Intertemporal Collective-Action Problems in Graduate-Worker Organizing

Introduction

Graduate-worker organizing is on a tear. In January 2023, Yale University graduate workers chose to unionize in a resounding victory with ninety-one percent of voting workers backing the union.[1] The month before, an election at Boston University (BU) was even more one-sided, as graduate workers unionized with only twenty-eight (out of over 1,400) votes against.[2] Yale and BU are just two examples in a string of elections in which graduate workers have displayed virtual unanimity in choosing to unionize.[3] In agency fiscal year 2023, the seven largest National Labor Relations Board (NLRB or Board) representation cases closed involved graduate workers,[4] and in each election, graduate workers voted to form a union.[5]

This series of victories is all the more impressive given that many of the workers who voted for the unions will likely never see any concrete benefit. A large subset of graduate students who are eligible to vote under current precedent are upper-year students.[6] If past trends are any indication, many of these workers will graduate or finish working before their unions can negotiate a first contract.[7] Conversely, some early-year students do not work and are consequently ineligible to vote in these landslide victories, even as they stand to benefit the most from unionization as future bargaining unit members.

Graduate-worker unions represent an especially pronounced case of what Brishen Rogers has labeled “an ‘intertemporal’ collective action problem.”[8] Intertemporal collective-action problems arise in union organizing when a set of current employees makes a unionization decision that affects a future set of workers, and the deciding workers therefore do not fully internalize the results of their decision.[9] In graduate organizing campaigns, current employees, who may be disproportionately upper-year students, decide whether to unionize, and their decision affects early-year students required to work in the future.[10] Moreover, these current employees do not internalize the full impact of their unionization decision on younger peers. This situation is perhaps the most extreme form of the problem Rogers described: in graduate-worker unions, the “union pioneers who bear the upfront costs of unionization” may not receive any “of the benefits” in the future.[11]

In this Note, I argue that early-year graduate students with degree work requirements should be allowed to participate in union representation procedures. Section 9(a) of the National Labor Relations Act (NLRA or Act) provides that “[r]epresentatives designated or selected for the purposes of collective bargaining by the majority of the employees in a unit appropriate for such purposes, shall be the exclusive representatives of all the employees in such unit . . . .”[12] This provision, as interpreted by the Board, leads to three key conditions for representation-election eligibility relevant to early-year graduate students. First, voters must be “employees.”[13] Second, voters who are employees but not currently working must “have a reasonable expectation of future employment” with the applicable employer.[14] Third, employee-voters must “share a community of interests” with their peers to support a bargaining unit.[15] Early-year graduate students easily meet the latter two criteria under the NLRB’s seasonal-worker and bargaining-unit doctrines. I argue that the Board and reviewing courts should find early-year graduate students with degree work requirements to be “statutory employees”[16] as well, with attendant voice rights in unionization decisions.

In my research, I have found no union that has submitted a representation petition including these early-year students. Graduate unions and Board officials should include these individuals in election units for strategic and policy reasons, respectively. For graduate unions, early-year students may have stronger incentives to choose union representation than students nearing graduation. Although recent graduate-union elections have generally been “lopsided” in favor of workers,[17] these margins were close in the past, and at least one petitioning union has lost in the last year.[18] Adoption of this Note’s proposal could make a difference if current momentum decreases. In addition, for similar reasons, early-year students can help graduate unions wield a strong hand in contract negotiations. Vote margin is important to bargaining position: An employer will feel more pressure after a ninety-five percent union victory than after a fifty-five percent vote.[19] Participation of early-year students would also signal durable union support and convince universities that they cannot wait out the graduation of union leaders. Finally, recognition of early-year students as employees for voting purposes might allow graduate unions to broaden the scope of contract bargaining. Currently, graduate unions cannot bargain over the stipends of students who are not yet working. Recognizing these students as employees may allow unions to negotiate over these stipends, raising standards for all individuals at a university.

For the Board, allowing early-year graduate students with degree work requirements to participate in representation decisions would further the Act’s goal of “employee choice.”[20] As Benjamin Sachs noted, “The good to be maximized by the rules governing employee decisionmaking on the union question is employee choice.”[21] Board officials have noted the need for flexible policies that secure employee choice in graduate-worker elections by adopting “lookback formulas” that have allowed “individuals who held a unit position during the prior academic year . . . to vote even where they did not currently hold a unit position.”[22] These formulas, while a good start, discriminate between employees with similar employment outlooks based on whether they happen to have worked in the past. The “look-forward” period that this Note proposes would remedy this distinction and better fulfill the Board’s goal to ensure that all “employees in a unit appropriate for” collective bargaining have a choice in whether a union will represent them.[23]

