JANUS v. AMERICAN FEDERATION OF STATE, COUNTY, AND MUNICIPAL EMPLOYEES, COUNCIL 31

NOTICE: This opinion is subject to formal revision before publication in the preliminary print of the United States Reports. Readers are requested to notify the Reporter of Decisions, Supreme Court of the United States, Washington, D. C. 20543, of any typographical or other formal errors, in order that corrections may be made before the preliminary print goes to press.

MARK JANUS, PETITIONER

v.

AMERICAN FEDERATION OF STATE, COUNTY, AND MUNICIPAL EMPLOYEES, COUNCIL 31, et al.

No. 16–1466

Supreme Court of United States.

on writ of certiorari to the united states court of appeals for the seventh circuit

[June 27, 2018]

Justice Alito delivered the opinion of the Court.

Under Illinois law, public employees are forced to subsidize a union, even if they choose not to join and strongly object to the positions the union takes in collective bargaining and related activities. We conclude that this arrangement violates the free speech rights of nonmembers by compelling them to subsidize private speech on matters of substantial public concern.

We upheld a similar law in Abood v. Detroit Bd. of Ed.,431 U. S. 209 (1977), and we recognize the importance of following precedent unless there are strong reasons for not doing so. But there are very strong reasons in this case. Fundamental free speech rights are at stake. Abood was poorly reasoned. It has led to practical problems and abuse. It is inconsistent with other First Amendment cases and has been undermined by more recent decisions. Developments since Abood was handed down have shed new light on the issue of agency fees, and no reliance interests on the part of public-sector unions are sufficient to justify the perpetuation of the free speech violations that Abood has countenanced for the past 41 years. Abood is therefore overruled.

I

A

Under the Illinois Public Labor Relations Act (IPLRA), employees of the State and its political subdivisions are permitted to unionize. See Ill. Comp. Stat., ch. 5, §315/6(a) (West 2016). If a majority of the employees in a bargaining unit vote to be represented by a union, that union is designated as the exclusive representative of all the employees. §§315/3(s)(1), 315/6(c), 315/9. Employees in the unit are not obligated to join the union selected by their co-workers, but whether they join or not, that union is deemed to be their sole permitted representative. See §§315/6(a), (c).

Once a union is so designated, it is vested with broad authority. Only the union may negotiate with the employer on matters relating to “pay, wages, hours[,] and other conditions of employment.” §315/6(c). And this authority extends to the negotiation of what the IPLRA calls “policy matters,” such as merit pay, the size of the work force, layoffs, privatization, promotion methods, and non-discrimination policies. §315/4; see §315/6(c); see gener- ally, e.g., Illinois Dept. of Central Management Servs. v. AFSCME, Council 31, No. S–CB–16–17 etc., 33 PERI ¶67 (ILRB Dec. 13, 2016) (Board Decision).

Designating a union as the employees’ exclusive representative substantially restricts the rights of individual employees. Among other things, this designation means that individual employees may not be represented by any agent other than the designated union; nor may individual employees negotiate directly with their employer. §§315/6(c)–(d), 315/10(a)(4); see Matthews v. Chicago Transit Authority, 2016 IL 117638, 51 N. E. 3d 753, 782; accord, Medo Photo Supply Corp. v. NLRB,321 U. S. 678, 683–684 (1944). Protection of the employees’ interests is placed in the hands of the union, and therefore the union is required by law to provide fair representation for all employees in the unit, members and nonmembers alike. §315/6(d).

Employees who decline to join the union are not assessed full union dues but must instead pay what is generally called an “agency fee,” which amounts to a percentage of the union dues. Under Abood, nonmembers may be charged for the portion of union dues attributable to activities that are “germane to [the union’s] duties as collective-bargaining representative,” but nonmembers may not be required to fund the union’s political and ideological projects. 431 U. S., at 235; see id., at 235–236. In labor-law parlance, the outlays in the first category are known as “chargeable” expenditures, while those in the latter are labeled “nonchargeable.”

Illinois law does not specify in detail which expenditures are chargeable and which are not. The IPLRA provides that an agency fee may compensate a union for the costs incurred in “the collective bargaining process, contract administration[,] and pursuing matters affecting wages, hours[,] and conditions of employment.” §315/6(e); see also §315/3(g). Excluded from the agency-fee calculation are union expenditures “related to the election or support of any candidate for political office.” §315/3(g); see §315/6(e).

Applying this standard, a union categorizes its expenditures as chargeable or nonchargeable and thus determines a nonmember’s “proportionate share,” §315/6(e); this determination is then audited; the amount of the “proportionate share” is certified to the employer; and the employer automatically deducts that amount from the nonmembers’ wages. See ibid.; App. to Pet. for Cert. 37a; see also Harris v. Quinn, 573 U. S. ___, ___–___ (2014) (slip op., at 19–20) (describing this process). Nonmembers need not be asked, and they are not required to consent before the fees are deducted.

After the amount of the agency fee is fixed each year, the union must send nonmembers what is known as a Hudson notice. See Teachers v. Hudson,475 U. S. 292 (1986). This notice is supposed to provide nonmembers with “an adequate explanation of the basis for the [agency] fee.”&�Բ��;Id., at 310. If nonmembers “suspect that a union has improperly put certain expenses in the [chargeable] category,” they may challenge that determination. Harris, supra, at ___ (slip op., at 19).

