The axe has finally fallen. In Janus v. American Federation of State, County, & Municipal Employees, Council 31, the Supreme Court struck down the major source of financial security enjoyed by public-sector unions, which represent nearly half of the nation’s fifteen million union members. Countless press stories, law review articles, and amicus briefs have criticized and defended this outcome.

This Article has a different aim. Rather than relitigating Janus, the questions it asks are instead forward looking: What comes next? Is there life for public-sector unions after Janus? And if so, what might it look like?

In engaging these questions, this Article has three goals. First, it pushes back on the narrative that public unions have no choice now but to struggle on within a national right-to-work environment. That is certainly one possibility, but pro-labor states have available a range of legislative responses that may soften Janus’s blow or even negate it altogether.

One response that has generated vigorous debate in policy circles would be for pro-labor states to require public employers to reimburse unions for their bargaining-related costs directly. The standard objection is that this will undercut unions’ ability to advocate for workers. The Article’s second goal is to confront this objection head-on, with an argument that draws on an unlikely source: an analogy between public unions and public defenders. As it turns out, America’s woeful experience with indigent criminal defense teaches some powerful lessons about how not to fund entities whose purpose is to contest the government’s narrow self-interest. But it also suggests funding approaches that would raise no independence concerns at all.

That leads to the Article’s final objective: to propose model legislation for state lawmakers to implement direct reimbursement of unions. The proposal is revenue neutral for public employers and unions, and it is wealth enhancing for workers in light of nuances in the federal income tax code. Readers interested in the nuts and bolts of the proposed legislation may wish to skip the first three parts of this Article, which make the case for why reimbursement is desirable, and start at Part IV. For convenience, a model bill is included in the Appendix.


  1. LIFE BEFORE JANUS        
    1. The State’s Interests in Adversarial Collective Bargaining
    2. The Fair-Share Fee Approach
  2. JANUS
  3. LIFE AFTER JANUS           
    1. Option #1: Right to Work (or, Do Nothing)
    2. Option #2: Tinkering with Right to Work
    3. Option #3: Members-Only Bargaining
    4. Option #4: Direct Government Reimbursement
      1. The Union-Independence Objection
      2. The Public Defender Analogy
      3. Lessons from Public Defenders: What Not to Do
      4. Lessons from Public Defenders: What Might Work
  4. STATE LABOR LAW AFTER JANUS              
    1. Model Legislation
      1. The Government-Reimbursement Exception
      2. Determining Reimbursement Amounts
    2. Open Questions
      1. When Should the Government Make Payments?
      2. Should Reimbursement Be Mandatory or Permissive?
      3. In Permissive States, What Rights Should Workers Have if Their Individual Public Employers Oppose Reimbursement?
    3. Implementing the Reimbursement Approach
    4. A Brief Response to Professors Fisk and Malin




Woe is the public-sector labor union. Already besieged on political, 1 See Josh Eidelson, Unions Are Losing Their Decades-Long ‘Right to Work’ Fight, Bloomberg Businessweek (Feb. 16, 2017), (on file with the Columbia Law Review) (noting that between 2012 and 2017, six states—Indiana, Kentucky, Michigan, Missouri, West Virginia, and Wisconsin—passed right-to-work legislation prohibiting compelled fair-share fees). technological, 2 See Richard C. Kearney & Patrice M. Mareschal, Labor Relations in the Public Sector 346 (5th ed. 2014) (“[T]echnological innovations have made it increasingly feasible to replace . . . [public] employees with machines or software.”). and demographic fronts, 3 See id. at 345, 347 (describing challenges from the “demographic and sociocultural changes in the workforce,” including the retirement “of staunch union advocates and leaders”). public unions have suffered a major legal blow—perhaps even a “death blow” 4 See, e.g., Michelle Chen, Will the Supreme Court Deal Public-Sector Unionism a Death Blow?, Nation (Feb. 23, 2018), []. —now that the Supreme Court has ruled against them in Janus v. American Federation of State, County, & Municipal Employees, Council 31. 5 138 S. Ct. 2448, 2459–60 (2018).

Janus involved a challenge to the fair-share-fee system (also known as the agency-fee system) used for decades in over twenty states to ensure the financial vitality of public-sector unions. 6 See infra notes 60–63 and accompanying text for a full list of states that permit fair-share fees. Under that system, every worker was required to pay a fee to support the union’s collective-bargaining costs, even if the union took positions to which some workers objected. The rationale was that because the union is legally obligated to represent all workers fairly (even those who oppose it), 7 See, e.g., Vaca v. Sipes, 386 U.S. 171, 181–82 (1967) (describing the duty of fair representation under the National Labor Relations Act, which governs private-sector labor relations); Lane v. I.U.O.E. Stationary Eng’rs, Local 39, 260 Cal. Rptr. 634, 637 (Ct. App. 1989) (holding that the same duty exists for public-sector unions). every worker must pay their “fair share” of the union’s costs. For if workers were allowed to opt out of their share of expenses and still retain the benefits of the union’s representation, even ardent union supporters would have reason to “free ride” rather than pay their fees. 8 See Abood v. Detroit Bd. of Educ., 431 U.S. 209, 222 (1977).

The Supreme Court upheld fair-share fees four decades ago in Abood v. Detroit Board of Education, 9 See id. at 232. but a five-Justice conservative majority invalidated them in Janus. 10 Janus, 138 S. Ct. at 2485–86. All that is left now, it seems, is to survey the wreckage and ask: Just how bad will things get for public unions once fair-share fees are gone? 11 See Adam Ashton, Court Case Will Cost California Unions Big Money Immediately. Then the Real Fight Begins., Sacramento Bee (June 28, 2018),
news/politics-government/the-state-worker/article213953319.html [] (detailing the impact that Janus will have on California public employee unions); Dana Goldstein & Erica L. Green, What the Supreme Court’s Janus Decision Means for Teacher Unions, N.Y. Times (June 27, 2018), (on file with the Columbia Law Review) (projecting that “teachers unions could lose up to a third of their members and funding”); Amelia Thomson-DeVeaux, Could the Supreme Court Really Bust Public-Sector Unions?, FiveThirtyEight (Feb. 26, 2018), [] (describing the then-upcoming Janus argument and analyzing its potential ramifications).
Evidence from states with right-to-work legislation suggests that anywhere from 20 to 71% of public-sector workers will choose to stop paying if given the choice. 12 See infra section III.A. A survey conducted by a leading national union reached a similar result, finding that between 15 and 65% of the workers covered by its contracts are at risk of opting out. 13 See Eidelson, supra note 1. The range of possible outcomes is thus vast: from the weakening of union influence to the end of public-sector unionism as we know it.

So what will life look like after Janus? This Article takes up that question with three core objectives. First, it pushes back against the prevailing assumption that public-sector unions have little recourse but to muddle on within a nationwide right-to-work regime in which unions negotiate contracts on behalf of entire units of workers but are financed by only voluntary dues-paying members. 14 See, e.g., Madeline Will, Are Teachers’ Unions on the Brink of Demise?, Educ. Wk. (Feb. 13, 2018), [] (arguing that teachers unions would see a decrease in membership if the Supreme Court ruled against fair-share fees in Janus); Thomson-DeVeaux, supra note 11 (predicting the future of public-sector unions by “comparing right-to-work and non-right-to-work states and by examining the results of states changing their policies on union fees”). To be sure, that outcome is a real possibility, and this Article considers what life might look like for unions if pro-labor states do not modify their labor laws after Janus. 15 See infra section III.A. There is some indication that states may tinker within a right-to-work regime to help unions attract members by giving employees more paid time to conduct membership drives, or shielding employee information from anti-union advocacy groups, for example. See, e.g., Lauren Rosenhall, California Unions Planning Next Steps if Janus Ruling Goes Against Them, S.F. Chron. (Mar. 4, 2018),
politics/article/California-unions-planning-next-steps-if-Janus-12726938.php [
But labor proponents are not so submissive, and labor law is not so static. 16 See infra Part III (discussing a range of anticipated changes to labor law after Janus).

One possible legislative response is for states to amend the accounting mechanism through which unions receive their current levels of funding. Now that the First Amendment forbids public employers to force workers to pay fees to a union, 17 Janus v. Am. Fed’n of State, Cty., & Mun. Emps., Council 31, 138 S. Ct. 2448, 2486 (2018). employers can just reimburse unions for their bargaining costs directly, recouping that amount through offsetting reductions to future wage raises. Three recent law review publications (and several op-eds) have advanced this government-reimbursement approach, and two Justices have inquired about it during oral argument. 18 See Transcript of Oral Argument at 23, Friedrichs v. Cal. Teachers Ass’n, 136 S. Ct. 1083 (2016) (No. 14-915), 2016 WL 344760 (Sotomayor, J.) (“[Could] the government . . . fund the union?”); id. at 27 (Scalia, J.) (“I suppose you can enact a statute that says the government will fund [the union]?”); Daniel Hemel & David Louk, Is Abood Irrelevant?, 82 U. Chi. L. Rev. Dialogue 227, 234 (2015),
viewcontent.cgi?article=1045&context=uclrev_online [] [hereinafter Hemel & Louk, Is Abood Irrelevant?]; Benjamin I. Sachs, Agency Fees and the First Amendment, 131 Harv. L. Rev. 1046, 1070 (2018); Aaron Tang, Public Sector Unions, the First Amendment, and the Costs of Collective Bargaining, 91 N.Y.U. L. Rev. 144, 150 (2016) [hereinafter Tang, Public Sector Unions]; see also Daniel Hemel & David Louk, How to Save Public Sector Unions, Slate (June 27, 2018),
2018/06/supreme-courts-janus-decision-how-blue-states-can-still-save-public-sector-unions.html []; Aaron Tang, The Janus Ruling Doesn’t Have to Be Fatal for Public Sector Unions. Here’s How., Sacramento Bee (June 27, 2018),
opinion/op-ed/soapbox/article213925464.html []; Benjamin Sachs & Sharon Block, How Democratic Lawmakers Should Help Unions Reeling from the Janus Decision, Vox (June 27, 2018),
public-unions-janus-reforms-fees-decline-reform-supreme-court-hope [].

The nascent literature on the government-reimbursement model is incomplete in one crucial respect, however. It has yet to grapple fully with a fundamental objection: Public-sector bargaining will not work without “independent and credible counterparties for . . . public employers to bargain with.” 19 Brief for the State of California Supporting Affirmance at 12, Janus, 138 S. Ct. 2448 (No. 16-1466), 2018 WL 496675. After all, workers must have confidence that the union truly represents their interests at the bargaining table. But that confidence may be destroyed if the union is beholden to the government employer—the very party against whom it is bargaining!—for its financial security. 20 Id. at 12–13 (arguing that a union may “not be funded by [a public] employer or the State and retain the credibility and independence needed to make the system work”).

This, then, is the Article’s second objective: to provide a full-throated response to the “union-independence” objection. 21 The literature thus far discloses brief mentions of institutional-design options for mitigating the union-independence objection. See Tang, Public Sector Unions, supra note 18, at 217–18 & n.300 (proposing that unions and employers let members or an independent third party serve as a check on funding levels to ensure union independence); see also Hemel & Louk, Is Abood Irrelevant?, supra note 18, at 239 (independent third-party check); Sachs, supra note 18, at 1073 (union-member check). The stakes of this argument are high, for if a government reimbursement model can be designed in a way that truly preserves union independence, the result would be to enable pro-labor public employers to continue pursuing labor peace in a simple manner, very close to the status quo. There will be modest transaction costs in switching to a direct-reimbursement approach. 22 See infra section IV.C. But in principle, the approach would be revenue neutral for employers and unions, 23 See Tang, Public Sector Unions, supra note 18, at 175–77 (describing the revenue neutrality of the direct-reimbursement model from the union and employer perspectives). and wealth enhancing for most union members in light of nuances in federal income tax law. 24 See Hemel & Louk, Is Abood Irrelevant?, supra note 18, at 230; infra notes 373–377 and accompanying text.

So what is the answer to the union-independence objection? The answer begins, perhaps surprisingly, in the world of indigent criminal defense. To see why, recall how the objection posits that an adversarial entity like a union would face an inherent conflict of interest if it received funding from the government. Such an entity, it is argued, would have to choose between advocating zealously against the government (at the risk of losing its funding) and subverting the best interests of its members (to stay in the government’s good graces). But this objection confronts a pervasive counterexample: Government expends vast sums, more than $5 billion nationwide according to one recent count, to support indigent-defense attorneys who also occupy positions adversarial to the government. 25 Holly R. Stevens et al., Ctr. for Justice, Law & Soc’y at George Mason Univ., State, County and Local Expenditures for Indigent Defense Services Fiscal Year 2008, at 7 (2010),
ls_sclaid_def_expenditures_fy08.authcheckdam.pdf [] (noting $5.3 billion in state spending alone in 2008).
Indigent defense in America certainly is not perfect; indeed that might be the furthest possible descriptor from the truth. 26 See, e.g., Erwin Chemerinsky, Lessons from Gideon, 122 Yale L.J. 2676, 2679–80 (2013) (drawing sharp contrast between the “presumption that the presence of counsel has insured adequate representation” and the practical reality); Carol S. Steiker, Gideon at Fifty: A Problem of Political Will, 122 Yale L.J. 2694, 2695–96 (2013) (“[D]ismal failures of representation are all too common.”). But virtually all of the critiques are rooted in the insufficiency of resources, not the distinct kind of charge implicated by the independence objection. 27 See infra notes 208–212 and accompanying text. The insufficient-funding complaint is, if anything, in tension with the independence objection insofar as more government support should exacerbate any supposed conflict. So if the idea of public defenders reaching eye-wink agreements with prosecutors to betray their clients’ interests in exchange for government largesse sounds foreign to you, there may be less to the union-independence objection than meets the eye.

However, the analogy should not be overstated. There are certainly ways in which, although they are both entities adversarial to the government, public unions and public defenders face distinct pressures within distinct legal orders. 28 See infra section III.D.2 (discussing the public union–public defender analogy). But examining the public defender analogy forces us to recognize that not every system for publicly financing an adversarial entity produces the same consequences.

An evaluation of the public defender analogy leads to the Article’s final objective: providing concrete guidance to state lawmakers interested in enacting a direct-reimbursement approach to union financing. Several variables complicate this task, including the diversity of state labor law and the strength of each state electorate’s views on the benefits of public-sector unions. 29 To clarify, the purpose of this Article is not to try to move anyone’s views on the latter front. The aim is instead more neutral and practical: Taking existing levels of support for public unions as a given, this Article seeks to offer institutional-design options for states interested in minimizing Janus’s disruptive effect on their existing systems of public-sector labor relations. The upshot is that the Article’s recommendations will include statutory changes that would be wise for any state that wishes to pursue a reimbursement model, along with some more nuanced statutory revisions that should vary depending on existing state law. For example, every state would benefit from an express exception to existing statutory prohibitions against employer support for unions, though the exact text of each state’s exception should vary in order to incorporate language from (and judicial and administrative interpretations regarding) existing state fair-share-fee laws. 30 See infra section IV.A. Furthermore, some states that previously mandated all public employers to agree to fair-share-fee clauses may choose to do the same for government reimbursement, while others may wish to delegate that choice to individual employers (as most states did with respect to fair-share fees). 31 See infra section IV.B. And in states that choose the latter route, further questions will arise, such as whether reimbursement clauses must or merely may be negotiated (and what should happen upon impasse). 32 See infra section IV.B.

The Article proceeds in four parts. Part I briefly describes life for public-sector unions before Janus and identifies the governmental interests that a union financing system should strive to accommodate. Part II is also cursory, recounting the Court’s reasoning in Janus itself. 33 The Article’s goal is thus not to relitigate Janus or deal with its implications for First Amendment doctrine more broadly. To the latter point, however, see generally Catherine L. Fisk, A Progressive Labor Vision of the First Amendment: Past as Prologue, 118 Colum. L. Rev. 2057 (2018), for a powerful argument for how labor may seek to repurpose Janus’s robust vision of the First Amendment to advance progressive aims.

That sets the table for Part III, which details the four primary options for what life may look like for public unions after Janus: the nationwide right-to-work regime that many seem to take as a given, a right-to-work environment with minor legislative tweaks to aid in union organizing, a “members-only bargaining” approach in which unions would negotiate only on behalf of dues-paying members, and direct union reimbursement. The last of these options occupies the bulk of Part III, which examines the strongest challenge to direct reimbursement: the union-independence objection. This is where this Article sketches out the analogy between public-sector unions and public defenders and attempts to draw out lessons for what pro-labor states ought (and perhaps more importantly, ought not) to do to facilitate independent union representation of worker interests.

Finally, for those who find direct reimbursement promising, Part IV considers state labor law after Janus. Drawing on interviews with a diverse set of public employers, union advocates, and other stakeholders, Part IV offers concrete model legislation for how states interested in undoing the effects of Janus may amend their laws to provide for adequate union financing while minimizing interference with union independence.

To preview the proposal: States would require public employers to bargain over provisions entitling public-sector unions to reimbursement for all bargaining-related expenses—the exact same set of expenses that were chargeable in the agency-fee model. 34 By making a reimbursement clause a mandatory subject of bargaining rather than an across-the-board requirement for every union, state law would give public-sector unions the flexibility to decline reimbursement and trade away that statutory right for some other wage or benefit for workers. To assure union independence, employers would be forbidden to threaten any reduction to reimbursement levels (for example, as a way to leverage union concessions on wages or benefits), just as employers in the agency-fee model could not unilaterally reduce agency-fee payments. And disputes over the reimbursability of particular union expenditures would be resolved using the same neutral administrative process that applied to challenges brought by objecting workers alleging that certain union expenses were nonchargeable under the agency-fee model. 35 See infra section IV.A.2.

This approach may offer a rare kind of win–win–win for public employ­ers, employees, and unions. Public employers could implement the reimbursement model in a revenue-neutral fashion by making parallel wage reductions. Public employees would actually earn more net take-home pay under the reimbursement model due to a reduction to their federal tax burden. 36 See infra notes 373–377 and accompanying text. And from the union’s perspective, direct reim­bursement would largely restore and mirror the pre-Janus agency-fee approach. Just as before, unions would set their budgets based on projected costs and collect member dues to fund their day-to-day activities. 37 Continuing to collect member dues would, among other things, provide liquidity for ongoing costs and prevent a union from becoming dependent on employer funding. For if an employer were to violate its statutory reimbursement duty (thereby kicking off protracted litigation by the union to recover what it is owed) or if a state were to eliminate the reimbursement duty altogether, the union would still be sustained by member dues that it had been collecting all along. See infra section IV.C. At the end of each year, unions would certify the amount of their costs that were germane to collective bargaining, and employers would reimburse that amount. Unions would then rebate a pro rata portion of the reimbursement payment back to each union member, refunding the bargaining-related portion of their member dues. In this sense, direct union reimbursement is really just union member reimbursement.

For that reason, direct reimbursement’s biggest benefit for labor may be its ability to assist unions with the vital task of organizing new members and increasing union responsiveness—a task many union proponents see as a defining challenge post-Janus. 38 See, e.g., Catherine Fisk & Martin H. Malin, After Janus, 107 Calif. L. Rev. (forthcoming 2019) (manuscript at 61–64), (on file with the Columbia Law Review) (explaining the importance of educating workers “about the benefits of membership and the level of membership solidarity in the bargaining unit”); Chris Brooks, Viewpoint: Boss Can’t Be Janus Fix, Labor Notes (July 25, 2018), [
4HAU-E7FS] (arguing that there are no “shortcuts to surviving Janus” and that labor must “get down to the hard work of organizing”).
After Janus, union organizers face the difficult burden of persuading new employees and on-the-fence workers to voluntarily pay full member dues, which can be upward of $1,000 per year, 39 See Alana Semuels, Why Are Unions So Worried About an Upcoming Supreme Court Case?, Atlantic (Jan. 8, 2016), [] (“In California, members pay annual dues that average about $1,000 a year . . . .”). even though they can receive the same benefits by paying nothing. If direct reimbursement is implemented, membership will cost far less because each member will receive a rebate for the bargaining-related portion of the member’s dues. In fact, under direct reimbursement, the price of joining the union—and having a voice in its important affairs—would be the same as the price of joining under the old agency-fee model (just the political portion of dues, often closer to $300 per year). 40 See id. (noting that in California, nonmembers in the agency-fee model still paid “about $600 or $650 for the agency fee alone”). For this reason, the direct-reimbursement approach is as effective a solution to the union collective action problem as the agency-fee model it would replace. As Professors Catherine Fisk and Martin Malin have forcefully argued, this cost reduction combats the union collective action problem by making it “an economically rational decision for an employee to decide to join the union.” 41 See Fisk & Malin, supra note 38, at 47.

One final preface: This Article seeks to offer a holistic account for both why and how the direct-reimbursement model can be a viable replacement for fair-share fees, thereby minimizing the disruption felt by public employers, employees, and unions after Janus. But  those  such  as  state  legislative  staff  members  or  union  advocates  who  are  most  interested  in  the  how — the  nitty-gritty  project  of  state  legislative  reform — may  wish  to  jump  straight  to  Part IV given its focus on design and practice. A model bill (which has already served as a template for draft legislation in a handful of states 42 See, e.g., S.B. 487, 13th Leg., Reg. Sess. (Haw. 2019); Max Parrott, Gottfried’s Janus Workaround Reopens Labor Debate, City & State N.Y. (July 10, 2018), [] [hereinafter Parrott, Gottfried’s Janus Workaround] (describing draft legislation in New York). ) follows in the Appendix.

I. Life Before Janus

To understand the world facing public-sector unions after Janus, one must first appreciate the interests that led certain states to implement the compelled fair-share-fee regime before Janus. 43 For a list of states that adopted this approach as of the Janus decision, see infra notes 60–63. It is the attainment of those interests, after all, that should serve as the ultimate measure of whether any proposed legislative response to Janus is worthy of adoption. This Part accordingly begins by briefly recounting pro-labor states’ underlying interests in public-sector collective bargaining before turning to a description of how fair-share fees became part of a standard approach for achieving them.

A. The State’s Interests in Adversarial Collective Bargaining

At least since Abood, the state interest in public-sector unionization has been described using a single, encompassing phrase: “labor peace.” 44 Abood v. Detroit Bd. Of Educ., 431 U.S. 209, 224 (1977). The phrase actually entails several interests best described successively.

First, the state has a broad interest in fostering worker satisfaction. Workers who feel they have a meaningful voice in the terms and conditions of their employment are happier than those who do not, and all other things equal, happier workers are more productive workers. 45 See Kenneth G. Dau-Schmidt & Arthur R. Traynor, Regulating Unions and Collective Bargaining, in 2 Encyclopedia of Law and Economics: Labor and Employment Law and Economics 96, 109 (Kenneth G. Dau-Schmidt et al. eds., 2009) (describing positive effects from employee organization and noting that “[e]mployees’ collective voice may also directly increase productivity”). Public employers may thus be reluctant to impose wage rates and conditions of employment unilaterally, preferring instead to treat those topics as subjects of back-and-forth negotiation. 46 See Brief for Respondents Lisa Madigan & Michael Hoffman at 38, Janus v. Am. Fed’n of State, Cty., & Mun. Emps., Council 31, 138 S. Ct. 2448 (2018) (No. 16-1466), 2018 WL 481593 (“With the participation of an exclusive representative, the government can establish employment terms in a more durable and stable manner than if it imposed those terms unilaterally.”). There are limits to this proposition, however. Workers would certainly feel happiest of all if they were all permitted to approach their public employers to request increased compensation and were always told “yes.” But at some point, public employers must say “no” given the press of resource constraints and the difficulty of negotiating separately with every individual employee.

That introduces a second, competing interest: the public employer’s concern for improving worker satisfaction efficiently. Workers will be more satisfied if given a say in important workplace decisions, but the employer lacks capacity to negotiate with hundreds—or even thousands—of individual workers on a constant basis. So the government employer may naturally seek to provide some input mechanism for its employees without overrunning its managerial prerogative.

This, of course, is where unions come in, tasked with the responsibility of representing the interests of some body of workers through a single, consolidated voice. But the government employer’s interest in efficiency carries further than to justify the creation of collective bargaining representatives; it also entails a limit of recognizing one and only one representative. For as the Court in Abood itself recognized, allowing multiple unions each to negotiate against a public employer would generate “inter-union rivalries,” “confusion . . . from attempting to enforce two or more agreements specifying different terms and conditions of employment,” and “the possibility of . . . conflicting demands from different unions,” to the detriment of worker satisfaction and employer efficiency alike. 47 Abood, 431 U.S. at 220–21.

From the employer’s perspective, having a single union with which to negotiate is good but not quite enough. The union must also be independent and adequately funded. A union perceived to be beholden to the employer’s own preferred policies might engender distrust among workers, reducing rather than improving workplace satisfaction. Likewise, a union lacking resources might be seen as too weak to advocate on behalf of worker interests and thus unworthy of workers’ participation. The government’s interest in pursuing (1) worker satisfaction in (2) an efficient manner thus requires not just the recognition of a single organization to represent workers’ interests but also that the organization enjoy (3) independence and (4) adequate financial resources.

B. The Fair-Share Fee Approach

How did states arrive upon fair-share fees as part of a system for achieving the just-described interests in public employment? They did not write on a blank slate. By the time public-sector collective bargaining began to emerge in the 1960s, there was already a robust private-sector union model on which to build—the federal National Labor Relations Act (NLRA). 48 See generally Joseph E. Slater, Public Workers: Government Employee Unions, the Law, and the State, 1900–1962, at 158–83 (2004). It was thus quite natural for states to look to this model when designing labor law to govern their public sectors. 49 See id. at 176–77; see also Tang, Public Sector Unions, supra note 18, at 162–67.

The core substantive feature of the private-sector system of labor relations is the duty employers have to bargain collectively and in good faith with their employees on “wages, hours, and other terms and conditions of employment.” 50 29 U.S.C. § 158(d) (2012). Employers are also forbidden from interfering with the right of employees to organize. See id. §§ 157, 158(a)(1). And the keystone structural feature of the private-sector model is exclusive representation, in which a union selected by a majority of workers in a given unit is recognized as the entire unit’s sole bargaining representative. 51 See id. § 159(a) (“Representatives 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 . . . .”).