This Note makes minor contributions to two lines of thinking. First, a series of scholars, notably Rogers,[24] Sachs,[25] and Mark Barenberg,[26] have pointed to intertemporal collective-action problems as part of the reason why there need to be changes to NLRB voting procedures. Sachs, for example, argued that workers face “asymmetric impediments to unionization” including intertemporal collective-action problems,[27] justifying reformed voting mechanisms that reduce “managerial interference” in representation election campaigns.[28] Rogers, noting “coordination problems that may otherwise hinder organizing,” favored a modified card check format with an opportunity for workers to secretly take back their signature.[29] Barenberg explored a switch to a union default, reasoning that “the default state of unionization would more accurately reflect the undominated long-term subjective preferences of employees.”[30] I go in a different direction, arguing that in some number of cases, intertemporal collective-action problems justify changes to NLRB voting populations. Early-year graduate students who are required to work for their universities within a couple of years bear the outcome of a union election but have no input in the decision itself. Including these future workers in the voter pool through a broadened employee status will help overcome the intertemporal collective-action problem in graduate organizing, is good policy, and, importantly, is consistent with the NLRA.

Second, several scholars, including George Feldman in the past,[31] and Diana Reddy in the present,[32] have explored the relationship of union members “as workers”[33] to broader social interests. Feldman discussed how the Supreme Court has constrained employee status to confine the “horizontal dimension” of union representation, preventing unions from representing a broader range of interests (e.g., the interests of retirees) in collective bargaining.[34] Reddy recently compared the growing view of unions as agents for the public good with continuing ambivalence about the actual mandate of these organizations to “help[] workers improve their jobs.”[35] Graduate unions have something to contribute to this discourse, as upper-year graduate students who vote for a union based on their personal experience as graduate workers, and despite the fact that they will graduate or finish working before a first contract, effectively endorse unions as agents of workplace change for their peers. In this way, the intertemporal collective action problem in graduate unions is a ripe setting to continue the conversation Reddy began about valuing unions for “doing exactly what they are statutorily designed to do.”[36]

Part I introduces graduate-worker organizing and the structure of graduate programs. Part II offers a short summary of the NLRB election process and traces the judicial and administrative construction of an unordinary, “ordinary meaning” of NLRA employment. Part III reviews literature on intertemporal collective-action problems, including in union organizing, and recognizes that organizing graduate workers face a particularly acute form of this problem. It then proposes allowing certain early-year graduate students to participate in union-representation procedures, finding that these students share a community of interests with upper-year peers, have a near-certain expectation of future employment, and can be statutory employees. Part IV briefly discusses support for graduate unions as agents of workplace change and concludes.

I. Introduction to Graduate Organizing and Programs

In 1935, Congress passed the NLRA, which protects the rights of “employees” to organize labor unions and “engage in other concerted activities for the purpose of collective bargaining or other mutual aid or protection.”[37] Section 10 of the Act assigns the NLRB the responsibility “to prevent any person from engaging in any unfair labor practice,”[38] which includes interference with employee organizing rights.[39] The NLRA does not cover public-sector workers[40]: federal employees have certain organizing rights under the federal civil-service statute,[41] and state and local government employees rely on state laws for labor-law protections, if any exist.[42] Among private-sector workers, the Act’s coverage is generally industry-agnostic. Other than specific exclusions for agricultural, domestic, and certain transportation workers, the text of the NLRA establishes a “uniform standard” for statutory employment.[43]

Graduate-student workers at the University of Wisconsin formed the first graduate union in 1969.[44] Throughout the late twentieth century, graduate organizing flourished at many state universities,[45] where local public-sector bargaining regimes governed labor relations between administrators and graduate assistants.[46] At the same time, graduate workers at private universities covered by the NLRA faced an uphill legal battle for union recognition.[47] In the 1970s, the NLRB successively “held that graduate assistants should be excluded from a bargaining unit of university faculty members because they did not share a community of interest with the faculty”[48] and “that certain university research assistants were ‘primarily students’ and thus not statutory employees.”[49] Organizing in an unfavorable legal regime and facing at-times bitterly hostile university administrations,[50] graduate unions were unable to gain a foothold at private universities throughout the 1990s.[51]

In 2000, the Board “opened the door”[52] for graduate organizing at private schools, finding in New York University “that certain university graduate assistants were statutory employees.”[53] In 2004, however, the NLRB slammed the door shut, reverting in Brown University to the previous NLRA exclusion.[54] In 2016, the NLRB switched positions once more, deciding in Columbia University “that student assistants who perform work at the direction of their university for which they are compensated are statutory employees.”[55] Despite fears of yet another reversal during the Trump Administration,[56] the NLRB did not get a chance to revisit Columbia between 2017 and 2020.[57] Columbia, and its holding that graduate workers are employees under the Act, “remains good law today.”[58]