As illustrated by the record in this case, unions charge nonmembers, not just for the cost of collective bargaining per se, but also for many other supposedly connected activities. See App. to Pet. for Cert. 28a–39a. Here, the nonmembers were told that they had to pay for “[l]obbying,” “[s]ocial and recreational activities,” “advertising,” “[m]embership meetings and conventions,” and “litigation,” as well as other unspecified “[s]ervices” that “may ultimately inure to the benefit of the members of the local bargaining unit.”&�Բ��;Id., at 28a–32a. The total chargeable amount for nonmembers was 78.06% of full union dues. Id., at 34a.

B

Petitioner Mark Janus is employed by the Illinois Department of Healthcare and Family Services as a child support specialist. Id., at 10a. The employees in his unit are among the 35,000 public employees in Illinois who are represented by respondent American Federation of State, County, and Municipal Employees, Council 31 (Union). Ibid. Janus refused to join the Union because he opposes “many of the public policy positions that [it] advocates,” including the positions it takes in collective bargaining. Id., at 10a, 18a. Janus believes that the Union’s “behavior in bargaining does not appreciate the current fiscal crises in Illinois and does not reflect his best interests or the interests of Illinois citizens.”&�Բ��;Id., at 18a. Therefore, if he had the choice, he “would not pay any fees or otherwise subsidize [the Union].”&�Բ��;Ibid. Under his unit’s collective-bargaining agreement, however, he was required to pay an agency fee of $44.58 per month, id., at 14a—which would amount to about $535 per year.

Janus’s concern about Illinois’ current financial situation is shared by the Governor of the State, and it was the Governor who initially challenged the statute authorizing the imposition of agency fees. The Governor commenced an action in federal court, asking that the law be declared unconstitutional, and the Illinois attorney general (a respondent here) intervened to defend the law. App. 41. Janus and two other state employees also moved to intervene—but on the Governor’s side. Id., at 60.

Respondents moved to dismiss the Governor’s challenge for lack of standing, contending that the agency fees did not cause him any personal injury. E.g., id., at 48–49. The District Court agreed that the Governor could not maintain the lawsuit, but it held that petitioner and the other individuals who had moved to intervene had standing because the agency fees unquestionably injured them. Accordingly, “in the interest of judicial economy,” the court dismissed the Governor as a plaintiff, while simultane- ously allowing petitioner and the other employees to file their own complaint. Id., at 112. They did so, and the case proceeded on the basis of this new complaint.

The amended complaint claims that all “nonmember fee deductions are coerced political speech” and that “the First Amendment forbids coercing any money from the nonmembers.” App. to Pet. for Cert. 23a. Respondents moved to dismiss the amended complaint, correctly recognizing that the claim it asserted was foreclosed by Abood. The District Court granted the motion, id., at 7a, and the Court of Appeals for the Seventh Circuit affirmed, 851 F. 3d 746 (2017).

Janus then sought review in this Court, asking us to overrule Abood and hold that public-sector agency-fee arrangements are unconstitutional. We granted certiorari to consider this important question. 582 U. S. ___ (2017).

II

Before reaching this question, however, we must con- sider a threshold issue. Respondents contend that the Dis- trict Court lacked jurisdiction under Article III of the Constitution because petitioner “moved to intervene in [the Governor’s] jurisdictionally defective lawsuit.” Union Brief in Opposition 11; see also id., at 13–17; State Brief in Opposition 6; Brief for Union Respondent i, 16–17; Brief for State Respondents 14, n. 1. This argument is clearly wrong.

It rests on the faulty premise that petitioner intervened in the action brought by the Governor, but that is not what happened. The District Court did not grant petitioner’s motion to intervene in that lawsuit. Instead, the court essentially treated petitioner’s amended complaint as the operative complaint in a new lawsuit. App. 110–112. And when the case is viewed in that way, any Article III issue vanishes. As the District Court recognized—and as respondents concede—petitioner was injured in fact by Illinois’ agency-fee scheme, and his injuries can be redressed by a favorable court decision. Ibid.; see Record 2312–2313, 2322–2323. Therefore, he clearly has Article III standing. Lujan v. Defenders of Wildlife,504 U. S. 555, 560–561 (1992). It is true that the District Court docketed petitioner’s complaint under the number originally assigned to the Governor’s complaint, instead of giving it a new number of its own. But Article III jurisdiction does not turn on such trivialities.

The sole decision on which respondents rely, United States ex rel. Texas Portland Cement Co. v. McCord,233 U. S. 157 (1914), actually works against them. That case concerned a statute permitting creditors of a government contractor to bring suit on a bond between 6 and 12 months after the completion of the work.Id., at 162. One creditor filed suit before the 6-month starting date, but another intervened within the 6-to-12-month window. The Court held that the “[t]he intervention [did] not cure th[e] vice in the original [prematurely filed] suit,” but the Court also contemplated treating “intervention . . . as an original suit” in a case in which the intervenor met the requirements that a plaintiff must satisfy—e.g., filing a separate complaint and properly serving the defendants. Id., at 163–164. Because that is what petitioner did here, we may reach the merits of the question presented.

