Harvard Law Review Harvard Law Review Harvard Law Review

First Amendment

Agency Fees and the First Amendment

The full text of this Article may be found by clicking on the PDF link to the left.

In the last five years, the United States Supreme Court has decided three cases involving agency fees — the mandatory payments that certain employees are required to make to unions. The Court will hear a fourth case this Term.1×1. See Janus v. AFSCME, 851 F.3d 746 (7th Cir. 2017), cert. granted, 198 L. Ed. 2d 780 (2017). All of these cases raise the question of whether agency fee agreements in the public sector are constitutional under the First Amendment. In a case argued in October Term 2015, Friedrichs v. California Teachers Ass’n,2×2. 136 S. Ct. 1083 (2016) (mem.) (per curiam) (affirming by an equally divided Court). the Court appeared poised to hold in the negative and to decide that agency fees amount to compelled speech and association.3×3. Id. at 1083. The death of Justice Scalia prevented the Court from reaching such a holding in Friedrichs, but the question has returned again in Janus v. American Federation of State, County and Municipal Employees, Council 31.4×4. See, e.g., Moshe Z. Marvit, Labor Opponents Already Have the Next “Friedrichs” SCOTUS Case Ready to Go Under Trump, In These Times (Jan. 4, 2017, 1:55 PM), http://inthesetimes.com/working/entry/19776/will_trumps_supreme_court_reverse_fair_share_fees_unions_foes_hope_so [https://perma.cc/NGW9-AMJK].

To those outside the field of labor law, the issue of agency fees might seem picayune.5×5. For expositional simplicity, this Article will use “agency fees” or simply “fees” interchangeably to refer to payments to unions that are required by a collective bargaining agreement. “Dues” will be used to refer to payments that go beyond what is required by a collective bargaining agreement; that is, payments that include both the mandatory component and a nonmandatory component. See, e.g., Catherine L. Fisk & Benjamin I. Sachs, Restoring Equity in Right-to-Work Law, 4 U.C. Irvine L. Rev. 857, 858–59 (2014). This Article’s analysis, like the Court’s jurisprudence, is concerned only with agency fees. But it is not, and it is not for reasons that explain the Court’s interest in the issue. Agency fees are the sole means through which unions have been permitted to overcome what otherwise would be an existential collective action problem. In brief, unions have a legal obligation to provide benefits to all workers in a given workplace or bargaining unit.6×6. For a summary, see, for example, id. Union-negotiated benefits therefore have the character of public goods: if a union negotiates a wage increase, or better benefits package, or enhanced safety and health protections, these improvements must be extended to all the workers covered by the collective agreement. If unions are required to rely for their financing on voluntary payments from these workers, then unions would face extensive free riding by all those workers who would rather receive benefits for free than pay for them.7×7. See, e.g., Harris v. Quinn, 134 S. Ct. 2618, 2657 (2014) (Kagan, J., dissenting). A decision holding agency fees unconstitutional would thus jeopardize the future of public sector unions.

Much has been written about agency fees, and the issue is among the most extensively adjudicated in American labor law. But this Article will argue that we have misconceived what agency fees are. The jurisprudence and scholarship on agency fees proceeds from the assumption that such fees are workers’ money that workers pay to a union. As Justice Scalia put it in a recent opinion, an agency fee agreement gives the union authority “to acquire and spend other people’s money.”8×8. Davenport v. Wash. Educ. Ass’n, 551 U.S. 177, 187 (2007). This understanding of agency fees sets up the First Amendment question: because the money that becomes fees belongs to workers, its compelled transfer to the union may constitute compelled speech or association.9×9. Of course, even on the traditional understanding — of fees as employee property paid by workers to unions — there are strong arguments that mandatory agency fees do not amount to unconstitutional compelled speech. See, e.g., Catherine L. Fisk & Erwin Chemerinsky, Political Speech and Association Rights After Knox v. SEIU, Local 1000, 98 Cornell L. Rev. 1023 (2013). That set of arguments is beyond the scope of this Article. But this is the wrong way to understand agency fees for two sets of reasons.