Graduate unions attempt to organize a large swath of workers. The number of graduate workers has grown rapidly in recent decades, with graduate employment increasing by about seventeen percent between 2005 and 2015.[59] Data from 2012 suggest that over ten percent of graduate students, including almost sixty percent of Ph.D. candidates, serve as teaching assistants or research assistants.[60] In 2016, graduate unions represented over 64,000 graduate workers at public universities.[61] The total number of unionized graduate workers has exploded in the years since graduate workers at private universities gained organizing rights: in NLRB fiscal year 2023, about 22,000 graduate workers unionized in the seven largest finalized elections alone.[62] These elections have certified notably large units that, in many cases, outnumber the median private-sector unit by several orders of magnitude.[63]

Graduate workers encompass master’s, professional, and doctoral students. But doctoral candidates have consistently driven graduate organizing and formed the bulk of bargaining units. As the NLRB noted in Columbia, Ph.D. programs may require students to perform teaching or research work.[64] Some institutions backload this work in the degree program. At Yale, for example, many doctoral students can expect to “teach in their third, fourth, and sixth years” of degree studies.[65] At Columbia, the Board similarly found that “[i]n most students’ second through fourth year, taking on teaching or research duties is a condition for full receipt of [doctoral] funding,” while “the first and fifth years are funded without a condition of service.”[66] This setup creates an uneven distribution of graduate workers by year at these schools: early-year students are relatively unlikely to be performing academic labor, whereas many mid-year and upper-year students work as graduate assistants. Importantly, even when upper-year students do not have to work, many still do so. Data from supplemental briefings following the Columbia decision suggested that, at Columbia, about thirty-four percent of fifth-year doctoral candidates worked as graduate assistants later in their studies—that is, in their fifth year or beyond.[67] As explored further below, this distribution of workers by year contributes to “an ‘intertemporal’ collective action problem”[68] that affects graduate organizing.[69]

To date, scholarship on graduate-worker unions and labor relations has generally followed one of two lines of inquiry. One set of scholars has developed arguments around the NLRA employee status of all graduate workers or certain classes of them.[70] For example, the employee status of student-athletes has been a particularly popular topic in recent discussions.[71] A second group of authors has performed historical and sociological research on the experiences of graduate workers and the repression of graduate organizing.[72] These pieces describe what concerted action has been like on the ground, and how universities and government officials have responded. This Note contributes to both lines of thinking. In Part III, the Note offers a legal argument about NLRA employment. I argue that early-year students who are not yet working are likewise NLRA “employees.” Then, in Part IV, the Note connects the experience of upper-year graduate workers to broader themes in labor organizing and bargaining. The Note therefore centers a population that has gone unaddressed in the literature so far and analyzes graduate organizing using a new framing: the intertemporal collective-action problem.

II. The Unordinary, “Ordinary Meaning” NLRA Employee

The NLRB election process begins when workers file an election petition with the Board.[73] The petition must demonstrate a minimum of thirty-percent support from the unionizing workforce,[74] which filers will often prove through signed union cards.[75] Assuming a valid petition, the Board “will seek an election agreement between the employer, union, and other parties setting the date, time, and place for balloting” and an “appropriate unit.”[76] A regional director resolves any disputes over unit contours and voter eligibility at a hearing, and the Board sets an election date.[77] After the election, “a union that receives a majority of the votes cast is certified as the employees’ bargaining representative and is entitled to be recognized by the employer as the exclusive bargaining agent for the employees in the unit.”[78]

The NLRA gives organizing rights to “employees,”[79] and defines the term to “‘include any employee,’ subject only to certain specifically enumerated exceptions.”[80] Given the definition’s “exquisite circularity,”[81] the Supreme Court has had to set the boundaries of the definition, with guidance and support from interstitial decisions by the NLRB.[82] In these cases, decisionmakers have created an “ordinary meaning”[83] test for employment far broader than the “dictionary definition of employee.”[84]

According to Court precedent, “[t]he ordinary dictionary definition of ‘employee’ includes any ‘person who works for another in return for financial or other compensation.’”[85] In Allied Chemical & Alkali Workers v. Pittsburgh Plate Glass Co., the Court applied a similar definition and found that retirees were not statutory employees, concluding that “[t]he ordinary meaning of ‘employee’ does not include retired workers; retired employees have ceased to work for another for hire.”[86] Similarly, in NLRB v. Town & Country Electric, the Court applied the definition quoted above to find that salts were statutory employees under a “broad, literal reading of the statute.”[87]