Public-sector labor law largely replicated these institutional designs. Forty-two states permit at least some form of collective bargaining across various sectors of public employment, 52 See Kearney & Mareschal, supra note 2, at 64–85 (describing policies in the forty-two states with at least some form of recognized public-sector collective bargaining). and all states with comprehensive collective bargaining regimes provide for exclusive representation. 53 See id. at 76 (noting that all states with comprehensive bargaining regimes “provide for exclusive recognition,” like the NLRA does). The commitment to listen to and negotiate with a union representing public-employee interests across an array of topics furthers the state’s worker-satisfaction objective, while the requirement that workers speak through a single voice ensures that this objective is pursued in an efficient manner.

The self-imposed duty to negotiate with an exclusive representative introduces a dilemma, however. Not every worker is going to agree with the positions and priorities taken by the union. For these workers, the system of exclusive representation may actually reduce satisfaction and lead to intraworkplace conflict. This concern is particularly acute given the possibility that, once selected by a majority of workers in a unit, a union may play favorites, privileging the interests of its supporters to the harm of those who oppose it. Here again, the NLRA’s approach to private-sector labor relations offered states a ready-made solution: the duty of fair representation. 54 See supra note 7 and accompanying text; see also Martin H. Malin, The Supreme Court and the Duty of Fair Representation, 27 Harv. C.R.-C.L. L. Rev. 127, 127–28 (1992) (“The duty of fair representation (DFR) is the primary legal vehicle for holding unions accountable to the individual employees they represent.”). Under that duty, a union is required to “exercise fairly the power conferred upon it” not just on behalf of some but rather on behalf of “all those for whom it acts, without hostile discrimination against [any].” 55 Steele v. Louisville & Nashville R.R., 323 U.S. 192, 203 (1944).

But the duty of fair representation creates another conundrum in turn. If workers are guaranteed fair representation by the union at the bargaining table and in grievances with their employers, what incentive does any individual worker have to pay for the union’s services at all? Taken together, the duty of fair representation and the union’s right to exclusive representation create an “incentive” for employees “to become ‘free riders’” who “refuse to contribute to the union while obtaining benefits of union representation that necessarily accrue to all.” 56 Abood v. Detroit Bd. of Educ., 431 U.S. 209, 222 (1977) (quoting Int’l Ass’n of Machinists v. Street, 367 U.S. 740, 761 (1961)). As Fisk and Malin convincingly explain, Janus and Abood’s characterization of the problem created by the duty of fair representation as a “free rider” problem actually misses the mark. See Fisk & Malin, supra note 38, at 8–10. The problem is more accurately described as a collective action problem because the very fact that “it is an economically rational decision for every employee to refrain from joining the union” in turn means that “the union will never be able to negotiate for . . . benefits in the first place.” Id. at 9. Put another way, the collective action problem is that if enough employees do not pay dues, there will be no union-wage premium or other union-negotiated benefit to “free ride” on at all.

Yet again, state lawmakers did not need to look far for an answer. Under the NLRA, private employers and unions may agree to fair-share-fee clauses that require as a condition of employment that all workers—even those who object to the union—share in the union’s bargaining, contract administration, and grievance costs. 57 See 29 U.S.C. § 158(a)(3) (2012) (permitting agreements conditioning employment on union “membership” following a thirty-day grace period); NLRB v. Gen. Motors Corp., 373 U.S. 734, 742 (1963) (stating that when section 8(a)(3) of the NLRA describes eventual union “membership” as a permissible condition of employment, it really means “‘[m]embership’ . . . whittled down to its financial core”). Such clauses directly eliminate the free-rider problem, ensuring adequate union financing. Moreover, by leaving the union’s finances in the hands of workers, the fair-share-fee system ostensibly preserves the value of union independence. 58 But see Tang, Public Sector Unions, supra note 18, at 215–16 (describing how incentives for unions to preserve fair-share-fee clauses in bargaining agreements may also lead to conflict-of-interest concerns).

At the time when Janus was decided, twenty-three jurisdictions (twenty-two states and Washington, D.C.) had adopted the NLRA’s fair-share fee solution to the union-financing problem in the public sector. 59 See infra notes 60–63 and accompanying text. Not every state did so identically, however, and one distinction is particularly worth noting. A few states made the fair-share-fee system mandatory for all public employers. For example, Hawaii law provided that a public “employer shall deduct” a fair-share fee amount “from the payroll of every nonmember employee in the appropriate bargaining unit.” 60 Haw. Rev. Stat. § 89-4(a) (2018) (emphasis added); see also N.Y. Civ. Serv. Law § 208(3) (McKinney 2018); 28 R.I. Gen. Laws § 28-9.3-7 (2018) (prescribing that nonmember public-school teachers “shall pay . . . a service charge as a contribution toward the collective bargaining procedures involved in securing a contract . . . in an amount” equal to union dues); 36 R.I. Gen. Laws § 36-11-2 (2018) (prescribing that nonmember state employees “shall pay . . . a service charge as a contribution toward the negotiation and administration of any collective bargaining agreement” equal to union dues).

Most of the twenty-three fair-share jurisdictions, by contrast, enacted permissive laws that authorized—but did not require—public employers to include fair-share-fee clauses in collective bargaining agreements (CBAs). These laws essentially delegated the choice of whether to implement a fair-share-fee system to individual state and local government employers, who negotiated the matter with unions in turn. Illinois, for instance, provided that “a collective bargaining agreement . . . may include a provision requiring employees covered by the agreement who are not members . . . to pay to the [union] a fair share fee for services rendered.” 61 115 Ill. Comp. Stat. Ann. 5/11 (West 2018) (emphasis added) (for teachers); see also 5 Ill. Comp. Stat. Ann. 315/6(a) (West 2018) (for other public employees). For other state statutes allowing public employers to include fair-share-fee clauses, see Alaska Stat. § 23.40.110(b) (2018); Colo. Rev. Stat. § 29-5-209 (2018) (for “peace officers and firefighters”); Del. Code tit. 19, § 1319 (2018); D.C. Code § 1-617.07 (2018); Me. Rev. Stat. Ann. tit. 26, § 979-B (2018); Md. Code Ann., Educ. § 6-407 (West 2018) (for public-school employees); Md. Code Ann., State Pers. & Pens. § 3-502 (West 2018) (for state employees); Mass. Gen. Laws Ann. ch. 150E, § 2 (West 2018); Minn. Stat. § 179A.06 (2018); Mont. Code Ann. § 39-31-203 (West 2017); N.H. Rev. Stat. Ann. §§ 273-A:1, 273-A:3 (2018); N.J. Stat. Ann. § 34:13A-5.5 (West 2018); N.M. Stat. Ann. § 10-7E-4 (West 2018); Ohio Rev. Code Ann. § 4117.09(C) (2018); Or. Rev. Stat. § 243.672(1)(c) (2017); 43 Pa. Cons. Stat. § 1102.3 (West 2018); Wash. Rev. Code §§ 41.59.100, 41.80.100, 47.64.160 (2018) (for school employees, state employees, and marine employees, respectively).

Three other states mandated fair-share fees for certain sectors of public employment while leaving other public employers to decide whether to include such clauses as a matter of discretion. Thus, for example, California’s Educational Employment Relations Act made clear that a public-school employer “shall deduct” a fair-share fee from the wages of all non-union-member teachers “upon receiving notice from the exclusive representative” that it has been certified by a given bargaining unit. 62 Cal. Gov’t Code §§ 3546(a), 3583.5(a) (2019). However, California teachers can vote to rescind a fair-share clause. See Cal. Gov’t Code § 3546(d). Other states that require mandatory fair-share fees for only certain sectors of government employment include Connecticut and Vermont. See Conn. Gen. Stat. § 5-280(a) (2018) (mandatory fair-share fees for state employees); Vt. Stat. Ann. tit. 16, § 1982(a) (2018) (mandatory fair-share fees for teachers). Other kinds of public employees in California, though, were subject to a permissive regime in which employers could agree to fair-share-fee arrangements but were not required to do so. 63 See, e.g., Cal. Gov’t Code § 3502.5(a) (permitting, but not mandating, fair-share fees for local public employees); id. §§ 3513(k), 3515, 3515.7(a) (same for state employees); id. § 3573 (same for higher-education employees); Cal. Pub. Util. Code §§ 99564, 99566.1(a) (2019) (same for transit employees); Conn. Gen. Stat. § 7-477 (2018) (same for Connecticut municipal employees); id. § 10-153a(c) (same for Connecticut teachers); Vt. Stat. Ann. tit. 3, § 962(10) (2018) (same for Vermont state employees).

Not every state legislature arrived at the same favorable view of fair-share fees, of course. Twenty-eight states—including five since 2012 64 Those states are Indiana, Kentucky, Michigan, West Virginia, and Wisconsin. See Right to Work States Timeline, Nat’l Right to Work Comm., [] (last visited Oct. 15, 2018). — concluded that the claimed benefits of public-sector unionization are not worth the costs of imposing a fair-share-fee regime. These states accordingly enacted so-called “right to work” laws that forbid public employers and unions to agree to fair-share-fee clauses at all. 65 For a current list of these states, see id. Union opponents hoped Janus would add the remaining states to this list in a kind of First Amendment big bang.

II. Janus

As just discussed, a number of pro-labor states made the conscious decision when designing their public-sector labor laws to import the NLRA’s fair-share-fee model into the public sector. 66 See supra notes 60–63 and accompanying text. That choice implicated more than just policy concerns; it raised constitutional questions too. 67 For a discussion of the historical context in which state legislatures made this choice, which arguably justified their belief that public-sector fair-share fees triggered no First Amendment concerns, see Tang, Public Sector Unions, supra note 18, at 167–72. After all, there are many things that private employers may do that public employers may not in light of constitutional prohibitions on state action. 68 See, e.g., Sophia Z. Lee, A Revolution at War with Itself? Preserving Employment Preferences from Weber to Ricci, 123 Yale L.J. 2964, 2992 (2014) (observing that “Title VII preserved space for voluntary preferences by private employers that were no longer available to public employers under equal protection”). A private employer may fire an at-will employee based on disagreement with the employee’s political speech, for instance, but the First Amendment limits the public employer’s ability to do the same. 69 Compare Sean Illing, Can the NFL Fire Players for Kneeling During the Anthem? 9 Legal Experts Say Yes., Vox (Sept. 25, 2017), [] (detailing the opinions of legal academics who agreed that NFL teams could fire players for kneeling during the national anthem), and Chris Kluwe, I Was an NFL Player Until I Was Fired by Two Cowards and a Bigot, Deadspin (Jan. 2, 2014), [] (describing the views of an NFL punter allegedly fired for advocating for gay rights), with Van Ooteghem v. Gray, 654 F.2d 304, 304–05 (5th Cir. 1981) (en banc) (per curiam) (holding that the First Amendment forbids a public employer from firing an employee for speech in favor of gay rights). So the question in Janus was whether compelling objecting workers to financially support a union on pain of their jobs runs afoul of those workers’ free speech rights. 70 Janus v. Am. Fed’n of State, Cty., & Mun. Emps., Council 31, 138 S. Ct. 2448, 2459–62 (2018).

Of course, Janus was not the first occasion the Supreme Court considered this question. The Court confronted and resolved it in Abood v. Detroit Board of Education, crafting a compromise that allowed public employers to force all workers to share in a union’s costs of “collective bargaining, contract administration, and grievance adjustment,” but not for its costs pursuing “political . . . [or] ideological causes not germane to its duties as collective-bargaining representative.” 71 431 U.S. 209, 225–26, 235 (1977). That basic dividing line—between chargeable bargaining-related costs and nonchargeable political costs—has served as the backbone of public-sector fair-share arrangements ever since Abood. Pro-labor states have relied on this line in drafting statutory provisions defining the scope of fair-share fees that may be collected from objecting workers, 72 See infra note 291. and a vast body of case law and state administrative law has emerged to clarify which expenses are chargeable and which are not. 73 See infra notes 272–273 and accompanying text.

In recent years, however, union opponents began to take aim against even the charging of bargaining-related expenditures. Conservative five-member majorities on the Court responded by criticizing fair-share fees as an “anomaly” in a 2012 case 74 See Knox v. Serv. Emps. Int’l Union, Local 1000, 567 U.S. 298, 311 (2012) (referring to the “justification for compelling nonmembers to pay a portion of union dues” as “something of an anomaly”). and then invalidating a fair-share-fee arrangement two years later in the context of home health care workers who were joint employees of a state and a private party. 75 See Harris v. Quinn, 134 S. Ct. 2618, 2644 (2014) (invalidating fair-share fees charged to home-care personal assistants). When the Court granted certiorari the following year to review the constitutionality of a run-of-the-mill fair-share-fee agreement in Friedrichs v. California Teachers Ass’n, the writing seemed to be on the wall. 76 See 135 S. Ct. 2933 (2015) (mem.); Friedrichs v. Cal. Teachers Ass’n, No. SACV 13-676-JLS (CWx), 2013 WL 9825479, at *1–2 (C.D. Cal. Dec. 5, 2013), aff’d, No. 13-57095, 2014 WL 10076847 (9th Cir. Nov. 18, 2014). Justice Scalia’s death shortly after oral argument in Friedrichs led to a brief reprieve for public-sector unions. 77 See Friedrichs v. Cal. Teachers Ass’n, 136 S. Ct. 1083 (2016) (per curiam) (mem.) (affirming by an equally divided Court). But after Justice Gorsuch was confirmed to Scalia’s vacant seat, labor opponents quickly filed another case: Janus.

The facts of Janus were unremarkable. Mark Janus was an Illinois child support specialist who was employed in a bargaining unit represented by the American Federation of State, County, and Municipal Employees (AFSCME), Council 31. 78 Janus v. Am. Fed’n of State, Cty., & Mun. Emps., Council 31, 138 S. Ct. 2448, 2461 (2018). Although Janus was not a member of the union, he was compelled to pay it a fair-share fee under the terms of a collective bargaining agreement between AFSCME and the state. 79 Id. The fee amounted to $44.58 per month. 80 Id. Unhappy with this requirement, Janus intervened in a lawsuit challenging the fair-share fee. 81 Id. at 2462. After the Seventh Circuit ruled against him, the Supreme Court granted certiorari. 82 See Janus v. Am. Fed’n, 138 S. Ct. 54 (2017) (mem.) (granting certiorari).

The Court ruled in his favor in a 5-4, party-line vote. 83 Janus, 138 S. Ct. at 2459–60; id. at 2487 (Sotomayor, J., dissenting); id. at 2487 (Kagan, J., dissenting) (joined by Breyer, Ginsburg & Sotomayor, JJ.). In an opinion written by Justice Alito, the conservative majority began by identifying the free speech values served by the Court’s general disdain for government-compelled speech. 84 Id. at 2462–64 (majority opinion) (“Compelling individuals to mouth support for views they find objectionable violates [a] cardinal constitutional command.”). And although Janus was technically required to financially support the speech of a union rather than to speak a particular message himself, the majority noted that “[c]ompelling a person to subsidize the speech of other private speakers raises similar First Amendment concerns.” 85 Id. at 2464 (emphasis omitted). The majority thus proceeded to apply heightened scrutiny to the fair-share fee requirement, asking whether the state possesses a “compelling” interest that “cannot be achieved through means significantly less restrictive of associational freedoms.” 86 Id. at 2465 (internal quotations marks omitted) (quoting Knox v. Serv. Emps. Int’l Union, Local 1000, 567 U.S. 298, 310 (2012)).

The majority concluded that the State of Illinois and the union failed to surmount that standard. The Court assumed without deciding that the state possessed sufficiently compelling interests in labor peace. 87 Id. (“We assume that ‘labor peace[ ]’ . . . is a compelling state interest . . . .”). But it held that compelled fair-share fee payments were not the least restrictive means for achieving those interests because unions can succeed even under a voluntary, right-to-work regime. For example, the majority noted that “[u]nder federal law, a union chosen by a majority vote” of a relevant unit of federal workers is “designated as the exclusive representative,” yet “federal law does not permit agency fees.” 88 Id. at 2466. Nonetheless, the majority observed, union members comprise “about 27% of the federal work force.” 89 Id. The same is true, the majority concluded, of “millions of public employees in the 28 States that have laws generally prohibiting agency fees.” 90 Id. In short, the majority concluded that fair-share fees are unconstitutional because a public-sector labor-relations system could succeed even if unions are funded only by voluntary union payments.

While this topline holding was unsurprising, the majority went one step further to address an argument buried on the last three pages of Janus’s opening brief: whether to invalidate the opt-out system that had operated in public-sector labor law for decades. 91 See Brief for Petitioner at 61–63, Janus, 138 S. Ct. 2448 (No. 16-1466), 2017 WL 5952674. Under that system, public employers use a default approach in which they automatically deduct the amount of a fair-share fee from a nonmember’s wages and then provide the nonmember a certain window during which they can opt out of the fee. 92 Janus, 138 S. Ct. at 2461; see also Davenport v. Wash. Educ. Ass’n, 551 U.S. 177, 181 (2007) (explaining that the First Amendment does not “mandate[ ] that a public-sector union obtain affirmative consent before spending a nonmember’s agency fees for purposes not chargeable under Abood”). Unlike in Friedrichs, 93 See Petition for a Writ of Certiorari at i, Friedrichs v. Cal. Teachers Ass’n, 136 S. Ct. 1083 (2016) (No. 14-915), 2015 WL 393856 (“Whether it violates the First Amendment to require that public employees affirmatively object to subsidizing nonchargeable speech by public-sector unions, rather than requiring that employees affirmatively consent to subsidizing such speech.”). the Court did not grant certiorari on that question in Janus—most likely because the union respondent in Janus itself already utilized an opt-in approach for payment of full union dues. 94 See Brief for Respondent American Federation of State, County, & Municipal Employees, Council 31 at 58, Janus, 138 S. Ct. 2448 (No. 16-1466), 2018 WL 509654 (explaining that “[f]ull union dues are collected [by the union] only from employees ‘who individually request it’” (quoting Joint Appendix at 58, Janus, 138 S. Ct. 2448 (No. 16-1466), 2017 WL 6887533)). Yet the Court ruled on the issue anyhow, declaring that no payment to the union may be collected from a worker “unless the employee affirmatively consents to pay.” 95 Janus, 138 S. Ct. at 2486.

Justice Kagan dissented, joined by Justices Ginsburg, Breyer, and Sotomayor. In addition to providing a defense of Abood, Justice Kagan took strongest exception to the majority’s stunted application of stare decisis. 96 See id. at 2497, 2499 (Kagan, J., dissenting) (“But the worse part of today’s opinion is where the majority subverts all known principles of stare decisis.”). “[T]housands of current contracts covering millions of workers,” the dissent lamented, will be upended as a result of Janus. 97 Id. at 2499. Justice Kagan further worried that this may just be the start, given that “almost all economic and regulatory policy affects or touches speech” in a way that the majority’s reasoning renders potentially dubious. 98 Id. at 2502. With characteristic verve, Justice Kagan concluded by arguing that “the majority’s road runs long. And at every stop are black-robed rulers overriding citizens’ choices[,] . . . including over the role of public-sector unions.” 99 Id.

III. Life After Janus

One critical aspect of the Court’s reasoning in Janus is its conclusion that a system of union financing based on dues voluntarily paid by members would be an adequate alternative for achieving the state’s interests in labor peace. 100 See id. at 2466 (majority opinion) (“Whatever may have been the case 41 years ago when Abood was handed down, it is now undeniable that ‘labor peace’ can readily be achieved ‘through means significantly less restrictive of associational freedoms’ than the assessment of agency fees.” (quoting Harris v. Quinn, 134 S. Ct. 2618, 2639 (2014))). Fair-share fees are unnecessary, in other words, because unions can function just fine within a right-to-work legal regime. That is an empirical proposition that may or may not turn out to be true. But from labor’s perspective, there is no inevitable reason to find out.

This Part accordingly considers what life might look like for public-sector unions after Janus under a range of legislative alternatives. It begins by reviewing the right-to-work default—that is, how public unions will likely be affected if states make no major legislative changes in response to Janus. The top-line conclusion is that unions should expect a significant reduction in voluntary membership ranging anywhere from 15 to 71%, with similar downstream impacts on their budgets. 101 See infra section III.A. This Part next considers legislative responses that accept the right-to-work baseline but tinker on the margins so as to encourage union membership. 102 See infra section III.B. It then turns to the members-only bargaining alternative that would allow unions to negotiate contracts that apply only to their members. 103 See infra section III.C. The Article describes its salutary attributes before explaining why both government employers and unions may be reluctant to pursue the legislative changes needed to implement it. 104 See infra section III.C. The final section explores the direct-reimbursement alternative. 105 See infra section III.D. It starts by advancing the primary objection, which is rooted in the concern for union independence. The Part concludes by drawing on experience from the analogous setting of indigent criminal defense to suggest how government may avoid that concern by funding union activities in particular ways.

A. Option #1: Right to Work (or, Do Nothing)

What would life look like for public unions if states make no material legislative changes in response to Janus? To answer this question is largely to engage in behavioral prediction. The key issue is how many public workers in the twenty-three formerly fair-share jurisdictions will continue to pay their union dues after they are freed of that obligation, given that they will retain the benefits of union representation no matter what.

There is already something of a natural experiment that can shed light on the question. Since 2000, eight states—Florida, Idaho, Iowa, Kansas, Nebraska, Nevada, North Dakota, and South Dakota—have implemented a public-sector labor-law environment that is materially similar to the statutory regime that the twenty-three former fair-share jurisdictions will find themselves in after Janus. 106 See Brief of Amicus Curiae Mackinac Center for Public Policy in Support of Petitioners at 31–38, Friedrichs v. Cal. Teachers Ass’n, 136 S. Ct. 1083 (2016) (No. 14-915), 2015 WL 5461532 [hereinafter Mackinac Center Friedrichs Amicus Brief]. Note that Iowa limited the scope of collective bargaining in 2017, but that does not affect the data discussed below (which is based on pre-2017 membership numbers). See Jason Noble & Brianne Pfannenstiel, Here Are the 5 Key Changes in Iowa’s Collective Bargaining Bill, Des Moines Reg. (Feb. 8, 2017),
2017/02/08/here-5-key-changes-iowas-collective-bargaining-bill/97658446/ [
YHA9-CCDA] (“In its broadest strokes, the 68-page legislation creates separate bargaining processes for public safety and non-public safety workers, largely preserving existing bargaining rights for law enforcement officers, firefighters and other public safety workers while stripping them away for everyone else.”).
That is, eight states have employed a comprehensive system of collective bargaining wherein a single union is granted the right to exclusively represent all workers in a unit subject to a duty of fair representation, but in which fair-share fees are prohibited by virtue of right-to-work legislation. 107 See Mackinac Center Friedrichs Amicus Brief, supra note 106, at 33–35.

In theory, then, the rates of voluntary union membership in these states should provide some clue as to what would befall public-sector unions in the twenty-three jurisdictions that formerly authorized fair-share fees. The comparison is not ideal; social, cultural, and political differences may lead to a more union-friendly environment in the twenty-three (largely blue) jurisdictions than in the eight legislative right-to-work (purple-reddish) comparators. 108 Of the eight states, only Florida, Iowa, and Nevada could plausibly be termed swing states, while the five remaining states—Idaho, Kansas, Nebraska, North Dakota, and South Dakota—lean heavily Republican. See State of the States, Gallup,
poll/125066/state-states.aspx [] (last visited Nov. 10, 2018) (aggregating polling data based on a variety of metrics, including percentage of each state that leans Republican and percentage of the state that leans Democratic).
So in considering the numbers below, one should bear in mind that the experience in the eight legislative-right-to-work states may oversell the extent of union decline.

What do the numbers look like? It turns out that the answer turns heavily on one’s choice of data sets. The shift to a right-to-work regime may reduce rates of union fee-paying by as little as 20% (per the Bureau of Labor Statistics Current Population Survey), 109 See infra note 117 and accompanying text. by a more significant 34% (per the National Center of Education Statistics Schools and Staffing Survey), 110 See infra note 125 and accompanying text. or by as much as 71% (according to a payroll deduction method). 111 See infra note 129 and accompanying text. A discussion of these competing methodologies follows.

One  of  the  early  forays  into  predicting  the  impact  of  a  ruling  against  public-sector  unions  was  presented  in  an  amicus  brief filed in Friedrichs v. California Teachers Ass’n, 112 136 S. Ct. 1083 (2016) (per curiam) (mem.) (affirming by an equally divided Court after Justice Scalia’s passing). the precursor case to Janus. There, the Mackinac Center for Public Policy, a free-market think tank, utilized data from the Bureau of Labor Statistics’ Current Population Survey (CPS). 113 See Mackinac Center Friedrichs Amicus Brief, supra note 106, at 14–38. The CPS is a “monthly sample survey of about 60,000 households that obtains information on employment” including union membership and coverage by a collective bargaining agreement. 114 Press Release, Bureau of Labor Statistics, Union Membership (Annual) News Release (Jan. 24, 2014), []. As relevant here, the survey asks respondents (A) whether they are a member of a labor union, and (B) if they are not a union member, whether they are covered by a collective bargaining agreement nonetheless. 115 See U.S. Census Bureau, Current Population Survey Interviewing Manual: June 2013, at B5-4 (2013), (on file with the Columbia Law Review).

The Mackinac Center used the total number of union members given in response to Question (A) as the numerator, adding that number to the number of nonmembers covered by a bargaining agreement in Question (B) to form a denominator. 116 See Mackinac Center Friedrichs Amicus Brief, supra note 106, at 35–36 & n.52. The result of this calculation was eye-opening: Mackinac found that union membership rates in the eight legislative right-to-work states hovered around 75–80% between 2000 and 2014, implying that just 20–25% of all workers opt out of paying dues when given the choice. 117 See id. at 35–37.

Two Harvard economists and a University of Wisconsin political scientist filed a forceful amicus brief responding to the Mackinac brief, challenging the methodology used therein. 118 Brief of Amici Social Scientists in Support of Respondents at 1–2, Friedrichs v. Cal. Teachers Ass’n, 136 S. Ct. 1083 (2016) (No. 14-915), 2015 WL 7252638 [hereinafter Brief of Amici Social Scientists]. The critical error in that brief, the social scientists observed, is that not all union members who respond “yes” to Question (A) on the CPS are covered by a collective bargaining agreement. 119 See id. at 4–5 (“Mackinac’s measure relies exclusively on the CPS, which does not ask self-identified union members if they are covered by CBAs. Mackinac converts this gap in CPS data into an assumption that all union members are covered by CBAs. This unjustified assumption is demonstrably false.”). Many public workers, it turns out, choose to join unions that have not negotiated collective bargaining agreements with their employers. 120 See id. at 22 (finding, for example, that between 2011–2012, 17% of teachers who are union members in states that mandate collective bargaining but prohibit fair-share fees are not covered by a collective bargaining agreement). The consequence is that the CPS method artificially inflates the numerator by including in it a sizable number of workers who are members of unions different than the unions whose covered workers are used to calculate the denominator. 121 See id. at 20–27 (“Mackinac relies exclusively on the CPS . . . [which] asks those not belonging to a union if they are covered by a CBA, but does not ask that question of union members. This gap invites the assumption that all union members are covered by a CBA.”).