First, the Court treats agency fees as employees’ money because those fees are paid, as a formal matter, out of employee paychecks. The money that becomes fees appears as wages on the income side of an employee’s paycheck and as a mandatory payment on the expense side of the paycheck. But this is an accounting formalism required by labor law. For purposes of First Amendment analysis, agency fees are — and should be treated as — a payment made by employers to unions.

As this Article will show, in order to eliminate the company unions that were prevalent in the 1930s, the National Labor Relations Act10×10. 29 U.S.C. §§ 151–169 (2012). (NLRA), and the state public sector labor laws that followed it, prohibit employers from paying any monies directly to unions. Instead, labor law establishes an accounting regime under which employers pay wages to workers, who then must pay fees to the union. Although the specific numbers are not important, an example from the contemporary labor market will help explain the operation of this accounting rule. The most recent estimates show, for example, that unionization results in a wage premium of about 11%.11×11. By collectivizing their bargaining power, unionization allows workers to increase the compensation they receive from employers. This union “premium” includes higher wages, and a good approximation of the current wage premium is 11%. See, e.g., Frank Manzo IV et al., The State of the Unions 2016, at 14–15 (2016). Of course, different studies report different specific premia and the 11% figure is used in this Article for illustrative purposes only. See, e.g., Jake Rosenfeld, What Unions No Longer Do 45 (2014) (reporting a 2009 premium of 22% for private sector workers, and 14% for public sector workers); Lawrence Mishel, Economic Policy Institute, Issue Brief No. 342, Unions, Inequality, and Faltering Middle-Class Wages (2012), http://www.epi.org/publication/ib342-unions-inequality-faltering-middle-class/ [https://perma.cc/C9DW-YYRQ] (finding premia to vary based on demographics and identifying in 2011 a 10.9% premium for whites, a 17.3% premium for African Americans, and a 23.1% premium for Hispanics). Moreover, the union premium also includes things like expanded paid leave, better health and retirement benefits, and heightened protections against discharge. See, e.g., John W. Budd, The Effect of Unions on Employee Benefits and Non-Wage Compensation: Monopoly Power, Collective Voice, and Facilitation, in What Do Unions Do? A Twenty-Year Perspective 160 (James T. Bennett & Bruce E. Kaufman eds., 2007). Because the 11% figure captures only wage gains, and does not include benefits, it significantly underestimates the actual value of the union premium. See, e.g., id. at 177. It is also true that, in certain contexts, the union premium amounts to a reduction in wage losses rather than an absolute wage increase: the 11% figure reflects an average across the entire labor market. Agency fees now average in the neighborhood of 2% of wages.12×12. Full union dues amount to approximately 3% of wages. See, e.g., Ben Casselman, Closer Look at Union vs. Nonunion Workers’ Wages, Wall St. J. (Dec. 17, 2012, 3:40 PM), http://blogs.wsj.com/economics/2012/12/17/closer-look-at-union-vs-nonunion-workers-wages/ [https://perma.cc/V77W-7KVG]. Agency fees, which include only the amount that unions spend on collective bargaining and contract administration, vary widely across unions and across collective bargaining agreements. The specific number is irrelevant to the analysis in this Article. For purposes of illustration only, I estimate agency fees as equaling two-thirds of full union dues, or 2% of wages. The agency fee at issue in the Friedrichs litigation was approximately two-thirds of full dues, and thus the figure seems appropriate for illustration purposes here. See, e.g., Alana Semuels, Why Are Unions So Worried About an Upcoming Supreme Court Case?, The Atlantic (Jan. 8, 2016), https://www.theatlantic.com/business/archive/2016/01/friedrichs-labor/423129/ [https://perma.cc/M94S-LD6Y] (“In California, members pay annual dues that average about $1,000 a year, while non-members pay about $600 to $650 for the agency fee alone.”); Complaint at 18, Friedrichs v. California Teachers Ass’n, No. 8:13-cv-00676 (C.D. Cal. Apr. 30, 2013) (“The total amount of annual dues generally exceeds $1,000 per teacher, while the amount of the refund received by nonmembers who successfully opt out of the non-chargeable portion of their agency fees is generally around $350 to $400.”). Observers have calculated agency fees for a variety of contexts, generally aligning with the estimates above. See, e.g., Jan Murphy, Should Non-Union Members Have to Pay a Fair-Share Fee? Supreme Court to Decide, Penn Live (Jan. 11, 2016, 7:45 AM), http://www.pennlive.com/politics/index.ssf/2016/01/should_non-union_members_have.html [https://perma.cc/G5TZ-PWMP] (showing that the agency fee paid by Pennsylvania teachers is 77% of full dues); Member vs. Fee Payer: What’s the Difference, Minn. Ass’n of Prof. Emps., https://www.mape.org/my-mape/our-union/member-vs-fee-payer-whats-difference [https://perma.cc/RNK7-79HA] (showing that the agency fee paid by workers in Minnesota’s largest public employee union was 85% of full dues). Labor law’s accounting rules require that employers pay all 11% of the premium to individual workers in the form of higher wages, and then allows employers to mandate that workers pay — on pain of discharge — 2% back to the union. Prohibited by labor law is a functionally equivalent regime in which workers receive a 9% wage increase and employers pay 2% directly to unions.