In these same cases, however, the Court has admitted that it has often failed to follow a “literal” standard.[88] As the Pittsburgh Plate Glass Court recognized,[89] the Court has found that job applicants can be statutory employees with a claim to the NLRA’s protections.[90] The same Court noted the NLRA’s solicitude for “registrants at hiring halls—who have never been hired in the first place.”[91] Neither job applicants nor hiring hall registrants meet the “ordinary dictionary definition of employee,”[92] because they do not “work[] for another in return for financial or other compensation.”[93] Nevertheless, Court precedent has extended statutory protections to these individuals “as a means towards the accomplishment of the main object of the legislation.”[94] On the whole, the Supreme Court’s test for NLRA employment can best be described as an “unordinary” ordinary meaning standard: the test covers both workers actively employed by a firm (ordinary employees) and workers who are not currently employed but are linked to a firm in a meaningful way (unordinary employees).[95]

In recent years, lower courts have continued to take a not-so-literal approach on the “borderland”[96] of NLRA employment. In International Alliance of Theatrical and Stage Employees (IATSE) v. NLRB, a panel of the D.C. Circuit concluded that hiring-hall registrants were “employee[s] who engaged in a strike” even though “a number of the registrants had never been referred to any of the Employers involved in [the] case.”[97] The court found the holding so obvious that it refused to give Chevron deference to the NLRB’s opposite conclusion,[98] finding it “unreasonable”[99] even though the registrants were clearly not “ordinary meaning” employees.

The Board has used similarly “unordinary” reasoning, even when excluding workers from employee status. The Board has consistently maintained a broader definition of “employee” than the Pittsburgh Plate Glass Court, stating that “employee” covers “members of the working class generally.”[100] In WBAI Pacifica Foundation, the Board, distinguishing cases like Phelps Dodge,[101] concluded that “unpaid staff are not employees within the meaning of Section 2(3) because there is no economic aspect to their relationship with the Employer, either actual or anticipated.”[102] In Amnesty International, the Board applied WBAI Pacifica to “unpaid interns,” finding that they were not statutory employees because they “did not receive or anticipate any economic compensation from the Respondent.”[103] The Amnesty Board majority cited the lack of “evidence that the Respondent ever hired interns as paid staff members following their internships” as an important basis for its determination.[104] The Board’s focus on “actual or anticipated” compensation and future hiring in these decisions would be incorrect under a literal “ordinary meaning” analysis; “anticipated” compensation is irrelevant to the Town & Country definition. Yet both decisions were unanimous judgments, and the concurrence in Amnesty might have used an even wider rule.[105]

Taken together, judicial and administrative precedents have effectively created an unordinary “ordinary meaning” NLRA employee that “broadly”[106] includes individuals with at least some “rudimentary economic relationship, actual or anticipated”[107] to an employer. While decisions repeatedly point to opinions referencing “ordinary meaning” as leading precedent, they construct the “employee” definition in a different way in practice.[108] The judicially created “ordinary meaning” NLRA employee can either work for an employer now or expect to do so in the future, opening the door to coverage for the early-year students discussed below.

III. Overcoming Intertemporal Collective-Action Problems in Graduate Organizing

A. Intertemporal Collective-Action Problems and Graduate Workers

In the last twenty-five years, scholars have begun to apply the teachings of behavioral economics to the study of law through the field of “behavioral law and economics.”[109] As leading authors in the field have described, “[t]he task of behavioral law and economics, simply stated, is to explore the implications of actual (not hypothesized) human behavior for the law.”[110] Labor-law scholars have begun to develop a “labor law and behavioral economics” that explores the ways in which “various cognitive distortions and biases” may depress unionization below a utility-maximizing standard.[111] Central to this effort has been an expanded understanding of the ways in which individual and collective behavior manifests in collective-action problems that affect worker organizing.

Scholars have highlighted several types of collective-action problems in union organizing. First, workers have the “classic collective action problem,” in which “workers may fail to organize due to the high short-term costs of organizing.”[112] Second, workers face the “intrapersonal collective action problem” due to internal discounting of the long-run benefits of unionization compared to the serious immediate risks of an organizing campaign.[113] Third, workers face the “intertemporal collective action problem,”[114] where “the workers who bear the upfront costs of unionization are not able in fact to recapture those costs through downstream rewards.”[115] In other words, workers who decide to unionize do not get the full benefits of their decision, while future workers bear the fruits of a decision in which they had no voice. Experimental behavioral-economics research suggests that these intertemporal problems seriously impede collective action in settings like worker organizing where “the benefits of cooperation are shifted into the future.”[116] For example, Felix Kölle and Thomas Lauer found “a significant and substantial decrease in” cooperation in an immediate cost, future benefit scenario akin to unionization.[117]

Graduate workers seeking to unionize face a particularly acute form of intertemporal collective-action problem. Many graduate assistants who are active workers, and therefore eligible to vote in NLRB elections under current law, are upper-year students. These students may graduate or finish working before the union can negotiate a first contract, meaning that they may receive virtually no tangible benefits from the decision to unionize. In contrast, early-year students who have work requirements attached to their degree program are almost certain to work (and consequently enjoy the benefits of the decision to unionize). These workers, however, do not have a say in the unionization decision under a traditional eligibility analysis if they are not yet working.

