemshould form a hierarchy where subcategoriesare nested within larger categories.3 Third, agood classification system must exhibit proper-ties that cause it to persist. A general concernabout any classification system is whetherchanges in the environment will soon render itobsolete. Categories exist and persist in naturethrough homeostatic properties—mechanismsthat essentially enforce the categories by ensur-ing that deviations have a low chance of persist-ing (Boyd, 1999), essentially enforcing the sys-tem and allowing it to persist over time.

The most basic concern with all prior classifi-cation systems of alternative work is that themembers within the created categories eithernever were or are no longer very similar to eachother. Nor are they distinct from members inother categories. They also present attributesthat are best seen as points on a continuum,such as job security, rather than discrete out-comes. A further concern is their lack of homeo-static properties, evident from the observationthat in all three cases, changes in employmentover time have blurred the differences betweencategories, eroding their usefulness in the con-text of modern workplaces. Each of these con-cerns represents a violation of an important cri-terion of successful classification systems,leading to problems for research and theory de-velopment. Moreover, these prior systemswere not designed to inform research issues as-sociated with management. We review thembriefly below.

Contingent Work

The rise of agency-based temporary help inthe late 1980s led to a descriptive term for sucharrangements—contingent work (Freeman,

1 The central debate in creating such systems is betweenrealism, the perspective that classifications should and canbe based on differences that occur in reality or nature, andconventionalism, the perspective that classifications inpractice are determined by the interests of those doing theclassifying—the equivalent to social construction argu-ments in the social sciences. These positions are not abso-lute, however, and especially within the conventionalist po-sition, there are gradations of views. Perhaps the mostmainstream view is that of causal dependence (Kukla,2000)—that while our classification systems may be basedon reality, the choices that we make when we create cate-gories and assign individual cases to them create an orderthat is not necessarily objective. Dupré (1993) advanced theidea that there may be many systems of classification for thesame phenomena, depending on what one’s interests are—asituation he refers to as promiscuous dependence. In biol-ogy, for example, we classify animals based on their evolu-tionary roots (phylogenetic systems) or on their attributes(e.g., mammals versus fish), as with the Linnaean system.

2 Organizing items based on color, for example, would beconsistent with this criterion. This criterion, however, is onlya necessary and not a sufficient condition for a good classi-fication system (Bird & Tobin, 2010). A grouping of otherwiseunrelated items by color, for example, would not tell usmuch that is useful about the members of that group be-cause those members could differ on so many otherproperties.

3 For example, all rational numbers can be expressed asa fraction of two integers, and integers and nonintegers aredistinct subcategories of rational numbers. Both share theproperty of being expressible as the fraction of two integers,but they are distinct in the sense that integers can be writtenwithout a fractional or decimal component whereas nonin-tegers must be expressed using a fractional or decimalcomponent.


1996). This reasonably simple distinction be-tween “regular jobs” and contingent work wasbased on the notion that temporary jobs allshared the common property of being insecureand contingent on short-term business needs.Over time, the contingent work classificationcame to include agency temps, direct hiretemps, independent contracting, part-time work,and, in some accounts, everything except full-time regular employment (Belous, 1989; Blank,1998). The lack of consensus as to what arrange-ments fall under the category of contingent workis a primary reason scholars have come to la-ment that “as a concept, contingent work lacks aclear definition” (Osnowitz, 2010: 10).

One problem with this classification is thatjob security is not a discrete variable. It existsalong a continuum from secure to insecure, pre-venting us from making clear distinctionsamong arrangements in their degree of security(e.g., when does a job move from “secure” to“insecure”?). Moreover, the workplace hasevolved such that layoffs and associated re-structurings have made many if not most regu-lar jobs less certain and secure. At the sametime, alternative arrangements in some contextsmay be reasonably secure and long term, espe-cially leased employment. The primary propertyused to distinguish between regular and contin-gent work arrangements—insecurity—no longerprovides a useful criterion to classify work.

Furthermore, those arrangements that fall inthe contingent category— contractors, temps,leased employees— do not necessarily havemuch in common with each other. Indeed, a re-cent review of the literature on contingent workarrangements led Connelly and Gallagher toconclude that although it is often treated assuch, “one compelling finding is that contingentwork is not a unitary concept” (2004: 979).

Nonstandard Work

Arguably, the most common classification ofwork arrangements is the distinction betweenstandard jobs, representing full-time employ-ment with long-term expectations, and nonstan-dard jobs, which include everything else. Theproblem here, as with contingent work, is thatnonstandard jobs have little in common witheach other, other than not being standard jobs.Here again, developments in the workplace mayhave outrun the classification system. Kalle-

berg, Reskin, and Hudson (2000: 257) originallydefined standard jobs as work performed on afixed schedule (usually full time), at the firm’splace of business, under the firm’s control, andwith the mutual expectation of continued em-ployment. But regular employees now oftenwork from home and with flexible schedules—indeed, often to meet their own needs—and ex-pectations of continued work can be just asgreat for contractors as for regular employees.

Standard jobs have generally been equatedwith good jobs (e.g., Kalleberg et al., 2000). Yetthe meaning of terms such as standard (Ashfordet al., 2007) and good (Smith & Neuwirth, 2008) issurprisingly fluid, leading others to observe thatit is probably no longer useful to attempt tocategorize jobs with such a simple dichotomy(Osnowitz, 2010; Smith, 2001a). Nonstandard ar-rangements, like the various forms of IT contractwork described by Barley and Kunda (2004), canbe good jobs in the sense that they pay very welland give the individual real control over theterms and conditions of work, as well as how thework is performed; at least some part-time workis voluntary and chosen because it meets thework schedule needs of the individual workers(Kalleberg, 2000; Polivka, 1996); and some full-time regular jobs are bad jobs, characterized bylow pay, low security, bad conditions, and so on.

Externalized Work

Pfeffer and Baron (1988) were among the firstto think carefully about alternative work ar-rangements, and their work continues to informmuch of the current thinking on the subject (e.g.,see Ashford et al., 2007). They focused specifi-cally on understanding the movement of workbeyond the boundary of the firm, especially out-sourcing, as opposed to understanding alterna-tive work arrangements per se, although theframework has been used as such. They pro-posed that externalization exists on a contin-uum along three dimensions: (1) the locationwhere the work is performed, (2) the expectedduration of the employment, and (3) the source orlocation of the administrative control overthe work.

As with contingent work and nonstandardwork, the evolution of practice has eroded theusefulness of this taxonomy. Again, locationmay no longer reveal much when employees areworking from home on a more or less permanent


basis and vendors and contractors have officesat their client’s location. Expectations about ten-ure are likely to be highly uncertain, and, asnoted above, such expectations would be a con-tinuum that does not allow clear distinctions.Their third criterion of control, however, is cen-tral to our classification as well.

Problems for Research andTheory Development

These shortcomings have led to confusion inthe existing literature and, as others have ob-served, have stunted our ability to develop, test,and extend management theory (Ashford et al.,2007; Connelly & Gallagher, 2004). Research onalternative arrangements has primarily con-sisted of applying theories and constructs devel-oped in the context of traditional employmentrelationships to alternative arrangements, oftenwithout careful consideration of whether theirmeaning and application should remain thesame in different contexts (Ashford et al., 2007;Connelly & Gallagher, 2004: 977). Moreover, inthe work that does consider such contextual dif-ferences, scholars have had to rely on the overlybroad distinctions provided by previous classi-fication systems—distinctions no longer appli-cable to modern workplaces.

The inability of existing classification sys-tems to distinguish among different types of al-ternative arrangements leads to three sets ofproblems. The first has to do with the externalvalidity of studies of alternative work arrange-ments. As noted above, recent empirical workhas tended to focus on a single alternative ar-rangement (Ashford et al., 2007). For example,Osnowitz (2010) detailed the experiences of in-dependent contractors, while Barley and Kunda(2004) focused primarily on contract workers af-filiated with brokers, and Bauer and Truxillo(2000) examined workers in temp-to-perm ar-rangements. Lacking an understanding of howthese (and other) alternative arrangements aretheoretically similar and distinct from eachother makes it difficult to evaluate the externalvalidity of such studies. To what extent, for ex-ample, should we expect Osnowitz’s (2010) find-ings about contractors to apply to direct hiretemporary workers or employees working underPEO arrangements?