III

In Abood, the Court upheld the constitutionality of an agency-shop arrangement like the one now before us, 431 U. S., at 232, but in more recent cases we have recognized that this holding is “something of an anomaly,”&�Բ��;Knox v. Service Employees,567 U. S. 298, 311 (2012), and that Abood’s “analysis is questionable on several grounds,”&�Բ��;Harris, 573 U. S., at ___ (slip op., at 17); see id., at ___–___ (slip op., at 17–20) (discussing flaws in Abood’s reasoning). We have therefore refused to extend Abood to situations where it does not squarely control, see Harris, supra, at ___–___ (slip op., at 27–29), while leaving for another day the question whether Abood should be overruled, Harris, supra, at ___, n. 19 (slip op., at 27, n. 19); see Knox, supra, at 310–311.

We now address that question. We first consider whether Abood’s holding is consistent with standard First Amendment principles.

A

The First Amendment, made applicable to the States by the Fourteenth Amendment, forbids abridgment of the freedom of speech. We have held time and again that freedom of speech “includes both the right to speak freely and the right to refrain from speaking at all.”&�Բ��;Wooley v. Maynard,430 U. S. 705, 714 (1977); see Riley v. National Federation of Blind of N. C., Inc.,487 U. S. 781, 796–797 (1988); Harper & Row, Publishers, Inc. v. Nation Enterprises,471 U. S. 539, 559 (1985); Miami Herald Publishing Co. v. Tornillo,418 U. S. 241, 256–257 (1974); accord, Pacific Gas & Elec. Co. v. Public Util. Comm’n of Cal.,475 U. S. 1, 9 (1986) (plurality opinion). The right to eschew association for expressive purposes is likewise protected. Roberts v. United States Jaycees,468 U. S. 609, 623 (1984) (“Freedom of association . . . plainly presupposes a freedom not to associate”); see Pacific Gas & Elec., supra, at 12 (“[F]orced associations that burden protected speech are impermissible”). As Justice Jackson memorably put it: “If there is any fixed star in our constitutional constellation, it is that no official, high or petty, can prescribe what shall be orthodox in politics, nationalism, religion, or other matters of opinion or force citizens to confess by word or act their faith therein.”&�Բ��;West Virginia Bd. of Ed. v. Barnette,319 U. S. 624, 642 (1943) (emphasis added).

Compelling individuals to mouth support for views they find objectionable violates that cardinal constitutional command, and in most contexts, any such effort would be universally condemned. Suppose, for example, that the State of Illinois required all residents to sign a document expressing support for a particular set of positions on controversial public issues—say, the platform of one of the major political parties. No one, we trust, would seriously argue that the First Amendment permits this.

Perhaps because such compulsion so plainly violates the Constitution, most of our free speech cases have involved restrictions on what can be said, rather than laws compelling speech. But measures compelling speech are at least as threatening.

Free speech serves many ends. It is essential to our democratic form of government, see, e.g., Garrison v. Louisiana,379 U. S. 64, 74–75 (1964), and it furthers the search for truth, see, e.g., Thornhill v. Alabama,310 U. S. 88, 95 (1940). Whenever the Federal Government or a State prevents individuals from saying what they think on important matters or compels them to voice ideas with which they disagree, it undermines these ends.

When speech is compelled, however, additional damage is done. In that situation, individuals are coerced into betraying their convictions. Forcing free and independent individuals to endorse ideas they find objectionable is always demeaning, and for this reason, one of our landmark free speech cases said that a law commanding “involuntary affirmation” of objected-to beliefs would require “even more immediate and urgent grounds” than a law demanding silence. Barnette, supra, at 633; see also Riley, supra, at 796–797 (rejecting “deferential test” for compelled speech claims).

Compelling a person to subsidize the speech of other private speakers raises similar First Amendment concerns. Knox, supra, at 309; United States v. United Foods, Inc.,533 U. S. 405, 410 (2001); Abood, supra, at 222, 234–235. As Jefferson famously put it, “to compel a man to furnish contributions of money for the propagation of opinions which he disbelieves and abhor[s] is sinful and tyrannical.” A Bill for Establishing Religious Freedom, in 2 Papers of Thomas Jefferson 545 (J. Boyd ed. 1950) (emphasis deleted and footnote omitted); see also Hudson, 475 U. S., at 305, n. 15. We have therefore recognized that a “ ‘significant impingement on First Amendment rights’ ” occurs when public employees are required to provide financial support for a union that “takes many positions during collective bargaining that have powerful political and civic consequences.”&�Բ��;Knox, supra, at 310–311 (quoting Ellis v. Railway Clerks,466 U. S. 435, 455 (1984)).

Because the compelled subsidization of private speech seriously impinges on First Amendment rights, it cannot be casually allowed. Our free speech cases have identified “levels of scrutiny” to be applied in different contexts, and in three recent cases, we have considered the standard that should be used in judging the constitutionality of agency fees. See Knox, supra; Harris, supra; Friedrichs v. California Teachers Assn., 578 U. S. ___ (2016) (per cu- riam) (affirming decision below by equally divided Court).