Graduate-worker organizing may therefore approach what Thomas Schelling described as “[i]ntergenerational [d]iscounting.”[118] As explained by Kölle and Lauer, intergenerational discounting includes “cases [where] benefits might not even be claimed by oneself but only by subsequent generations.”[119] In graduate-worker organizing, the benefits of the decision to unionize by upper-year students may accrue only to future “generations” of graduate workers. The outcomes of classic intergenerational discounting cases like climate change[120] suggest how serious an obstacle intertemporal collective-action problems may be for graduate organizers.

This Note argues that labor law already provides a solution to the intertemporal collective-action problem in graduate-worker organizing. Specifically, it argues that early-year students who are not yet working but have degree work requirements should be eligible to participate in NLRB representation procedures. Section 9(a) of the Act provides that “[r]epresentatives designated or selected for the purposes of collective bargaining by the majority of the employees in a unit appropriate for such purposes, shall be the exclusive representatives of all the employees in such unit . . . .”[121] This provision, as interpreted by the Board, leads to three key conditions for representation-election eligibility relevant to early-year graduate students. First, voters must be “employees”;[122] I outlined the “unordinary” NLRA employee definition in Part II.[123] Second, voters who are employees but not currently working must “have a reasonable expectation of future employment” with the applicable employer.[124] Third, employee voters must “share a community of interests” with their peers to support a bargaining unit.[125] Early-year graduate students with degree work requirements pass all three tests. Although commenters may point to the so-called “prework rule” as barring early-year graduate students from voting, NLRB regional directors have already approved an eligibility formula that replaces the prework rule in graduate-worker elections.[126] These officials could extend these formulas through a “look-forward” period that captures early-year graduate students with degree work requirements.[127]

The rest of this Part analyzes early-year graduate students on each condition. I take the conditions in reverse order based on ease of resolution. I conclude by discussing modifications to voter-eligibility formulas that would implement this Note’s proposal.

B. Community of Interests

First, Supreme Court and NLRB precedent requires voters to “share a community of interests” with their peers.[128] The Court and the Board have emphasized the importance of “mutuality of interest” in bargaining unit determination;[129] “[i]f the petitioned-for employees have a sufficient mutuality of interests, then the unit is, absent countervailing considerations, appropriate for collective bargaining.”[130] Under the Board’s recent American Steel Construction decision covering “subdivision” units,[131] “the petitioned-for unit must be (1) homogeneous, (2) identifiable, and (3) separate or sufficiently distinct.”[132] Graduate students as a whole are clearly both “identifiable”[133] and “sufficiently distinct,”[134] so this Section will focus on homogeneity.

Homogeneity “simply reflects the principle . . . that petitioned-for employees must share a community of interest that renders the unit suitable for collective bargaining.”[135] American Steel Construction outlines numerous factors for the community-of-interest analysis.[136] Setting aside when they will work, early-year students meet most, if not all, of these factors, as they will do the same work as their older peers only later in time.[137] Critics may argue that the bargaining interests of early- and later-year students are different; perhaps later-year students place greater value on family-oriented benefits like childcare than their (on average) younger peers. As a legal matter, relatively minor differences in bargaining priorities are not enough to “negate” an otherwise valid community.[138] In Columbia, for example, the Board found a bargaining unit including doctoral, graduate, and undergraduate workers to “constitute a readily identifiable grouping of employees within the university’s operations that share a community of interest.”[139]

Given that the temporal element is the main distinction, a comparison to the Supreme Court’s community-of-interest discussion in the Pittsburgh Plate Glass retiree case may be most insightful. In Pittsburgh Plate Glass, the Court found that “active and retired employees . . . plainly do not share a community of interests broad enough to justify inclusion of the retirees in the bargaining unit.”[140] The Court found that the interests of retirees were fundamentally different from active workers: retiree “interests extend only to retirement benefits, to the exclusion of wage rates, hours, working conditions, and all other terms of active employment.”[141] The Court worried that, in the words of one author, “a union would sell out its retirees to benefit its active workers.”[142]

Setting aside that final point,[143] early-year graduate students have much more in common with active workers than retirees. Within one or two years, early-year students will be subject to the same exact “wage rates, hours, working conditions, and all other terms of active employment”[144] as their older peers. In fact, early-year graduate students may be working by the time a contract settling those matters is in place, giving them strong motivation to care about bargaining on those topics.[145] Early-year students have essentially the same concerns as their older peers despite the temporal distinction. Consequently, early-year students “share a community of interests broad enough to justify inclusion . . . in the bargaining unit”[146] and can share a bargaining unit with their upper-year peers.