The social scientists proceeded to run their own calculation using a different data source: the National Center for Education Statistics Schools and Staffing Survey (SASS) from the National Center for Education. 122 Schools and Staffing Survey (SASS), Nat’l Ctr. for Educ. Statistics, [] (last visited Oct. 16, 2018). Unlike  the  CPS,  the  SASS  does  ask  whether  union  members  are  covered  by  a  bargaining  agreement  so  as  to  identify  the  correct  numerator,  although  the  survey  is  limited  to  the  public  education sector. 123 See Brief of Amici Social Scientists, supra note 118, at 22 (“The SASS data . . . permit precise specification of the total CBA-covered population and its components: dues paying union members, agency-fee paying nonmembers (which Mackinac effectively ignores), and free-riders.”). In  any  event,  the  result  of  the  SASS  methodology  is  to  find  a  much  lower  rate  of  teacher  union  membership — 66% — in  right-to-work  states. 124 See id. at 23 tbl.3. Note that the social scientists include three additional states (Indiana, Oklahoma, and Tennessee) in their group of relevant comparators with an employer duty to bargain collectively, exclusive representation, a duty of fair representation, and a legislative right to work. See id. at 18 tbl.1. It is unclear whether or how the inclusion of these states affects the SASS-calculated free-rider rate of 34%. Id. at 23 tbl.3. That implies that the post-Janus right-to-work jurisdictions may expect a significantly larger free-rider rate of around 34%. 125 Id. at 23 tbl.3.

To its credit, the Mackinac Center filed another amicus brief during the Janus litigation acknowledging the possible bias in its CPS methodology and proposing a third method based on state employer payroll deduction data. 126 See Brief of Amicus Curiae Mackinac Center for Public Policy in Support of Petitioner at 28, Janus v. Am. Fed’n of State, Cty., & Mun. Emps., Council 31, 138 S. Ct. 2448 (2018) (No. 16-1466) (acknowledging that the CPS method may “lead to more respondents being counted as union members” than appropriate). Under this payroll-deduction methodology, the Mackinac Center leveraged the fact that many government employers deduct union fee payments directly out of worker paychecks. 127 See id. at 10–11. Note that the Mackinac Center was able to obtain only statewide public employee data. Because employers also know how many workers are covered by their contracts, 128 Id. employer-provided data may yield the most accurate picture of union membership rates in right-to-work states, given that even the SASS method­ology relies ultimately on surveys of workers who may not accurately report their union membership status or coverage under a collective bargaining agreement.

The final figures are again eye-opening, only this time for the opposite reason as the Mackinac Center’s initial findings. In the eight states with comprehensive bargaining, exclusive representation, a duty of fair representation, and a legislative right-to-work, the payroll-deduction method revealed a meager 29% union membership rate. 129 See id. at 17–18. Put another way, as many as 71% of the public workers in these states who were covered by a collective bargaining agreement free rode on the dues paid by their colleagues. Mackinac points out that this is likely higher than the true free-rider rate insofar as some public workers may choose to pay their union dues via cash, check, or credit card, rather than automatic payroll deduction. 130 See id. at 11. But the key point is that any organization that faced a sudden loss of a significant portion of its membership (when its entire funding stream comes from those members) would find itself in dire straits indeed. 131 By way of comparison, total K–12 public-education spending fell nationwide by roughly 4% after the Great Recession, with significant effects on reading and math scores as well graduation rates. See Matt Barnum, Less Money for Schools After the Recession Meant Lower Test Scores and Graduation Rates, Study Finds, Chalkbeat (Jan. 12, 2018), []; see also C. Kirabo Jackson et al., Do School Spending Cuts Matter? Evidence from the Great Recession 22 (Jan. 9, 2018) (unpublished manuscript),
c_kirabo_jackson/35/ (on file with the Columbia Law Review) (“Unfortunately, we find students that experienced reduced public school spending had both lower test scores but also less high school completion.”).
After all, unions facing such steep membership declines can do little to reverse the accompanying budgetary effects since any increase to membership dues would serve only to exacerbate workers’ incentive to opt out of joining the union in the first place.

Some more data points may be of interest. First, as noted above, the just described range of free-rider rates—from 20 to 71%, depending on  the data source—is strikingly similar to data from one major national union’s own internal surveys. 132 See supra notes 113–130 and accompanying text. The American Federation of State, County, and Municipal Employees conducted an impressive set of 600,000 one-on-one interviews with workers covered by its contracts and learned that approximately 15% of them “would likely not pay dues under right-to-work” and that around an additional 50% would be “on the fence.” 133 See Eidelson, supra note 1. This implies a potential range of between 15 to 65% who would free ride.

Second, in the three to five years since Wisconsin and Michigan enacted their public-sector right-to-work laws, the state affiliates of the National Education Association saw decreases in membership of 52% and 21%, respectively. 134 See Bradley D. Marianno & Katharine O. Strunk, After Janus, Educ. Next, Fall 2018, at 18, 21. However, much of this decline in both states may be attributable to limits on union collective bargaining rights, which further reduced teachers’ incentive to join. See, e.g., Steven Greenhouse, Wisconsin’s Legacy for Unions, N.Y. Times (Feb. 22, 2014), (on file with the Columbia Law Review). For comparison, the National Education Association expects a 14% membership loss nationwide in just the first two years following Janus, although that percentage includes membership in legislative right-to-work states and thus understates the likely loss in the states affected by Janus. See Madeline Will, The Largest Teachers’ Union Predicts a 14 Percent Membership Loss over Two Years, Educ. Wk.: Tchr. Beat (July 1, 2018),
court_ruling.html [].
The evidence thus suggests that public-sector unions will suffer a significant membership and financial decline under right-to-work. Indeed, there is some reason to think the real-life implications of Janus may fall on the more disruptive side of the projections just given, since Janus invalidates the preexisting opt-out rule for union fee payments. 135 See supra notes 91–95 and accompanying text. As behavioral economists have long noted, persons with equal interest in paying for a good or service will show radically different levels of participation depending on whether they are automatically enrolled with the right to opt out or affirmatively required to opt in. 136 See Brigitte C. Madrian & Dennis F. Shea, The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior, 116 Q.J. Econ. 1149, 1149–50 (2001) (describing substantial increases in participation in 401(k) savings plans under an automatic-enrollment approach).

Not all unions will be similarly affected, of course. As the Wisconsin and Michigan data show, unions may be hit harder in some states than others. 137 See supra note 134 and accompanying text. And unions in some sectors of employment may be more resilient. 138 See Press Release, Bureau of Labor Statistics, Union Members—2017, at 2 (Jan. 19, 2018), [] (noting that divergent union density rates depend on occupational group and sector of employment). Certain workplace cultures may be more suited to overcoming the free-rider problem than others by virtue of the nature of the profession. For instance, a recent study of New York City public unions found a wide range of rates of nonmembers under the fair-share model. 139 See Ruth Milkman & Stephanie Luce, Joseph S. Murphy Inst. for Worker Educ. & Labor Studies, The State of the Unions 2017: A Profile of Organized Labor in New York City, New York State, and the United States 5 tbl.B1 (2017),
CUNY_GC/media/CUNY-Graduate-Center/PDF/Press%20Room/Union_Density_2017_C.pdf [].
Whereas only 4.8% of teachers covered by contracts negotiated by the United Federation of Teachers declined to join the union (in turn triggering their duty to pay fair-share fees), the rate of nonmember fee payers was 8% among nurses, 11.7% among transport workers, and 43.2% among staff analysts. 140 Id. This suggests that while public-sector unions as a whole will be worse off under right-to-work, the impact will likely be felt unevenly in different kinds of public employment.

B. Option #2: Tinkering with Right to Work

Faced with the inevitability of substantial membership losses, public-sector unions have already succeeded in pushing for some minor legislative reforms. However, unlike the legislative alternatives discussed in the next two sections, which radically change the rules of public-sector labor relations, these more modest reforms take right-to-work as the baseline. The reforms thus aim to tinker with the right-to-work environment in which organizers persuade public workers to join the union, with the hope of stemming the decline of membership as much as possible. Four kinds of reforms characterize this approach.

First, some states have passed laws with the aim of narrowing the time period in which objecting workers may opt out of paying dues. For example, Hawaii enacted H.B. 1725, which gives objecting workers a thirty-day window in which they can notify their union of their desire to cease payments. 141 See H.B. 1725, 29th Leg., Reg. Sess. (Haw. 2018). Delaware enacted H.B. 314, which would let unions set the terms for when objecting workers can revoke their payments and would otherwise impose a default period of fifteen to thirty days. 142 See H.B. 314, 149th Gen. Assemb., Reg. Sess. (Del. 2018). These laws, however, may be susceptible to constitutional challenge to the extent they allow unions to continue deducting fees from workers who have never consented. For, as noted earlier, Janus held that workers must affirmatively consent before any fee may be deducted from their wages. 143 See supra note 95 and accompanying text.

Second, a number of states have passed laws approaching the problem from a different direction by giving unions more opportunities to meet with public-sector workers to convince them to become members in the first place. New Jersey’s Workplace Democracy Enhancement Act is a good example. 144 See N.J. Stat. Ann. § 34:13A (West 2018). That law grants unions the right to meet with newly hired employees and requires public employers to turn over employee contact information. 145 Id. Two new Maryland laws contain similar employee-access provisions. 146 See H.B. 1017, 2018 Gen. Assemb., Reg. Sess. (Md. 2018) (state employees); H.B. 811, 2018 Gen. Assemb., Reg. Sess. (Md. 2018) (public-school employees). Such laws are likely to help unions, at least incrementally, in their efforts to sign up new members, even if they do not address the free-rider problem.

Third, some states have taken the protective measure of insulating employee contact information from anti-union groups that may try to persuade workers to quit the union. New York Governor Andrew Cuomo signed an executive order ensuring the privacy of public-sector worker contact information. 147 N.Y. Exec. Order No. 183 (Aug. 6, 2018), [].
California Senate Bill 866 requires the time and place of new employee union orientations to be kept confidential. 148 S.B. 866, 2018 Leg., 2017–2018 Reg. Sess. § 16 (Cal. 2018) (“The date, time, and place of the orientation shall not be disclosed to anyone other than the employees, the exclusive representative, or a vendor that is contracted to provide a service for purposes of the orientation.”). Such efforts seem likely to help stem the tide of membership losses, although they, too, have no effect on the basic incentive for workers to free ride.

Finally, New York has taken the more significant step of modifying unions’ obligation to fairly represent nonmembers: Although unions would continue to be required to represent such workers fairly during contract negotiations, unions would be free to refrain from representing nonmembers during grievance proceedings. 149 See Brian Young, Eight Ways New York Is Strengthening Unions Before Janus, UCOMM Blog (Apr. 19, 2018), [] (“[T]he union shall not be obligated to represent the person in any investigation of employee . . . discipline. . . . The union shall only represent the non-dues paying member as it relates to the negotiation of terms of a collective bargaining agreement or enforcement of same.”).
International Union of Operating Engineers Local 150 in Illinois recently filed a lawsuit seeking similar relief. Plaintiffs in that case ask that, in the event that Janus invalidates fair-share fees, the First Amendment should likewise grant them a right not to represent objecting workers in contract-administration matters. See Complaint for Declaratory Relief at 6–7, Sweeney v. Madigan, No. 1:18-cv-01362 (N.D. Ill. Feb. 6, 2019), (“[I]f Janus is decided as most commentators expect, the unions and their members would also suffer . . . . Forcing unions to advocate on behalf of non-members who object to the very reasons they exist is a severe violation of unions’ First Amendment rights to association.”).
This “members-only representation” or “fee-for-service” approach would reduce some of the economic incentive for workers to free ride, since the refusal to pay would result in reduced services. This change would most directly affect the calculus for workers who reasonably anticipate disputing an employer’s disciplinary actions, since that is the circumstance in which the right to a union representative holds great value. 150 For this reason, it may be that the members-only representation approach will encourage the lowest-performing workers who are at risk of termination or disciplinary action to join the union, while having little influence on higher-performing workers. In any case, many union officials worry that the members-only representation model will be counterproductive. According to AFSCME spokesman Anders Lindall, the four largest public-sector unions—AFSCME, the American Federation of Teachers (AFT), the National Education Association (NEA), and the Service Employees International Union (SEIU)—have “jointly taken the position that when employees in a workplace vote for union representation, all workers” must be “represented equally” in grievance proceedings because “[a]nything less can pit workers against each other and fuel a race to the bottom.” 151 Kim Geiger, After Losing Landmark Supreme Court Case, Labor Splits on What to Do Next, Chi. Trib. (June 28, 2018), (on file with the Columbia Law Review); see also Max Parrott, Cuomo’s Anti-Janus Labor Law May Divide Workers, Critics Contend, City & State N.Y. (June 29, 2018), [] (“In New York, the state with the highest union-density of any state’s workforce, . . . the change could significantly weaken unions. In preparation for this moment, the New York state budget included an amendment to the Taylor Law.”). For a detailed critique of members-only representation in the wake of Janus, see Fisk & Malin, supra note 38, at 17–29.

Ultimately, each of these proposals to tinker with the right-to-work environment may well dampen the blow to unions to some degree. But they fail to alter the basic economic dilemma facing many public workers (in particular, those who do not anticipate disciplinary action by the employer). Why pay anything to a union when you can get the same contractual benefits for nothing? 152 Another recent proposal, offered by Professor Samuel Estreicher, would be to require that every public worker make a payment to the union, with an option for objecting workers to direct their payment to a charity of their choosing. See Samuel Estreicher, How Unions Can Survive a Supreme Court Defeat, Bloomberg (Mar. 2, 2018), (on file with the Columbia Law Review). This approach would minimize the incentive for workers to opt out, since pure economic free riders may be ambivalent between their money going to the union or some charity. Estreicher’s approach is, however, constitutionally uncertain. The Supreme Court explained in United States v. United Foods, Inc. that “First Amendment values are at serious risk if the government can compel a . . . discrete group of citizens[ ] to pay special subsidies for speech on the side that it favors.” 533 U.S. 405, 411 (2001). A law compelling persons to give some amount of money to charitable ends would arguably violate the rights of those who object to giving any money to any charitable organization at all. As discussed next, one major alternative proposal would take aim at this fundamental asymmetry. 153 One additional possibility that has attracted attention is for states to privatize entire sectors of their public workforce, thereby enabling fair-share fees under the NLRA’s private-sector model. Oregon recently did so when it reclassified 35,000 home-healthcare workers as private employees of a privately run third-party corporation. See Josh Kelety, Union-Friendly Home Healthcare Worker Bill Approved by State Legislature, Reflector (Mar. 7, 2018), [].
There are serious reasons to doubt the scalability of this approach from the home-healthcare-worker sector, however. Unlike in that sector, in which individual workers are decentralized without a strong identity as public employees, government workers in other sectors (such as public education) may be mistrustful of some third-party corporation having full control over their employment. Government employers may similarly dislike the idea of ceding to a private entity control over pay rates, hiring and firing decisions, and other terms of employment. And any attempt by the government to somehow reassert such control by way of the third-party management company would give rise to a plausible argument of state action, recreating the very First Amendment violation that the privatization move would aim to circumvent. See Brentwood Acad. v. Tenn. Secondary Sch. Athletic Ass’n, 531 U.S. 288, 298 (2001) (holding that state action exists when “[t]he nominally private character” of an entity is “overborne by the pervasive entwinement of public institutions and public officials in its composition and workings”).

C. Option #3: Members-Only Bargaining

The right-to-work regime functions on what Professors Catherine Fisk and Benjamin Sachs describe as a “double bind”: As the exclusive representative of all workers in a unit, a public-sector union is “obligated to provide free services for workers who don’t want a union,” but those workers are permitted to opt out of paying union fees. 154 Benjamin Sachs & Catherine Fisk, Why Should Unions Negotiate for Workers Who Don’t Pay Their Fair Share?, L.A. Times (July 9, 2014),
opinion/op-ed/la-oe-sachs-unions-supreme-court-20140710-story.html (on file with the Columbia Law Review).
Fisk and Sachs suggest what is accordingly an intuitive and powerful response: relax the bind. “Where unions are unable to require objecting workers to pay fees,” they argue, states should “get rid of the rule of exclusive representation.” 155 Id.

In practice, this would mean that workers who opt out of the union would not pay the union any fees and would no longer be covered by the terms of the union-negotiated collective bargaining agreement. Objecting workers would then be free to negotiate against the employer for terms and conditions of employment individually (or in concert with some other collective agent), though they could not use the union’s personnel for the purposes of pursuing grievances or defending disciplinary actions.

Fisk and Sachs make a compelling argument for this solution. After all, once a worker is permitted to opt out of paying for the union’s services, the “logically consistent and fair conclusion” is to permit the union to stop providing them. 156 Catherine L. Fisk & Benjamin I. Sachs, Restoring Equity in Right-to-Work Law, 4 U.C. Irvine L. Rev. 857, 867 (2014). Fairness is one thing, but desirability is another. And on that score it is not obvious that either of the major stakeholders in this policy debate—pro-labor public employers and union supporters—would prefer members-only bargaining over a right-to-work regime.

Start with public employers that wish to give their workers a voice in workplace decisions, thereby improving morale and satisfaction. As noted above, this interest is important, but only to the extent it can be achieved efficiently. 157 See supra section I.A. Worker satisfaction might be immense in a system in which any employee can knock on the employer’s door to request a raise, benefit, or policy change, and in which the employer always capitulates. But the flood of such requests would quickly overwhelm the employer, to say nothing of the costs; employers would face the prospect of individual demands by any of the hundreds—or even thousands—of workers in the bargaining unit. 158 Consider, for example, the plight of the school-district employer in a place like Los Angeles, which employed 26,558 teachers in 2016 and 2017. See L.A. Unified Sch. Dist., Fingertip Facts 1,
32/NewlyUpdatedFingertip%20Facts2017-18_English.pdf [] (last updated Oct. 2017).
The Abood Court thus seems prescient when it observed more than forty years ago that exclusive representation eliminates the “confusion and conflict that could arise” from competing demands from employees regarding wages and all sorts of workplace policies. 159 Abood v. Detroit Bd. of Educ., 431 U.S. 209, 224 (1977).

If the members-only approach is problematic for employers, it is arguably just as bad for unions. That is because the only thing potentially worse for a union than ambivalence among workers to paying dues is affirmative competition over dues from other unions who claim to be better able to represent worker interests. Yet as Shaun Richman, a former organizing director for the American Federation of Teachers, observes, “make no mistake,” once a union agrees to represent only “a portion of a bargaining unit . . . another organization will come along to recruit the workers who are left out by promising better benefits or an alternative approach to seeking improvements on the job.” 160 Shaun Richman, The Promise and the Peril of Members-Only Unions, In These Times (Nov. 4, 2015), []. The negative experience of the Tennessee Education Association (TEA) after Republican lawmakers implemented a labor-law reform recognizing competing unions is a testimony to this concern: After the reform was enacted, one local TEA affiliate’s lead negotiator observed that “[t]he free[riders] were something we griped about for years . . . . [T]hat was a big complaint, but looking back, I would take the free[riders] in a heartbeat to have the right to exclusive representation again.” 161 Chris Brooks, The Cure Worse than the Disease: Expelling Freeloaders in an Open-Shop State, New Lab. F. (Aug. 2017),
24/the-cure-worse-than-the-disease/ [] (internal quotation marks omitted) (quoting Carol Hambright).

Competition among multiple unions to serve worker interests might or might not be a good thing for workers and society more broadly, but that is a policy judgment that lies outside the scope of this Article. 162 On the one hand, unions may become more responsive to worker interests; on the other hand, the conflicts among workers and their competing unions might produce troubling intraworkplace strife and severely complicate the public employer’s ability to maintain a cohesive, supportive environment among its workforce. What is clear, however, is that incumbent unions may be skeptical of calls for eliminating their right to exclusive representation. That skepticism is especially understandable once union proponents look around the room to see who is actually advancing members-only bargaining reforms in state legislatures: conservative organizations like the American Legislative Exchange Council (ALEC) and the State Policy Network. 163 See Cynthia Estlund, Are Unions a Constitutional Anomaly?, 114 Mich. L. Rev. 169, 231 (2015) (noting both sides’ interest in members-only bargaining); Chris Brooks, Why We Shouldn’t Fall for the Members-Only Unionism Trap, In These Times (Dec. 22, 2017), []. ALEC, for example, has proposed a model “Public Employee Freedom Act” that would forbid “impos[ing] representation by an employee organization on public employees who are not members of that organization” and recognize the right of public employees to “represent themselves in their relations with the public employer.” 164 Public Employee Freedom Act, Am. Legislative Exch. Council,
model-policy/public-employee-freedom-act/ [] (last visited Oct. 16, 2018).
That, of course, is members-only bargaining in a nutshell.

It is worth pausing for a moment to take stock of the public-sector union’s plight. The right-to-work default that exists after Janus threatens the loss of anywhere from 15 to 70% of the union’s membership and funding stream. And the principal alternative discussed so far—relaxing the union’s duty to represent non-fee payers—is arguably no better. What, then, can the pro-labor employer and union do to preserve the current system of worker representation? The next section considers such an alternative.

D. Option #4: Direct Government Reimbursement

From a certain vantage point, there is something odd about the fair-share-fee financing model that dominated public-sector labor law until Janus. State governments frequently identify services that they believe will redound to the benefit of a sizable group of people or society more broadly. When they do, the usual approach is to use general tax revenues to provide the service rather than concentrating the costs of the service on just the most directly affected persons. 165 For example, when the federal government funds a public-health initiative such as the provision of contraceptive services through payments to entities like Planned Parenthood, it does so out of general treasury funds, not a targeted exaction on the likeliest beneficiaries of contraceptive services, (such as women of child-bearing age). See Tang, Public Sector Unions, supra note 18, at 173; Danielle Kurtzleben, Fact Check: How Does Planned Parenthood Spend That Government Money, NPR: It’s All Politics (Aug. 5, 2015), []. For an argument that targeted exactions and general taxes should be treated equivalently for constitutional purposes, see generally Nikolas Bowie, The Government Could Not Work Doctrine, 105 Va. L. Rev. (forthcoming 2019).

Now  that  Janus  has  made  it  unconstitutional  for  public  employers  to  fund  unions  using  fair-share  fees,  it  should  be  asked  whether  the  usual  tax-and-spend  approach  makes  sense.  Under  this  approach,  the  government  would  reimburse  a  union  for  its  bargaining  costs  directly,  reducing  future  wages  or  benefits  by  an  offsetting  amount. 166 See supra note 18 and accompanying text. Employers and unions would accordingly be in an identical financial position as before Janus, and workers would actually be better off due to a reduced federal income tax burden. 167 See supra notes 23–24 and accompanying text. Critically, this approach, as many have argued, eliminates any First Amendment concerns expressed in Janus because no objecting worker would be required to make any payment to a union to which they object. 168 See supra note 18 and accompanying text.

None of this is to suggest that direct reimbursement of union costs is without controversy. It is just that the major objection is a policy concern, not a constitutional one. Direct government reimbursement of the union’s costs of collective bargaining, it is argued, would pose a problem for union independence. This Article confronts that objection now, proceeding in four steps. First, it presents the union-independence objection in its strongest form. 169 See infra section III.D.1. It next describes the heretofore unrecognized analogy between public-sector unions and public defenders. 170 See infra section III.D.2. The aim is to show that we are not in uncharted waters when it comes to government-funded entities whose purpose is to advocate against the government’s own narrow self-interest.

The Article then canvases the more specific lessons we might learn from the public defender analogy. 171 See infra section III.D.3. It starts with lessons of the “what not to do” variety, playing out two models of public defender financing—adversarial negotiation and political negotiation—that would likely do more harm than good if imported to the public-sector union context. More constructively, it describes two models of governmental funding for criminal defense that could work in the union setting: “independent determination” and “principal control.” The Article takes up those models in Part IV, which presents concrete legislative proposals reflecting the strongest possible institutional designs and organizational practices for bringing the direct government reimbursement alternative into effect. 172 See infra Part IV.

1. The Union-Independence Objection. — Here is the biblical version of the best argument against government itself paying for a union’s collective bargaining costs: “No one can serve two masters . . . .” 173 Matthew 6:24 (World English Bible). The union’s fundamental duty, after all, is to serve the interests of the workers it represents. So they, and only they, should be the union’s master. Yet allowing the government to fund the union would interject the government as a second master and an especially pernicious one at that. For the government is not just some neutral bystander in the union’s business; it is the direct counterparty against whom the union is supposed to advocate on behalf of workers.

The concern for union independence is not merely theoretical. It arises out of a deep history of so-called “company unions” in the private sector. After Congress enshrined the right of private-sector employees to organize and bargain collectively under the National Industrial Recovery Act of 1933, 174 See National Industrial Recovery Act, ch. 90, 48 Stat. 195 (1933). a number of private employers took solace in the age-old adage, “if you can’t beat ‘em, join ‘em.” In this context, “joining” employees in the enterprise of workplace organizing meant setting up company unions whose entire nature—from leadership to bargaining positions to source of financing—was reliant upon the employer itself. The National Labor Relations Board (NLRB) describes a classic example of company unionism in the Republic Steel Corp. case, in which an employer threatened and coerced workers to “stick with the company union” instead of “belong[ing] to the outside union,” 175 9 N.L.R.B. 219, 329 (1938). and in which the employer paid its preferred workplace representatives to accommodate the company’s preferences on a host of workplace issues such as wage disputes. 176 Id. at 230–31 (explaining how “collective bargaining” regarding a modest wage rate took place essentially through unilateral employer decisionmaking with little pushback from the company union, and noting how employee representatives “were paid by the [employer]” for attendance at meetings). By the time the National Labor Relations Act was enacted in 1935, company unions of this ilk already boasted some three million members—a number approaching the four and a half million workers who were members of independent, outside unions. 177 Mark Barenberg, The Political Economy of the Wagner Act: Power, Symbol, and Workplace Cooperation, 106 Harv. L. Rev. 1381, 1386 (1993) [hereinafter Barenberg, Political Economy].