In Knox, the first of these cases, we found it sufficient to hold that the conduct in question was unconstitutional under even the test used for the compulsory subsidization of commercial speech. 567 U. S., at 309–310, 321–322. Even though commercial speech has been thought to enjoy a lesser degree of protection, see, e.g., Central Hudson Gas & Elec. Corp. v. Public Serv. Comm’n of N. Y.,447 U. S. 557, 562–563 (1980), prior precedent in that area, specifically United Foods, supra, had applied what we characterized as “exacting” scrutiny, Knox, 567 U. S., at 310, a less demanding test than the “strict” scrutiny that might be thought to apply outside the commercial sphere. Under “exacting” scrutiny, we noted, a compelled subsidy must “serve a compelling state interest that cannot be achieved through means significantly less restrictive of associa- tional freedoms.”&�Բ��;Ibid. (internal quotation marks and altera- tions omitted).

In Harris, the second of these cases, we again found that an agency-fee requirement failed “exacting scrutiny.” 573 U. S., at ___ (slip op., at 33). But we questioned whether that test provides sufficient protection for free speech rights, since “it is apparent that the speech compelled” in agency-fee cases “is not commercial speech.”&�Բ��;Id., at ___ (slip op., at 30).

Picking up that cue, petitioner in the present case contends that the Illinois law at issue should be subjected to “strict scrutiny.” Brief for Petitioner 36. The dissent, on the other hand, proposes that we apply what amounts to rational-basis review, that is, that we ask only whether a government employer could reasonably believe that the exaction of agency fees serves its interests. See post, at 4 (Kagan, J., dissenting) (“A government entity could reasonably conclude that such a clause was needed”). This form of minimal scrutiny is foreign to our free-speech jurisprudence, and we reject it here. At the same time, we again find it unnecessary to decide the issue of strict scrutiny because the Illinois scheme cannot survive under even the more permissive standard applied in Knox and Harris.

In the remainder of this part of our opinion (Parts III–B and III–C), we will apply this standard to the justifications for agency fees adopted by the Court in Abood. Then, in Parts IV and V, we will turn to alternative rationales proffered by respondents and their amici.

B

In Abood, the main defense of the agency-fee arrangement was that it served the State’s interest in “labor peace,” 431 U. S., at 224. By “labor peace,” the Abood Court meant avoidance of the conflict and disruption that it envisioned would occur if the employees in a unit were represented by more than one union. In such a situation, the Court predicted, “inter-union rivalries” would foster “dissension within the work force,” and the employer could face “conflicting demands from different unions.”&�Բ��;Id., at 220–221. Confusion would ensue if the employer entered into and attempted to “enforce two or more agreements specifying different terms and conditions of employment.”&�Բ��;Id., at 220. And a settlement with one union would be “subject to attack from [a] rival labor organizatio[n].”&�Բ��;Id., at 221.

We assume that “labor peace,” in this sense of the term, is a compelling state interest, but Abood cited no evidence that the pandemonium it imagined would result if agency fees were not allowed, and it is now clear that Abood’s fears were unfounded. The Abood Court assumed that designation of a union as the exclusive representative of all the employees in a unit and the exaction of agency fees are inextricably linked, but that is simply not true. Harris, supra, at ___ (slip op., at 31).

The federal employment experience is illustrative. Under federal law, a union chosen by majority vote is designated as the exclusive representative of all the employees, but federal law does not permit agency fees. See 5 U. S. C. §§7102, 7111(a), 7114(a). Nevertheless, nearly a million federal employees—about 27% of the federal work force—are union members.[1] The situation in the Postal Service is similar. Although permitted to choose an exclusive representative, Postal Service employees are not required to pay an agency fee, 39 U. S. C. §§1203(a), 1209(c), and about 400,000 are union members.[2] Likewise, millions of public employees in the 28 States that have laws generally prohibiting agency fees are represented by unions that serve as the exclusive representatives of all the employees.[3] 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. Harris, supra, at ___ (slip op., at 30) (internal quotation marks omitted).

C

In addition to the promotion of “labor peace,”&�Բ��;Abood cited “the risk of ‘free riders’ ” as justification for agency fees, 431 U. S., at 224. Respondents and some of their amici endorse this reasoning, contending that agency fees are needed to prevent nonmembers from enjoying the benefits of union representation without shouldering the costs. Brief for Union Respondent 34–36; Brief for State Respondents 41–45; see, e.g., Brief for International Brotherhood of Teamsters as Amicus Curiae&�Բ��;3–5.

Petitioner strenuously objects to this free-rider label. He argues that he is not a free rider on a bus headed for a destination that he wishes to reach but is more like a person shanghaied for an unwanted voyage.

Whichever description fits the majority of public employees who would not subsidize a union if given the option, avoiding free riders is not a compelling interest. As we have noted, “free-rider arguments . . . are generally insufficient to overcome First Amendment objections.”&�Բ��;Knox, 567 U. S., at 311. To hold otherwise across the board would have startling consequences. Many private groups speak out with the objective of obtaining government action that will have the effect of benefiting nonmembers. May all those who are thought to benefit from such efforts be compelled to subsidize this speech?