In a similar vein, Connelly and Gallagher(2004) suggested that the lack of consistent find-

ings in research on contingent work about out-comes such as commitment, job satisfaction,organizational citizenship behavior, and knowl-edge sharing may be explained by the hetero-geneity among arrangements lumped togetherunder the umbrella of contingent or nonstan-dard work. Moreover, given that existing frame-works tend to group all such alternative ar-rangements together, comparisons amongalternative arrangements are few and far be-tween, although ethnographic studies suggestthat that there are significant differences amongsuch arrangements. Addressing similar con-cerns, Ashford and colleagues (2007: 99) havecalled for researchers to be clearer about howtheir sample of workers in alternative arrange-ments differs theoretically from workers in otheralternative arrangements, yet existing classifi-cation systems provide no guidance for identi-fying such differences.

Second, the inability to distinguish theoreti-cally among alternative arrangements limitsour ability to understand how organizations de-cide which arrangements to use and in whichcombinations (Huselid & Becker, 2011). Existingresearch has tended to focus on the use of alter-native arrangements in general, primarily at-tributing the use of such arrangements to thedesire for internal and external flexibility (e.g.,Cappelli & Neumark, 2004; Kalleberg, 2000;Smith, 1997). However, the few studies wherescholars have asked organizations directlyabout their use of different alternative arrange-ments (Abraham, 1990; Houseman, 2001; Kalle-berg, Reynolds, & Marsden, 2003) suggest thatdifferent rationales drive the use of each ar-rangement. Lepak and Snell (1999) provided themost general theoretical model for understand-ing when organizations will select among avail-able employment strategies, although it too ismore useful for determining when organizationswill choose employment versus its alternatives,rather than how organizations decide amongthe alternatives themselves.

The third issue has to do with our ability tounderstand how managers handle the basictasks of managing workers under different ar-rangements, especially when those arrange-ments overlap in the same organization. Groupsof individuals often work alongside each otherwhile performing essentially the same tasksand yet have very different arrangements gov-erning their work (e.g., see Bidwell, 2009; Bro-


schak & Davis-Blake, 2006; Davis-Blake, Bro-schak, & George, 2003; Smith, 2001b). As Smith’s(2001b) study of workers from temp agenciesworking alongside employees in a high-techfirm revealed, this significantly complicates thejobs of managers, in part because even whenpolicies are in place to differentiate betweenworkers in different arrangements, such workersoften end up being highly integrated (Barley &Kunda, 2004; Bidwell, 2009).

For example, while the majority of registerednurses (RNs) working in hospitals are full-timeemployees, roughly 20 percent work part time,doing the same work only for fewer hours perweek. Hospitals also use per diem nurses fromagencies; these nurses work on a per shift basislike temps. Traveling nurses, yet another ar-rangement, are more like “leased” employees,employed by agencies and provided to hospitalsfor engagements lasting as long as a year. Thepercentage of nurses providing care under thesealternative arrangements has nearly tripledover the past decade and is expected to increase(KPMG, 2011). A nontrivial percentage of nursesalso work as independent contractors, negotiat-ing their own contracts and maintaining theirown liability and indemnity insurance. RNsworking in smaller health care facilities are of-ten employed through a PEO. It is common fornurses engaged through these different chan-nels to work side by side and report to the samemanager. And the overlap of nursing arrange-ments in hospitals and among nurses is notunique, since such situations are increasinglycommon in settings as diverse as over-the-roadtrucking companies (Belman, Monaco, & Brooks,2005) and law firms (Sherer & Lee, 2002).

In addition to shaping the interactions be-tween workers and managers, the context of thework arrangement is also likely to shape howworkers interact and develop relationships withthose under similar and different arrangements(Barley & Kunda, 2004; Lautsch, 2002; Osnowitz,2010). The use of alternative arrangements thusalters the social context of work, which, in turn,plays a critical role in shaping the behaviorsand experiences of workers (Grant & Parker,2009: 322) and managers. Our inability to distin-guish between employment and alternatives toemployment, as well as among the alternativesthemselves, prevents us from developing a com-prehensive set of theories as to how such ar-rangements shape the social context of work

and how workers are managed under those dif-ferent arrangements.

The above discussion suggests that a theory-driven classification system ensuring more sim-ilarity within categories than between them,while also providing clear, discrete distinctionsamong categories, should go a long way in ad-dressing these shortcomings. And given thatchanges in employment over time have largelyrendered prior systems obsolete, a classificationsystem that is likely to persist over time despitechanges in the environment is highly desirable.We develop such a system in the followingsection.

AN ALTERNATIVE CLASSIFICATION SYSTEM

The place to begin constructing any classifi-cation system is with the attributes of the indi-vidual members being mapped, with the keyidea being that the members of each categoryshare properties and there will be much greatersimilarity of individuals within categories thanacross categories. Table 2 provides descriptionsof the wide variety of arrangements currentlyused by organizations to engage employees ineconomic work. These are compiled from awidespread review of previous research (pri-marily in the fields of management, economics,sociology, and legal studies), in addition to le-gal statutes, government reports, various indus-try reports, and our own extensive experiencesworking, teaching, and consulting with a largevariety of organizations.

For our purposes here, the shared propertiesshould be those that inform concerns in the field ofmanagement. And for this reason, our approach toclassification is based on the concept of control. Inthe context of economic work, control is generallydefined as the ability of organizations and/ormanagers to obtain desired behavior from work-ers (Edwards, 1979: 17). Control is arguably themost fundamental issue in the workplace, withScott and Davis asserting that nearly all topics ofinterest to management scholars “relate more orless directly to the subject of control” (2006: 202).This is especially true in the context of work rela-tionships, where a long line of literature predatingeven Marx (Marx & Engles, 1992/1888) has exploredwho has the power to make decisions in the work-place (Bendix, 2001/1956). And the extent of controlemployers have over workers, the source of andconstraints on that control, and how control (or


TABLE 2Work Arrangements Available to Organizations

Arrangement Description

Full-time employees • Work for the organization that both employs them and uses their services• Organization controls both the work process and work outcome• Organization is responsible for most regulatory requirements, including payroll andemployment taxes, in addition to managing the screening, hiring, wage setting, andtermination

Part-time employees • Work for the organization that both employs them and uses their services• Organization controls both the work process and work outcome• Organization is responsible for most regulatory requirements, including payroll andemployment taxes, in addition to managing the screening, hiring, wage setting, andtermination• Distinction between full and part time is based on number of hours worked, withpart-time workers generally working less than 35 hours per week; benefits aretypically less, as are career prospects• Typically stay in the same job, often the same job as full-time employees

On-call employees • Work for the organization that both employs them and uses their services• Organization controls both the work process and work outcome• Organization is responsible for most regulatory requirements, including payroll andemployment taxes, in addition to managing the screening, hiring, wage setting, andtermination• Do not have a regular schedule for work; typically called into work only whenneeded, although they can be scheduled to work for several days or weeks in a row• Required to be available during certain on-call periods; not treated as a “regular”employee

Direct hire temporary employees • Work for the organization that both employs them and uses their services• Organization controls both the work process and work outcome• Organization is responsible for most regulatory requirements, including payroll andemployment taxes, in addition to managing the screening, hiring, wage setting, andtermination• Perform work as a short-term supplement to a firm’s workforce (includes seasonalworkers)• No expectation of continuing relationship

PEO • PEOs are administrative organizations that engage in a contractual relationshipwith their client organizations• PEOs do not provide labor to work sites; rather, an organization transfers its currentworkforce to the employ of a third-party agency: the workers stay put, typically doingthe same tasks with similar terms and conditions of employment, but legalresponsibility for the employees is transferred to the PEO• PEOs maintain an employment relationship with their employees that is intendedto be long term and not temporary• PEOs share/allocate employer responsibilities in a manner consistent withmaintaining client responsibility for their product or service• PEOs typically provide benefit plans for the workers and provide human resourceservices to the work-site employer

Leased employees • Employed by a staffing firm or agency• Longer engagements than agency workers, typically working alongside employeesin skilled roles• Client controls the work outcome, while the leasing firm reserves the right to directand control workers• Client’s direction and control over the work process is limited to the extentnecessary to conduct the client’s business; in general, workers must follow the workrules and regulations stipulated by the client in contracts for which services arebeing provided• Workers return to the leasing firm for reassignment after completion of their workwith the client• Agency is responsible for most regulatory requirements, including payroll andemployment taxes, in addition to managing the screening, hiring, wage setting, andtermination

(Continued)


lack thereof) affects managers and workers con-tinue to be subjects of extensive theorizing inmanagement and sociology (Kunda, 1992; Mars-den, 1999; Perlow, 1998; Wright & McMahan, 1992).Moreover, control emerges as an important con-sideration within nearly every study and review ofalternative work arrangements (e.g., Kalleberg,2000). Yet there has been no systematic analysis ofhow control differs across the various arrange-ments for governing economic work, the mecha-nisms imposing such variation, or how such vari-ations may shape the attitudes, behaviors, andoutcomes for the individuals, managers, and or-

ganizations engaged in arrangements character-ized by variations in control.