One  of  the  NLRA’s  key  objectives  was  thus  to  eradicate  company  unions  altogether. 178 See Mark Barenberg, Democracy and Domination in the Law of Workplace Cooperation: From Bureaucratic to Flexible Production, 94 Colum. L. Rev. 753, 773 (1994) [hereinafter Barenberg, Democracy and Domination]; see also Barenberg, Political Economy, supra note 177, at 1385 n.16 (describing the typical company union). The primary mechanism was section 8(a)(2), which prohibits employers from “dominat[ing] or interfer[ing] with the . . . administration of any labor organization or contribut[ing] financial or other support to it.” 179 29 U.S.C. § 158(a)(2) (2012). Senator Robert Wagner, the principal draftsman of the NLRA, justified the ban on financial support in particular by comparing company unions to the “practice whereby the person on one side of the bargaining table pays the attorney of those with whom he deals.” 180 79 Cong. Rec. 7565, 7570 (1935) (statement of Sen. Wagner).

Section 8(a)(2) achieved its aim. Company unions are now a relic of history. 181 See Barenberg, Political Economy, supra note 177, at 1385–86. Yet there is a growing call among labor-law scholars to recognize that, in light of rapid changes in our economy and the nature of our workplaces, the NLRA’s total ban on company support for unions may actually thwart healthy employee–management relations more than it serves them. Professor Estreicher has argued, for example, that “[t]he ‘company union’ prohibition of the 1930s imposes significant costs on our society” and that “Congress should [therefore] partially repeal section 8(a)(2).” 182 Samuel Estreicher, Employee Involvement and the “Company Union” Prohibition: The Case for Partial Repeal of Section 8(a)(2) of the NLRA, 69 N.Y.U. L. Rev. 125, 126–27, 150 (1994) (arguing that Congress should restrict section 8(a)(2)’s application to entities that bargain with their employer over terms and conditions of employment). This misalignment between section 8(a)(2)’s ban on employer support and the reality in American workplaces, Professor Samuel Estreicher explains, is rooted in changes in “what workers want and how work is organized” that date back into the 1970s. 183 Id. at 134. Others have offered their own proposals to pare down section 8(a)(2)’s broad effect and allow employers and employees to cooperate in pursuit of shared aims. 184 See, e.g., Barenberg, Democracy and Domination, supra note 178, at 768–69 (arguing for an amendment to section 8(a)(2) to “allow in-house [union] options that fulfill . . . new standards of non-domination,” including “strategic representative councils”); Heather M. Whitney, Rethinking the Ban on Employer-Labor Organization Cooperation, 37 Cardozo L. Rev. 1455, 1509–19 (2016) (offering legislative and judicial approaches to narrowing section 8(a)(2)’s coverage).

Yet despite the diminished concern over company unions that has emerged in the eighty years since the NLRA’s passage—and despite the academy’s pleas for relaxing section 8(a)(2)—there are some who continue to sound the union-independence alarm. The State of California argued in Janus, for example, that “[a]ny ‘bargaining’ arrangement in which the employees’ representative was funded . . . by the employer, rather than by the represented employees themselves, would be (and be perceived as) little more than a return to the setting of employment terms by unilateral management action.” 185 Brief for the State of California Supporting Affirmance, supra note 19, at 13. Or as Professor Estreicher himself argued alongside several other prominent labor-law scholars in an amicus brief, “[i]f the collective-bargaining route is truly to provide an independent employee voice, the affected employees must fund their own representative. Otherwise, that representative, however well-intentioned it may be, will be reasonably viewed . . . as an agent of management . . . .” 186 Brief of Professors Cynthia L. Estlund et al. as Amici Curiae in Support of Respondents at 4, Janus v. Am. Fed’n of State, Cty., & Mun. Emps., Council 31, 138 S. Ct. 2448 (2018) (No. 16-1466), 2018 WL 527955; see also id. at 14 (“If the representative were to be funded by the [government employer], it would in short order be reasonably viewed by the employees as an organ of management, a human resources department of the government employer in all but name.”). Critically, the union-independence objectors appear to see the problem as categorical in nature. As California argues, unions simply can “not be funded by the employer or the State and retain the credibility and independence needed to make the system work.” 187 Brief for the State of California Supporting Affirmance, supra note 19, at 12–13; see also Lee Saunders, A Union Response to the Supreme Court’s Janus Decision, Am. Prospect (July 9, 2018), [] (“The union must belong to the workers, and they must pay for their union so they can own it.”).

Several union proponents interviewed in the course of preparing this Article echoed this concern, from high-level officials in the national offices of large public-sector unions to members and organizers at the local affiliate level. One national level official may have put it best when he explained that “it’s hard to fathom how a union can be an independent voice on behalf of all workers if the union is dependent for its funding and livelihood on the entity sitting across the bargaining table.” 188 Interview with Teague Paterson, Deputy Gen. Counsel, Am. Fed’n of State, Cty., & Mun. Emps., and Matthew Blumin, Assoc. Gen. Counsel, Am. Fed’n of State, Cty., & Mun. Emps. (on file with the Columbia Law Review).

2. The Public Defender Analogy. — One way to see through the union-independence objection is with an analogy. Public-sector unions occupy the same kind of adversarial position vis-à-vis the government as another commonplace entity: public defenders. Yet for all the criticisms that public defenders have received, worries of dependence on government financial support are not high among them. This section begins with a defense of the claim that the two entities are similarly situated before exploring the analogy’s payoffs.

Three basic attributes about the public-sector union leave it susceptible in theory to charges of dependence on the government. First, the public-sector union is the agent in a principal–agent relationship. 189 See Wallace Corp. v. NLRB, 323 U.S. 248, 255 (1944) (“By its selection as bargaining representative, [the union] has become the agent of all the employees . . . .”). The employees who are covered by the union-negotiated collective bargaining agreement are the principal, and it is their interests that the union must strive ultimately to serve. 190 See id. But just as in any principal–agent relationship, the union possesses its own self-interest, too, which may not always be aligned with that of its principal. 191 See generally John W. Pratt & Richard J. Zeckhauser, Principals and Agents: An Overview, in Principals & Agents: The Structure of Business 1, 1 (John W. Pratt & Richard J. Zeckhauser eds., 1985) (describing the classical principal–agent dilemma colorfully).

Second, the public union is fundamentally engaged in an adversarial relationship with a unit of government—namely, a government employer. Unions demand things from employers—wage increases, workplace policy changes, and so on—that the employers often prefer not to grant. As recent events in West Virginia and Oklahoma attest, sometimes the disagreement between workers and management can escalate, resulting in labor unrest. 192 See Jess Bidgood, West Virginia Raises Teachers’ Pay to End Statewide Strike, N.Y. Times (Mar. 6, 2018), (on file with the Columbia Law Review) (“The [West Virginia teachers’] strike ground the state’s public schools to a halt for nine days . . . .”); Dana Goldstein, Their Pay Has Stood Still. Now Oklahoma Teachers Could Be the Next to Walk., N.Y. Times (Mar. 20, 2018), (on file with the Columbia Law Review) (“[I]n Oklahoma, where teachers have not had a raise from the state in a decade, they have vowed to go on strike . . . if the Legislature does not act to increase pay and education budgets.”).

Third, public-sector unions need resources to function. The project of communicating with the union’s principal (covered employees) to learn about workplace demands is time consuming, and the task of presenting the strongest case to the employer regarding those demands can be challenging and costly. So too the efforts involved with administering the terms of a complex collective bargaining agreement and providing representation to workers in disputes with the employer.

It is the confluence of these qualities that leads to the purported union-independence problem. For if the government funds the union, the union may treat the preservation of that funding stream as its utmost priority, abandoning the pursuit of employee interests during adversarial negotiations against the government.

Each of these attributes describes the public defender as well. The public defender is the agent to her client, the principal. 193 See Deborah A. DeMott, The Lawyer as Agent, 67 Fordham L. Rev. 301, 301 (1998) (noting that “the lawyer-client relationship is a common[ ] illustration of agency”). But see Robert E. Toone, The Absence of Agency in Indigent Defense, 52 Am. Crim. L. Rev. 25, 27 (2015) (arguing that indigent criminal suspects lack a core aspect of agency—the right to fire their lawyers). The public defender is also engaged in a classically adversarial relationship with a unit of government—namely, a government prosecutor. 194 See Polk Cty. v. Dodson, 454 U.S. 312, 318 (1981) (“In our system a defense lawyer characteristically opposes the designated representatives of the State.”). Finally, public defenders need resources to operate. So the conditions for a public defender independence problem would seem to exist here, too: If the government provides financial support to the public defenders, that very act could lead the defenders to abandon their clients’ interests for fear that zealous representation will jeopardize their funding moving forward.

Shifting our lens back from theory, however, the public defender “independence problem,” as it were, seems more imaginary than real. 195 See, e.g., Charles J. Ogletree, Jr., Beyond Justifications: Seeking Motivations to Sustain Public Defenders, 106 Harv. L. Rev. 1239, 1286–89 (1993) (describing the successful work done in one public defender service); Eve Brensike Primus, Culture as a Structural Problem in Indigent Defense, 100 Minn. L. Rev. 1769, 1769 (2016) (arguing that “[m]any defenders perform valiantly under trying circumstances and manage to provide zealous, client-centered defense advocacy”); Jonathan A. Rapping, Reclaiming Our Rightful Place: Reviving the Hero Image of the Public Defender, 99 Iowa L. Rev. 1893, 1895–98 (2014) (describing the history of heroic social portrayals of public defenders). But see Jonathan D. Casper, Did You Have a Lawyer When You Went to Court? No, I Had a Public Defender., Yale Rev. L. & Soc. Action, Spring 1971, at 4, 6 (quoting one prisoner as saying that “[a] public defender is just like the prosecutor’s assistant” (internal quotation marks omitted)). Attacks against our indigent criminal defense systems in America are common, as discussed below, but there is little sustained argument that public defenders are systematically and categorically incapable of repre­senting their clients loyally simply because they receive government funding. 196 Some have argued separately that certain attorney-assignment systems create conflicts of interest, for example, when judges (rather than defender offices) are responsible for selecting indigent defense counsel in particular cases. See Understanding Judicial Interference with the Defense Function—ABA Principle 1, Sixth Amendment Ctr., [https://] (last visited Oct. 16, 2018) (“Fearing the loss of their jobs by not pleasing the judge who hired or supervises them, defenders will often take on more cases than they can ethically handle . . . delay working on a case, and . . . triage their hours available in favor of some clients . . . to the detriment of others.”). This is problematic, but the problem is avoided altogether in the public-union setting because employees themselves get to choose their unions, not some government officer. See, e.g., Cal. Gov’t Code § 3544(b) (2019) (stating that unions are recognized upon “proof of majority support” from members).
Yet that is precisely what the union-independence objector must show—that, to paraphrase California’s argument in Janus, no public defender could “be funded by the . . . State and retain the credibility and independence needed to make the system work.” 197 Brief for the State of California Supporting Affirmance, supra note 19, at 12–13.

It is not just decades of experience with public defenders that renders this suggestion implausible; the Supreme Court has forcefully rejected it too. In Polk County v. Dodson, the Court considered the argument that the “State’s funding of criminal defense[ ] makes it a ‘joint participant’ in that enterprise, locked in a ‘symbiotic relationship’ with individual public defenders.” 198 454 U.S. at 322 n.12. The technical issue in the case was whether public defenders act “under color of state law” for purposes of a claim against them under 42 U.S.C. § 1983. 199 42 U.S.C. § 1983 (2012); Dodson, 454 U.S. at 314. In rejecting that contention, the Court left little doubt about its view of public defender independence. Because public defenders do not answer to some “administrative superior” in the government and because it would be unlawful for the government to interfere with a defender’s “professional independence,” the Court denied the claim that the government, by virtue of simply “having employed [and paid] public defenders,” has necessarily “attempted to control their action in” some impermissible manner. 200 Dodson, 454 U.S. at 322.

The same is true of public-sector unions. Union leaders do not answer to any “administrative superior” in the government; they obtain approval for the positions they take at the bargaining table from their members. 201 See Cal. Gov’t Code § 3520.5 (“[E]mployee organizations formally recognized as exclusive representatives pursuant to a vote of the employees may be revoked by a majority vote of the employees . . . .”); William B. Gould IV, A Primer on American Labor Law 178–80 (4th ed. 2004) (discussing processes through which unions elect local, regional, and national officers). And even if it were permissible for a union to receive payment from the government to offset its costs (like a defender), it would in all events remain unlawful for the public employer to “[d]ominate or interfere with the . . . administration of any [union].” 202 Cal. Gov’t. Code § 3519(d); see also, e.g., 5 Ill. Comp. Stat. Ann. 315/10(a)(1) (West 2018) (outlining a similar prohibition on interference with unions). The takeaway, then, is that the union-independence objector must explain either why public defenders really are hopelessly conflicted in virtue of their government funding, contra the Supreme Court, or why there is something materially different about public-sector unions.

To the latter point, one possible contention is that public defenders may suffer less from the dependence problem because indigent defense services are constitutionally required under the Sixth Amendment. 203 See U.S. Const. amend. VI; Gideon v. Wainwright, 372 U.S. 335, 344 (1963). Public-sector unions, by contrast, function as a matter of legislative convenience in pursuit of state interests in labor peace and efficiency. The distinction is true, of course, but the conclusion does not follow. Why should government funding compromise public defender independence less (when all states are compelled to do it) than it compromises public union independence (when states choose to do it of their own accord)? The fact that states choose freely to support public unions should suggest that they are more likely to want to do it right—that is, by protecting union independence. Or to put it differently, why would a state go through the trouble of using public funding as a subversive tool for co-opting public-sector unions when the easiest (and cheapest) way to defang the unions would be to refuse to recognize or bargain with them at all?

A second argument for distinguishing public defenders is that various state rules of professional conduct provide an important antidote to the dependence problem for lawyers, 204 See Dodson, 454 U.S. at 321 (“[A] public defender works under canons of professional responsibility that mandate his exercise of independent judgment on behalf of the client.”). whereas no similar licensing authority or enforceable norm of conduct exists in the union context. It is unclear, however, why attorney disciplinary proceedings should be considered the gold standard for ensuring zealous representation. As scholars of professional responsibility have long recognized, attorneys are rarely held to account for ethical violations. 205 See Leslie C. Levin, The Emperor’s Clothes and Other Tales About the Standards for Imposing Lawyer Discipline Sanctions, 48 Am. U. L. Rev. 1, 8–9 (1998) (finding that only 5 to 6% of attorney complaints result in any sanctions against lawyers); see also id. at 9 & n.33 (noting only 6,411 reports of lawyer sanctions being imposed in 1996—well less than 1% of the total number of licensed attorneys—and that, even then, sanctions are “often light and inconsistent”). In any case, a norm of good faith representation does apply to public-sector unions under state labor law, the violation of which triggers liability for the union. 206 See, e.g., Cal. Gov’t Code § 3571.1 (making it an unfair labor practice for “an employee organization to . . . [f]ail to represent fairly and impartially all the employees in the unit for which it is the exclusive representative”); N.Y. Civ. Serv. Law § 209-a(2) (McKinney 2018) (making it improper for an employee organization to deliberately “breach its duty of fair representation to public employees”). And what is more, perhaps the greatest check that deters an agent from disloyal action (whether a public-sector union or a public defender) is the principal’s power to terminate the agent’s employment upon discovering the agent’s self-dealing. Yet unlike the public defender, who may not be fired at will even when the client suspects a lack of zealous representation, a public-sector union can always be decertified by vote of its members. 207 Compare Toone, supra note 193 (discussing the indigent defendant’s inability to “select his representative, negotiate the terms of representation, or direct the lawyer’s conduct during the case”), with Cal. Gov’t Code § 3520.5 (“[E]mployee organizations formally recognized as exclusive representatives pursuant to a vote of the employees may be revoked by a majority vote of the employees . . . .”).

A final critique of the public defender analogy takes a different tack. Even if the similarities between public unions and defenders are admitted, the critique runs, drawing on the public defender model to bolster some ideal vision of government-funded public-sector unions is self-defeating because the public defender system is itself irredeemable. 208 See Mary Sue Backus & Paul Marcus, The Right to Counsel in Criminal Cases, A National Crisis, 57 Hastings L.J. 1031, 1035 (2006) (“Some attorneys went to trial without ever meeting their clients outside the courtroom. Some neglected to interview obvious alibi witness. Some accepted without question reports from prosecutors’ medical and forensic experts that were ripe for challenge.”); supra note 26 and accompanying text. This is a powerful charge, in large part because of recent attention to the many ways in which underresourced and overworked public defenders are unable to serve the needs of indigent defendants. 209 See, e.g., Oliver Laughland, The Human Toll of America’s Public Defender Crisis, Guardian (Sept. 17, 2016), []. But it is important to be clear about why existing systems for providing indigent defense are failing so many Americans: the woeful insufficiency of funding. 210 See id. (“Around the US, defenders routinely report an increase in overburdening and underfunding, caused by a variety of structural, political and economic drivers.”). A report issued by the Department of Justice found, for example, that in 2007 “[m]ore than 7 in 10 county-based [public defender] offices had an insufficient number of attorneys to meet the professional guidelines.” 211 Donald J. Farole Jr. & Lynn Langton, U.S. Dep’t of Justice, Census of Public Defender Offices, 2007, at 8 (2010), []. To meet those guidelines, the median public defender office would need to increase its number of litigating attorneys by a staggering 57%. 212 See id. at 10 (noting that median offices would have to add four more attorneys to their existing seven full-time staff to reach guideline levels).

In view of these statistics, it is virtually unassailable that most public defender offices are underfunded. But notice how this is a very different charge than the union-independence concern we started with. And if anything, calls for government to provide greater resources to public defenders undermine the view that one can never have a loyal agent who is paid by the government. That is because the sense of obligation such an agent might feel toward her government sponsor would be exacerbated, not mitigated, if the government were to pay the agent even more.

In the end, there are many problems with the American approach to indigent defense. But the limited point here is that the worry that lawyers representing indigent defendants have an inescapable conflict of interest that leads them to betray their clients out of fear of losing government funding is not one of them. Moreover, the major problems with indigent defense in America owe to the fact that funding levels are politically negotiated, 213 See Primus, supra note 195, at 1807–08 (describing how politics influence budget setting process); Steiker, supra note 26, at 2700 (“Although the need for greater resources for indigent defense services may be obvious, it is here that political will falters most, for equally obvious reasons.”). which  is  itself  an  important  lesson  that  pro-labor  states  should  learn  when  designing  their  models  for  offsetting  public  union  bargaining  costs. 214 See infra section III.D.3 (describing problems with the political-negotiation model for funding public defenders and unions). Politically negotiated funding levels are not inevitable; the United Kingdom stands as an important example of a model for funding indigent defense that is largely insulated from political pressures. 215 See infra section III.D.4; see also Norman Lefstein, In Search of Gideon’s Promise: Lessons from England and the Need for Federal Help, 55 Hastings L.J. 835, 869–71, 879–80 (2004). The moral of the public defender analogy, in other words, is not simply that public unions can continue to enjoy independence after receiving public funding but also that there are better (and worse) ways for states to structure their funding systems. Those options will be explored now.

3. Lessons from Public Defenders: What Not to Do. — The two most apparent lessons to draw from the public defender analogy are how not to structure a public union financing model. Such a model would be doomed to failure if it left decisions regarding funding levels up to adversarial negotiation or political negotiation.

a. Adversarial Negotiation. — The worst way to set up a union financing system would be to leave funding levels to the process of adversarial negotiation. Such an approach would lead to the very union-independence concerns that critics fear. Although examples of this kind of approach are rare in the analogous public defender context, one story from that setting can help to illustrate the approach’s flaws.

In 2004, the ACLU filed a class action lawsuit against Grant County, Washington alleging endemic problems with the County’s indigent defense services. 216 See Beth Colgan, A Well-Grounded Fear: Civil Reform of Criminal Justice, 41 Clearinghouse Rev. J. Poverty L. & Pol’y 232, 232–33 (2007). The county eventually settled the suit. Id. at 237. In addition to identifying chronic levels of underfunding, the lawsuit made jarring allegations regarding how money was allocated to different indigent defense lawyers: at the behest of the prosecutor. “The prosecutor participated in contract negotiations with public defenders,” “drafted public defense contracts,” and even had the power to “approve certain payments for public defense.” 217 Id. at 234. Public defenders were, as a result, “beholden to the prosecutor” for everything from being selected for assignments to the amount of money they would receive for each representation. 218 Id. That led, in turn, to a “complete breakdown of the adversarial system,” including one remarkable story in which a defender was so fearful of resisting the prosecutor that he “failed to object when [the] prosecutor compared his client to Adolph Hitler in his closing argument.” 219 Id.

The Grant County experience shows precisely the kinds of problems one might encounter with any system that asks an adversarial agent to advocate the best interests of its principal against the very government official who is responsible for deciding how much the agent gets paid. In the union setting, one might imagine similar encroachments on independence were the union required to bargain against a government employer simultaneously for the union’s own reimbursement levels and for wage and benefit increases for its workers. Such a union would face the inherent pressure to sacrifice what is best for its workers to retain what is best for it—continued access to government funding. Indeed, this sort of an arrangement might pressure the union to turn into a kind of employer-controlled company union that plagued the private sector in the 1930s. 220 See supra section III.D.1. So however states may choose to have union funding levels determined, this much is clear: The amount cannot be left up to adversarial negotiation with the public employer.

b. Political Negotiation. — Stories like Grant County’s prosecutor-directed public defender funding system are rare. As a general matter, choices over public defender budgets are made not by the prosecutor but by local and state legislatures through the process of political negotiation. 221 See supra note 213. Alas, this approach to determining public defender financing levels is itself deeply flawed. The problem is not that defenders face some intractable conflict of interest under the political-negotiation model; in fact, defenders are often glorified as valiant public servants who fight for justice on behalf of their clients despite impossible conditions. 222 See supra note 195 and accompanying text. The problem is instead that indigent criminal defendants are among the least influential political actors in our society, which leaves little political incentive for officials to appropriate adequate public defense budgets. 223 See Steiker, supra note 26, at 2700 (“With clamoring demand for dwindling public funds . . . it is not surprising that more money for lawyers representing alleged criminals is not high on anyone’s list.”); see also Chemerinsky, supra note 26, at 2691–92.

The failure of the state’s political branches to respond to a dire underfunding problem in Missouri is instructive. In 2005, a state report found that Missouri had the lowest per capita spending of any statewide public defender system in the nation. 224 State of Mo. Pub. Def. Comm’n, Fiscal Year 2012 Annual Report 1 (2012), []. In 2008, the Missouri Senate finally moved a bill to increase public defender financing and set maximum caseload standards, but the bill never reached the House floor. 225 Id. at 2. Undeterred, the Senate moved separately to limit caseloads, but the Governor vetoed that bill in 2009. 226 Id. at 3. A breakthrough seemed to occur when the Governor and legislature agreed to increase public defender funding by more than $2 million in 2009, but the Governor ultimately decided to withhold all but $500,000 of those funds after the economic recession. 227 Id. After the defenders continued to face crushing caseloads (leading eight offices to take the controversial step of refusing new cases), the legislature approved an additional $1.163 million in spending in 2012, only for the Governor to withhold nearly two-thirds of that amount. 228 Id. at 5.

Given the many pressures on government budgets, it is easy to see why the political-negotiation approach offers little financial security for public defenders. And a similar fate may befall public-sector unions were their funding levels left up to the vagaries of local and statewide budgetary processes. Under such a system, a union would make its case for its desired amount of funding not to the employer as part of negotiations at the bargaining table but rather to a local or state legislature. 229 Just as is the case with public defender financing, the state could choose to have public-sector unions financed either on a statewide basis or by each locality. See Stevens et al., supra note 25, at 5 (noting that twenty-three states’ indigent defense services are fully state funded, whereas the remainder use a combination of state and county funding). Union funding requests would thus be weighed against competing budgetary priorities and evaluated in light of the government’s overall fiscal health.

To be sure, public-sector unions in jurisdictions with more favorable political climates (and in more influential sectors of government) may wind up with sufficient resources under this kind of an approach, at least while economic conditions are positive. Unions as a group, after all, are hardly as politically powerless as indigent criminal defendants. But support for labor is not constant in every locality, and economies wax and wane. The risk with a political-negotiation approach is thus as much a matter of insufficient funding levels as it is a matter of unpredictability: It is hard for any agent to serve its principal when the resources needed to provide effective representation may dissipate with any given electoral cycle.

4. Lessons from Public Defenders: What Might Work. — Two different models for funding criminal defense show more promise for public-sector unions. Government may provide more effective advocacy by leaving funding levels either to independent determination by a body of experts or to the control of the principals themselves.

a. Independent Determination. — This Article has noted how the adversarial and political negotiation of public defender budgets can lead to risks of public defender dependence and insufficient resources, respectively. But what if indigent defense funding levels were determined independently based on need by some body with deep expertise in the underlying subject matter? That kind of an approach likely sounds foreign to the reader, as no American jurisdiction employs it. 230 A number of states have created independent statewide boards or commissions of experts in public defense, but these bodies have only the power to recommend budget appropriations, not to determine them unilaterally. See, e.g., La. Stat. Ann. § 15:146(B)(2) (2018) (creating the Louisiana Public Defender Board, a body of persons with “significant experience in the defense of criminal proceedings”); id. § 15:147(B)(3)–(4) (granting the Board the power to review and make reports to the state legislature regarding public defender budgets); Me. Rev. Stat. Ann. tit. 4, § 1801 (2018) (creating the “independent” Maine Commission on Indigent Legal Services); id. § 1804(3)(I) (giving the Commission power to “submit a biennial budget request” to state budgetary authorities). But that is essentially the system used to fund indigent defense in the United Kingdom.

Under the 1999 Access to Justice Act, Parliament established the Legal Services Commission (LSC), a body that includes “members who . . . have experience in . . . the provision of [criminal defense] services.” 231 Access to Justice Act 1999, c.22, § 25.3 (Eng.). The LSC’s criminal defense functions include two points of significance. First, the LSC promulgates a “general criminal contract” that includes a detailed schedule of the “hourly fees that solicitors cumulate for a variety of defense activities.” 232 Lefstein, supra note 215, at 876–79. Note that in 2013, the United Kingdom amended the fee schedule by statute. See The Criminal Legal Aid (Remuneration) Regulations 2013, SI 2013/435, sch. 4, ¶ 2 (Eng.). The discussion in this section accordingly focuses on the United Kingdom’s pre-2013 approach of leaving funding decisions to the independent LSC. For example, solicitors are paid £69.05 per hour for providing advice and assistance at police stations; that rate is increased to £80 per hour for advice given on certain serious offenses. 233 See The Criminal Legal Aid (Remuneration) Regulations 2013, SI 2013/435, sch. 4, ¶ 2 (Eng.). Solicitors then receive £66.30 per hour for further work in defense of a client who is detained for an additional period. 234 See id. ¶ 3 tbl.A. Rates are also provided for expenses such as expert fees and travel expenses. 235 Lefstein, supra note 215, at 880. The guiding principle behind the fee schedule is that “remuneration” should take into account “the need to secure the provision of [defense] services . . . by a sufficient number of competent persons” as well as “the cost to public funds.” 236 Access to Justice Act § 25.3.