Suppose that a particular group lobbies or speaks out on behalf of what it thinks are the needs of senior citizens or veterans or physicians, to take just a few examples. Could the government require that all seniors, veterans, or doctors pay for that service even if they object? It has never been thought that this is permissible. “[P]rivate speech often furthers the interests of nonspeakers,” but “that does not alone empower the state to compel the speech to be paid for.”&�Բ��;Lehnert v. Ferris Faculty Assn.,500 U. S. 507, 556 (1991) (Scalia, J., concurring in judgment in part and dissenting in part). In simple terms, the First Amendment does not permit the government to compel a person to pay for another party’s speech just because the government thinks that the speech furthers the interests of the person who does not want to pay.[4]

Those supporting agency fees contend that the situation here is different because unions are statutorily required to “represen[t] the interests of all public employees in the unit,” whether or not they are union members. §315/6(d); see, e.g., Brief for State Respondents 40–41, 45; post, at 7 (Kagan, J., dissenting). Why might this matter?

We can think of two possible arguments. It might be argued that a State has a compelling interest in requiring the payment of agency fees because (1) unions would otherwise be unwilling to represent nonmembers or (2) it would be fundamentally unfair to require unions to provide fair representation for nonmembers if nonmembers were not required to pay. Neither of these arguments is sound.

First, it is simply not true that unions will refuse to serve as the exclusive representative of all employees in the unit if they are not given agency fees. As noted, unions represent millions of public employees in jurisdictions that do not permit agency fees. No union is ever compelled to seek that designation. On the contrary, designation as exclusive representative is avidly sought.[5] Why is this so?

Even without agency fees, designation as the exclusive representative confers many benefits. As noted, that status gives the union a privileged place in negotiations over wages, benefits, and working conditions. See §315/6(c). Not only is the union given the exclusive right to speak for all the employees in collective bargaining, but the employer is required by state law to listen to and to bargain in good faith with only that union. §315/7. Designation as exclusive representative thus “results in a tremendous increase in the power” of the union. American Communications Assn. v. Douds,339 U. S. 382, 401 (1950).

In addition, a union designated as exclusive representative is often granted special privileges, such as obtaining information about employees, see §315/6(c), and having dues and fees deducted directly from employee wages, §§315/6(e)–(f ). The collective-bargaining agreement in this case guarantees a long list of additional privileges. See App. 138–143.

These benefits greatly outweigh any extra burden imposed by the duty of providing fair representation for nonmembers. What this duty entails, in simple terms, is an obligation not to “act solely in the interests of [the union’s] own members.” Brief for State Respondents 41; see Cintron v. AFSCME, Council 31, No. S–CB–16–032, p. 1, 34 PERI ¶105 (ILRB Dec. 13, 2017) (union may not intentionally direct “animosity” toward nonmembers based on their “dissident union practices”); accord, 14 Penn Plaza LLC v. Pyett,556 U. S. 247, 271 (2009); Vaca v. Sipes,386 U. S. 171, 177 (1967).

What does this mean when it comes to the negotiation of a contract? The union may not negotiate a collective-bargaining agreement that discriminates against nonmembers, see Steele v. Louisville & Nashville R. Co.,323 U. S. 192, 202–203 (1944), but the union’s bargaining latitude would be little different if state law simply prohibited public employers from entering into agreements that discriminate in that way. And for that matter, it is questionable whether the Constitution would permit a public-sector employer to adopt a collective-bargaining agreement that discriminates against nonmembers. See id., at 198–199, 202 (analogizing a private-sector union’s fair-representation duty to the duty “the Constitution imposes upon a legislature to give equal protection to the interests of those for whom it legislates”); cf. Rumsfeld v. Forum for Academic and Institutional Rights, Inc.,547 U. S. 47, 69 (2006) (recognizing that government may not “impose penalties or withhold benefits based on membership in a disfavored group” where doing so “ma[kes] group membership less attractive”). To the extent that an employer would be barred from acceding to a discriminatory agreement anyway, the union’s duty not to ask for one is superfluous. It is noteworthy that neither respondents nor any of the 39 amicus briefs supporting them—nor the dissent—has explained why the duty of fair representation causes public-sector unions to incur significantly greater expenses than they would otherwise bear in negotiating collective-bargaining agreements.

What about the representation of nonmembers in grievance proceedings? Unions do not undertake this activity solely for the benefit of nonmembers—which is why Illinois law gives a public-sector union the right to send a representative to such proceedings even if the employee declines union representation. §315/6(b). Representation of nonmembers furthers the union’s interest in keeping control of the administration of the collective-bargaining agreement, since the resolution of one employee’s grievance can affect others. And when a union controls the grievance process, it may, as a practical matter, effectively subordinate “the interests of [an] individual em- ployee . . . to the collective interests of all employees in the bargaining unit.”Alexander v. Gardner-Denver Co.,415 U. S. 36, 58, n. 19 (1974); see Stahulak v. Chicago, 184 Ill. 2d 176, 180–181, 703 N. E. 2d 44, 46–47 (1998); Mahoney v. Chicago, 293 Ill. App. 3d 69, 73–74, 687 N. E. 2d 132, 135–137 (1997) (union has “ ‘discretion to refuse to process’ ” a grievance, provided it does not act “arbitrar[ily]” or “in bad faith” (emphasis deleted)).

In any event, whatever unwanted burden is imposed by the representation of nonmembers in disciplinary matters can be eliminated “through means significantly less restrictive of associational freedoms” than the imposition of agency fees. Harris, 573 U. S., at ___ (slip op., at 30) (internal quotation marks omitted). Individual nonmembers could be required to pay for that service or could be denied union representation altogether.[6] Thus, agency fees cannot be sustained on the ground that unions would otherwise be unwilling to represent nonmembers.