As a result of this accounting regime, the money that becomes agency fees does in fact pass through employee paychecks en route from employer to union. But, for purposes of First Amendment analysis, the fact that these monies pass through employee paychecks is irrelevant: the payment must be treated as one made by the employer to the union. The Court established this rule in a cognate area of First Amendment law, where it resolved the question of whether government payments made to individual families and then paid, by those families, to religious schools violate the Establishment Clause. The answer to this question turns on whether the payments to the religious schools are treated as coming from the families or from the government. The Court has held that this answer, in turn, depends on whether the families have a “genuine choice” about paying the money to a religious school or a secular school.13×13. Zelman v. Simmons-Harris, 536 U.S. 639, 662 (2002). Where there is “true private choice,”14×14. Id. at 650. the money is treated as coming from the family, breaking the “circuit between government and religion” and there is no First Amendment violation.15×15. Id. at 652. Where such choice is lacking, the fact that the monies pass through the hands of families becomes irrelevant and the monies must be treated as “a government program of direct aid to religious schools.”16×16. Mitchell v. Helms, 530 U.S. 793, 842 (2000) (O’Connor, J., concurring).

In the agency fees context, employees of course have no “genuine choice” about whether or not they pay fees to a union — that is the essence of every constitutional challenge to agency fees. Because there is no employee choice on the matter, however, the fact that employer payments pass through employee paychecks does not “break the circuit” between the employer and the union. For purposes of First Amendment analysis, therefore, the payments should be treated as if they flowed directly from the employer to the union.17×17. As also discussed infra note 97, although the First Amendment would ignore labor law’s accounting formalism and treat agency fee payments as flowing directly from employer to union, the NLRA — and analogous state statutes — are entitled to impose this accounting system and to distinguish between payments made directly by employers to unions and payments that flow from employers, through employee paychecks, and on to unions. Compare 29 U.S.C. §§ 158(a)(2), 186 (2012), with id. §§ 158(a)(3), 186(c)(4). Thus, the fact that the First Amendment would view agency fees as a payment made by employers to unions does not invalidate a statutory regime that prohibits one kind of accounting system and enables the other. And monies that flow from the employer to the union do not create compelled speech or association problems for employees.

Second, irrespective of the accounting regime, there are more foundational reasons for treating agency fees as union property rather than as the property of individual workers. Unionization increases the bargaining power of workers by allowing them to negotiate collectively rather than individually with employers. This increased bargaining power enables workers, as a collective, to transfer more wealth from the employer to themselves than the workers would have been able to secure as individuals.18×18. See generally Richard B. Freeman & James L. Medoff, What Do Unions Do? 6–7 (1984). Agency fees amount to a small fraction of this transfer — a small fraction of the union premium. The question is whether the union premium, and the fees that come out of it, ought to be treated as property of the union in the first instance or as property of individual workers.