The second crucial point is that the amount of funding budgeted for indigent defense in the United Kingdom is not limited by some external source, such as an annual, politicized appropriations process. The Access to Justice Act is instead clear that “[t]he Lord Chancellor shall pay to the [Legal Services] Commission such sums as are required to meet the costs of any advice, assistance, and representation funded by the Commission as part of the Criminal Defence Service.” 237 Id. § 18.1 (emphasis added). This bears re-emphasis: “[C]riminal legal aid in England [is] a ‘demand-led’ program, in which the government [is] duty bound to find the requisite funds to cover its costs.” 238 Lefstein, supra note 215, at 870 (quoting Lee Bridges, The Right to Representation and Legal Aid, in The Handbook of the Criminal Justice Process 137, 139 (Mike McConville & Geoffrey Wilson eds., 2002)).

The result is that total criminal defense spending in the United Kingdom is largely a function of the fee schedule set by the Legal Services Commission. Defense attorneys keep detailed records of their hours and are paid periodically by the LSC according to the schedule, subject to reconciliation at the end of each year to determine any over- or underpayment. 239 Id. at 881. The LSC may also reassess a claim and adjust payment at any time within two years of a claim submission. 240 Id. at 881–82. In the end, the result of this approach is unsurprising: The United Kingdom spends considerably more on the public defense function than the United States. A 2011 report found, for instance, that whereas the United States spends .0002% of its per capita GDP on public defense per person, that rate was one-thousand times higher in the United Kingdom. 241 See Justice Policy Inst., Finding Direction: Expanding Criminal Justice Options by Considering Policies of Other Nations 51 (2011), [].

How might an independent determination approach work for union fees? At its core, states would commit to leaving final decisions over union funding levels in the hands of an independent board with experience in public-sector labor law. Conveniently, many states already have some form of Public Employee Relations Board (PERB) tasked with administering state laws governing public-sector collective bargaining. 242 For a current list of state Public Employee Relations Boards, see National Right to Work: Public Employee Relations Boards, Nat’l Right to Work Legal Def. Found., [
YQK9] [hereinafter PERB List] (last visited Oct. 17, 2018).
These boards are independent bodies that might reasonably be entrusted with the power to determine whether individual locals have made appropriate funding requests. 243 As discussed below, the PERBs are a natural candidate insofar as they have already long been involved in the business of adjudicating disputes over proper union-funding requests under the fair-share-fee system. See infra section IV.A.2.a. The nuts and bolts of this process could vary. For instance, the approach that would most closely resemble the U.K. indigent defense system would be to have the PERBs set standard reimbursement rates for different kinds of tasks that unions commonly carry out on behalf of workers during negotiations, contract administration, and grievances. 244 One union leader interviewed for this Article stated in reaction to this idea that this kind of approach would convert union financing into a kind of “ratemaking” process already used in other settings, such as with public utilities. I defer until the next subsection a discussion of what kind of activities should be properly paid for by the employer. See infra section IV.A.1. Unions  would  then  keep  track  of  their  hours  spent  on  the  relevant  tasks  and  submit  a  final  tally  for  purposes  of  reimbursement  by  the  public  employer. 245 As discussed below, one could also envision a statewide reimbursement process rather than reimbursement by local employers. See infra note 318 and accompanying text.

That kind of a system would represent a significant departure from the prior norm in which individual locals set their budgets based on their particular circumstances, including their internal staff structures and pay levels and the responsibilities imposed by their contracts. 246 For an example of how one local set its annual budget and thus monthly dues amounts, see Knox v.  Serv. Emps. Int’l Union, Local 1000, 567 U.S. 298, 303–05 (2012). The rate-making approach would arguably be an improvement over this status quo, as standardizing reimbursement levels might rein in spending by less efficient unions. 247 On the other hand, setting standard reimbursement rates may make it harder for unions to retain high-quality personnel to represent and bargain on behalf of workers, as has arguably been the case after the United Kingdom reduced reimbursement levels. Cf. Lefstein, supra note 215, at 901–02 (describing concerns that recent law graduates will not be able to work in indigent criminal defense due to low pay rates). On the other hand, there are costs to changing the system, including the risks that the announced rates may be too high for employers to bear, too low for unions to subsist, or an inaccurate representation of the actual cost of various activities, thereby incentivizing unions to spend more time on more profitable—yet less desirable—tasks. As this Article describes later, a less disruptive approach would be for locals to track their actual costs of bargaining, contract administration, and grievance processing and submit them to the employer for reim­bursement. 248 See infra section IV.A.1. The public employer would then pay the requested amount if it deems the costs appropriate or challenge any expenditure it thinks improper. It would then be up to the PERB, which could delegate initial determination to an administrative law judge, as is common presently, 249 See, e.g., Cal. Code Regs. tit. 8, § 32997 (2019) (“It shall be an unfair practice for an exclusive representative to collect agency fees in violation of these regulations.”); id. § 32680 (stating that the PERB may order a hearing if informal conferencing fails to settle a dispute); id. § 32168 (mandating that “[h]earings shall be conducted by a Board agent designated by the Board,” that is, an administrative law judge). to decide whether the union’s requested reimbursements are permissible under state law—much like PERBs adjudicated challenges brought by objecting workers to fair-share fees charged under the prior regime. 250 See id. § 32994 (governing the procedure for adjudicating fair-share fee challenges in California); infra note 281. These design questions will be discussed at greater length below. 251 See infra section IV.B.

b. Principal Controlled. — A second approach to solving the inde­pendence problem would be to leave funding-level decisions to the relevant principal (criminal defendants or public workers) instead of a government body. There is no obvious analogy to this concept in the world of indigent defense, in which the whole premise is that clients lack sufficient funds to choose their own counsel, let alone to make choices about how much to spend on defense resources. 252 See, e.g., Phyllis E. Mann, Ethical Obligations of Indigent Defense Attorneys to Their Clients, 75 Mo. L. Rev. 715, 731 (2010) (“Some jurisdictions obtain extensive financial information from defendants seeking counsel . . . before deeming a defendant indigent. . . . Often this income cut-off level is quite low.”). But the realm of privately retained defense is actually a decent comparator, in which individual defendants voluntarily decide how much of their resources to spend on their defense, trading off against other goods. 253 See, e.g., Andrew Cain, Bob McDonnell Owes $10 Million in Legal Fees, Rich. Times-Dispatch (Sept. 9, 2016), [] (noting that former Governor Bob McDonnell had placed his “5,099-square-foot home in the Wyndham section of western Henrico County . . . on the market in March 2015” to help pay for his defense bills). It goes without saying that this private system is hardly as maligned as our approach to indigent defense. 254 See supra note 26 and accompanying text.

In the union context, choices about what level of resources to allocate to a public-sector union could be left in the hands of the very public workers whom the union serves. A union might propose a budget each year to its members, who would then decide whether to approve it with full knowledge of the tradeoff that each dollar given to the union is a dollar that members will not have for their personal use. In that way, the principals themselves would dictate their agent’s spending levels, ensuring that the agent is loyal to their interests rather than the employer’s.

Of course, the approach just described is actually a portrayal of the fair-share-fee system, too. Under that system, after all, it was public workers (union members and objectors alike) who financed the union directly out of their paychecks, and who policed union budget levels either directly (via membership referenda on budgets) or indirectly (via membership selection of union representatives who in turn approve budgets). 255 The choice between these two approaches depends on each union’s bylaws. See Tang, Public Sector Unions, supra note 18, at 217 & n.299. But now that the fair-share fee approach is unconstitutional, the union’s funds may not come from objecting worker paychecks. So the principal-controlled approach would need to be modified so that after union members approve a proposed budget, the money would come not from worker paychecks but rather from the government directly. 256 The fact that the government would fund the union’s budget distinguishes this approach from privately retained criminal defense, in which the defendant funds the defense attorney’s budget. But one could build a hypothetical that eliminates this difference without any effect on the attorney’s independence. For example, suppose a criminal defendant is a government employee. Suppose, further, that instead of paying her defense attorney out of her wages, the employee asks the government to pay the attorney directly in amounts set by the employee (with an offsetting reduction to wages). Such employer-paid litigation costs are not unheard of; some employers offer private litigation insurance. Does the defense attorney suddenly face an inherent conflict of interest because the government is writing the check rather than the defendant? The answer is “no” for a good reason: It is the employee, not the government, who is in control of the defense attorney and budget at the end of the day. So too with a member-controlled union-funding approach. Union members would remain aware, however, that there is not an unlimited pot of resources: Every dollar directed to the union is in theory a dollar that the employer could otherwise direct to public employee wages or benefits. So members would continue to have the incentive to police against excessive union budget requests.

Notwithstanding the continued existence of this policing incentive, one might naturally wonder whether injecting the government employer as the ultimate source of union funding may make the union more dependent on the employer than was the case under the fair-share-fee system. It is a legitimate concern, but the answer is to realize that the fair-share-fee system itself jeopardized union independence in a similar way. Under that approach, too, the government employer was the ultimate source of union funding (since the employer paid the very wages that were channeled to the union), the union was dependent on employer cooperation (since objecting workers only paid the union because of the employer’s promise to fire anyone who did not), and the employer could threaten to eliminate the fair-share-fee clause during subsequent negotiations. 257 See Tang, Public Sector Unions, supra note 18, at 215–16 (“[A] direct-reimbursement regime poses no greater an agency problem than the fair-share-fee system it would replace because in both cases the ultimate power over the union’s affairs lies in the hands of the union’s members, not the employer.”); infra note 357.

The similarity between fair-share fees and direct employer reim­bursement has long been recognized. One company union representative testified in 1935 that distinguishing between the two for purposes of union independence is “absurd” because the “ultimate source of the money paid in by the members of a labor organization is from the employer and I cannot see what difference it makes whether the [employer] turns over a lump sum each year, according to a fixed arrangement, or whether the men pay a check-off.” 258 National Labor Relations Board: Hearings on S. 1958 Before the S. Comm. on Educ. & Labor, 74th Cong. 412 (1935) (statement of Jack Larkin, General Chairman Employees’ Reps., Weirton Steel Co.), reprinted in 2 NLRB, Legislative History of the National Labor Relations Act, 1935, at 1798 (1949). An approach in which union members determine the amount to be paid by the public employer to their union therefore poses no greater risk to union independence than the system it would replace.

* * *

This Article has argued that among the four major legislative approaches that pro-labor states can take after Janus—the right-to-work default regime, a right-to-work regime with minor changes, a members-only bargaining approach, and the direct government reimbursement approach—the reimbursement model has the greatest potential to preserve the existing system of public-sector labor relations. It has further suggested that the union-independence objection to this model can be answered by examining a close analog: indigent criminal defense. That setting demonstrates certain approaches that would not work, but it also points out two possible approaches that could: leaving union reimbursement levels to (1) an independent final determination by a neutral expert or (2) the control of the union’s members. Several issues of legislative design and practical implementation arising out of these alternatives are explored next.

IV. State Labor Law After Janus

This Part presents model legislative text that pro-labor states may draw on to implement a government reimbursement approach to public-sector union financing. In addition to offering a menu of options for how reimbursement amounts may be determined, this Part considers several questions that arise out of the proposed legislation, such as whether government reimbursement should be mandatory or permissive for local public employers, whether the reimbursement should be financed by the employer itself or some statewide source, and how disputes over reimbursement requests should be resolved. This Part then analyzes how the new labor-law regime might be carried into practice, focusing on the union’s internal process for recruiting members and administering the reimbursement provision as well as the timing of when the reimbursement approach should be implemented. 259 The suggestions that follow rely heavily on interviews conducted with a range of stakeholders in public-sector union financing, including officials who represent public employers, union organizers and officers, and persons affiliated with state public employee relations boards. I am heavily indebted to all who shared their time with me, including but not limited to the persons listed in the acknowledgments. Finally, it considers two objections to the reimbursement proposal advanced by Professors Fisk and Malin in an important forthcoming work. 260 See generally Fisk & Malin, supra note 38.

A. Model Legislation

In its most basic form, amending state labor law to authorize govern­ment reimbursement of union bargaining costs entails two steps: Step one is to carve out an exception for such reimbursement from existing prohibitions against government support of unions. Step two is to describe how reimbursement amounts would be determined.

1. The Government-Reimbursement Exception. — Starting with the first step, one must recognize that—absent legislative amendment—attempts by public employers to reimburse unions would likely violate existing unfair labor practice laws in the twenty-three former fair-share jurisdictions. Reimbursement would most clearly violate the law in states that explicitly forbid public employers to “contribute financial or other support to [a union].” 261 See Cal. Gov’t Code § 3543.5(d) (2019) (prohibiting public-school employers from contributing financial support to a teachers union); 5 Ill. Comp. Stat. Ann. 315/10(a)(1) (West 2018) (deeming it an unfair labor practice for an employer to contribute financially to its employees’ labor organization); Md. Code Ann., State Pers. & Pens. § 3-306(a)(2) (West 2018) (prohibiting the state from contributing financial support to a labor organization); Minn. Stat. § 179A.13 subdiv. 2(2) (2018) (same); Ohio Rev. Code Ann. § 4117.11(A)(2) (2018) (same); Vt. Stat. Ann. tit. 21, § 1621(a)(2) (2018) (prohibiting employers from contributing financially to labor organizations); Wis. Stat. § 111.84(1)(b) (2018) (same). These prohibitions mimic the NLRA. See 29 U.S.C. § 158(a)(2) (2012) (“It shall be an unfair labor practice for an employer . . . to . . . contribute financial or other support to [any labor organization] . . . .”). It would also be problematic in states without such express bans. Many of these states forbid public employers to “assist in the . . . existence or administration of” a union, 262 E.g., Del. Code tit. 19, § 1307(a)(2) (2018); Haw. Rev. Stat. § 89-13(a)(2) (2018); Mass. Gen. Laws Ann. ch. 150E, § 10(a)(2) (West 2018); Or. Rev. Stat. § 243.672(1)(b) (2018). and one can see how direct reimbursement might run afoul of such language. And while some states impose more general prohibitions that make it improper for employers to “dominate or interfere with” the administration of a union, there is precedent in these states suggesting that a direct employer payment to the union would be unlawful, too. 263 See N.Y. Civ. Serv. Law § 209-a(1)(b) (McKinney 2018) (making it an improper practice for an employer to “dominate or interfere with the formation or administration of any employee organization”); Croton Police Ass’n, 15 PERB ¶ 4644 (N.Y. Nov. 5, 1982) (ordering withdrawal of a union’s demand for “a monetary payment to the [union] itself” because the demand “raises the possibility of the [employer’s] impermissible financial support of the [union]”); Police Ass’n of the City of Yonkers, 14 PERB ¶ 4516 (N.Y. Jan. 20, 1981) (ordering withdrawal of a union’s request for a $50,000 donation from the employer in lieu of paid release time for members because such a payment “would raise questions of improper employer support of an employee organization”). So unless some exception is enacted to these unfair labor practice prohibitions, state law would likely preclude a government attempt to reimburse a union directly for its collective bargaining costs. 264 To be sure, there are some states with more general bans that lack precedent equating direct union payments with impermissible employer domination or interference. Even so, it may be wise for state lawmakers to enact an express statutory framework for reimbursement both to ensure that it is a mandatory subject of bargaining and to prevent employers from using reimbursement as a subversive tool for extracting concessions from the union. See supra section III.D.3.

Fortunately, there is a great deal of legislative precedent for exceptions to the general prohibition. Fair-share-fee clauses, for example, were themselves permitted by way of an exception from the general bar against employer support. 265 See, e.g., Alaska Stat. § 23.40.110 (2018) (“Nothing in this chapter prohibits a public employer from making an agreement with an organization to require as a condition of employment . . . payment by the employee to the [union] of a service fee to reimburse the [union] for the expense of representing the members of the bargaining unit.”); Cal. Gov’t Code § 3546(a) (“Notwithstanding any other provision of law . . . the employer shall deduct the amount of the fair share service fee . . . and pay that amount to the [union].”); N.J. Stat. Ann. § 34:13A-5.5(a) (West 2018) (“[T]he majority representative and the public employer of public employees in an appropriate unit shall . . . negotiate concerning the subject of requiring the payment by all nonmember employees in the unit to the majority representative of a representation fee in lieu of dues for services rendered by the majority representative.”); N.Y. Civ. Serv. Law § 208(3) (“[E]very employee organization that has been recognized . . . as the . . . representative of employees of the state . . . shall be entitled to have deducted from the wage . . . of the employees . . . the amount equivalent to the dues levied by such employee organization . . . . [T]he state . . . shall make such deductions and transmit the sum.”). For without such an exception, a government employer’s act of forcing objecting workers to pay fees to the union on pain of being fired could easily be considered financial support or assistance from the employer to the union. 266 Other kinds of exceptions permeate state labor law as well. Illinois, for example, provides that “an employer shall not be prohibited from permitting [union-affiliated] employees to confer with him during working hours without loss of time or pay,” even though that act could just as easily constitute financial support to a union. 5 Ill. Comp. Stat. Ann. 315/10(a)(1); see also 29 U.S.C. § 158(a)(2) (creating the same exception in the NLRA).

Now that Janus has invalidated these fair-share fee exceptions to the state law prohibitions against support for a union, what is needed is a new kind of “government reimbursement exception” to replace them. There are strong reasons for the exact text of this exception to vary across different states. But just to illustrate what a common version of the exception might look like, consider the following:

Notwithstanding any other provision of law, a public employer may reimburse an employee organization for costs germane to the collective bargaining process, contract administration, and pursuing matters affecting wages, hours and other conditions of employment.

As should be apparent, the critical legislative drafting question is how to define the domain of costs that may be reimbursed to the union. 267 Of course, even after the domain of reimbursable costs is delineated, there is still the question of how the union and public employer will go about deciding which costs fit within that domain in any given period. That is the subject of the next legislative drafting question regarding a PERB-determination model, a membership-determination model, and a combination of the two. See infra section IV.A.2. In theory, a state could impose any kind of limit it wishes. The above language would limit reimbursement to only those costs that are germane to three categories of activities: collective bargaining; contract administration; and pursuing matters affecting wages, hours, and conditions of employment (including, for example, grievance adjustment). 268 See Kearney & Mareschal, supra note 2, at 305 (arguing that “a responsive and effective grievance procedure” is the “primary means for living peacefully with the contract”). Such a limit would mimic the law that has been in force in several pro-labor states for decades after Abood first allowed unions and public employers to charge fair-share fees “for the purposes of collective bargaining, contract administration, and grievance adjustment.” 269 Abood v. Detroit Bd. of Educ., 431 U.S. 209, 225–26 (1977); see also, e.g., Cal. Gov’t Code § 3583(a)(1) (mandating a fair-share fee to “cover the cost of negotiation, contract administration, and other activities of the employee organization that are germane to its functions as the exclusive bargaining representative”); Colo. Rev. Stat. § 29-5-209 (2018) (“A collective bargaining agreement may contain provisions requiring all members of the bargaining unit, as a condition of employment, to pay necessary fees and expenses germane to collective bargaining . . . .”); Del. Code tit. 19, § 1302(q) (2018) (entitling an employee to a rebate of the “portion of [her] pro rata share of the expenditures of the exclusive representative which are not made for purposes pertaining to the collective bargaining process, contract administration or pursuit of matters affecting wages, hours and other conditions of employment”); 5 Ill. Comp. Stat. Ann. 315/6(a) (“Employees may be required, pursuant to the terms of a lawful fair share agreement, to pay a fee which shall be their proportionate share of the costs of the collective bargaining process, contract administration and pursuing matters affecting wages, hours and other conditions of employment . . . .”); Md. Code Ann., Educ. § 6-407 (West 2018) (providing for a “requirement of a reasonable service or representation fee, to be charged nonmembers for representing them in negotiations, contract administration, including grievances, and other activities”); Wash. Rev. Code § 41.80.100 (2018) (providing that fair-share fees may be charged for “purposes germane to the collective bargaining process, to contract administration, or to pursuing matters affecting wages, hours, and other conditions of employment”). Codifying statutory language to reflect both existing state law as well as pre-Janus judicial precedent governing the costs properly chargeable to objecting employees serves three important and interlocking aims.

First,  and  most  importantly,  drafting  the  new  government  reimburse­ment  exception  to  incorporate  existing  statutory  and  case  law  governing  fair-share  fees  serves  the  value  of  stability. Public employers and unions have been cooperating in pursuit of labor peace in the twenty-three former fair-share jurisdictions for decades, and these jurisdictions forcefully avow that their systems of labor relations have been effective. 270 See Brief for the States of New York et al. as Amici Curiae in Support of Respondents at 2, Janus v. Am. Fed’n of State, Cty., & Mun. Emps., Council 31, 138 S. Ct. 2448 (2018) (No. 16-1466), 2018 WL 529834 (arguing on behalf of twenty states and Washington, D.C., that the “success of state labor-relations systems” based on fair-share-fee financing “is evidence that . . . states and local governments [have] adopt[ed] policies best tailored to meet their needs in achieving labor peace”). A core part of that system is that unions have enjoyed some financial security, a settled expectation of access to the resources they need to engage in collective bargaining, contract administration, and grievance adjustment. State fair-share-fee laws protected that expectation, 271 See supra note 269 and accompanying text; infra note 284 and accompanying text. and a long line of case law has settled particularized disputes regarding the sorts of expenses unions can permissibly charge to all workers and therefore anticipate for budgetary planning purposes. 272 See, e.g., Locke v. Karass, 555 U.S. 207, 210 (2009) (holding that the First Amendment permits a local union to charge nonmembers for qualifying litigation expenses incurred by a national union affiliate); Lehnert v. Ferris Faculty Ass’n, 500 U.S. 507, 532 (1991) (upholding a local union’s decision to charge nonmembers a fair-share-fee covering costs of state and national affiliates but not lobbying activities); Abood, 431 U.S. at 222. Importantly, that case law also encompasses state court and administrative rulings regarding the kind of union costs that may be properly charged to all workers under each state’s fair-share clause 273 See, e.g., Cumero v. Pub. Emp’t Relations Bd., 778 P.2d 174, 176 (Cal. 1989) (deciding whether certain challenged costs are permissible under the state fair-share fee law, even if not forbidden by the First Amendment); Melvin H. Dalan, 8 PERI ¶ 1078 (Ill. Educ. Labor Relations Bd. July 10, 1992), 1992 WL 12647341 (applying state law and Lehnert’s chargeability rules to decide a host of fact-bound challenges to a union’s fair-share fee); James J. Burton, 29 PERB ¶ 3074 (N.Y. Nov. 20, 1996) (upholding a challenge to an agency fee under state law); In re Raymond A. Heitger, SERB 95-004 (Ohio State Emp’t Relations Bd. Jan. 26, 1995), 1995 WL 17827475, at *5 (applying state law to dismiss a fair-share-fee challenge). —rulings that can help minimize the number and scope of disagreements over requested payments under a reimbursement regime. 274 See infra section IV.B (describing dispute-resolution procedures designed to ensure that disagreements over reimbursement requests under the direct-government-payment approach are resolved in a manner that is materially identical to disagreements brought by objecting workers under Chi. Teachers Union, Local No. 1 v. Hudson, 475 U.S. 292 (1986)). In short, state lawmakers can lock in these established understandings by using the exact language from state fair-share laws to define the domain of expenditures for which unions can receive reimbursement from public employers.

Adopting a government reimbursement exception to permit employers to reimburse unions for the same costs that unions have long charged workers under fair-share statutes—nothing more, nothing less—is also neutral. The purpose of the new exception, in other words, would be to allow the existing system of public-sector labor relations to continue uninterrupted, not to decrease or increase union strength as compared to the pre-Janus baseline. Neutrality is important because it allows this project of legislative reform to focus not on value-laden questions such as the desirability of more or less influential unions but rather institutional design issues such as how best to minimize switching costs and unintended consequences—costs that are reasonably viewed as negative regardless of one’s a priori views on unionization.

Finally, a stable and neutral version of the government reimbursement exception depoliticizes the issue to the greatest extent possible. That is not to say that politics can be expected to go out the window; labor proponents and opponents will assuredly debate the merits of a government reimbursement bill with fervor. But when the aim of such a bill is to preserve a settled expectation regarding the kinds of costs unions have long been allowed to charge in carrying out their core bargaining activities rather than to expand or curtail the financial resources available to unions, lawmakers may reasonably view a vote on the government reimbursement approach as a vote for the status quo instead of a vote for some sweeping change. 275 By comparison, a vote for members-only bargaining or to do nothing (and thereby permit Janus to impose a new right to work) would seem rather revolutionary. To illustrate, consistent with the rule announced in Abood, existing state fair-share laws forbid unions to charge objecting workers for the cost of political activities, such as contributions to political campaigns. 276 See Abood, 431 U.S. at 235–36; infra notes 283–285 and accompanying text. By likewise prohibiting government reimbursement for union political expenses, lawmakers can help preempt arguments that the bill is simply a handout to a preferred political cause.

Because the guiding principle is to peg the scope of permissible government reimbursement to the expenses that have long been chargeable under state fair-share-fee laws, thereby preserving settled employer and union expectations, each state will naturally need to adopt different language for its reimbursement exception in line with its existing fair-share statute. The model provision above defines reimbursable expenses in positive terms, authorizing public employers to offset costs germane to a union’s collective bargaining, contract administration, and grievance adjustment activities. 277 See supra section IV.A. Eleven  state  fair-share-fee  laws  use  some  kind  of  positive  formulation,  seven  of  which  include  those  three  categories  or  some  variation  thereof. 278 See supra note 269 and accompanying text (listing California, Colorado, Delaware, Illinois, Maryland, and Washington); see also 28 R.I. Gen. Laws § 28-9.3-7 (2018) (permitting a fair-share fee for costs “involved in securing a contract and the administration of any collective bargaining agreement”); 36 R.I. Gen. Laws § 36-11-2 (2018) (requiring a fair-share fee “toward the negotiation and administration of any collective bargaining agreement”). Each of these seven states would therefore enact an exception that looks substantially like the model given above, though their exact language would vary in line with differences in their fair-share fee formulations. 279 Compare, e.g., Colo. Rev. Stat. § 29-5-209 (2018) (permitting chargeability of expenses “germane to” collective bargaining and contract enforcement), with, e.g., 5 Ill. Comp. Stat. Ann. 315/6(a) (West 2018) (permitting chargeability of “costs of the collective bargaining process, contract administration, and pursuing matters affecting wages, hours, and other conditions of employment”).