C. Expectation of Future Employment

Second, Board and Circuit case law requires individuals who are employees but not currently working for an employer to “have a reasonable expectation of future employment.”[147] The NLRB and lower courts have developed frameworks for elections involving “seasonal” and “intermittent employees.”[148] “In deciding whether seasonal employees’ [sic] are eligible voters, the Board assesses their expectation of future employment.”[149] The Board performs this analysis using the Maine Apple Growers framework, including “the size of the area labor force, the stability of the Employer’s labor requirements and the extent to which it is dependent upon seasonal labor, the actual reemployment season-to-season of the worker complement, and the Employer’s recall or preference policy regarding seasonal employees.”[150] Circuit courts have consistently cited these factors when reviewing NLRB decisions.[151]

As discussed above, early-year graduate students with degree work requirements have a near-certain expectation of future employment. Universities hire graduate assistants from a limited pool of labor, i.e., graduate and professional students. Universities are highly dependent upon the labor of graduate assistants in the fall and spring semesters.[152] Most importantly, many universities have a “requirement” policy, going beyond a “recall or preference policy,” which compels students to work in certain degree years. Early-year graduate students have much more than a reasonable expectation of future employment and pass the seasonal or intermittent worker test.

D. Employee Status

Third, the NLRA requires voters to be “employees.”[153] As noted above, courts and the NLRB have constructed an unordinary, “ordinary meaning” definition of employment that covers both current and expected employment relationships.[154] Distilling the case law from Part II, individuals can be employees if they have an “actual or anticipated” “economic relationship” with their employer.[155]

Early-year students in degree programs with work requirements have an anticipated economic relationship with their universities. As the dissent in Columbia emphasized, graduate students attend universities to obtain degrees.[156] The students in question must work for their university in order to receive funding and/or graduate.[157] In Columbia, the NLRB held that the work those students will perform makes them “statutory employees.”[158] Early-year students therefore have a cognizable anticipated economic relationship that can make them statutory employees too, even before they begin working for the university.

Early-year graduate students are much more like the hiring-hall registrant employees in IATSE than the unpaid staff nonemployees in WBAI Pacifica and Amnesty International. The crux of the unpaid staff cases was the lack of any economic relationship whatsoever.[159] WBAI Pacifica and Amnesty International repeatedly emphasized that the staff at issue do not “receive compensation” either now or in the future.[160] Unlike those staff, early-year graduate students with work requirements anticipate an economic relationship in the relatively near future. Those students, with rare exceptions, must become statutory employees to graduate, which is the main reason they attend university.

Early-year graduate students are in some ways similar to the hiring hall registrants in IATSE. Like hiring-hall registrants, graduate workers form a pool of labor from which administrators can assign workers to open positions. The D.C. Circuit panel in IATSE held that actual connection with an employer was not necessary to support employee status;[161] neither unattached registrants nor early-year students have such a connection. While early-year graduate students are not available to work on an ongoing basis like registrants, registrants can refuse a position more easily than graduate assistants can. The graduate employment relationship, moreover, is much more permanent than that for registrants, who may “often be referred to an [e]mployer for short periods of time, sometimes only a single day.”[162] The registrants in IATSE had a relatively low probability of a relatively short-term economic relationship with the employers in question. Early-year graduate students with degree work mandates, in contrast, have an extremely high probability of a more durable economic relationship with their universities.

Overall, early-year graduate students with work requirements for degree completion can be employees under the NLRA. At first glance, this Note’s argument may seem to fly in the face of the Court’s “ordinary meaning” test for NLRA employment. It is true that early-year graduate students do not yet “work for another for hire”[163] (or at least do not work for their universities as graduate assistants). However, analysis of judicial and administrative precedent shows that the courts and the Board have constructed an “ordinary meaning” of NLRA employee status that encompasses both current and expected economic relationships. Certain early-year graduate students meet the second prong of this definition and can be statutory employees with the right to participate in union-representation decisions.

E. The Prework Rule and Eligibility Formula

As the above sections show, early-year graduate students with degree work requirements are statutory employees who share a community of interests with their peers. This fact is all that the text of the NLRA requires to vote in a representation election. As quoted above, Section 9(a) of the Act grants recognition to “[r]epresentatives designated or selected for the purposes of collective bargaining by the majority of the employees in a unit appropriate for such purposes.”[164] In other words, the Act explicitly lists two criteria for voter eligibility: employee status and membership in an appropriate unit. Early-year graduate students with degree work requirements meet both criteria.