Directive Control

Control within formal organizations is, atleast in part, structural (Brass, 2002; Scott & Da-vis, 2006). In the work context, structure emergesfrom the way work is organized (Goldberg, 1980).Edwards (1979) identified three dimensions ofcontrol arising from the organization of work:direction, evaluation, and discipline. Directivecontrol refers to the “mechanism by which the

TABLE 2(Continued)

Arrangement Description

Agency temporary workers • Employed by a temporary agency• Agency provides the worker to client organizations for what are usually short-termassignments• Client controls the work outcome, while the agency reserves the right to direct andcontrol workers• Client’s direction and control over the work process are limited to the extentnecessary to conduct the client’s business• Work is typically performed at the client’s location• Workers return to the agency for reassignment after completion of their work withthe client• Agency is responsible for most regulatory requirements, including payroll andemployment taxes, in addition to managing the screening, hiring, wage setting, andtermination

Independent contractors • Work for themselves—are not employees• Contract directly with clients• Work outcome is specified by the client in the contract, but the work process iscontrolled by the worker• Typically hired on a project basis, although the scope and length of projects mayvary considerably• May work for multiple clients simultaneously and for various lengths of time• Independent contractors are responsible for paying their own taxes and securingbenefits

Day laborers • Day laborers are technically contractors hired on a one-job basis, typically lastingbetween one and three days, although they can be hired repeatedly by the sameorganization• Pay and tasks are often negotiated after the worker is selected, with informalnegotiation sometimes continuing at the job site depending on circumstances• Absence of formal contracts• Payment is often in cash, evading taxes

Vendor-on-premises • An organization contracts with a vendor to perform work• Individuals who perform the work are either employed by or are engaged asindependent contractors by the vendor• The work process is directed by the vendor• A key element of such contracts is the notion of a “work product”; the vendor iscontracted to provide a service, but how the service is provided is up to the vendor• Vendor is responsible for most regulatory requirements, including payroll andemployment taxes, in addition to managing the screening, hiring, wage setting, andtermination• Relevant context is where work is performed at client’s location while engagingwith client’s employees and management


employer directs work tasks, specifying whatneeds to be done, in what order, with what de-gree of precision or accuracy, and in what pe-riod of time” (Edwards, 1979: 18). It is about con-trol over the work process, which managersmight describe as equivalent to supervision. Anorganization’s ability to decide and direct de-tailed work assignments has repeatedly beenidentified as the defining feature of the employ-ment relationship (e.g., Marsden, 1999: 3).

The extent of directive control also appears tobe the dimension that differs most across thevarious arrangements for organizing economicwork. As we describe below, under certain ar-rangements the organization that engages theworker completely determines how the workgets done; in others the control over the workprocess is shared with a third party and, thus, islimited to the extent necessary to conduct theorganization’s business; and in still others theworker retains complete control over how thework gets done. As we describe below, differ-ences in the evaluation and disciplinary aspectsof control follow in almost all cases from theextent of directive control exerted by the organi-zation.

Consistent with its importance in practice,control over the work process and who exercisesit is also of central concern in the legal system.The legal framework for employment is surpris-ingly consistent across the world. A recent re-view of national employment laws in over sixtycountries conducted by the International LabourOrganization concluded that

the employment relationship is a legal notionwidely used in countries around the world to referto the relationship between a person called anemployee and an employer for whom the em-ployee performs work under certain conditions inreturn for remuneration. It is through the employ-ment relationship, however defined, that recipro-cal rights and obligations are created betweenthe employee and the employer (2006: 3).

Employment law virtually everywhere distin-guishes employment relationships from nonem-ployment arrangements based on the extent towhich the employer is able to control the workprocess (Befort, 2003; Stone, 2006). This distinc-tion is clearest in developed economies, such asthose of the United States and European Unioncountries, where enforcement is most vigilantand the courts regularly settle employment dis-putes. New legislation and its enforcement and

interpretation by the courts have led to furtherdistinctions among different types of employ-ment relationships, as well as distinctionsamong the various alternatives to employment.As a result, although there is no single set ofemployment laws in the United States and othercommon law countries, consistent principlesacross legislation and case law exist that informa robust classification system grounded in theconcept of directive control.

Most important, the law makes two key dis-tinctions that influence directive control. Thefirst relates to the presence or absence of a thirdparty. Economic work arrangements typicallyinvolve two parties, but new arrangements inthe workplace sometimes involve three parties.The law treats these triangular arrangementsdifferently from direct relationships, effectivelyconstraining organizations’ ability to exert di-rective control within certain third-party ar-rangements (Pfeffer & Baron, 1988). The secondis the contractual nature of the work arrange-ment. Employment relationships are governedby employment law and involve an open-endedcontract where few terms are specified in ad-vance. Employers have tremendous latitude todirect virtually everything employees do in theworkplace. In contrast, many of the new ar-rangements are not employment and insteadrely on closed contracts. Because these contractsare governed by contract law, most if not allterms are negotiated in advance, either fullyproscribing directive control or limiting thescope of control over the work process (Muhl,2002). The number of parties involved and theemployment law versus contractual law natureof the work arrangement help build the founda-tion of our classification system (Figure 1).

Employment

Within the context of an employment relation-ship, the law grants the organization a greatdeal of control over the work process. The em-ployer may tell the employee not only what to dobut how to do it.4 In all other arrangements theorganization’s ability to direct the work process

4 The key point we make here is that organizations areable to exert full directive control. To what extent and underwhat conditions they actually exert full control is a differentquestion that we leave to others (i.e., Coase, 1939; Marsden,1999; Simon 1947, 1951).


is completely proscribed, limited to specifyingthe end product or final result of the work as partof a contractual agreement. The organizationmay dictate what to do, but the worker retainscomplete control over how to do it (Muhl, 2002).Control over the work process thus provides abright line distinction between employment andall other arrangements. Our classification sys-tem therefore begins by distinguishing employ-ment from contract work, which we discuss inmore detail below.

An employment relationship may involve athird party, such as an agency that employs aworker who then performs work for a client ofthe agency. In such arrangements the organiza-tion shares control over the work process withthe third party. The presence or absence of athird party thus creates subcategories withinthe broader category of employment— directemployment and coemployment. The distinc-tions within types of direct employment andwithin types of coemployment are more modestthan the distinctions between direct employ-ment and coemployment— hence both thenested and hierarchical aspects of the system.We begin by discussing the subcategories ofdirect employment and then move on to discuss-ing coemployment and its subcategories.

Direct employment. As noted above, directemployment creates a relationship where theemployer has almost complete control over whatand how work is done. Employers retain thepower to change standard operating proceduresand daily tasks or to reconfigure the productionprocess (Burawoy, 1979; Edwards, 1979). An em-ployer can tell an electrician employee to stopdoing electrical work and paint a fence. If theywant, employers can tell a computer program-mer to write code in a manner that violatesstandard protocols.

In return for such a wide scope of directivecontrol, employers must comply with an exten-sive array of legislation and regulations thatgovern other aspects of their relationship withthe employee. The direct and administrativecosts of these requirements are considerable.One estimate suggests that the paperwork costsalone of complying with the legal obligations ofthe employment relationship are more than$1,700 per employee each year (Crain & Johnson,2001). In part because the costs to both employerand employee of establishing and ending theserelationships make them costly to break, there is

also a presumption that most direct employmentrelationships will be relatively long-lived, andthey are therefore characterized by open-endedcontracts where few terms are negotiatedex ante.

The expectation of continued employment,which emerges in large part from the trade-offsassociated with employers’ assuming directivecontrol, provides employers with a great manytools to get employees to comply with theirwishes that are unavailable in other arrange-ments. For example, they can screen applicantson a wide range of issues before hiring them asemployees, including such attributes as intelli-gence and conscientiousness, as well as knowl-edge, skills, and ability.5 They are able to so-cialize new hires into the norms and values ofthe firm, transferring the culture of the organi-zation, and can train, develop, and observe in-dividuals before putting them in charge of anestablishment. They can also motivate andmanage employees with long-term rewards andpromotion prospects. Performance appraisals,which in part are also forward looking, are an-other practice unique to employment (see be-low). From the perspective of the organization,direct employment is the most responsive ar-rangement for managing work.