Three other positive state laws—Alaska, Maine, and Pennsylvania—use more general language, allowing public unions to charge all workers for “expenditures that are germane to the [union’s] representational activities.” 280 Me. Rev. Stat. Ann. tit. 26, § 979-B (2007); see also Alaska Stat. § 23.40.110(b) (2018) (allowing employers to require “payment by the employee to the exclusive bargaining agent of a service fee to reimburse the exclusive bargaining agent for the expense of representing the members of the bargaining unit”); 43 Pa. Cons. Stat. § 1102.2 (2018) (permitting reimbursement for activities “reasonably employed to implement or effectuate the duties of the [union] as exclusive representative”). So those states would adopt the same language in their new exceptions allowing public employers to reimburse unions directly, thereby ensuring that future disputes over the reimbursability of certain union expenditures will be resolved under settled state law precedents. 281 See supra note 273 and accompanying text. To the extent that a state worries that such an open-ended formulation may expand the scope of reimbursable activities beyond the pre-Janus baseline, the state can enact an express limitation incorporating federal court precedents as used in New Mexico and as in the model given below. See infra notes 290–291 and accompanying text. Finally, New Mexico takes a different approach. While it identifies collective bargaining, contract administration, and grievance adjustment as permissible costs, the state also explicitly references “expenditures by a [union] which are permissibly chargeable to all employees in the appropriate bargaining unit under United States and New Mexico statutes and case law.” 282 N.M. Stat. Ann. § 10-7E-4(J) (West 2018). This kind of an express formulation offers a promising approach for states with fair-share-fee laws that are silent as to the categories of properly chargeable expenditures.

In  contrast  to  the  state  fair-share  laws  that  positively  enumerate  categories  of  chargeable  expenditures,  several  others  use  a  negative  approach  that  permits  unions  to  charge  for  all  activities  other  than  certain  categories  that  are  expressly  excluded.  New York’s law is a good example; it permits unions and public employers to enforce a fair-share fee so long as the union has “a procedure providing for the refund . . . of expenditures by the organization in aid of activities or causes of a political or ideological nature only incidentally related to terms and conditions of employment.” 283 N.Y. Civ. Serv. Law § 208(3) (McKinney 2018). Seven additional state laws use such an approach, 284 See Cal. Gov’t Code § 3515.8 (2019) (permitting, for state employees, reimbursement of “expenditures by the recognized employee organization that is either in aid of activities or causes of a partisan political or ideological nature only incidentally related to the terms and conditions of employment”); Haw. Rev. Stat. § 89-4(a) (2018) (requiring refund of “expenditures made by the exclusive representative for activities of a political and ideological nature unrelated to terms and conditions of employment”); 115 Ill. Comp. Stat. Ann. 5/11 (West 2018) (“The amount certified by the exclusive representative shall not include any fees for contributions related to the election or support of any candidate for political office.”); N.J. Stat. Ann. § 34:13A-5.5 (West 2018) (requiring a refund of “any part of [a] fee” that is “in aid of activities or causes of a partisan political or ideological nature only incidentally related to the terms and conditions of employment”); Ohio Rev. Code Ann. § 4117.09(C) (2018) (entitling workers to a “rebate of expenditures in support of partisan politics or ideological causes not germaine [sic] to the work of employee organizations in the realm of collective bargaining”); see also Mass. Gen. Laws Ann. ch. 150E, § 12 (West 2018) (enumerating five categories of nonchargeable expenditures); Minn. Stat. § 179A.06 (2018) (excluding “cost of benefits . . . available only to members of the [union]”). which is grounded in the other half of Abood’s holding: that objecting workers cannot be required to share the union’s costs “on behalf of political candidates, or toward the advancement of other ideological causes not germane to its duties as collective-bargaining representative.” 285 Abood v. Detroit Bd. of Educ., 431 U.S. 209, 235 (1977). Any state that has chosen to define the chargeable expenses that unions can incur by excluding impermissible costs (rather than enumerating permissible ones) should do the same for its government reimbursement exception so as to preserve all relevant state law rulings delineating the chargeable–nonchargeable line. 286 See supra note 273 and accompanying text.

Finally, nine jurisdictions’ labor laws neither expressly include certain union activities as chargeable nor exclude others as nonchargeable. 287 See Cal. Gov’t Code § 3502.5 (local public employees); Conn. Gen. Stat. § 5-280(a) (2018) (authorizing fees in “an amount equal to the regular dues, fees and assessments that a member is charged,” without describing what activities are chargeable or limits thereon); D.C. Code § 1-617.07 (2018) (failing to describe what is chargeable); Md. Code Ann., State Pers. & Pens. § 3-502 (West 2018); Mont. Code Ann. § 39-31-204 (West 2019); N.H. Rev. Stat. Ann. §§ 273-A:1, 273-A:3 (2018); Or. Rev. Stat. §§ 243.650(10), 243.672(1)(c) (2017); Vt. Stat. Ann. tit. 3, § 962 (2018); Vt. Stat. Ann. tit. 16, §§ 1981, 1982 (2018); Wash. Rev. Code § 41.59.100 (2018) (focusing on teachers).
Note that the total number of state laws surveyed in this Part exceeds the number of fair-share jurisdictions because some states use different definitions for different sectors of public employment. Compare Cal. Gov’t Code § 3546(a) (providing a positive definition for California teachers), with id. § 3515.8 (providing a negative definition for California state employees), with id. § 3502.5 (providing no guidance as to chargeable expenses for nonteacher local public employees).
These states were able to stay silent as to chargeable union expenses in their fair-share laws because the First Amendment itself limited the scope of costs that could be charged to objecting workers; authorizing an undefined fair-share service fee was thus tantamount to permitting fees to the limits of the First Amendment. 288 See supra notes 269, 272 and accompanying text. Now that Janus has forbidden all fair-share fees, however, statutory silence will not do—a state interested in authorizing direct government reimbursement for union activities must provide some guidance as to what may be reimbursed. 289 Moreover, simply recycling language from the previously silent fair-share laws would not provide future decisionmakers sufficient indication that lawmakers intended future disputes over reimbursement amounts to be resolved by reference to the existing body of state and federal case law that developed in Abood’s aftermath.

Fortunately, New Mexico’s fair-share provision provides an exemplar for this last group of states. 290 See N.M. Stat. Ann. § 10-7E-4(J) (West 2018); supra note 282 and accompanying text. Here is how such a provision would look:

Notwithstanding any other provision of law, a public employer may reimburse an employee organization for expenditures which were permissibly chargeable to all employees in the appropriate bargaining unit under [Name of State] and United States statutes and case law prior to the Supreme Court’s decision in Janus v. AFSCME, Council 31. 291 States often reference specific Supreme Court decisions to fix some universe of legal rights for purposes of public-sector labor law. See, e.g., Cal. Gov’t Code § 3546(f) (“The employer of a public school employee shall provide the exclusive representative of a public employee with [certain information] . . . so that the exclusive representative can comply with the notification requirements set forth by the United States Supreme Court in Chicago Teachers Union v. Hudson.” (citation omitted)).

There are two final points to mention regarding the government reimbursement exception. Professor Sachs has astutely recognized that some states may need to enact an additional statutory exception to avoid an unintended consequence that would otherwise arise under state campaign finance law. 292 See Sachs, supra note 18, at 1074. To explain, government contractors in some states are forbidden to make political contributions to officials out of concern that such contributions will be viewed as corrupt bribes intended to persuade lawmakers to award the entity some desired contract. 293 See, e.g., Conn. Gen. Stat. § 9-612(f)(1)(D), 9-612(f)(2) (2018) (prohibiting state contractors from making contributions to various types of political candidates). To avoid converting unions into contractors whose political expression would be restricted under such rules, states would need to clarify that the “receipt of government reimbursement for a union’s collective bargaining activities shall not render the union a ‘government contractor’ for purposes of state campaign finance law.” Again, the point here is not to confer some special treatment on unions but rather to preserve a preexisting system of labor relations that states believe has operated successfully. 294 Indeed, there is little reason to think that the contractor prohibitions were intended to cover unions in the first place, since workers (not elected officials) choose which unions to certify and thus which unions will receive funding. See supra note 207 and accompanying text. Furthermore, to the extent that political contributions may be viewed as an attempt to curry more favorable union treatment and financial security from government officials, that same concern would have existed under the fair-share regime in which government officials had the freedom to decline to enter into fair-share clauses. See supra note 62 and accompanying text.

Finally, for those who continue to worry about the union-independence concern, 295 See supra section III.D.1. it is important to recognize the state law prohibitions aimed at preserving union independence that would remain in force even after an exception permitting government reimbursement is enacted. For one thing, every state would continue to have on the books bans against public employer “dominat[ion] or interfere[nce] with the . . . admin­istration of any” union. 296 See Cal. Gov’t Code § 3519(d); supra note 202 and accompanying text. So any employer that actually threatened to use the reimbursement mechanism to control or subvert the union’s ability to zealously represent workers would commit an unfair labor practice. 297 See supra note 202 and accompanying text. As described next, it is hard to see how such opportunities would arise once state lawmakers are clear on how reimbursement is to function in practice, free of employer influence. For another, public employers would also continue to be required to recognize the unions that a majority of workers select as exclusive representative, further limiting the ability of the government to foist upon workers some disloyal bargaining agent. 298 See Barenberg, Democracy and Domination, supra note 178, at 790 (noting that in the NLRA context, like the public-sector context, employers are required to “recognize the union chosen by a majority of the bargaining unit members,” a fact that “formally forecloses discriminatory recognition of a company union and nonrecognition of a nondominated, majority union”).

2. Determining Reimbursement Amounts. — The second step in enacting a government-reimbursement regime is to specify how individual unions and public employers are to settle on the amount of reimbursement to be awarded. In a perfect world, unions and employers would always see eye-to-eye on the sum total of expenses that fit within the categories permitted under a state’s particular reimbursement exception. But that is unrealistic. And importantly, states cannot leave the dispute-resolution process unspecified because doing so would introduce the risk that unions and employers might leave reimbursement amounts as a matter for adversarial or political negotiation, choices that would jeopardize union independence and financial predictability. 299 See supra section III.D.3.

Accordingly, states should enact a clear statutory procedure for finally determining the amount of government reimbursement to which a union is entitled each year. 300 For the sake of simplicity, the Article assumes that the union makes annual reimbursement requests to the employer. But unions and employers may well prefer, for budgeting and liquidity reasons, to square up more frequently, through monthly or quarterly reimbursement requests and accountings. As the preceding section made clear, there are two promising options: a model in which disputes over reimbursement requests are resolved by an independent, neutral body such as a PERB, 301 See supra section III.D.4.a. and a model in which union members have the final say. 302 See supra section III.D.4.b. A third option would be some hybrid of the two. This Article proposes statutory language to implement these three options presently, considering important design questions along the way.

a. The PERB Model. — The least disruptive model, and hence the default model suggested here, is for states to enact a procedure for determining union-reimbursement amounts that essentially mimics the process that has been used under the fair-share fee regime. Under that process, unions set their annual budgets, allocating one portion to activities that are germane to collective bargaining (and thus chargeable to all workers) and another portion to nongermane activities such as political contributions (which are not chargeable to objecting workers). 303 See, e.g., Knox v. Serv. Emps. Int’l Union, Local 1000, 567 U.S. 298, 303–04 (2012) (describing the budgeting process for one local public-sector union). Workers who disagreed with their union’s proposed allocation could then challenge their fair-share fee assessment through a legally prescribed procedure in which the power to decide on the chargeability of particular expenditures would be generally left to the state PERB. 304 For a list of state PERBs, see PERB List, supra note 242. For example, Hawaii’s public-sector fair-share law provided that “[i]f a nonmember employee objects to the amount [charged by the union], the nonmember employee may petition the [Hawaii Labor Relations Board] for review thereof within fifteen days.” 305 Haw. Rev. Stat. § 89-4(a) (2018); see also, e.g., Cal. Gov’t Code § 3515.8 (2019) (providing that the PERB “may compel the recognized employee organization to return that portion of a fair share fee which the board may determine to be subject to refund under the provisions of this section” governing state public employees); id. § 3541.5 (granting jurisdiction to the PERB over unfair-labor-practice charges involving teachers unions); 115 Ill. Comp. Stat. Ann. 5/15 (West 2018) (granting the Illinois Educational Labor Relations Board jurisdiction to decide unfair-labor-practice charges); Ohio Rev. Code Ann. § 4117.09(C) (2018) (“[A] challenge to the [union’s internal determination regarding the chargeability of a fair-share fee] may be filed with the state employment relations board within thirty days of the determination date . . . .”); Pub. Emps. Fed’n v. Pub. Emp’t Relations Bd., 461 N.Y.S.2d 924, 925 (App. Div. 1983) (“Included within the ambit of PERB’s authority to prevent unfair practices is control over agency shop fee deductions.” (citation omitted)); Cal. Code Regs. tit. 8, § 32997 (2019) (providing that improper collection of agency fees constitutes an unfair labor practice).
Note that some states chose not to vest power to resolve fair-share-chargeability disputes in PERBs, instead leaving claims to private arbitration, the judiciary, or some other state entity. See, e.g., Minn. Stat. § 179A.06(3) (2018) (“The commissioner [of the bureau of mediation services] shall hear and decide all issues in [fair-share-fee] challenges.”); N.J. Stat. Ann. § 34:13A-5.6 (West 2018) (providing that fair-share challenges be reviewed by a specially constituted public employment relations commission appeals board). Such states would encounter greater switching costs to implement a new PERB-based model and so may wish to continue their existing approach or adopt a membership-based determination model instead. See, e.g., 43 Pa. Cons. Stat. § 1102.5(c) (West 2018) (providing that fair-share-fee challenges “shall be resolved . . . by an impartial arbitrator” and that “[t]his subsection does not preclude a constitutional challenge being filed in a court”).

States can continue this same PERB-based process for adjudicating disagreements over the permissibility of union expenditures, only now with the public employer filing challenges instead of objecting workers. By doing so, states can ensure that employers and unions alike have a fair, established, and expert forum for resolving disputes. Such an approach is also responsive to concerns over union independence. No union member, after all, should worry that their bargaining representative will soft-pedal worker interests at the bargaining table when the amount of funding to which the union is entitled is ultimately determined not by the employer but by an independent body through an even-handed adjudicatory process.

At a first cut, then, a state wishing to employ the PERB-based model could enact a provision along the following lines:

(a) When a collective bargaining agreement includes a provision requiring a public employer to reimburse an employee organization for expenditures permitted under [the state’s reimbursement exception], 306 See supra section IV.A.1 (proposing text for state-reimbursement exceptions). the employee organization shall annually certify to the employer the amount of permissible expenditures it has actually incurred.

(1) If the employer agrees that the identified expenditures are reimbursable under [the state’s reimbursement exception], it shall promptly reimburse the employee organization for the certified amount.

(2) If the employer disagrees with any of the identified expenditures, the employer must file within thirty days an objection with the [state PERB] identifying the specific expenditures to be challenged. The employer shall set aside the challenged expenditures in an escrow account maintained by the employer, employee organization, or [state PERB], 307 Such escrow accounts are already common. See, e.g., Del. Code tit. 19, § 1319 (2018) (union-maintained escrow); 115 Ill. Comp. Stat. Ann. 5/11 (PERB-maintained escrow); Minn. Stat. § 179A.06(3) (employer-maintained escrow). and promptly reimburse the employee organization for all other expenditures.

(b) [Provision incorporating a state’s administrative, alternative dispute, and/or judicial resolution mechanisms for resolving fair-share fee challenges].

Four major follow-up questions to this PERB-based model legislation warrant exploration. First, as the bracketed placeholder provision in subsection (b) suggests, it is important for the legislation to identify the administrative (or other) review procedures, if any, that should precede the PERB’s final decision. After all, it may be impractical (especially in larger states) for PERBs to serve as an initial fact-finder and decisionmaker with respect to every dispute over union reimbursement.

Fortunately, states need not reinvent the wheel: They can simply codify the same procedural mechanism that preceded PERB determination of fair-share challenges brought by objecting workers. 308 Note that some states have codified these procedures by statute, whereas others have left the procedures to be established by the PERB itself. See infra notes 309, 311, 313 (listing procedural mechanisms established by statute and by PERB regulation); see also, e.g., Cal. Gov’t Code § 3546(a) (“Agency fee payers shall have the right, pursuant to regulations adopted by the Public Employment Relations Board, to receive a rebate or fee reduction upon request . . . .”).
Note that the placeholder in subsection (b) could either explicitly restate the terms of the relevant state statutes and regulations, or it could incorporate them by reference. For an example of language along the latter lines, consider: “The procedures governing review of challenges brought pursuant to section (a)(2) shall be the same as those that governed challenges brought by objecting workers to fair-share fees under all applicable state laws prior to the Supreme Court’s decision in Janus v. AFSCME, Council 31.”
To that point, several states left initial decisionmaking in the hands of administrative law judges (ALJs) who adjudicated disputes over fair-share fees with some frequency. 309 See, e.g., 5 Ill. Comp. Stat. Ann. 315/11(a) (authorizing a “hearing before the Board . . . or before a qualified hearing officer designated by the Board” for unfair-labor-practice charges); N.Y. Civ. Serv. Law § 209-a(4)(a) (McKinney 2018) (“A party filing an improper practice charge under this section may petition the board to obtain injunctive relief, pending a decision on the merits of said charge by an administrative law judge . . . .”); Cal. Code Regs. tit. 8, § 32168(a) (“Hearings shall be conducted by a Board agent designated by the Board, except that the Board itself or a Board member may act as a hearing officer.”). That approach would make great sense in the reimbursement context, too, in which ALJs can serve a fact-finding role with decisions appealable to the PERB. A similar answer resolves questions such as the proper burden of proof in reimbursement challenges; states can simply allocate the burden in the same way as before. 310 To this point, some states placed the burden of establishing the permissibility of a challenged expenditure on the union, while others required the challenger to show that a given expenditure was arbitrary or capricious. Compare, e.g., Cal. Code Regs. tit. 8, § 32994(b)(6) (“The exclusive representative bears the burden of establishing the reasonableness of the amount of the chargeable expenditures.”), with Ohio Rev. Code Ann. § 4117.09(C) (requiring a challenger to show that the fee was “arbitrary or capricious”).

To be sure, not all state fair-share laws utilized ALJs for initial determination. Some states used a private initial dispute resolution model, where rulings were first given by arbitrators rather than ALJs. 311 See, e.g., Cal. Code Regs. tit. 8, § 32994(b)(3)–(4) (“Upon receipt of an agency-fee challenge, the exclusive representative shall within 45 days of the last day for filing a challenge request a prompt hearing regarding the agency fee before an impartial decisionmaker . . . selected by the American Arbitration Association or the California State Mediation Service.”). Although the ALJ model may serve valuable benefits such as providing useful information to the public about the law, 312 See Judith Resnik, The Privatization of Process: Requiem for and Celebration of the Federal Rules of Civil Procedure at 75, 162 U. Pa. L. Rev. 1793, 1797 (2014). a state could reasonably prefer to continue with a private dispute model to avoid disturbing settled practices and expectations regarding the body vested with initial decisionmaking authority. 313 Note that many state fair-share laws also required unions to establish internal procedures (including, in some cases, recourse to arbitration, see supra note 305) for objecting workers to challenge fair-share-fee determinations prior to recourse to the administrative law process. See, e.g., Cal. Gov’t Code § 3515.8 (requiring “procedures established by the recognized employee organization” for rebate); N.J. Stat. Ann. § 34:13A-5.6 (West 2018) (“[T]he [fair-share fee] shall be available only to a majority representative that has established and maintained a demand and return system which provides pro rata returns . . . .”); Ohio Rev. Code Ann. § 4117.09(C) (“Any public employee organization representing public employees pursuant to this chapter shall prescribe an internal procedure to determine a rebate, if any, for nonmembers . . . .”). These internal, union-operated systems made some sense when disputes over fees involved unions and the workers they represented, but they are less sensible when the dispute is between a union and employer. So states could reasonably institute a dispute resolution mechanism that requires public employers to first challenge a reimbursement request before an ALJ and then appeal to the PERB itself.

Discussion of the pre-PERB review process for challenges to union reimbursement points up the need to confront the back-end right to judicial review as well. Again, the guiding principle should be consistency; states should provide the same right to aggrieved parties to pursue judicial review of PERB decisions that existed under the fair-share regime. 314 See, e.g., 5 Ill. Comp. Stat. Ann. 315/16 (“After the exhaustion of any arbitration mandated by this Act . . . suits for violation of agreements . . . between a public employer and a labor organization representing public employees may be brought by the parties to such agreement in the circuit court in the county . . . .”); N.Y. Civ. Serv. Law § 213 (establishing judicial review procedures for final orders of the state PERB). There is one wrinkle, though, which may justify limiting a public employer’s right to judicial review. Under the fair-share-fee regime, the objecting worker’s claim was constitutional in nature, whereas under the reimbursement regime the objecting employer merely brings a statutory challenge. A state could plausibly conclude that public employers have a less-weighty interest in additional levels of review beyond ALJ and PERB determinations.

Second, some may worry that the PERB-based model suffers from an intrinsic flaw in that PERB members are themselves political appointees. Several union-side officials whom the author interviewed regarding the government-reimbursement alternative shared this concern, pointing out that some state PERBs have been less friendly to labor than others over certain periods.

One response, although likely not a comforting one for union proponents, is that this responsiveness to the changing political preferences of the electorate is a feature, not a bug. If state voters believe that PERB members should take more of a pro-public-employer stance, then that may itself be a legitimate outcome. Moreover, the political contingency of PERB-member identity does not distinguish this approach to union reimbursement from the state of affairs that existed in the fair-share fee regime, in which a more pro-employer PERB could in theory also narrow the scope of chargeability in challenges brought by objecting workers. In other words, union proponents might prefer a model in which a perpetually pro-labor entity decides challenges to reimbursement requests, but that is a call to increase union influence relative to the pre-Janus baseline. The aim here is merely to offer designs that can minimize Janus’s disruptive effect while remaining neutral as to union strength.

In any case, the PERB politicization critique may end up proving little. For one thing, PERBs are generally conceived of as independent agencies whose members are protected by for-cause-removal restrictions. 315 See, e.g., Cal. Gov’t Code § 3541(a) (“[T]he Public Employment Relations Board . . . shall be independent of any state agency and . . . [a] member of the board may be removed by the Governor upon notice and hearing for neglect of duty or malfeasance in office, but for no other cause.”); 5 Ill. Comp. Stat. Ann. 315/5(d) (providing that members of the Illinois Labor Relations Board “may be removed from office . . . for inefficiency, neglect of duty, misconduct or malfeasance in office, and for no other cause, and only upon notice and hearing”). As the Supreme Court has made clear, such restrictions on an executive’s power to remove can go a long way to insulate an agency’s independence from political pressures. 316 Humphrey’s Ex’r v. United States, 295 U.S. 602, 629 (1935) (“The authority of Congress . . . to require [agencies] to act in discharge of their duties independently of executive control cannot well be doubted; and that authority includes, as an appropriate incident, power to . . . forbid their removal except for cause . . . .”). For another, a state that is moved by the critique can simply draft around it when setting its reimbursement procedure. Rather than leaving final decisionmaking authority in the PERB, such a state could create a separate body with the stated purpose of remaining neutral as to the labor–employer divide. Indeed, New Jersey did exactly that with respect to its fair-share fee challenge procedure, creating a special appeal board consisting of three members—one “representative of public employers,” one “representative of public employee organizations,” and one “who shall represent the interest of the public as a strictly impartial member”—who are ultimately responsible for resolving chargeability disputes. 317 N.J. Stat. Ann. § 34:13A-5.6. The “impartial member” requirement under New Jersey’s law is quite rigorous; the member must not have had “more than a casual association or relationship with any public employers, public employer organizations or public employee organizations in the 10 years prior to appointment.” Id. A similar board could be constituted for resolving union reimbursement requests.

Third, it is worth discussing the source of the funds that are used to reimburse the union. In the model provision above, each individual public employer is ultimately responsible for providing funds to reimburse qualifying union expenditures. This mirrors the fair-share approach, in which individual employers are likewise the final source of union funding (albeit channeled through worker paychecks). This symmetry is again intentional, as it ensures that the reimbursement approach continues the preexisting system of labor relations as closely as possible. That said, one might imagine a system in which the state itself administers a central fund to reimburse unions for eligible expenses, financed by revenues collected on a statewide basis. In fact, a bill was introduced to establish just such a statewide fund for union expenses in Hawaii, although it has remained in committee since January 2017. 318 See H.R. 923, 29th Leg., Reg. Sess. (Haw. 2017). The Oregon legislature recently introduced a similar bill. See H.B. 2643, 80th Leg. Assemb., Reg. Sess. (Or. 2019). Still, such a mechanism would lead to additional switching costs (insofar as no state currently uses a statewide funding source for all unions 319 See supra notes 60–63. ) and could pose a greater threat to union independence (insofar as unions might cease collecting dues and thus become vulnerable to changing political winds 320 See infra notes 369–370 and accompanying text. ), so it is not the default approach suggested here.

Fourth, one could consider a much more involved role for the PERB in setting union reimbursement levels. Consistent with the PERB’s role under the fair-share-fee regime, the approach suggested here treats the PERB as a final adjudicator that issues binding opinions on whether specific reimbursement requests are permissible under a state’s statutes and case law. But one could also envision a more proactive regulatory role in which PERBs would act as ratemakers tasked with fixing appropriate reimbursement rates for different common union tasks. Just as the United Kingdom’s indigent defense agency sets hourly reimbursement rates for different client-counseling and investigatory activities, for instance, a state PERB could announce a given reimbursement rate for each hour spent on contract negotiations, contract administration, grievance processing, and so forth.