However, as a policy matter, the Board has adopted the so-called “prework rule” for voter eligibility.[165] According to this “well settled” doctrine, “in order to be eligible to vote, an individual must be employed and working on the established eligibility date, unless absent for one of the reasons set out in the Direction of Election.”[166] The Board “defines ‘working’ as the actual performance of bargaining unit work.”[167] Early-year graduate students with future work requirements would appear to fail this criterion, as they are admittedly not yet performing teaching or research work before the election.

Nonetheless, the Board has recognized that the prework rule is not appropriate in certain “industries [that] do not have the kind of steady employment that is characteristic of the mainstream of industrial enterprise.”[168] In these cases, the Board has found it “necessary to devise an eligibility formula in those industries which will best be tailored to their special needs.”[169] The eligibility formula acts as a substitute for the prework rule and allows workers who are linked to an employer but not currently working to participate in representation elections.[170]

The Board has used these formulas in numerous graduate-worker elections already. In these elections, NLRB regional directors have approved eligibility formulas with “a lookback period” to capture “students [who] do not necessarily work in consecutive semesters but retain a continuing interest in the terms and conditions of employment of the unit.”[171] As discussed by an NLRB regional director in a decision involving Northeastern University graduate workers, Board “regional directors found lookback formulas appropriate in” graduate worker elections at Columbia University, Duke University, University of Chicago, University of Pennsylvania, Vanderbilt University, and Yale University.[172] Public records show several other examples of graduate-worker elections conducted with lookback formulas, including elections at Harvard University and Johns Hopkins University.[173] NLRB officials have already determined that the prework rule can be inappropriate for graduate-union elections and instated “lookback” periods to ensure all interested employees have a voice in the choice of their bargaining representative.

Regional directors should recognize a “look-forward” period in graduate-worker eligibility formulas as well. Current lookback formulas discriminate between similarly situated workers based on something like historical accident. Jane Doe, who worked last year, is not working this year, and will work again next year, can vote. John Smith, who has a degree work requirement for next year but started his degree program a year later, cannot. Both Jane and John may know with certainty that they will perform bargaining-unit work in the near future. Their voter eligibility, however, is different based on whether they happen to have worked in the past. The situation may actually be worse in some cases. Imagine Jane was a sixth-year doctoral student graduating in the spring. In that case, the lookback formula would allow someone who will not work in the future to vote while excluding someone who almost certainly will.

Eligibility formulas are meant to avoid arbitrary outcomes in industries with variable employment.[174] Unfortunately, the Board’s current “lookback” formula for graduate-worker elections does just this: it discriminates between workers with identical future-employment expectations based on whether they performed work in the past. NLRB regional directors supervising graduate union elections should instate a “look-forward” period to capture early-year graduate students with degree work requirements and allow these workers to vote with their peers.[175]

IV. The “Horizontal Dimension” of Graduate-Worker Organizing

Recognizing the intertemporal collective-action problem in graduate organizing highlights how graduate workers offer something larger for the modern labor movement. This Note briefly concludes by considering how graduate unions provide an example of individuals valuing unions for, in the words of Diana Reddy, “doing exactly what they are statutorily designed to do.”[176]

Thirty years ago, George Feldman identified the “horizontal dimension” of union representation, that is, “the legitimate scope of the union’s activities, and the closely related issue of for whom it is legally proper for the union to speak.”[177] Feldman found that the judiciary has taken a narrow view of this horizontal dimension, confining collective bargaining to the interests of unit members as workers.[178] Last year, Reddy resurfaced the connection between union collective bargaining and public interests in a piece contrasting glowing public support for unions as agents of “economic populism,” “antidiscrimination,” and “the common good,”[179] with “the fact that framing unions as helping workers improve their jobs decreases support for them.”[180] Reddy suggested that “labor unions should and must establish . . . that what unions do to improve the working lives of their own members is itself a common good.”[181]

An intertemporal collective-action problem exists in graduate organizing because the pool of beneficiaries is broader and different than the pool of voters. In other words, graduate unions stretch Feldman’s “horizontal dimension.”[182] All unions, in some sense, benefit future workers along a third, temporal dimension. The future workers that graduate unions benefit, however, may be identifiable ex ante: some active workers can be near-certain that their negotiations will benefit many of their friends and peers. At the same time, a sizeable number of active workers can also be near-certain that they will never benefit from the contract they negotiate. Placing these facts side-by-side, upper-year students are both embedded worker-voters and quasi-members of the general public, and their votes can represent a view on the merits of unionization at least partially detached from their personal interests.