Nested within the category of direct employ-ment are subcategories of relationships thathave been treated as unique in most prior re-search. While they all share the properties men-tioned above, they differ in part on the expecta-tion of continuous employment and consistentterms and conditions of work. Part-time workersare in many ways like full-time regular employ-ees, but they typically do not have career ad-vancement prospects and tend to work in thesame job throughout their tenure. Managementpractices associated with development and pro-motion rarely apply here, so we might expectthese workers’ connection to the organization, aswell as their attitudes and behaviors, to be dif-ferent than for full-time employees (at leastthose who have advancement prospects).

On-call workers may have a reasonably long-term relationship with their employer but haveless predictable schedules and, overall, may not

5 A client could try a similar approach before engagingcontractors, but it would run the risk of turning the contrac-tual relationship into an employment relationship the moreit applied traditional employment practices to the role.


work many days for the employer. We mightexpect them to have a different relationshipthan regular or even part-time employees. Hir-ing practices are less rigorous than for full-timeor part-time employees doing similar work, butthe jobs may be more specific and central to theoperation than temp work, therefore requiringsomething of a continuing relationship. So wemight see more training and more managementthan for temp jobs.

Substitute teachers are a prime example ofon-call workers, since they are employed byschool districts to replace full-time teachers onan as-needed basis for reasons such as personalillness, personal emergencies, attendance atworkshops, and the like. Substitute teachers arerarely able to predict when they will be neededand are called upon with little advance notice;one Texas school district proudly announcesthat “approximately 50% of all [substitute teach-ing] assignments are scheduled with at leasttwo days advance notice” (Texarkana Indepen-dent School District, 2012: 2). While school dis-tricts tend to have training and hiring programsin place for substitutes, they lack the consis-tency and sophistication of those programs inplace for full-time teachers (Ostapczuk, 1994).Substitute teachers can have long-term relation-ships with a particular school district, however.

Direct hire temps typically have short tenurewith an employer and work in simple, standard-ized jobs with few expectations of a continuingrelationship, such as call center workers hiredon a short-term basis to meet temporary in-creases in demand (Chambel & Castanheira,2006). Seasonal jobs, such as retail clerks hiredfor the Christmas holidays, are perhaps the bestknown direct hire temp role. For the most part,the employees arrive at the job with the skillsthey need to perform it. We see few managementpractices here except for direct supervision.

Coemployment. Coemployment adds a thirdparty to the traditional employment relation-ship. In the most common arrangement theworker is hired and employed by the third party,often a staffing agency, which, in turn, contractswith a client organization to provide the workerfor a specific assignment. Agency temps andleased employees are the most common exam-ples. The client negotiates the specific terms ofthe arrangement (start date, end date, termina-tion clause, pay, etc.) with the agency, and these

terms are used to establish a contractual rela-tionship between the client and the agency.

The agency is the employer of record and as-sumes a series of legal obligations, includingresponsibility for most regulatory requirements,such as payroll and employment taxes, in addi-tion to managing the screening, hiring, wagesetting, and termination of employment (Bidwell& Fernandez-Mateo, 2008). In exchange for pass-ing these largely administrative obligations tothe third party, the client cedes some controlover the work process, sharing directive controlwith the agency.

Although the third party reserves the right todirect workers, it seldom exercises this rightonce the worker has been assigned to the client(Pfeffer & Baron, 1988). Rather, the extent of di-rective control granted to the client is negotiatedex ante, with directive control limited to thatwhich is necessary to conduct the client’s busi-ness—in other words, within the broad goals setout in the contract. This shared directive controlis accompanied by a shared responsibility toprovide workers with a variety of protectionsassociated with the immediate workplace:health and safety, antidiscrimination, and fairtreatment as interpreted by the courts. Workerscan bring suit against both the agency and theclient organization for violations of employmentregulations.

The difference between the various arrange-ments in the coemployment hierarchy centers onthe nature of the work assignments. Agencytemporary workers are typically provided forshort assignments where the work is reasonablystandardized (e.g., clerical positions) or highlyskilled (e.g., computer programmers), and thetemps come in with the skills needed to do thejob. Leased employees are engaged in longer-term projects, some many years, and are morelikely to work on projects unique to the clientand central to its business. For example, Barleyand Kunda (2004: 69–71) chronicled how com-puter program developers are often engaged byfirms, through agencies, to work on large-scaleprojects lasting long periods of time, such asdesigning and implementing entire informationsystems or providing technical support for criti-cal applications on a continuing basis.

PEO arrangements transfer the employmentrelationship for one’s current workforce to athird-party agency. These relationships are de-signed to be long term—indeed, permanent.


Small businesses are both the most frequentand most satisfied users of PEO arrangements(Klaas, Yang, Gainey, & McClendon, 2005). Theworkers stay put, typically doing the same taskswith similar terms and conditions of employ-ment, and their previous supervisors continue todirect their work. But PEOs are not simply in thebackground. They set and communicate all em-ployment practices to the employees, includingthe terms and conditions of work, and handlemost human resource issues.

Contract Work Arrangements

As noted above, the law provides and enforcesa bright line distinction between employment,including all its subcategories, and all otherarrangements. Under contract work arrange-ments the client for which the work is beingperformed is prohibited from directing the workprocess; it cannot exercise directive control.Rather, it is limited to specifying the work out-come as part of a contractual agreement. As aresult, the terms of the contract must be negoti-ated ex ante. The contract specifies the work tobe completed, when the work arrangement willend, and what the rewards and incentives forthe worker will be (Benner, 2002). Such agree-ments are enforced by the courts (Marsden, 2004:668) and—with few exceptions—are governedby contract law rather than employment law.With employment relationships, in contrast, thecourts will not interfere with the day-to-day su-pervisory decisions of an employer and its per-formance evaluations. Organizations need to becautious to avoid exerting directive control overcontract workers because doing so risks turningthe contractor into an employee in terms of legalstatus and requirements. As a result, instead ofmanagement and supervision practices drivingemployee attitudes and behavior, we have theterms and conditions of the contracts drivingcontractor behavior. Instead of performance ap-praisal systems to handle performance evalua-tion and discipline issues, we have contractsand the courts to adjudicate them (de Haas, 2013;Muhl, 2002).

Just as in the broad category of employment,the presence or absence of a third party createssubcategories within the broader category ofcontract work: direct contracting andsubcontracting.

Direct contracting. Direct contracting involvesa direct relationship between a worker and theorganization for which he or she performs thework. A contract, usually in the form of a “state-ment of work,” is negotiated and agreed upon,and this document governs the relationship forits duration. Independent contractors typicallynegotiate the terms of the contract directly withthe client. On-the-spot negotiating occurs insome contexts, perhaps best illustrated with daylaborers. The defining feature of direct contract-ing is that the worker retains full control overthe work process.6

Independent contractors work in a variety ofindustries but tend to be members of well-defined occupations (Barley & Kunda, 2006; Os-nowitz, 2010). Osnowitz (2010), for example, stud-ied freelance writers and editors, as well ascomputer programmers and engineers.O’Mahony and Bechky (2006) observed a highincidence of independent contracting amongfilm production workers. Even day laborers tendto be hired for occupationally oriented jobs inthe construction and light-manufacturing indus-tries (Valenzuela, 2003). Perhaps the reason whycontracting focuses on clear occupations is thatall parties have a common understanding ofperformance standards for many aspects ofthose work processes, reducing the need to spec-ify them in contracts.

Subcontracting. The second subcategory ofcontract work—subcontracting—introduces aclear third party. Here the client organizationcontracts with a vendor to provide a service. Theindividuals who perform the work are either em-ployed by or are engaged as independent con-tractors by the vendor, who then directs the

6 An interesting twist on the independent contracting re-lationship is when agents represent workers (as in high-endentertainment)—relationships where workers pay theagents to find opportunities for them and negotiate the termsof the assignments on their behalf (Bielby & Bielby, 1999).While these relationships do involve a third party in thesearch for and negotiation of assignments, the agent is typ-ically acting as a broker and is not involved in the ongoingrelationship. Therefore, these arrangements are best classi-fied as a direct relationship between the worker and clientorganization. (In team-based professional sports, the work-ers are actual employees of their teams. But they also haveemployment contracts that delimit most of their workingarrangement with the team. In that sense it might be betterto think of them as contractors than employees.) Executivesearch may represent something like the reverse arrange-ment, where clients pay agents to find employees for them.


work. A key element of contracts in the realm ofsubcontracting is the legal notion of a “workproduct” or the nature of the service or outputbeing produced by the vendor. How that workproduct is produced is up to the vendor. Theclient organization exerts no control over thework process. For example, a contract mightspecify the menu that a food provider offers inthe client’s cafeteria but not all the details ofhow that food will be prepared.