Given this Article’s stated preference to prioritize continuity and thereby minimize disruptions to the existing system of public-sector relations, the model provision above implements the preexisting PERB-as-adjudicator approach. That is not to deny the potential virtues of a PERB ratemaking role, and a state could quite reasonably consider that route. In doing so, however, a state should take care in setting reim­bursement levels in an evenhanded manner consistent with stakeholder input so as to avoid the unintended consequence of incentivizing unions to expend more resources on activities that yield higher reimbursement rates, even if those activities are not in workers’ best interests. For current purposes, this Article simply notes that the PERB-as-adjudicator model avoids those risks while drawing on decades’ worth of case law concerning the chargeability of various union expenses. 321 See supra note 273 and accompanying text.

b. The Membership Model. — For states that wish to avoid a PERB-based model for settling union reimbursement amounts, there is a reasonable alternative: placing final responsibility for reimbursement levels in the hands of union members. Under this approach, union officials would still prepare an accounting of their reimbursable bargaining-related expenses that would be sent to the public employer. But rather than resolving challenges to expenditures using PERB-adjudication on the back end, state law could authorize unions and employers to agree to a membership-centered enforcement mechanism on the front end. More specifically, state law could authorize collective bargaining agreements that stipulate that public employers will reimburse unions for statutorily authorized expenses after they have been approved by the union’s members. Here’s how such a law might look:

(a) When a collective bargaining agreement includes a provision requiring a public employer to reimburse an employee organization for expenditures permitted under [the state’s reimbursement exception], 322 See supra section IV.A.1. the employee organization shall certify to its membership the amount of permissible expenditures it has actually incurred. A public employer shall promptly reimburse the employee organization for expenditures that the organization’s members approve and confirm as consistent with [the state’s reimbursement exception].

(b) The membership approval and confirmation process in section (a) may occur via direct membership referendum or via representative vote as authorized by the employee organization’s bylaws.

The  basic  approach  here  is  to  vest  final  accountability  over  union  reimbursement  requests  in  the  members  themselves  by  requiring  them  to  confront  the  reality  that  the  skin  in  the  union  reimbursement  game  is  their  own,  no  less  than  in  the  fair-share  regime.  In fact, there is a reasonable analog to this membership-vote approach under the private-sector fair-share model, under which members have the right to vote on dues increases under the Labor Management Reporting and Disclosure Act of 1959. 323 See 29 U.S.C. § 411(a)(3) (2012). A similar dynamic exists here, in which every dollar that a public employer sends to reimburse a union is a dollar that could have gone to the workers’ paychecks. So union members have a natural incentive to police the union to ensure that its expenditures are both reasonable on an absolute basis and in conformance with the statutorily permissible categories of reimbursement. Or to put it another way, a public employer that continually receives large reimbursement requests to pay for outlandish union-staff salaries or overt political contributions will naturally be reluctant to grant significant future wage increases on top of the large union reimbursement tab. And public workers should have little trouble recognizing this dynamic. 324 Indeed, a public employer would likely be allowed to inform its workers of this reason for its bargaining position, further increasing the workers’ interest in monitoring excessive-reimbursement requests. See Americare Pine Lodge Nursing & Rehab. Ctr. v. NLRB, 164 F.3d 867, 875 (4th Cir. 1999) (explaining that “[a]n employer may speak freely to its employees about . . . the status of negotiations, outstanding offers, its position, [and] the reasons for its position . . . as long as those communications do not contain some sort of express or implied quid pro quo offer that is not before the union”).

Three follow-up questions come to mind. First, as subsection (b) expresses, the process by which union members formally approve the union’s requested reimbursement amount and confirm its consistency with the statutorily permitted categories is reasonably left up to the union itself. Some (often smaller) unions already require full-membership approval of budgets via a referendum process, 325 See AFSCME Local 688 Bylaws, AFSCME, [] (last visited Oct. 17, 2018) (“The Executive Board shall review and approve or amend the budget before it is submitted to the Membership for adoption.”). and it is clear how this kind of an approach would help align union leaders’ budget-setting incentives with the best interests of members. Unions with larger membership bases, however, may find it difficult to administer a full-body vote process and so may prefer to empower members to elect representatives who in turn scrutinize and vote on budget requests. 326 See Chi. Teachers Union, Constitution and By-Laws 16 (2009),
grievances/CONSTITUTION-and-BYLAWS-11-12-10.pdf?1294855346 [] (requiring the House of Delegates—a group of union representatives elected by the union membership—to approve annual budgets).
To impose a one-size-fits-all solution would disrupt the settled practice in many unions and produce definite switching costs that could well exceed the theoretical benefits of a universal all-member (or elected-representative) approval process.

Second, one common reaction to the membership-based approval process is to wonder whether members will actually serve the kind of policing function this Article has hypothesized. If one’s intuition is “no,” then the related question arises why any public employer would ever agree to such a reimbursement regime. To the skeptic, the membership-based model for determining reimbursement levels may seem like a blank check—something no employer in its right mind would want to sign over.

The truth is almost certainly that local units will display a hetero­geneous level of interest in checking union expenditures. Locals have different relationships with their members; some have earned greater deference over a long period of representation but others less so. In the end, there is reason to think that workers are generally not so passive when it comes to monitoring their unions’ behavior. One piece of circumstantial evidence is that the NLRB continues to see several hundred decertification petitions filed against private-sector unions each year. 327 See Decertification Petitions – RD, NLRB, [] (last visited Oct. 16, 2018) (identifying between 300 and 600 decertification petitions filed each year from 2008 to 2017). Although it is hard to generalize to the public sector, the sense is that if workers care enough about union efficiency to throw out their representatives that often, then it may not be naïve to think they will care enough to monitor against indulgent or unnecessary union expenditures.

Moreover, public employers hold the trump card at the end of the day because they can offset high reimbursement requests relatively cheaply during future contract negotiations. More specifically, unions generally spend around 2% of employee wages on bargaining-related activities, 328 See Sachs, supra note 18, at 1048. so even a union that overstates its reimbursement request by 10% will cost the employer a fraction of a percent of its total labor costs. Because the employer can just reduce future wage increases for workers represented by such a union by the same offsetting fraction of a percent (that is, 0.2%) during subsequent contract negotiations, the employer retains full control over its financial health. 329 There is another answer to the concern that some public employers (especially in more politically conservative localities) will prefer not to agree to a membership-based reimbursement regime: Many public employers in those same localities harbored the same preference against, and chose not to enter into, fair-share-fee clauses. Public-employer antipathy for unions, in other words, does not distinguish the reimbursement approach from the regime that preceded it.

Third and finally, it is fair to wonder whether public employers will raise a different kind of objection to member-determined levels of reimbursement. What if members approve expenditures that violate the terms of the state’s reimbursement exception, for example, by granting reimbursement for contributions to political campaigns? It is possible some public employers may not care, since they can just take those improvident expenditures out of future wage raises. But as a political matter, others may be less sanguine. Suppose, for instance, that a given public employer comprises elected officials (for example, a local school board) and that union reimbursement payments are used as political contributions to those officials’ opponents (for example, challenger school board candidates). As a statutory matter, such a reimbursement would fall into an odd gray area, as it would constitute (a) an unfair labor practice as employer financial support to a union that falls outside the scope of the reimbursement exception, yet also (b) one that the union’s own members authorized. Practically, then, these political and statutory conundrums may require adding a back-end adjudicatory process to ferret out impermissible union expenditures in the membership-based model too. This approach is considered next.

c. The PERB–Membership Hybrid Model. — It may be possible to capture the benefits of both PERB and membership-based monitoring of union reimbursement payments. Requiring both independent agency and member-based review of reimbursement requests could increase union compliance with the statutorily permitted categories of reimbursement while further ensuring that the governmental source of funds poses no risk to union independence. Here is a model provision:

(a) When a collective bargaining agreement includes a provision requiring a public employer to reimburse an employee organization for expenditures permitted under [the state’s reimbursement exception], the employee organization shall annually certify to the employer the amount of permissible expenditures it has actually incurred.

(1) If the employer agrees that the identified expenditures are reimbursable under [the state’s reimbursement exception], it shall promptly reimburse the employee organization for the certified amount.

(2) If the employer disagrees with any of the identified expenditures, the employer must file within thirty days an objection with the [state PERB] identifying the specific expenditures to be challenged. The employer shall set aside the challenged expenditures in an escrow account maintained by the employer, employee organization, or [state PERB], and promptly reimburse the employee organization for all other expenditures.

(b) [Placeholder for incorporating a state’s administrative, alter­native dispute, and/or judicial resolution mechanisms for resolving fair-share fee challenges].

(c) If the expenditures submitted by an employee organization to a public employer for reimbursement pursuant to section (a) have been approved by the organization’s members and determined by them to be consistent with [the state’s reimbursement exception], such determination shall be conclusive on the parties absent a showing to the [state PERB] that the determination was arbitrary or capricious.

(1) The membership approval and determination process in section (c) may occur via direct membership referendum or via representative vote as authorized by the employee organization’s bylaws.

Here is how this combined approach would work: The union would begin the reimbursement process by certifying the amount of expenditures it believes to be consistent with the state’s reimbursement exception. The employer would then pay the portion it agrees with and challenge any it does not before the PERB, following the state’s existing procedural mechanisms (including, for example, initial decisionmaking by an ALJ). 330 In a state where the PERB did not finally resolve fair-share-fee challenges prior to Janus, the state could instead continue with its existing private-dispute mechanism. See, e.g., 43 Pa. Cons. Stat. § 1102.5(c) (2018).

But there is a wrinkle. If the union has presented the reimbursement request to its members and obtained both their approval of the absolute amount requested and their confirmation that the requested amounts fit within the statutorily permitted categories of reimbursement, that membership finding creates a presumption before the PERB that the requested expenditures are in fact permissible. This presumption has the salutary goal of creating an incentive for unions to obtain approval for their reimbursement requests from their members. But the public employer would not be powerless; the presumption of permissibility can be rebutted upon a showing that the member-determination process was arbitrary or capricious. 331 Review of a union-funding determination is not foreign to this area of law: Ohio’s fair-share-fee regime, for example, explicitly requires its PERB to review union fee calculations under this standard of review. See Ohio Rev. Code Ann. § 4117.09(C) (2018). For example, if the union withheld, or the members were aware of yet ignored, clear evidence that the union’s requested reimbursement included money contributed to a political candidate’s campaign for office, that would constitute arbitrary or capricious action that the PERB would overturn. Public employers would therefore have a safety valve to ensure the propriety of union expenses, unions would have an incentive to obtain member approval of the same, and the arbitrary or capricious standard of review triggered by such member approval could reduce litigation of frivolous or marginal cases by public employers. 332 Note that a state could also delegate the choice among the PERB, member, and hybrid models for deciding reimbursement amounts to the individual employer and union. Such an option would enable unions and public employers to experiment to see which approach to reimbursement is most efficacious.

B. Open Questions

There is a common theme in the answers to each of the specific design questions considered so far: State lawmakers should adopt, to the greatest extent possible, the same substantive and procedural rules that were used for decades to administer the fair-share-fee model for union financing. The end goal is stability and continuity. Little should change under the new reimbursement model except that public-sector unions would receive the funding needed to represent workers through direct payments from the public employer. Several more general questions arise, however, now that the nuts and bolts of the reimbursement model have been specified.

1. When Should the Government Make Payments? — Under the fair-share system, unions projected their annual budgets in advance, calculating the pro rata portion of that budget to be paid by objecting workers based on the proportion of the prior year’s budget that was spent on chargeable activities. 333 See, e.g., Knox v. Serv. Emps. Int’l Union, Local 1000, 567 U.S. 298, 303 (2012) (describing the budgeting process used by the SEIU, Local 1000). Objecting workers then paid that chargeable amount in a monthly fair-share fee, while union members paid a larger amount in full member dues (comprising the pro rata portion of both chargeable and nonchargeable amounts). 334 See id. From the union’s perspective, then, each month saw a reliable ex ante stream of income that would be used to pay for the union’s ongoing activities.

In theory, a government payment system could proceed in the same way. Locals could project their annual budgets in advance, identify the anticipated costs that are germane to bargaining activities and thus properly reimbursable by the government, and divide by twelve to reach a monthly amount owed by the public employer. In keeping with the argument so far, this approach would have the virtue of familiarity, as it would most closely resemble the settled practice (at least from the union’s viewpoint) before Janus. So a state would have good reason to codify a law authorizing similar ex ante payments from public employers before the union actually incurs the relevant bargaining-related costs. 335 The model provisions proposed above would need to be revised to accomplish this end. For example, the term “reimburse” could be replaced with “pay,” and references to permissible expenditures “actually incurred” could be replaced by expenditures that are “projected.” See supra section IV.A.

But there are three substantial reasons to prefer an ex post system of payment in which public employers reimburse unions for permissible expenditures after they are incurred. One reason is a response to the union-independence objection. A union member might be less worried about their representative’s loyalty when the public employer merely reimburses the union for a set of concrete, statutorily defined, and already-incurred bargaining costs than when the employer pays the union for some collection of roughly anticipated, future activities. Whereas the former seems closer to the United Kingdom’s ratemaking system for reimbursing indigent defense attorneys, which does not trigger conflict of interest concerns, 336 See supra section III.D.4.a. the latter may have the appearance of an open-ended inducement for the union to do the employer’s bidding. 337 To be clear, even an ex ante system of government payment would be free of union-independence concerns, or at least as free as the fair-share system it would replace. That is because the union would not be reliant on the public employer’s graces for continued funding. Such funding would be required under state law (at least when a bargaining agreement includes the relevant provision, which is also true of fair-share clauses), and disputes over funding levels would be resolved by a PERB or union members, not the employer.

A second reason to prefer a reimbursement approach comes from the employer’s perspective. Put simply, public employers may find reimbursement more palatable financially and politically than payment up front. Financially, employers may worry that a union’s projections as to its expected bargaining-related costs will overstate the expenditures it actually incurs, to the detriment of the employer’s own budget. Politically, agreeing to reimburse the union for bargaining-related costs after they have been incurred converts the government payment model of union financing into a kind of actuarial exercise, limiting concerns that government payments are actually handouts to unions for use on contentious political and ideological activities. Indeed, at least one state already recognized in the fair-share context, even if only implicitly, that describing union financing as a matter of “reimbursement” is the more politically prudent approach. 338 Alaska Stat. § 23.40.110(b)(2) (2018) (permitting public employers to agree to provisions requiring “payment . . . of a service fee to reimburse the exclusive bargaining agent for the expense of representing the members of the bargaining unit” (emphasis added)).

Finally, a reimbursement approach may serve benefits for unions too. This is a counterintuitive suggestion in one sense, since the obvious answer to the question whether an entity would prefer to have some money sooner or later is “sooner.” But as discussed at greater length below, a system in which the public employer pays the union in advance for its expected bargaining costs would introduce a complication regarding how a union is to identify its members. 339 See infra section IV.C. Under the fair-share system, after all, a worker who wishes to become a union member must simply follow the local’s process for joining and pay full member dues. Requiring payment of dues is important in part because it delimits who is not a member and therefore not eligible to vote on union affairs. 340 See NLRB v. Gen. Motors Corp., 373 U.S. 734, 737 (1963) (“Employees choosing not to join [the union] . . . would not be entitled to attend union meetings, vote upon ratification of agreements negotiated by the union, or have a voice in the internal affairs of the union.”). Yet that leads to some uncertainty under the reimbursement approach: If a union’s budget (or at least a great bulk of it) comes not from voluntary payments by members but rather from the government, 341 See Sachs, supra note 18, at 1048 n.12 (collecting sources indicating that fair-share fees amounted to roughly two-thirds of full member dues). how is the union to distinguish between members and nonmembers when it comes time to decide who participates in internal union governance?

When public employers reimburse union expenses (rather than paying them ex ante), a convenient solution to the membership problem surfaces: Unions should keep collecting full dues from members just as they have always done. Then, when the public employer eventually reimburses the union for its statutorily permitted bargaining expenses, the union can rebate the pro rata amount of that reimbursement payment back to its individual members. Unions would thus be able to organize and identify their members in the same way they have done for decades, they would continue to have access to resources pending receipt of reimbursement from the employer, and workers would be no worse for wear after receiving a rebate for the chargeable portion of their dues each year. 342 Note that states may also wish to permit unions and public employers to agree to more frequent reimbursements as a way of ensuring union liquidity. See infra notes 367–368 and accompanying text.

A reimbursement approach does raise one potential complication for unions that organize on a statewide or council basis and that employ state- or nationwide staff who service local bargaining units on an as-needed basis. 343 See, e.g., Lehnert v. Ferris Faculty Ass’n, 500 U.S. 507, 522–23 (1991) (describing this affiliated structure as one “under which many unions . . . operate”). In Lehnert, the Supreme Court held that a local union could charge all workers “for their pro rata share of the costs associated with otherwise chargeable activities of its state and national affiliates, even if those activities were not performed for the direct benefit of the objecting employees’ bargaining unit.” 344 Id. at 524. That approach effectively authorized unions to operate a kind of insurance risk pool, in which all locals would share in the aggregate cost of state and national staff who may be called upon for assistance with particular kinds of grievances or disputes.

There is no apparent reason why that approach should be forbidden. States should be free to clarify that their reimbursement exceptions permit payment for the kinds of statewide or national services approved in Lehnert. 345 To this end, a state could add a clarification to its reimbursement exception that would permit reimbursement to the employee organization for costs germane to the collective bargaining process, contract administration, and matters affecting wages, hours, and other conditions of employment, “including costs associated with otherwise chargeable activities of its state and national affiliates, even if those activities were not performed for the direct benefit of the [relevant] bargaining unit.” Id. On the other hand, if a state chose not to authorize the Lehnert cost-sharing approach, it is possible the effect would actually be salutary from the perspective of some unions. For if a public employer were required to pay all of the representational costs incurred by the union in grieving a certain employer action (rather than just a pro rata share of the statewide affiliate’s total costs), the employer might face stronger pressure to abide by the terms of the contract. That is because aggressive employer actions in tension with the contract would lead directly to a greater reimbursement obligation when the union files a challenge. In this sense, the reimbursement approach may actually create incentives for public employers to be more respectful of worker rights rather than less.

2. Should Reimbursement Be Mandatory or Permissive? — Another question for state lawmakers to consider is whether reimbursement for union bargaining expenses should be a mandatory duty placed on all public employers or instead a commitment that employers are permitted (but not required) to undertake. The decision is of obvious significance insofar as a permissive approach would allow public employers in less labor-friendly localities to opt out of the regime altogether.

Given these stakes, it is clear that the question cannot be answered in the abstract without first adopting a normative view on the desirability of public-sector unionization. But this is not an abstract exercise: States have already decided in the fair-share fee context whether public-sector union financial security should be protected as a matter of obligatory state policy or if that choice should instead be delegated to individual public employers. As noted above, the vast majority of states adopted permissive fair-share-fee laws, 346 See supra note 61 and accompanying text. a handful adopted mandatory laws, 347 See supra note 60 and accompanying text. and a still smaller number took different approaches for different sectors of public workers. 348 See supra notes 62–63 and accompanying text. So the most neutral approach would be for the states to enact government reimbursement laws that mirror their pre-Janus fair-share counterparts—that is, mandating public employer reimbursement when state fair-share laws were also mandatory and merely permitting it when state laws did the same. 349 One noteworthy exception that may apply to this general presumption is states that enacted mandatory fair-share laws but that also have state-law prohibitions against unfunded mandates. For instance, California’s labor law for K–12 public-education employees requires school districts to agree to fair-share-fee clauses. See Cal. Gov’t Code § 3546(a) (2019). Meanwhile, however, article XIII B, section 6(a) of the California Constitution provides that “[w]henever the Legislature . . . mandates a new program or higher level of service on any local government, the State shall provide a subvention of funds to reimburse that local government for the costs of the program or increased level of service.” Cal. Const. art. XIII B, § 6(a). Thus, mandating California school districts reimburse public-sector unions for permissible bargaining expenses would arguably trigger the state’s duty to reimburse the local government in turn. So the state could plausibly choose to make the reimbursement approach permissive, or contingent on public workers opting in via member vote instead. See infra notes 350, 359–361 and accompanying text. That said, the application of the unfunded-mandates provision to a reimbursement statute is not a forgone conclusion; there is a decent argument that requiring public employers to reimburse unions directly in lieu of enforcing fair-share-fee clauses is not a “new program or higher level of service” but instead the continuation of the preexisting fair-share approach to ensuring union financial security. See Tang, Public Sector Unions, supra note 18, at 175–76 (describing the revenue neutrality of the direct-reimbursement model from the public employer’s perspective). That would be especially true if the reimbursement law explicitly authorized public employers to reduce future wages or benefits to offset the new cost of the reimbursement requirement.

Alternatively, even a state that had a mandatory fair-share requirement might prefer a law that mandates bargaining over reimbursement but does not compel every union to accept it. Doing so would give unions a choice whether to negotiate a reimbursement clause or instead trade away that valuable provision as leverage for some other benefit for workers. Lawmakers in Hawaii and New York appear to have chosen precisely this approach. 350 See S.B. 487, 13th Leg., Reg. Sess. (Haw. 2019) (“An employee organization may choose to bargain for other concessions regarding wages, hours, benefits, or other terms or conditions of employment in lieu of the reimbursement the employee organization is entitled to . . . .”); Parrott, Gottfried’s Janus Workaround, supra note 42 (noting that New York Assembly Member Gottfried proposed a bill that would “leave[ ] it up to the union leadership to decide whether to institute the new policy, so unions are not forced to adopt it if they believe it would make them look compromised in the eyes of their membership”).

3. In Permissive States, What Rights Should Workers Have if Their Individual Public Employers Oppose Reimbursement? — For states that choose merely to permit public employers to reimburse unions for bargaining-related expenses, the natural follow-up question is how exactly the public employer may be allowed to decide on the matter. Must the employer at least negotiate with the union over a reimbursement provision at the bargaining table? There is a vast difference between treating reimbursement provisions as a mandatory subject for bargaining, such that all employers must negotiate the issue in good faith, and allowing employers to decide unilaterally against reimbursement provisions with no input from a union at all. It should go without saying that unions will prefer the former while employers will prefer the latter. 351 See Kearney & Mareschal, supra note 2, at 77 (“As a general rule, management prefers a narrow scope of bargaining, whereas unions seek a broad scope.”).

Again, the best response is to resolve this issue just as it was resolved under the pre-Janus fair-share model. Many states, in keeping with the NLRA’s approach of making fair-share fee provisions a mandatory subject of collective bargaining in the private sector, 352 See NLRB v. Gen. Motors Corp., 373 U.S. 734, 737–38, 744– 45 (1963) (“We hold that the employer was not excused from his duty to bargain over the [agency-shop agreement] on the theory that his acceding to it would necessarily involve him in an unfair labor practice.”). imposed the same mandatory bargaining obligation on public-sector employers. 353 See, e.g., Md. Code Ann., Educ. § 6-407(c)(1) (West 2018) (“[T]he public school employer shall negotiate [a fair-share-fee provision] with the employee organization designated as the exclusive representative for the public school employees in a unit . . . .”); N.J. Stat. Ann. § 34:13A-5.5(a) (West 2018) (requiring a union and public employer, “where requested by the majority representative,” to “negotiate concerning the subject of requiring the payment [of a fair-share fee] by all nonmember employees in the unit to the [union]”). Other states, however, took a different approach, declaring fair-share fees to be only a permissive matter for collective bargaining. 354 See, e.g., Md. Code Ann., State Pers. & Pens. § 3-502(b)(1) (West 2018) (“Collective bargaining may include negotiations relating to the right of an employee organization to receive service fees from nonmembers.”); N.M. Stat. Ann. § 10-7E-9(G) (West 2018) (“The issue of fair share shall be left a permissive subject of bargaining by the public employer and the exclusive representative of each bargaining unit.”). The distinction matters because unions have far greater influence with respect to mandatory subjects, for which employers cannot unilaterally set their preferred policies and must instead resolve disagreements using the impasse procedures prescribed by state law. 355 Compare, Cal. Gov’t Code §§ 3548–3548.8 (2019) (setting forth procedures for resolving an impasse over mandatory subjects of bargaining, as defined in Cal. Gov’t Code § 3543.2(a)(1)), with id. § 3543.2(a)(3) (providing unions a more limited right to “consult” over nonmandatory subjects of bargaining). See generally Kearney & Mareschal, supra note 2, at 77 (“Collective bargaining items may be mandatory, permissive, or prohibited . . . . [M]anagement [typically] prefers a narrow scope of bargaining, whereas unions seek a broad scope . . . . In public employment, . . . strikes are . . . prohibited or subject to restrictive conditions. Thus, [there is] the need for . . . procedures to resolve impasses.”). The key, though, is that one need not make a fresh value judgment as to whether mandatory or permissive bargaining of reimbursement provisions would be “better.” There is already a settled answer to that question under each former fair-share jurisdiction’s fair-share law—and that answer can be directly imported into the proposed reimbursement approach.

To union proponents, however, even declaring reimbursement a mandatory subject of collective bargaining may not be enough. One may worry that anti-union public employers will use the negotiable nature of reimbursement clauses as a wedge in future negotiations. Such an employer might, for example, threaten to eliminate a reimbursement clause from a future bargaining agreement in exchange for concessions on wages, benefits, or other terms or conditions of employment. 356 Such a choice by the employer would not necessarily constitute an unfair labor practice, so long as the employer negotiates in good faith. See, e.g., Int’l Bhd. of Teamsters Local 700, 29 PERI ¶ 179 (Ill. Labor Relations Bd. May 13, 2013) (“We are unable to find any precedent for the proposition that an employer’s refusal to make a concession on a fair share or similar proposal is a sufficient basis, without more, for a finding that the employer has engaged in illegal surface bargaining.”). Union negotiators would then face a difficult choice between prioritizing their own financial security or the interests of public workers.

This is certainly a legitimate concern for those worried about union independence. But it is a concern that afflicted the fair-share-fee system as well, in which employers in permissive regimes were also perfectly free to condition their agreement to a fair-share provision on other concessions by the union. 357 See Mt. Vernon Sch. Dist. No. 80, 11 PERI ¶ 1013 (Ill. Educ. Labor Relations Bd. Jan. 6, 1995) (upholding a contract in which the school district agreed to a fair-share clause on the condition that the union waive its right to negotiate over any matter during the life of the parties’ contract); see also, e.g., Ill.-Am. Water Co., 301 N.L.R.B. 218, 222 (1991) (holding that a private employer’s threat to eliminate a union-security clause from a previous contract was not an unfair labor practice); Challenge-Cook Bros., 288 N.L.R.B. 387, 389 (1988) (same). Union proponents may therefore wish for more favorable treatment of reimbursement clauses, but those are arguments for increasing union strength relative to the pre-Janus baseline, not for preserving the system of public-sector labor relations that pro-labor states argue operated so successfully before Janus. 358 See supra note 270 and accompanying text. For present purposes, it suffices to say that reimbursement clauses are no more susceptible to public employer threats than fair-share clauses.