Consider the viewpoint of a sixth-year graduate student voting in a unionization election. This voter has had ample experience with the university workplace. They may have significant things that they would want to change about their job or thoughts on what their priorities would be in bargaining. At the same time, they know without a doubt that they will move on before the union negotiates a first contract or establishes a grievance procedure. When this worker considers their experience on the job and chooses to unionize, they are, in a real sense, endorsing the project of unions as agents of workplace change for their peers who will work in the future. Their vote says, “I want a union so that my workplace will be better for those who follow me.”

Graduate unions provide an example for labor leaders considering the challenge Reddy raised. Other unions are responding to the challenge too, as exemplified by the fight against “two-tier contracts” in recent bargaining at employers like Kellogg, Nabisco, Kaiser Permanente, and John Deere.[183] In the 1980s and 1990s, many unions agreed to tiered contracts that sorted workers into different wage and benefit trajectories depending on when they entered the workforce.[184] Older workers remained at higher tiers with legacy wages and benefits, while new entrants started at lower tiers with reduced wages and benefits.[185] Multi-tier contracting takes advantage of an intertemporal collective-action problem: current workers trade better terms for themselves for worse terms that bind future bargaining-unit members.[186]

Over the last few years, workers have demonstrated increased labor militancy through phenomena like “Striketober,”[187] and the election of reform candidates to lead the United Autoworkers (UAW) and the International Brotherhood of Teamsters (Teamsters).[188] Several unions have challenged tiered contracts at the bargaining table and on the picket line,[189] emphasizing how two-tiered agreements violate an intertemporal responsibility owed from one generation of union workers to the next. In 2021, Bakery, Confectionary, Tobacco Workers and Grain Millers (BCTGM) union members struck Kellogg and Nabisco over, among other issues, tiered contracts.[190] Union leadership explained how members owe an intertemporal duty of solidarity to future employees; in the words of BCTGM secretary-treasurer David Woods, “These future workers are like unborn babies . . . . They don’t have a face or a name yet, but we know they’re coming behind us.”[191] UAW members similarly made two-tier contracts a flash point in their 2023 “Stand Up Strike”[192] campaign against General Motors, Ford, and Stellantis.[193] And while UAW President Shawn Fain focused on how tiered contracts violate an innate sense of fairness,[194] other members, like Ford worker Steven Summers, highlighted the intertemporal dimension of tiered bargaining.[195] The BCTGM and UAW have been at the forefront of a movement that has rolled back tiered contracting in a series of successful negotiations and labor actions.[196]

Both unions organizing graduate workers and unions challenging tiered contracts have achieved significant gains despite serious intertemporal collective-action problems. In doing so, they have broadened Reddy’s thesis in an important way. Graduate unions are convincing upper-year students to vote for unions as agents of workplace change for their peers. Workers challenging tiering have foregone gains for themselves to fight for future generations of union siblings. In both cases, unions have stretched the bounds of labor action along the temporal dimension and affirmed that union activity matters both for current employees and for their successors. These unions therefore point toward an intertemporal extension of Reddy’s conclusion, emphasizing “that what unions do to improve the working lives of their own members,” and of their future members, “is itself a common good.”[197]

This Note’s legal proposal raises a way to better align the interests of graduate union voters and beneficiaries. Juxtaposing the recent successes in graduate organizing with the intertemporal collective-action problem underneath this proposal, however, suggests that graduate workers are already convincing their peers that unions have a role to play in American life for what they do on the job.[198] Upper-year graduate students voting to unionize are a model of the individuals that Reddy suggested the labor movement needs to support a resurgence in union membership.[199]

As Rogers noted, intertemporal collective-action problems are present in any union-organizing campaign.[200] Graduate unions, facing a particularly extreme case of intertemporal dissonance, have managed to win election after election. This outcome suggests that organizers in other industries may be able to take lessons in addressing these challenges from the success of graduate organizers. Whether we call it altruism or reflection on lived experience, conversations about the role that unions have for both current and future workers have been persuasive in organizing higher education and challenging tiered contracting.[201] Although a full analysis of the strategies involved in these campaigns is beyond the scope of this Note, these cases are worth further study for labor strategists in other contexts.[202]

Graduate unions are in some ways unique institutions in American labor relations due to the structure of their workforce. Few, if any, sets of workers face such an extreme intertemporal collective-action problem in organizing. And yet, over the past couple of years (and the past decades at public universities), graduate workers have shown resounding support for unionization.[203] Graduate organizers, alongside other members of the labor movement, are already solving intertemporal collective-action problems, in part by convincing upper-year peers “that what unions do to improve the working lives of their own members is itself a common good.”[204] The labor movement should take note of these victories and their voters as it seeks to turn enhanced public support for unions into increased union density.

*     *     *     *     *