Nor can such fees be justified on the ground that it would otherwise be unfair to require a union to bear the duty of fair representation. That duty is a necessary concomitant of the authority that a union seeks when it chooses to serve as the exclusive representative of all the employees in a unit. As explained, designating a union as the exclusive representative of nonmembers substantially restricts the nonmembers’ rights. Supra, at 2–3. Protection of their interests is placed in the hands of the union, and if the union were free to disregard or even work against those interests, these employees would be wholly unprotected. That is why we said many years ago that serious “constitutional questions [would] arise” if the union were not subject to the duty to represent all employees fairly. Steele, supra, at 198.

In sum, we do not see any reason to treat the free-rider interest any differently in the agency-fee context than in any other First Amendment context. See Knox, 567 U. S., at 311, 321. We therefore hold that agency fees cannot be upheld on free-rider grounds.

IV

Implicitly acknowledging the weakness of Abood’s own reasoning, proponents of agency fees have come forward with alternative justifications for the decision, and we now address these arguments.

A

The most surprising of these new arguments is the Union respondent’s originalist defense of Abood. According to this argument, Abood was correctly decided because the First Amendment was not originally understood to provide any protection for the free speech rights of public employees. Brief for Union Respondent 2–3, 17–20.

As an initial matter, we doubt that the Union—or its members—actually want us to hold that public employees have “no [free speech] rights.”&�Բ��;Id., at 1. Cf., e.g., Brief for National Treasury Employees Union as Amicus Curiae in Garcetti v. Ceballos, O. T. 2005, No. 04–473, p. 7 (arguing for “broa[d]” public-employee First Amendment rights); Brief for AFL–CIO as Amicus Curiae in No. 04–473 (similar).

It is particularly discordant to find this argument in a brief that trumpets the importance of stare decisis. See Brief for Union Respondent 47–57. Taking away free speech protection for public employees would mean overturning decades of landmark precedent. Under the Union’s theory, Pickering v. Board of Ed. of Township High School Dist. 205, Will Cty.,391 U. S. 563 (1968), and its progeny would fall. Yet Pickering, as we will discuss, is now the foundation for respondents’ chief defense of Abood. And indeed, Abood itself would have to go if public employees have no free speech rights, since Abood holds that the First Amendment prohibits the exaction of agency fees for political or ideological purposes. 431 U. S., at 234–235 (finding it “clear” that “a government may not require an individual to relinquish rights guaranteed him by the First Amendment as a condition of public employment”). Our political patronage cases would be doomed. See, e.g., Rutan v. Republican Party of Ill.,497 U. S. 62 (1990); Branti v. Finkel,445 U. S. 507 (1980); Elrod v. Burns,427 U. S. 347 (1976). Also imperiled would be older precedents like Wieman v. Updegraff,344 U. S. 183 (1952) (loyalty oaths), Shelton v. Tucker,364 U. S. 479 (1960) (disclosure of memberships and contributions), and Keyishian v. Board of Regents of Univ. of State of N. Y.,385 U. S. 589 (1967) (subversive speech). Respondents presumably want none of this, desiring instead that we apply the Constitution’s supposed original meaning only when it suits them—to retain the part of Abood that they like. See Tr. of Oral Arg. 56–57. We will not engage in this halfway originalism.

Nor, in any event, does the First Amendment’s original meaning support the Union’s claim. The Union offers no persuasive founding-era evidence that public employees were understood to lack free speech protections. While it observes that restrictions on federal employees’ activities have existed since the First Congress, most of its historical examples involved limitations on public officials’ outside business dealings, not on their speech. See Ex parte Curtis,106 U. S. 371, 372–373 (1882). The only early speech restrictions the Union identifies are an 1806 statute prohibiting military personnel from using “ ‘contemptuous or disrespectful words against the President’ ” and other officials, and an 1801 directive limiting electioneering by top government employees. Brief for Union Respondent 3. But those examples at most show that the government was understood to have power to limit employee speech that threatened important governmental interests (such as maintaining military discipline and preventing corruption)—not that public employees’ speech was entirely unprotected. Indeed, more recently this Court has upheld similar restrictions even while recognizing that government employees possess First Amendment rights. See, e.g.,Brown v. Glines,444 U. S. 348, 353 (1980) (upholding military restriction on speech that threatened troop readiness); Civil Service Comm’n v. Letter Carriers,413 U. S. 548, 556–557 (1973) (upholding limits on public employees’ political activities).

Ultimately, the Union relies, not on founding-era evidence, but on dictum from a 1983 opinion of this Court stating that, “[f]or most of th[e 20th] century, the unchallenged dogma was that a public employee had no right to object to conditions placed upon the terms of employment—including those which restricted the exercise of constitutional rights.”&�Բ��;Connick v. Myers,461 U. S. 138, 143; see Brief for Union Respondent 2, 17. Even on its own terms, this dictum about 20th-century views does not purport to describe how the First Amendment was understood in 1791. And a careful examination of the decisions by this Court that Connickcited to support its dictum, see 461 U. S., at 144, reveals that none of them rested on the facile premise that public employees are unprotected by the First Amendment. Instead, they considered (much as we do today) whether particular speech restrictions were “necessary to protect” fundamental government interests.Curtis, supra, at 374.