Because it is the union that produces the premium, and because individual workers would never earn the premium as individuals, it is not at all clear why we ought to treat the premium as the property of individual workers. To the contrary, it makes better sense to treat the excess wealth transferred by unionization — and the fees that come out of this transfer — as the product of collectivization that belongs to the collective that produced it. Agency fees, in other words, should be treated as the property of the union that secures them.

Although as yet unidentified in the literature, there is both doctrinal and theoretical support for this intuition. On the doctrinal side, two areas of takings jurisprudence are illuminating. One concerns the law of Interest on Lawyers’ Trust Accounts (IOLTA) programs, which use interest generated by lawyers’ trust accounts to fund legal services programs. As the Article will discuss, IOLTA programs are funded by interest that is generated solely through the collectivizing effect of the program itself: but for the aggregating effect of the program, individual clients would earn no interest.19×19. See infra pp. 1063–66. In light of this fact, courts have held either that individual clients have no property in the interest that the aggregated principal generates — they do not own it — or, alternatively, that the value of their property in the interest is zero. A similar rule emerges from a line of eminent domain cases: where value is created solely through the “union” of individual lots that results from the government’s action in taking the land, the individual lot owners have no property interest in the value thus created through the union of the lots.20×20. Indeed, “union” is the term the Court uses in these cases. See infra pp. 1066–67. In both these areas of law, then, where value is created through collectivization, it is the collective, not the individual, that owns the value thus created. And, as the Article will also show, treating value created collectively as the property of the collective has a home in a strand of economic theory that traces back at least to the work of John Stuart Mill.21×21. See infra section III.B, pp. 1068–69.

The Article thus provides two sets of reasons for concluding the Court is wrong that agency fee agreements give the union authority to “acquire and spend other people’s money.”22×22. Davenport v. Wash. Educ. Ass’n, 551 U.S. 177, 187 (2007) (emphasis omitted). And both arguments have the same fundamental implication for the jurisprudence of agency fees: if we understand that agency fees ought not to be treated as payments from employees to unions, the First Amendment problem with agency fees — the idea that such fees amount to compelled speech or association — disappears.

After providing a quick overview of the Supreme Court’s agency fees cases, this Article endeavors to do four things. First, to show that although the money that becomes agency fees does pass through employee paychecks en route from employers to unions, for purposes of First Amendment analysis this does not convert the employer’s money into employee property before the union receives it. Second, to show that there are strong justifications in precedent outside the labor context for treating the union premium, and the fees that come out of it, as the property of the union, and not as the property of individual workers. Third, to demonstrate the implications of no longer treating agency fees as payments from employees to unions: namely, that the First Amendment problem with agency fees evaporates. And, fourth, to note that states can also address this issue by changing public sector labor law’s accounting regime to formally permit direct payments from employers to unions, and to discuss the implications beyond the First Amendment of such a change in policy.23×23. Two recent articles argue that states could amend their public sector bargaining laws and permit direct employer payments to unions as a substitute for the current system, and both articles discuss what they see as the First Amendment implications of that point. See Daniel Hemel & David Louk, Is Abood Irrelevant?, 82 U. Chi. L. Rev. Dialogue 227 (2016); Aaron Tang, Public Sector Unions, the First Amendment, and the Costs of Collective Bargaining, 91 N.Y.U. L. Rev. 144 (2016). Neither piece argues, as this Article does, that the current system is already — for constitutional purposes — a system of direct payments from employers to unions. Nor does either of the earlier articles contend that, for the reasons developed here, agency fees ought to be treated as the property of the union that generates them.


* Kestnbaum Professor of Labor and Industry, Harvard Law School. The author thanks Kate Andrias, Craig Becker, Yochai Benkler, Sharon Block, Catherine Fisk, Matthew Ginsburg, Karl Klare, Daryl Levinson, John Manning, Casey Pitts, Judith Scott, Joseph Singer, and Andrew Strom for helpful comments and conversation. The author also thanks Niko Bowie, Daniel Crossen, Stephanie Jimenez, Madeleine Joseph, Feyilana Lawoyin, Alexander Miller, and Maia Usui for excellent research assistance.