In any case, states are free to adopt other solutions to any perceived fear of public employers refusing to negotiate fairly over reimbursement provisions. One particularly low-cost remedy is a direct-employee-referendum provision, which might be borrowed from California’s Meyers–Milias–Brown Act (MMBA) governing local public employers. Under that provision, 359 Cal. Gov’t Code § 3502.5(b). public workers were authorized to directly implement a fair-share-fee arrangement (notwithstanding the refusal of their employer to agree to such a clause during contract negotiations) if a majority of the unit’s employees voted in favor of it. 360 See id. The MMBA also required an initial petition calling for a vote on the fair-share issue signed by at least 30% of the employees in the bargaining unit. See id. A state interested in giving workers a direct voice on the reimbursement issue could implement a similar provision in its new reimbursement law. 361 A model provision to this effect is included as an option in the Appendix, infra.

c. Implementing the Reimbursement Approach

Settling on a statutory model for government reimbursement is a major piece of the puzzle. Yet statutory reimbursement provisions are not self-executing; 362 This is especially true of permissive regimes in which employers and unions must negotiate whether to include a reimbursement provision at all. But even in mandatory regimes, there are questions regarding how unions will determine membership and when the employer and union will enter into the new arrangement. they require work by unions and public employers to carry them into effect. Two major implementation questions are addressed here, although there are assuredly others that this Article will not have the space to discuss. 363 For example, one important implementation issue the Article does not discuss here is the need for locals to develop and maintain more fine-tuned accounting systems to keep track of staff time and other expenses incurred on reimbursable activities for the purpose of certifying an annual reimbursement amount to the public employer.

First, how should unions revise their membership requirements in light of government reimbursement? Direct reimbursement complicates existing union-membership requirements, which turn on payment of union dues. 364 See supra notes 339–340 and accompanying text. Those dues represent both chargeable expenses and nonchargeable expenses. Yet if the government covers the former category of costs (roughly two-thirds of total union budgets 365 See Sachs, supra note 18, at 1048 n.12 (estimating agency fees at two-thirds of total union dues). ), what should unions ask their members to pay moving forward?

One answer would be for unions just to collect the nonchargeable amount. Under that approach, public-sector employees could become union members simply by following their local’s stated procedure for joining and paying the smaller portion of union dues associated with nonchargeable political and ideological expenditures. That approach produces one possible benefit for unions, in that the “sticker price” for workers deciding whether to join the union (roughly one-third of the former full dues amount) would appear lower than it might have seemed under the fair-share approach (the full dues amount). Of course this appearance is deceiving, since in the fair-share regime nonmembers were required to pay the chargeable portion anyhow. But people respond powerfully to differently framed choices with identical economic effects. So it is plausible to think more workers would agree to join a union at the price of ⅓X (say, $500 per year) than would agree to join at the price of X ($1,500), when the implied cost of not joining is ⅔X ($1,000). 366 See generally Amos Tversky & Daniel Kahneman, The Framing of Decisions and the Psychology of Choice, 211 Science 453, 453 (1981) (“Because of imperfections of human perception and decision . . . changes of perspective often reverse the relative apparent size of objects and the relative desirability of options.”).

There are at least two downsides to this approach, however. One is a matter of liquidity. Collecting just the chargeable portion of member dues each month is not enough to finance ongoing union activities, including the staff time and associated costs of contract administration, grievance processing, and bargaining. So locals would have to look elsewhere for funding to cover those costs as they await a reimbursement check from the employer at the end of the relevant fiscal period. The model legislation proposed above envisions an annual reimbursement check, which aligns with the annual budgeting and accounting process used in the fair-share model. 367 See supra text following note 306. A state could choose to implement a shorter reimbursement period (or leave employers and unions free to negotiate such a term), but the tradeoff is three-fold. More frequent reimbursement requests will require more union staff time spent on accounting and the actual process of seeking payment—costs that will increase union expenditures in their own right. Relatedly, public employers will also have to devote more time scrutinizing union expenditures and deciding whether to pay them. Finally, more frequent requests will increase the opportunities for litigation and thus may result in more burdensome proceedings before a state’s PERB. 368 In theory, states could minimize litigation frequency by providing for an annual roll-up period in which all disputes over union reimbursement requests from the preceding year can be challenged within a certain time period. But that would not reduce the burden on unions and public employers to perform more frequent audits and calculations regarding reimbursable expenses.

A second, more significant problem with asking members to pay only the nonchargeable portion as member dues is that it leaves unions more exposed if public employers refuse to agree to a reimbursement clause in future contracts. Union officials the author spoke with about the reimbursement approach commonly cited this concern. For once the union stops collecting full member dues—and workers develop an expectation of paying just a fraction of their previous dues amounts—a choice by an employer to eliminate a reimbursement provision from a future contract may be financially ruinous. In such a case, the union would have to go back to its members and ask them to dramatically increase their monthly deductions (by a factor of roughly three times 369 See Sachs, supra note 18, at 1048 n.12 (estimating that nonchargeable union expenditures comprise roughly one-third of total union dues). ) in order to maintain similar levels of funding. Needless to say, workers might not be likely to agree to that request, and some might leave the union altogether. 370 There is a separate yet equally troubling worry if the public employer simply reneges on payment. That would constitute an unfair labor practice, at least under the model language proposed here, which requires prompt employer payment of all properly assessed expenses. But a public employer may fail to pay nonetheless (perhaps due to budget shortfalls), leading to protracted litigation before the union actually receives its reimbursement check. In that circumstance, the union would be without its primary source of funding for some indefinite period of time, and the alternative—asking members to voluntarily contribute a substantial bridge payment with the hope of some future refund—is not especially appealing.

The alternative, then, is for unions to continue collecting the full dues amount up front. That would serve as a backstop against a public employer’s threat to eliminate reimbursement, and it would resolve union liquidity concerns. Collecting full dues up front would also allow unions to continue their existing approach to membership in which workers must pay the full dues amount as a condition of joining. Then, once the public employer makes its reimbursement payment to the union, the union would rebate a pro rata portion of the payment to each individual worker, reimbursing them for their initial contribution to cover the union’s chargeable bargaining activities.

There is a downside to this approach too, however. To the fastidious public-sector worker, the combined transition to the reimbursement model plus the collection of a full dues amount will feel like a net financial loss, if only temporarily. That is because public employers interested in implementing the reimbursement model in a fiscally neutral manner will reduce future wage or benefit increases to offset the new obligation to reimburse the union. 371 See Tang, Public Sector Unions, supra note 18, at 175–76. So wages or benefits will go down (in comparison to a counterfactual scenario in which there is no reimbursement provision), while union dues amounts will remain the same. Thus, until the government actually reimburses the union and the union rebates a pro rata portion to its members (at which point all workers will be made whole), individual workers will be relatively worse off. 372 However, they will not be worse off by a lot. If one estimates that unions charge roughly 3% in dues and the chargeable portion is two-thirds of that amount, the workers will temporarily lose 2% of their pretax income until they are reimbursed by the union. See Sachs, supra note 18, at 1048 n.12. Not every worker will be so meticulous as to observe this relative difference—which amounts to giving the union a short-term loan in the amount of 2% of one’s monthly paycheck—but it stands to reason that some will.

So what is a union to do? The best answer is that any short-term loss experienced by workers is outweighed by an important benefit under the reimbursement model, which unions would do well to emphasize to their members. As Daniel Hemel and David Louk have incisively observed, most public-sector workers—especially low- and middle-class workers—take the standard deduction on their federal income tax returns. 373 See Hemel & Louk, Is Abood Irrelevant?, supra note 18, at 236–37 ([M]any low- and middle-income public sector employees opt to take the standard deduction, in which case the deductibility of union dues and agency fees provides no benefit to them whatsoever.”). That  proportion  will  be  even  higher  now  that  Congress  has  doubled  the  standard  deduction. 374 See Shaina Mishkin, Here’s How to Tell Whether the New GOP Tax Plan Will Simplify Your Taxes, Time (Dec. 20, 2017), [] (noting an estimate that the number of U.S. taxpayers who will itemize their deductions will fall from 46.5 million to 19.3 million, or from roughly 33% to 14%). Public  workers  who  take  the  standard  deduction  are  unable  to  avail  themselves  of  the  miscellaneous  itemized  deduction  for  union  fee  payments. 375 See Hemel & Louk, Is Abood Irrelevant?, supra note 18, at 236–37. What is more, the December 2017 tax law actually suspends all miscellaneous itemized deductions—including the deduction for union fees—for the next eight years. 376 See Individual Tax Reform and Alternative Minimum Tax, Pub. L. No. 115-97, § 11045, 131 Stat. 2054, 2088 (2017) (codified as amended at 26 U.S.C. § 67(g) (2018)) (“[N]o miscellaneous itemized deduction shall be allowed for any taxable year beginning after December 31, 2017, and before January 1, 2026.”). The net effect is that under the preexisting fair-share model, most public-sector workers would be taxed on the portion of their wages spent on chargeable union activity.

By contrast, when the government reimburses the union directly for its chargeable expenses and thus never routes that money through employee paychecks, each employee’s taxable wage (and therefore amount of tax owed) is reduced accordingly, even while their net pay remains the same. To give a concrete sense of this impact, consider a non-married worker who would have paid $1,000 of her $60,000 salary to the union for its chargeable expenses under the fair-share model. For that worker, holding all else equal, the switch to a reimbursement model would mean a net after-tax increase of roughly $300 in annual take-home pay. 377 See 2018 Federal Tax Rates, Personal Exemptions, & Standard Deductions, U.S. Tax Ctr., [] (last visited Oct. 17, 2018) (basing the calculation on the 2018 income tax rate of 22% for income earned in the $38,701 to $82,500 bracket); Social Security Wage Base Increases to $128,700 for 2018, Thomson Reuters Tax & Accounting (Oct. 16, 2017), [] (noting “the FICA tax rate for employers is 7.65%”). The worker would pay $220 less in federal income taxes (22% x $1,000) and $76 less in FICA taxes (7.65% x $1,000), for a total of $296 less. That is not a trivial amount and should be enough from a framing perspective to make it palatable for union members to loan the union 2% of their monthly paychecks until the employer reimburses the union, which will rebate that payment back to its members in turn.

A second important implementation question concerns the timing of when unions and employers should switch to the reimbursement model. The reason this is a problem is not self-evident, so this Article begins by building out the dilemma.

Janus can be thought of as a judicially imposed, right-to-work “big bang” in which all existing fair-share clauses in public-sector collective bargaining agreements are immediately invalidated. The government reimbursement approach is the closest thing to a seamless alternative that would preserve existing expectations among public employers, unions, and workers. 378 See supra section III.D. But there is one notable bump: In order for public employers to implement the reimbursement model in a revenue-neutral way, they would have to reduce wages (or benefits) to offset the new budget outlay for payments to the union. Workers, however, would respond with understandable resentment were a public employer to demand mid-contract wage concessions as a condition for agreement to a reimbursement provision. 379 See supra notes 371–372 and accompanying text. True, workers would be made whole in the end because the wage decrease would be accompanied by an equivalent reduction to their union dues given that the employer would ultimately pay for the union’s bargaining expenses. Appearances matter, though, not least of all when one is asked to take a haircut on one’s annual salary. 380 See, e.g., Tversky & Kahneman, supra note 366, at 454 (“The displeasure associated with losing a sum of money is generally greater than the pleasure associated with winning the same amount . . . .”). So unions should expect some opposition were they to go to their members with the following ask in the aftermath of Janus: “The employer has agreed to reimburse the union for bargaining-related expenses. We will rebate a pro rata portion of that reimbursement back to you, so long as you agree to an equivalent cut of roughly 2% of the salary that you are entitled to under our existing contract.” 381 This assumes that the relevant bargaining agreement has a provision permitting the union and employer to reopen negotiations mid-contract when another provision is rendered illegal by virtue of a judicial ruling. Such provisions are fairly common. See, e.g., 2015–2019 Collective Bargaining Agreement, SEIU Local 503, at 5, [] (last visited Oct. 18, 2018)(“In the event that any provision of this Agreement is at any time declared invalid by any court of competent jurisdiction . . . [t]he invalidated provision shall be subject to renegotiation by the parties within a reasonable period of time from such request.”). However, note that many such provisions limit renegotiations to the subject of the invalidated provision. Under such provisions, Janus may trigger the parties’ right to negotiate over alternative approaches to union financial security, including, for example, a direct-reimbursement approach, but the parties would have to mutually agree to also reopen negotiations regarding wages and benefits to afford an offsetting reduction requested by the employer.

The best solution, then, is for unions and public employers to wait to negotiate a reimbursement provision until the next full contract opener, which would most likely occur at the expiration of the existing collective bargaining agreement. At that point, the union and employer could agree to a reimbursement provision alongside a wage or benefit increase that is more modest than it otherwise would have been, in which the marginal reduction is equal to the employer’s expected annual cost of reimbursing the union. From a behavioral psychology perspective, the key is that workers would have no settled expectation as to their future salary levels. That difference is crucial because it affects the worker’s perception of the reimbursement provision, transforming it from the cause of an economic loss (which would be true if the worker’s salary were cut from a previously defined level) into a part of a package of benefits offered by the employer (since the reimbursement provision would free workers of the responsibility of funding most union activities). 382 See Tversky & Kahneman, supra note 366, at 456 (“Because the value function is steeper for losses than for gains, a difference between options will loom larger when it is framed as a disadvantage of one option rather than as an advantage of the other option.”). Put slightly differently, workers should be more satisfied with a reimbursement provision that accompanies a wage increase that is marginally smaller than it would have been in some unobservable counterfactual than with a reimbursement provision that accompanies a pay cut.

This is not to say that states that choose to operate a permissive reimbursement regime should require the parties to await a full contract negotiation before implementing a reimbursement provision. Insofar as permissive regimes leave reimbursement up to negotiation, the timing of any reimbursement provision is naturally left to unions and public employers to debate based on local conditions. In mandatory states, by contrast, it makes sense for state law to impose a uniform timeframe for implementing the new reimbursement regime. For the reasons given, mandatory states should specify that the duty to reimburse begins upon the expiration of existing agreements. 383 Indeed, an approach requiring mandatory reimbursement immediately upon passage of the state law would cause public employers to suffer an economic loss insofar as unions would feel no pressure to negotiate an offsetting wage decrease. Nor could the employer unilaterally reduce existing wages to recoup the offsetting amount. That is because any effort by public employers to do so, no less than fair-share fees themselves, would likely run afoul of the First Amendment. See Tang, Public Sector Unions, supra note 18, at 207–08.

From the union’s perspective, waiting until the expiration of an existing agreement before receiving government reimbursement may be less than ideal. There may be a lean year or two when the union is left to voluntary dues, and membership will likely decrease in the meantime. 384 See supra section III.A. But so long as one views a short-term decline in union resources as less destructive of public-sector labor relations than a permanent erosion, the reimbursement alternative is worthy of consideration.

D. A Brief Response to Professors Fisk and Malin

In a forthcoming California Law Review article, Professors Fisk and Malin oppose the reimbursement approach advanced here for two reasons. 385 See Fisk & Malin, supra note 38, at 33–36. First, they argue that it “makes unions financially dependent on [public] employers,” thereby leaving unions’ “financial health vulnerable to changes in the political climate.” 386 Id at 33. Second, they worry whether direct reimbursement of union bargaining expenses will “substantially diminish the motivation for [local] union officials to engage the workforce and involve them in the union.” 387 Id. at 38. Both concerns are unquestionably important. But the former ignores important details in the proposal advanced in this Article, and the latter gets things backward for reasons revealed by Fisk and Malin’s own logic.

Starting with the union dependence concern, this Article has already acknowledged the dangers that might arise if unions were to stop collecting member dues in reliance on a government reimbursement policy that is susceptible to later political shifts. 388 See supra notes 369–370. But as the Article has also noted, public-sector unions have an easy solution: to continue collecting member dues in case that shift ever happens. In addition to providing a financial backstop against changing political winds, continuing to collect member dues would have the added benefit of providing the union a consistent revenue stream that can be used to finance bargaining and grievance costs on an ongoing basis prior to the union’s receipt of reimbursement. As the Article has explained, the union would then rebate a pro rata portion of the reimbursement back to each member, leaving members in the precisely same financial position as the agency-fee model invalidated by Janus. At its best, then, direct reimbursement would restore unions and union members to the same financial position as in the agency-fee model; at its worst, unions would still have access to the same voluntary dues payments that they would collect under the right-to-work environment that Fisk and Malin prefer.

Fisk and Malin’s second concern regarding unions’ motivation to engage and respond to members is also misplaced. Recall how Fisk and Malin persuasively argue that the unions face a collective action problem: From the perspective of each prospective member, the benefits of joining are outweighed by the costs given that (1) each worker is already entitled to most union benefits regardless of whether they voluntarily pay dues, and (2) after Janus, the marginal cost of joining the union is now the full pro rata portion of all of the union’s expenses, including both bargaining-related and political costs. 389 See Fisk & Malin, supra note 38, at 6–16 (“Like other large groups, union employees face a collective action problem . . . .”). Fisk and Malin helpfully articulate ways in which unions can improve the benefits side of the equation by offering additional value to prospective members. 390 See id. at 52–60 (“Membership becomes uneconomically rational when the employee is assured that most coworkers will also join the union. Here we explore several systems that might address the collective action problem by removing the incentive not to join and creating incentives to join.”). But they also recognize the importance of reducing the costs side. That is why they support the agency-fee system that existed before Janus, which required all workers to share in the union’s bargaining costs, such that the only added cost of membership was the small portion of costs associated with union political expenses. As Fisk and Malin put it, agency fees “deal[t] with the collective action problem by making the marginal cost of union membership small, such that it bec[ame] an economically rational decision for an employee to decide to join the union.” 391 Id. at 47.

But of course, that is exactly what direct reimbursement does, too. Once the union member receives the rebated pro rata portion of the government’s reimbursement payment for the union’s bargaining-related costs, the “marginal cost of membership” is the identical pro rata share of the union’s political expenditures that was true under the agency-fee model. The result is to increase, not decrease, workplace solidarity and democracy because union organizers can more easily persuade prospective members to join once the cost of doing so is just the small portion of member dues associated with political expenses—a net savings for each member that is made possible precisely because the union has fought for the reimbursement of member bargaining expenses.

Fisk  and  Malin  aggressively  defend  the  agency-fee  model’s  ability  to  reduce  the  marginal  cost  of  membership  as  a  powerful  antidote  to  the  collective  action  problem. 392 See id. (“Agency fees . . . deal with the collective action problem by making the marginal cost of membership small . . . .”). So  how  do  they  reach  the  opposite  conclusion  for  direct  reimbursement?  In  a  footnote,  they  suggest  that  it  is  because  a  prospective  member  in  the  reimbursement  regime  must  affirmatively  choose  to  pay  the  political  portion  of  member  dues,  whereas in  the  agency-fee  model  all  workers  were required  to  pay  full  dues  by  default  and  had  to  opt  out  if  they  wished  to  avoid  paying  the political portion. 393 Id. at 38 n.140. But that is an attack on Janus’s decision to switch the default opt-out rule, not an attack on reimbursement itself. 394 See supra notes 91–95 and accompanying text. Indeed, if Fisk and Malin truly think Janus’s new default rule makes such a big difference, then their support for the agency-fee model would wind up surprisingly thin.

Consider the following hypothetical. Suppose Janus had come out much more favorably to public-sector unions, reaffirming Abood’s core rule allowing the collection of agency fees. Suppose further, however, that the majority had decided to overrule the opt-out default rule for agency-fee collection, perhaps in a kind of compromise ruling. Thus,  while  all  workers  could  still  be  required  to  pay  their  pro  rata  share  of  bargaining-related  expenses  (the  agency  fee  itself),  the  union  would  be  allowed  to  collect  the  political  portion  of  member  dues  from  only  those  workers  who  affirmatively  consent. If Fisk and Malin believe that public-sector unions will “see it in their personal best interest to keep membership levels low” once this default is switched, as they have suggested in the course of critiquing the direct-reimbursement approach, 395 Fisk & Malin, supra note 38, at 38 n.140. the implication is that they would respond to this hypothetical pro-labor outcome by calling for all states to abolish agency fees by statute. After all, once workers are treated as nonmembers by default, unions would bear the burden of “engag[ing] workers and educat[ing] them about the benefits of union membership and the importance of solidarity” to convince them to join. 396 Id. Fisk and Malin claim that unions will not bother with such efforts once they are guaranteed a stable stream of funding via direct reimbursement. 397 Id. But that same critique would also apply to the identical stream of funding guaranteed via agency fees.

The reality is that unions would have at least three powerful incentives under the direct-reimbursement approach to continue engaging new members, even after Janus’s decision to implement an affirmative consent default. First, any union that wishes to fight on behalf of its members via issue advocacy, lobbying, or supporting candidates for office would need to engage a vibrant membership base to do so, since only the political portion of member dues can be used toward those ends. Second, in the reimbursement model, member dues are necessary to fund the union’s day-to-day activities because reimbursement only happens periodically, after the actual costs of bargaining and grievance are incurred. And third, those same dues are an important safeguard for the union’s long-term health, protecting the union from financial difficulty in the event a state or public employer backslides or reneges on its reimbursement duty.

Finally, even if one harbors a distrust of unions’ motivations to organize and engage additional members, the best answer would be to resolve that concern via state law. The simplest approach would be to require unions to continue collecting full member dues as a condition of obtaining reimbursement—perhaps reimbursement could be offered only to unions that reach a certain threshold percentage of membership, or perhaps reimbursement amounts themselves could be pegged to membership levels. Either approach would eliminate both the dependency concern (since unions would continue to have an active member base even if reimbursement were later eliminated) and the member-responsiveness concern (because vibrant union membership organizing would be a condition precedent to reimbursement itself). Ultimately, if one thinks the economic incentives for unions to engage additional members are not powerful enough on their own, the answer is not to relegate unions to the right-to-work baseline. The answer is to align the union’s interest in adequate funding with its interest in workplace solidarity and democracy as a matter of law.


Among the menu of legislative options available to pro-labor states after Janus, one statutory fix stands out. If the goal is to preserve these states’ pre-Janus system of labor relations—characterized by independent and adequately resourced unions negotiating exclusively and efficiently with public employers 398 See supra section I.A. —then direct government reimbursement of union bargaining costs can achieve that end.

As this Article has tried to show, how states implement a reimbursement regime matters a great deal. As any student of American indigent criminal defense knows, there are many bad ways to fund an adversarial entity in pursuit of some broader public objective. 399 See supra section III.C.3. But there are also ways to do it right, including by leaving funding determinations in the ultimate hands of neutral experts or the union members themselves. 400 See supra sections III.C.4, IV.A. There will be modest switching costs in moving to a reimbursement system, of that there is little doubt. But the costs seem minimal in comparison to the disruption wrought by Janus’s right-to-work big bang.

Appendix: Model Bill


Notwithstanding any other provision of law, a public employer may 401 This language reflects the majority position in former fair-share fee jurisdictions, which adopt a permissive stance toward fair-share-fee clauses. See supra notes 61, 63 and accompanying text. However, states with mandatory fair-share provisions may prefer to implement a mandatory approach to reimbursement as well. See supra notes 60, 62 and accompanying text. reimburse an employee organization for costs germane to the collective bargaining process, contract administration, and pursuing matters affecting wages, hours and other conditions of employment. 402 As explained above, see supra section IV.A.1, the exact language of a state’s reimbursement exception in section 1 should mirror the language used in the state’s preexisting fair-share fee statute in order to incorporate by reference the full body of state court and administrative precedent regarding the scope of permissible union expenses.


(a) An agreement by which a public employer reimburses an employee organization for expenses consistent with Section 1 shall be a [mandatory/permissive 403 This choice should mirror each state’s existing choice as to the mandatory–permissive nature of bargaining on fair-share-fee provisions. See supra notes 353–354 and accompanying text. ] subject of bargaining by the public employer and the exclusive representative of each bargaining unit.

(b) When a collective bargaining agreement includes a provision requiring a public employer to reimburse an employee organization for expenditures permitted under Section 1, 404 This language reflects a permissive approach in which individual public employers may, but need not, agree to a reimbursement provision. States with mandatory fair-share laws may wish to use similar mandatory language here instead. See supra note 60 and accompanying text. For states that choose a mandatory approach to reimbursement, note that section 2(a) would be unnecessary. the employee organization shall annually certify to the employer the amount of permissible expenditures it has actually incurred.

(1) If the employer agrees that the identified expenditures are reim­bursable under Section 1, it shall promptly reimburse the employee organization for the certified amount.

(2) If the employer disagrees with any of the identified expend­itures, the employer must file within thirty days an objection with the Public Employment Relations Board identifying the specific expenditures to be challenged. The employer shall set aside the challenged expenditures in an escrow account maintained by the employer, employee organization, or Board, and promptly reimburse the employee organization for all other expenditures. 405 This model provision implements a PERB-based approach to resolving disputes over the permissibility of union expenditures. Two alternatives are presented above: one approach that would leave disputes over the permissibility of certain union expenses to determination by the union’s own members, see supra section IV.A.2.b, and a hybrid model that would retain a role for both a PERB and a union’s membership, see supra section IV.A.2.c.

(c) The procedures governing review of challenges brought pursuant to Section 2(b)(2) shall be the same as those that governed challenges brought by objecting workers to fair-share fees under all applicable state laws prior to the Supreme Court’s decision in Janus v. AFSCME, Council 31.


[Placeholder for exception to state campaign finance restrictions on political expenditures by government contractors.] 406 See supra notes 292–294 and accompanying text.


[In addition to the procedure prescribed in Section 2(a), an agreement by which a public employer reimburses an employee organization for expenses consistent with Section 1 shall be placed in effect, without a negotiated agreement, upon (1) a signed petition of 30% of the employees in the applicable bargaining unit requesting a reimbursement agreement and an election to implement such an agreement, and (2) the approval of a majority of employees who cast ballots and vote in a secret ballot election in favor of the reimbursement agreement. The petition may be filed only after the recognized employee organization has requested the public employer to negotiate on a reimbursement agreement and, beginning seven working days after the public agency received this request, the two parties have had thirty calendar days to attempt good faith negotiations in an effort to reach agreement.] 407 This is an optional provision that would allow employees in a permissive state to add a reimbursement clause to their collective bargaining agreement even if the public employer refuses to agree to one. See supra notes 359–361 and accompanying text.