The Union has also failed to show that, even if public employees enjoyed free speech rights, the First Amendment was nonetheless originally understood to allow forced subsidies like those at issue here. We can safely say that, at the time of the adoption of the First Amendment, no one gave any thought to whether public-sector unions could charge nonmembers agency fees. Entities resembling labor unions did not exist at the founding, and public-sector unions did not emerge until the mid-20th century. The idea of public-sector unionization and agency fees would astound those who framed and ratified the Bill of Rights.[7] Thus, the Union cannot point to any accepted founding-era practice that even remotely resembles the compulsory assessment of agency fees from public-sector employees. We do know, however, that prominent members of the founding generation condemned laws requiring public employees to affirm or support beliefs with which they disagreed. As noted, Jefferson denounced compelled support for such beliefs as “ ‘sinful and tyrannical,’ ”&�Բ��;supra, at 9, and others expressed similar views.[8]

In short, the Union has offered no basis for concluding that Abood is supported by the original understanding of the First Amendment.

B

The principal defense of Abood advanced by respondents and the dissent is based on our decision in Pickering,391 U. S. 563, which held that a school district violated the First Amendment by firing a teacher for writing a letter critical of the school administration. Under Pickering and later cases in the same line, employee speech is largely unprotected if it is part of what the employee is paid to do, see Garcetti v. Ceballos,547 U. S. 410, 421–422 (2006), or if it involved a matter of only private concern, see Connick, supra, at 146–149. On the other hand, when a public employee speaks as a citizen on a matter of public concern, the employee’s speech is protected unless “ ‘the interest of the state, as an employer, in promoting the efficiency of the public services it performs through its employees’ outweighs ‘the interests of the [employee], as a citizen, in commenting upon matters of public concern.’ ”&�Բ��;Harris, 573 U. S., at ___ (slip op., at 35) (quoting Pickering, supra, at 568). Pickering was the centerpiece of the defense of Abood in Harris, see 573 U. S., at ___–___ (slip op., at 17–21) (Kagan, J., dissenting), and we found the argument unpersuasive, see id., at ___–___ (slip op., at 34–37). The intervening years have not improved its appeal.

As we pointed out in Harris, Abood was not based on Pickering. 573 U. S., at ___, and n. 26 (slip op., at 34, and n. 26). The Abood majority cited the case exactly once—in a footnote—and then merely to acknowledge that “there may be limits on the extent to which an employee in a sensitive or policymaking position may freely criticize his superiors and the policies they espouse.” 431 U. S., at 230, n. 27. That aside has no bearing on the agency-fee issue here.[9]

Respondents’ reliance on Pickering is thus “an effort to find a new justification for the decision in Abood.”Harris, supra, at ___ (slip op., at 34). And we have previously taken a dim view of similar attempts to recast problematic First Amendment decisions. See, e.g., Citizens United v. Federal Election Comm’n,558 U. S. 310, 348–349, 363 (2010) (rejecting efforts to recast Austin v. Michigan Chamber of Commerce,494 U. S. 652 (1990)); see also Citizens United, supra, at 382–385 (Roberts, C. J., concurring). We see no good reason, at this late date, to try to shoehorn Abood into the Pickering framework.

Even if that were attempted, the shoe would be a painful fit for at least three reasons.

First, the Pickering framework was developed for use in a very different context—in cases that involve “one employee’s speech and its impact on that employee’s public responsibilities.”&�Բ��;United States v. Treasury Employees,513 U. S. 454, 467 (1995). This case, by contrast, involves a blanket requirement that all employees subsidize speech with which they may not agree. While we have sometimes looked to Pickering in considering general rules that affect broad categories of employees, we have acknowledged that the standard Pickering analysis requires modification in that situation. See 513 U. S., at 466–468, and n. 11. A speech-restrictive law with “widespread impact,” we have said, “gives rise to far more serious concerns than could any single supervisory decision.”&�Բ��;Id., at 468. Therefore, when such a law is at issue, the government must shoulder a correspondingly “heav[ier]” burden, id., at 466, and is entitled to considerably less deference in its assessment that a predicted harm justifies a particular impingement on First Amendment rights, see id., at 475–476, n. 21; accord, id., at 482–483 (O’Connor, J., concurring in judgment in part and dissenting in part). The end product of those adjustments is a test that more closely resembles exacting scrutiny than the traditional Pickering analysis.

The core collective-bargaining issue of wages and benefits illustrates this point. Suppose that a single employee complains that he or she should have received a 5% raise. This individual complaint would likely constitute a matter of only private concern and would therefore be unprotected under Pickering. But a public-sector union’s demand for a 5% raise for the many thousands of employees it represents would be another matter entirely. Granting such a raise could have a serious impact on the budget of the government unit in question, and by the same token, denying a raise might have a significant effect on the performance of government services. When a large number of employees speak through their union, the category of speech that is of public concern is greatly enlarged, and the category of speech that is of only private concern is substantially shrunk. By disputing this, post, at 13–14, the dissent denies the obvious.