Subcontracting relationships generally takeone of two forms, depending on where the workis performed. In vendor-on-premises arrange-ments, the vendor provides the service at theclient’s place of business, with examples includ-ing Xerox running an onsite copy center andAramark managing a corporate dining hall. Thehigher end of the skill spectrum might includeauditing and other accounting work, where em-ployees of accounting firms work at the client’slocation to produce analyses for the client. Thelower end of the skill spectrum might includeproviding janitorial and landscaping services.

The reason these arrangements count as al-ternative work rather than simply a supplierrelationship is because the employees of thevendor work directly with and on behalf of theclient. In many cases the employees of the ven-dor and the client work side by side. Responsi-bilities are often unclear in such contexts andare negotiated on the spot, as Rubery, Cooke,Earnshaw, and Marchington (2003) repeatedlyobserved in their study of overlapping work ar-rangements in a large airport. In contrast, whenthe vendor’s work is offsite and its employeesdo not come in contact with the client’s employ-ees, the arrangement is really a sourcing ar-rangement— buying components from a sup-plier rather than making them internally, forinstance—and not an alternative workarrangement.

Homeostatic Properties

As noted above, a central concern about anyclassification system is whether changes in theenvironment will render it obsolete. A key short-coming of previous classifications of economicwork was that, over time, environmental pres-sures eroded the boundaries between catego-ries and otherwise altered the set of propertiesassociated with those categories they created,reducing their usefulness. The responsibilitiesand restrictions imposed by the law, outlined inFigure 2, form the homeostatic properties help-ing to ensure that the fundamental categories inour classification system will persist.

The law is primarily concerned with whichparty assumes the administrative requirementsassociated with economic work and which partyis responsible for protecting workers from a va-riety of potential abuses (Coens & Storres, 2006).In terms of administrative requirements, the lawidentifies the party responsible for paying a va-riety of employment taxes and retirement bene-fits. In terms of worker protections, the law iden-tifies the party (or parties) who assume liabilityfor injuries or disabilities sustained during thecourse of work, for discrimination, and for thepayment of minimum wages and overtime.These rights and responsibilities have assumedgreater importance through the evolution ofcase law, further reinforcing the distinctionsamong the major categories of economic work inour classification system.

These properties can be viewed as homeo-static because historical accounts of employ-ment law as far back as the Roman Empire(Hicks, 1969) reveal that major changes to em-ployment law occur slowly over considerableperiods of time and only after extensive politicaland legal contests (Feinman, 1976; Gonos, 1997:89). Changes become even more difficult as caselaw builds over time. As a result, once in place,

FIGURE 2Homeostatic Properties of Employment Law


the law and, through it, the enforcement mech-anisms associated with the court system enforcethe boundaries of the classification system weoutlined above. It rules out arrangements thatexist outside the boundaries of these legal cat-egories. Put differently, these properties permitthe development of alternatives within somesubcategories but not others, as revealedthrough the prevalence of blank cells in Fig-ure 2. As Gleason notes, “Each type of nonstan-dard employment exists in its current form be-cause there is either a relative absence of aregulatory environment or a regulatory environ-ment that frames its use” (2006: 8). While wewould not be surprised to see new arrange-ments emerge, these new arrangements willemerge within the categories identified by ourclassification system.

RESEARCH IMPLICATIONS

As noted earlier, a good classification systemshould reveal what is important about the mem-bers of the various categories created and there-fore should help suggest both important re-search questions and how to address them.What should researchers do differently basedon the classification system we outlined above?First, the categories of alternative arrangementswe identify differ quite a bit from each other.When they are lumped together, as they havebeen in previous classification systems, the het-erogeneity makes it almost impossible to findconsistent relationships (Connelly & Gal-lagher, 2004).

For example, studies of agency temps proba-bly have little relevance to direct hire temps,despite the common name. Our classificationsystem explains why: the level of directive con-trol and who exercises it, the contractual natureof the work relationship, and the number of par-ties involved are different in the two cases. Theformer is really a coemployment relationshipwhere two parties share directive controlaround a closed contract, whereas the latter is adirect employment arrangement where the em-ployer has full directive control in an open-ended contract between two parties. On theother hand, studies of independent contractorsand the way they are managed by the organiza-tion for which they are performing the work mayhave a great deal to say about subcontractingarrangements, despite the fact that the names

are quite different. The reason is because inboth cases the client has no directive controland the arrangements are governed bycontracts.

Applying Theories and Constructs Developedin the Context of Traditional Employment

A place for research to start, therefore, is withsome purely descriptive questions based on theabove classification system. It is important toknow how each of the alternative arrangementshandles the basic tasks of managing workers.The problem is that the standard model we havefor understanding the management of workersis based on regular, full-time, direct employ-ment, and that model does not apply well toworkers in alternative arrangements (e.g., Gal-lagher & McLean Parks, 2001).

To see how the traditional model fails to workin these contexts, consider the first step in thetypical employee management model—the ba-sic topic of recruitment and selection. It appliesonly to direct employment, and even there wesee limitations to its usefulness where alterna-tive arrangements are involved. For example,employees who are hired into regular jobs afterhaving temp assignments structured by agen-cies are not recruited in any sense: the tempagency decides which worker will take whichassignment. Nor are they selected in any sophis-ticated sense of the term: local hiring managerspresumably rely on outcomes associated withactual work experience—not a work sample—todetermine which individuals to hire.

For arrangements other than direct employ-ment, related questions arise, and they are notidentical to traditional hiring issues. For exam-ple, how do agencies decide which workers toplace in which assignments? Do they try tomake matches based on fit between individualvalues and client organizational culture? Theseare unlike employment matches because theyare only one-sided; unlike a candidate for em-ployment, an agency temp or leased employeedoes not always get a say as to which assign-ment he or she takes. Do brokers who makematches between clients and contractors trysomething similar? With direct contracting, doclients and contractors choose each other in adifferent manner than do employers andemployees?


Issues associated with performance manage-ment for employees are almost unrecognizablein the context of contractual arrangements. Inthe latter, appraisals and performance com-plaints can flow not just from the client to thecontractor but from the contractor to the client.The manifestation of such complaints is contractviolations. We might think of such complaints asthe equivalent of grievances in union contracts,a formal charge that one party or the other vio-lated the contract. Unlike in employment, wherethe employer adjudicates performance com-plaints unilaterally, in a contractual context anysolutions have to be bilateral or they end upin court.

Some traditional employment topics have noobvious equivalent in contractual arrange-ments. Employee turnover, for example, does notapply to contracting arrangements. In the con-text of coemployment, its meaning changes con-siderably (e.g., it is possible to quit an assign-ment and not quit the agency). It has beendifficult to think about how to approach thesequestions, in part, because we lacked a frame-work for clearly distinguishing between em-ployment and its alternatives, as well as amongthe alternatives themselves.

Identifying New and InterestingResearch Topics

The classification system also suggests anumber of new issues worthy of study. We knowrelatively little about the actual behaviors ofworkers in arrangements other than direct em-ployment (Ashford et al., 2007: 103). Given thatdirective control is about obtaining desired be-havior from workers (Edwards, 1979), we wouldbe surprised if worker behaviors did not differacross categories. Some suggestive evidence forbehavioral differences can be found in Osnow-itz’s (2010) study of independent contractors. Be-cause their contract with the client only speci-fies the work outcome and not the work process,independent contractors are more likely to ac-tively seek feedback and solicit buy-in from cli-ent managers since the client is otherwise notdirectly engaged in the work process. Free fromdirective control, they are also more likely toexercise voice than direct employees.

The issue of organizational identification(Dutton, Dukerich, & Harquail, 1994; Mael & Ash-forth, 1995) is of special interest in triangular

arrangements. How workers see themselves andwhich organization they identify with when theyare employed by one but working on behalf ofanother is an important and unexamined ques-tion. The relative status of the two organizationsmay well drive the outcome. For example, weknow that the status of employing agencies af-fects the supply of screenwriters (Bielby &Bielby, 1999), but what happens when the statusof the agency and the status of the client aredifferent? When an accountant from a lower-status firm works for a higher-status client, doesthe worker come to identify more with the clientthan otherwise? The same questions arise in thecontext of subcontracting arrangements. Moregenerally, it would be interesting to know therelative influence of clients versus employers inthese contexts on workplace attitudes like jobsatisfaction and subsequent careers.

Indeed, the biggest deviations from tradi-tional research topics associated with employ-ment appear to come with these triangular ar-rangements. Consider, for example, the basicissue of assessing job performance: In coem-ployment, is good performance defined as thatwhich helps the client or that which helps theagency who employs the worker? In many casesthey may conflict. Helping out the client in waysthat go beyond the contract at a minimum rep-resents an opportunity cost for the agency, whounder normal circumstances would expect to bepaid for such deviations. The notion of organi-zational citizenship no longer makes clear sensein such contexts: Does being a good citizen im-ply helping a client answer emergency calls, inthe process depriving the agency of additionalhours of pay? Or does it mean billing the clientfor every little extra one does for them, lookingout for the agency’s interests?

More generally, triangular relationships aremore complicated, with workers, agencies orvendors, and organizations assuming obliga-tions to multiple parties (Marsden, 2004; McLeanParks, Kidder, & Gallagher, 1998). Issues of dualallegiance, first explored in the context of union-ized workplaces where employees had ties bothto their employer and to their unions, reemergein nonunion settings and have been recast asdual commitment issues in the context of trian-gular employment relationships (Linden,Wayne, Kraimer, & Sparrowe, 2003). Earlier stud-ies showed that the two relationships oftenpulled individuals in different directions (Angle


& Perry, 1986; Purcell, 1954), demonstrating theneed to consider the trade-offs and competingdemands imposed on workers in triangular re-lationships, something that has received consid-erably less attention. Identifying which aspectsof these various relationships are more or lesslikely to be important depending on the outcomeof interest is especially critical.

How workers manage the demands of thesepossibly competing relationships is worth con-sidering. Bidwell and Fernandez-Mateo (2010)have shown that the length of relationships af-fects the economic returns to each party in co-employment arrangements, although it remainsto be seen how these multiple relationships andattributes of the parties influence other workerattitudes and behaviors. Prior research suggeststhat technical workers in independent contract-ing arrangements may have allegiance to proj-ects independent from allegiance to clients or tothe agents that may have arranged for theirwork (Barley & Kunda, 2004; O’Mahony & Bechky,2006), suggesting another competing demandthat has to be managed.

Managing Workers

Perhaps the most interesting questions thatstill need to be asked concern the interplay ofthe different arrangements, when contractors,leased employees, vendor’s employees, and reg-ular employees all come together in a clientorganization. Grant and Parker (2009) note thatrelationships between workers have becomemore important as a result of the greater use ofteams and empowerment practices that giveworkers the ability to solve problems on theirown. This creates some of the pressure to inter-act. As Smith (2001b) found, the mixing of tempand regular employees is almost inevitable, andsuch situations raise real challenges for manag-ers, who have a great deal of influence over howthey play out. Here the classification systemsuggests that conflicts for managers will be thegreatest when the organization blends employ-ment arrangements and contract arrangements.Managers are able to control the work process ofboth direct and coemployees, while contractworkers will be less susceptible to guidance orpressure from the client managers. Managersface constraints in their ability to direct the workof coemployees, limiting their flexibility. Thissuggests that considerations of employment

flexibility at the organizational level, where al-ternative arrangements are introduced, differfrom employment flexibility at the manageriallevel (Smith, 2001b), a topic that has beenlargely overlooked despite the fact that it is in-creasingly common for managers to superviseindividuals working side by side under differentarrangements (Bidwell, 2009).

External Validity and Quasi-Experiments

The structure of our classification system alsohelps inform the external validity of existingstudies, based on which arrangements are themost similar to each other. For example, despitetheir different names, studies of agency tempsmay be quite relevant to understanding PEOsbecause of the attributes they share as subsetsof coemployment.

A final and potentially most promising use forour classification system is identifying quasi-experiments that become possible when makingcomparisons across different categories of ar-rangements. Some of these quasi-experimentsmight involve looking across categories. For ex-ample, we would expect the influence of man-agers to grow when moving from subcontract-ing, where even contract management isindirect, to direct contracting, where perfor-mance/contract management is negotiated di-rectly with the worker, to coemployment, wheredirective control is limited to execution issues,to direct employment, where directive control isunlimited. Comparisons across these categoriessuggest how the degree of management influ-ence and control affects worker outcomes. Thesame comparisons could be used to separate theeffects of coworker influence, which exist ineach arrangement, from the effects of supervi-sors just described. As another example, work-ers in PEO and agency temp arrangements bothhave three-way relationships, but the ties to theformer are much weaker than to the latter (PEOsmainly handle personnel issues while agenciesalso make work assignments, do training, andmanage performance). We should expectgreater evidence of dual allegiance in the latter.

One of the more robust and useful quasi-experiments comes when individuals move upthe hierarchy of arrangements in the same or-ganization—for example, from direct hire tempto on-call worker to part-time employee to full-time employee. This would allow fixed-effect


studies that could show in a powerful way theconsequences for employee attitudes and be-havior of moving toward greater attachment tothe employer while controlling for other factors.

Similarly, moving to the right across the directemployment– coemployment– direct contract-ing–subcontracting categories moves towardmore permeable firm boundaries, where more ofthe human capital competencies are comingfrom outside. It would be important for strategyand organizational theory researchers to seewhat else changes about organizations whenthe organization of economic work moves in thatdirection. Baron, Hannan, and Burton (2001)showed how the choice of attributes related tothe governance of work affects outcomes suchas organizational culture and values. For firms,an important attribute likely to be impacted isthe formation and sustainability of competitivestrategies (Matusik & Hill, 1998).

How Organizations ChooseAmong Arrangements

Perhaps the most obvious of the macroques-tions this classification system helps to addressis how organizations decide which arrange-ments to use and in which combinations(Huselid & Becker, 2011; Lepak & Snell, 1999). AsBaron (2004) notes, relationships governing workare arguably the most fundamental to an orga-nization’s identity. Each arrangement comeswith its own costs and benefits. The real choicefor organizations is how to balance the costs andbenefits of these different approaches to best fitthem to their context (Lepak & Snell (2002).

The boundary of the firm literature on con-tracting versus internalized arrangements asso-ciated with transaction costs (Williamson, 1985)is an obvious place to start examining this ques-tion, since there are clearly contexts where thedirective control associated with employment ismore valuable (e.g., Sherer, Rogovsky, & Wright,1998) and other contexts where contractingmakes more sense. While others have high-lighted the importance of taking into accountthe contractual nature of the work relationship(Benner, 2002; Lepak & Snell, 1999), our classifi-cation system draws attention to the heteroge-neity across different arrangements involvingcontracts. The relative value of having a secondmanagement party and the extent of directivecontrol no doubt factor into the choice as well.

CONCLUSION

In the arguments above we asserted that al-ternatives to full-time regular employment arenow so prominent that all research based on theworkplace needs to consider them. Traditionalworkplace research topics and explanationsrely on the full-time employment model as theirbasis, and there are good reasons for thinkingthat these topics and especially these explana-tions do not apply to alternative arrangements.In order to understand and study these arrange-ments, it is necessary to have a classificationsystem that can identify what is important abouteach arrangement and how each differs fromother arrangements.

Our systematic approach to developing aclassification system clearly distinguishes be-tween employment and its alternatives, as wellas among the alternatives themselves, bygrouping work arrangements into categoriesthat share common properties and that are dis-tinct from each other in ways that matter forpractice and for research. These properties cen-ter on control over the work process and includethe contractual nature and number of partiesinvolved in the work relationship. Our classifi-cation system is both informed by and reflectsthe legal distinctions among these categories.The legal arrangements have common elementsacross countries and create homeostatic proper-ties that enforce the categories and reduce therisk our classification system will be outrun bychanges in the workplace.

A significant proportion of the workplace al-ready operates under these alternative arrange-ments for economic work, and as the use of thesearrangements expands, they will surely becomea more central part of management. Given thesize and impact of these arrangements, they aredefinitely understudied at present, a conclusionconsistent with assessments that the work con-text per se has been understudied (Cappelli &Sherer, 1991; Humphrey et al., 2007). As a result,it is essential that researchers understand them.To that end, the classification system developedhere should facilitate progress in understandingthe implications of alternative work arrangementsfor organizations, managers, and workers.

REFERENCES

Abraham, K. G. 1990. Restructuring the employment relation-ship: The growth of market-mediated work arrange-


ments. In K. G. Abraham & R. B. McKersie (Eds.), Newdevelopments in the labor market: Toward a new insti-tutional paradigm: 85–119. Cambridge, MA: MIT Press.

Angle, H. L., & Perry, J. L. 1986. Dual commitment and labor-management relationship climates. Academy of Man-agement Journal, 29: 31–50.

Ashford, S., George, E., & Batt, R. 2007. Old assumptions, newwork: The opportunities and challenges of research onnonstandard employment. Academy of ManagementAnnals, 1: 65–117.

Barley, S. R., & Kunda, G. 2004. Gurus, hired guns, and warmbodies: Itinerant experts in a knowledge economy.Princeton, NJ: Princeton University Press.

Barley, S. R., & Kunda, G. 2006. Contracting: A new form ofprofessional practice. Academy of Management Per-spectives, 20(1): 45–66.

Baron, J. N. 2004. Employing identities in organizational ecol-ogy. Industrial and Corporate Change, 13: 3–32.

Baron, J. N., Hannan, M. T., & Burton, D. M. 2001. Labor pains:Change in organizational models and employee turn-over in young, high tech firms. American Journal ofSociology, 106: 960–1012.

Bauer, T. N., & Truxillo, D. M. 2000. Temp-to-permanent em-ployees: A longitudinal study of stress and selectionsuccess. Journal of Occupational Health Psychology, 5:337–346.

Befort, S. F. 2003. Revisiting the black hole of workplaceregulation: A historical and comparative perspective ofcontingent work. Berkeley Journal of Employment andLabor Law, 24: 153–178.

Belman, D. L., Monaco, K. A., & Brooks, T. J. 2005. Sailors of theconcrete sea: A portrait of truck drivers’ work and lives.Lansing: Michigan State University Press.

Belous, R. S. 1989. The contingent economy: The growth of thetemporary, part-time, and contingent workforce. Wash-ington, DC: National Planning Association.

Belzer, M. H. 2002. Technological innovation and the truckingindustry: Information revolution and the effect on thework process. Journal of Labor Studies, 23: 375–395.

Bendix, R. 2001. (First published in 1956.) Work and authorityin industry: Managerial ideologies in the course of in-dustrialization (2nd ed.). New Brunswick, NJ: New Trans-action.

Benner, C. 2002. Work in the new economy: Flexible labormarkets in Silicon Valley. Malden, MA: Blackwell.

Bidwell, M. J. 2009. Do peripheral workers do peripheralwork? Comparing the use of highly skilled contractorsand regular employees. Industrial and Labor RelationsReview, 62: 200–225.

Bidwell, M. J., & Fernandez-Mateo, I. 2008. Three’s a crowd?Understanding triadic employment relationships. In P.Cappelli (Ed.), Employment relationships: New modelsof white-collar work: 142–178. Cambridge: CambridgeUniversity Press.

Bidwell, M. J., & Fernandez-Mateo, I. 2010. Relationship du-ration and returns to brokerage in the staffing sector.Organization Science, 21: 1141–1158.

Bielby, W. T., & Bielby, D. D. 1999. Organizational mediationof project-based labor markets: Talent agencies and thecareers of screenwriters. American Sociological Review,64: 64–85.

Bird, A., & Tobin, E. 2010. Natural kinds. In E. N. Zalta (Ed.),The Stanford encyclopedia of philosophy. Palo Alto, CA:Center for the Study of Language and Information, Stan-ford University. Available at http://plato.stanford.edu/entries/natural-kinds/.

Blank, R. M. 1998. Contingent work in a changing labormarket. In R. Freeman & P. Gottschalk (Eds.), Generatingjobs: How to increase demand for less-skilled workers:258–294. New York: Russell Sage Foundation.

Boyd, R. 1999. Homeostasis, species, and higher taxa. In R.Wilson (Ed.), Species: New interdisciplinary essays: 141–186. Cambridge, MA: MIT Press.

Brass, D. J. 2002. Intraorganizational power and dependence.In J. A. C. Baum (Ed.), The Blackwell companion to orga-nizations: 138–157. Oxford: Blackwell.

Broschak, J. P., & Davis-Blake, A. 2006. Mixing standard workand nonstandard deals: The consequences of heteroge-neity in employment arrangements. Academy of Man-agement Journal, 49: 371–393.

Burawoy, M. 1979. Manufacturing consent: Changes in thelabor process under monopoly capitalism. Chicago: Uni-versity of Chicago Press.

Cappelli, P., & Neumark, D. 2004. External churning andinternal flexibility: Evidence on the functional flexibilityand core-periphery hypotheses. Industrial Relations, 43:148–182.

Cappelli, P., & Sherer, P. D. 1991. The missing role of contextin OB: The need for a meso-level approach. Research inOrganizational Behavior, 13: 55–110.

Chambel, M. J., & Castanheira, F. 2006. Different temporarywork status, different behaviors in organization. Journalof Business and Psychology, 20: 351–367.

CIETT. 2011. The agency work industry around the world.Brussels: International Confederation of Private Em-ployment Agencies. Available at http://www.ciett.org/index.php?id�155.

Coase, R. H. 1939. Rowland Hill and the Penny Post. Eco-nomica, 6: 423–435.

Coens, T. A., & Storres, A. L. 2006. No safe harbor: A review ofsignificant laws affecting contingent workers. In S. E.Gleason (Ed.), The shadow workforce: Perspectives oncontingent work in the United States, Europe, and Japan:135–168. Kalamazoo, MI: W.E. Upjohn Institute.

Connelly, C. E., & Gallagher, D. G. 2004. Emerging trends incontingent work research. Journal of Management, 30:959–983.

Crain, W. M., & Johnson, J. 2001. Compliance costs of federalworkplace regulations: Survey results for U.S. manufac-turers. Fairfax, VA: Regulatory Stuides Program, GeorgeMason University.

Davis-Blake, A., Broschak, J. P., & George, E. 2003. Happytogether? How using nonstandard workers affects exit,


voice, and loyalty among standard employees. Acad-emy of Management Journal, 46: 475–485.

de Haas, R. M. 2013. Vizcaino v. Microsoft. Berkeley Technol-ogy Law Journal, 13: 483–502.

Dupré, J. 1993. The disorder of things: Metaphysical founda-tions of the disunity of science. Cambridge, MA: HarvardUniversity Press.

Dutton, J. E., Dukerich, J. M., & Harquail, C. V. 1994. Organi-zational images and member identification. Administra-tive Science Quarterly, 39: 239–263.

Dwyer, C. J. 2011. Contingent labor management: The evolu-tion of the contemporary contingent labor workforce.Boston: Aberdeen Research Group.

Edwards, R. 1979. Contested terrain. New York: Basic Books.

Feinman, J. M. 1976. The development of the employment atwill rule. American Journal of Legal History, 20: 118–135.

Freeman, A. 1996. Contingent work and the role of labormarket intermediaries. In S. A. Levitan, G. L. Mangum, &S. L. Mangum (Eds.), Of heart and mind: Social policyessays in honor of Sar A. Levitan: 177–199. Kalamazoo,MI: W.E. Upjohn Institute.

Gallagher, D. G., & McLean Parks, J. 2001. I pledge thee mytroth . . . contingently: Commitment and the contingentwork relationship. Human Resource Management Re-view, 11: 181–208.

Gleason, S. E. 2006. The shadow workforce. In S. E. Gleason(Ed.), The shadow workforce: Perspectives on contingentwork in the United States, Europe, and Japan: 1–28. Kala-mazoo, MI: W.E. Upjohn Institute.

Goldberg, V. P. 1980. Bridges over contested terrain: Explor-ing the radical account of the employment relationship.Journal of Economic Behavior and Organization, 1: 249–274.

Gonos, G. 1997. The contest over “employer” status in thepostwar United States: The case of temporary help firms.Law and Society Review, 31: 81–110.

Grant, A., & Parker, S. 2009. Redesigning work design theo-ries: The rise of relational and proactive perspectives.Academy of Management Annals, 3: 317–375.

Hicks, J. 1969. A theory of economic history. Oxford: OxfordUniversity Press.

Houseman, S. N. 2001. Why employers use flexible staffingarrangements: Evidence from an establishment survey.Industrial and Labor Relations Review, 55: 149–170.

Humphrey, S. E., Nahrgang, J. D., & Morgeson, F. P. 2007.Integrating motivational, social, and contextual workdesign features: A meta-analytic summary and theoret-ical extension of the work design literature. Journal ofApplied Psychology, 92: 1332–1356.

Huselid, M. A., & Becker, B. E. 2011. Bridging micro and macrodomains: Workforce differentiation and strategic humanresource management. Journal of Management, 37: 395–403.

International Labour Organization. 2006. The agency workindustry around the world. Brussels: International La-bour Organization.

Kalleberg, A. L. 2000. Nonstandard employment relations:Part-time, temporary and contract work. Annual Reviewof Sociology, 26: 341–365.

Kalleberg, A. L., Reskin, B. F., & Hudson, K. 2000. Bad jobs inAmerica: Standard and nonstandard employment rela-tions and job quality in the United States. AmericanSociological Review, 65: 256–278.

Kalleberg, A. L., Reynolds, J., & Marsden, P. V. 2003. External-izing employment: Flexible staffing arrangements inU.S. organizations. Social Science Research, 32: 525–552.

Klaas, B. S., Yang, H., Gainey, T., & McClendon, J. A. 2005. HRin the small business enterprise: Assessing the impactof PEO utilization. Human Resource Management, 44:433–448.

KPMG. 2011. 2011 U.S. hospital nursing labor costs study. NewYork: KPMG.

Kukla, A. 2000. Social construction and the philosophy ofscience. London: Routledge.

Kunda, G. 1992. Engineering culture. Philadelphia: TempleUniversity Press.

Lautsch, B. A. 2002. Uncovering and explaining variance inthe features and outcomes of contingent work. Industrialand Labor Relations Review, 56: 23–43.

Lepak, D. P., & Snell, S. A. 1999. The human resource archi-tecture: Toward a theory of human capital allocationand development. Academy of Management Review, 24:31–48.

Linden, R. C., Wayne, S. J., Kraimer, M. L., & Sparrowe, R. T.2003. The dual commitments of contingent workers: Anexamination of contingents’ commitment to the agencyand the organization. Journal of Organizational Behav-ior, 24: 609–625.

Lodahl, J. B., & Gordon, G. 1972. The structure of scientificfields and the functioning of university graduate depart-ments. American Sociological Review, 37: 57–72.

Luo, T., Mann, A., & Holden, R. 2010. The expanding role oftemporary help services from 1990 to 2008. Monthly La-bor Review, 133(8): 3–16.

Mael, F. A., & Ashforth, B. E. 1995. Loyal from day one: Bio-data, organizational identification, and turnover amongnewcomers. Personnel Psychology, 48: 309–333.

Marsden, D. W. 1999. A theory of employment systems: Micro-foundations of societal diversity. New York: Oxford Uni-versity Press.

Marsden, D. W. 2004. The “network economy” and models ofthe employment contract. British Journal of IndustrialRelations, 42: 659–684.

Marx, K., & Engles, F. 1992. (First published in 1888.) Thecommunist manifesto. New York: Bantam Classics.

Matusik, S. F., & Hill, C. W. L. 1998. The utilization of contin-gent work, knowledge creation, and competitive advan-tage. Academy of Management Review, 23: 680.

McLean Parks, J., Kidder, D. L., & Gallagher, D. G. 1998.Fitting square pegs into round holes: Mapping the do-main of contingent work arrangements onto the psycho-


logical contract. Journal of Organizational Behavior,19(S1): 697–730.

Muhl, C. J. 2002. What is an employee? The answer dependson the federal law. Monthly Labor Review, 125: 3–11.

O’Mahony, S., & Bechky, B. A. 2006. Stretchwork: Managingthe career progression paradox in external labor mar-kets. Academy of Management Journal, 49: 918–941.

Osnowitz, D. 2010. Freelancing expertise: Contract profes-sionals in the new economy. Ithaca, NY: ILR Press.

Ostapczuk, E. D. 1994. What makes effective secondary edu-cation substitute teachers? Literature review. Washing-ton, DC: Educational Resources Information Center, U.S.Department of Education.

Peck, J., & Theodore, N. 2007. Flexible recession: The tempo-rary staffing industry and mediated work in the UnitedStates. Cambridge Journal of Economics, 31: 171–192.

Perlow, L. A. 1998. Boundary control: The social ordering ofwork and family time in a high-tech corporation. Admin-istrative Science Quarterly, 43: 328–357.

Pfeffer, J. 1993. Barriers to the advance of organizationalscience: Paradigm development as a dependent vari-able. Academy of Management Review, 18: 599–620.

Pfeffer, J., & Baron, J. N. 1988. Taking the workers back out:Recent trends in the structuring of employment. Re-search in Organizational Behavior, 10: 257–303.

Polivka, A. E. 1996. Into contingent and alternative employ-ment: By choice? Monthly Labor Review, 18: 55–74.

Purcell, T. V. 1954. Dual allegiance to company and unionpackinghouse workers: A Swift-UPA study in a crisissituation (1949–1952). Personnel Psychology, 7: 45–58.

Quine, W. V. 1969. Ontological relativity and other essays.New York: Columbia University Press.

Rubery, J., Cooke, F. L., Earnshaw, J., & Marchington, M. 2003.Inter-organizational relations and employment in amulti-employer environment. British Journal of Indus-trial Relations, 41: 265–289.

Salancik, G. R., Staw, B. M., & Pondy, L. R. 1980. Administra-tive turnover as a response to unmanaged organiza-tional interdependence. Academy of Management Jour-nal, 23: 422–437.

Scott, W. R., & Davis, G. F. 2006. Organizations and organiz-ing: Rational, natural and open systems perspectives.Upper Saddle River, NJ: Pearson Prentice Hall.

Sherer, P. D., & Lee, K. 2002. Institutional change in large lawfirms: A resource dependency and institutional perspec-tive. Academy of Management Journal, 45: 102–119.

Sherer, P. D., Rogovsky, N., & Wright, N. 1998. What drivesemployment relationships in taxicab organizations?Linking agency to firm capabilities and strategic oppor-tunities. Organization Science, 9: 34–48.

Simon, H. A. 1947. Administrative behavior. New York: Mac-millan.

Simon, H. A. 1951. A formal theory of the employment rela-tionship. Econometrica, 19: 293–305.

Smith, V. 1997. New forms of work organization. Annual Re-view of Sociology, 23: 315–339.

Smith, V. 2001a. Crossing the great divide: Worker risk andopportunity in the new economy. Ithaca, NY: ILR Press.

Smith, V. 2001b. Teamwork vs. tempwork: Managers and thedualisms of workplace restructuring. In D. Cornfield, K.Campbell, & H. McCammon (Eds.), Working in restruc-tured workplaces: New directions for the sociology ofwork: 7–28. Thousand Oaks, CA: Sage.

Smith, V., & Neuwirth, E. B. 2008. The good temp. Ithaca, NY:ILR Press.

Stone, K. 2006. Legal protections for atypical employees:Employment law for workers without workplaces andemployees without employers. Berkeley Journal of Em-ployment and Labor Law, 27: 251–286.

Texarkana Independent School District. 2012. Substituteteaching at TISD. http://www.txsid.net/foremployment/subteacher.asp, accessed January 12, 2012.

Valenzuela, A. 2003. Day labor work. Annual Review of So-ciology, 29: 307–333.

Williamson, O. E. 1985. The economic institutions of capital-ism. New York: Free Press.

Wright, P. M., & McMahan, G. C. 1992. Theoretical perspec-tives for strategic human resource management. Journalof Management, 18: 295–320.

Peter Cappelli ([email protected]) is the George W. Taylor Professor ofManagement at the Wharton School, University of Pennsylvania, director of Wharton’sCenter for Human Resources, and a research associate at the NBER. He received hisPh.D. in labor economics from the University of Oxford. His recent research examineschanges in employment relations and their implications.

JR Keller ([email protected]) is a doctoral candidate in the Management De-partment at the Wharton School, University of Pennsylvania. His research interestsinclude employee mobility, executive search and selection, the evolving role of labormarket intermediaries, and changing employment relations.


classifying work in the new economy - wharton faculty · zations with temporary, leased, and...

Documents