Second, the Pickering framework fits much less well where the government compels speech or speech subsidies in support of third parties. Pickering is based on the insight that the speech of a public-sector employee may interfere with the effective operation of a government office. When a public employer does not simply restrict potentially disruptive speech but commands that its employees mouth a message on its own behalf, the calculus is very different. Of course, if the speech in question is part of an employee’s official duties, the employer may insist that the employee deliver any lawful message. See Garcetti, 547 U. S., at 421–422, 425–426. Otherwise, however, it is not easy to imagine a situation in which a public employer has a legitimate need to demand that its employees recite words with which they disagree. And we have never applied Pickering in such a case.

Consider our decision in Connick. In that case, we held that an assistant district attorney’s complaints about the supervisors in her office were, for the most part, matters of only private concern. 461 U. S., at 148. As a result, we held, the district attorney could fire her for making those comments. Id., at 154. Now, suppose that the assistant had not made any critical comments about the supervisors but that the district attorney, out of the blue, demanded that she circulate a memo praising the supervisors. Would her refusal to go along still be a matter of purely private concern? And if not, would the order be justified on the ground that the effective operation of the office demanded that the assistant voice complimentary sentiments with which she disagreed? If Pickering applies at all to compelled speech—a question that we do not decide—it would certainly require adjustment in that context.

Third, although both Pickering and Abood divided speech into two categories, the cases’ categorization schemes do not line up. Superimposing the Pickering scheme on Abood would significantly change the Aboodregime.

Let us first look at speech that is not germane to collective bargaining but instead concerns political or ideological issues. Under Abood, a public employer is flatly prohibited from permitting nonmembers to be charged for this speech, but under Pickering, the employees’ free speech interests could be overcome if a court found that the employer’s interests outweighed the employees’.

A similar problem arises with respect to speech that is germane to collective bargaining. The parties dispute how much of this speech is of public concern, but respondents concede that much of it falls squarely into that category. See Tr. of Oral Arg. 47, 65. Under Abood, nonmembers may be required to pay for all this speech, but Pickering would permit that practice only if the employer’s interests outweighed those of the employees. Thus, recasting Abood as an application of Pickering would substantially alter the Abood scheme.

For all these reasons, Pickering is a poor fit indeed.

V

Even if we were to apply some form of Pickering, Illinois’ agency-fee arrangement would not survive.

A

Respondents begin by suggesting that union speech in collective-bargaining and grievance proceedings should be treated like the employee speech in Garcetti, i.e., as speech “pursuant to [an employee’s] official duties,” 547 U. S., at 421. Many employees, in both the public and private sectors, are paid to write or speak for the purpose of furthering the interests of their employers. There are laws that protect public employees from being compelled to say things that they reasonably believe to be untrue or improper, see id., at 425–426, but in general when public employees are performing their job duties, their speech may be controlled by their employer. Trying to fit union speech into this framework, respondents now suggest that the union speech funded by agency fees forms part of the official duties of the union officers who engage in the speech. Brief for Union Respondent 22–23; see Brief for State Respondents 23–24.

This argument distorts collective bargaining and grievance adjustment beyond recognition. When an employee engages in speech that is part of the employee’s job duties, the employee’s words are really the words of the employer. The employee is effectively the employer’s spokesperson. But when a union negotiates with the employer or represents employees in disciplinary proceedings, the union speaks for the employees, not the employer. Otherwise, the employer would be negotiating with itself and disputing its own actions. That is not what anybody understands to be happening.

What is more, if the union’s speech is really the employer’s speech, then the employer could dictate what the union says. Unions, we trust, would be appalled by such a suggestion. For these reasons, Garcetti is totally inapposite here.

B

Since the union speech paid for by agency fees is not controlled by Garcetti, we move on to the next step of the Pickering framework and ask whether the speech is on a matter of public or only private concern. In Harris, the dissent’s central argument in defense of Abood was that union speech in collective bargaining, including speech about wages and benefits, is basically a matter of only private interest. See 573 U. S., at ___–___ (slip op., at 19–20) (Kagan, J., dissenting). We squarely rejected that argument, see id., at ___–___ (slip op., at 35–36), and the facts of the present case substantiate what we said at that time: “[I]t is impossible to argue that the level of . . . state spending for employee benefits . . . is not a matter of great public concern,”&�Բ��;id., at ___ (slip op., at 36).

Illinois, like some other States and a number of counties and cities around the country, suffers from severe budget problems.[10] As of 2013, Illinois had nearly $160 billion in unfunded pension and retiree healthcare liabilities.[11] By 2017, that number had only grown, and the State was grappling with $15 billion in unpaid bills.[12] We are told that a “quarter of the budget is now devoted to paying down” those liabilities.[13] These problems and others led Moody’s and S&P to downgrade Illinois’ credit rating to “one step above junk”—the “lowest ranking on record for a U. S. state.”[14]

The Governor, on one side, and public-sector unions, on the other, disagree sharply about what to do about these problems. The State claims that its employment-related debt is “ ‘squeezing core programs in education, public safety, and human services, in addition to limiting [the State’s] ability to pay [its] bills.’ ” Securities Act of 1933 Release No. 9389, 105 S. E. C. Docket 3381 (2013). It therefore “told the Union that it would attempt to address th[e financial] crisis, at least in part, through collective bargaining.” Board Decision 12–13. And “the State’s desire for savings” in fact “dr[o]ve [its] bargaining” positions on matters such as health-insurance benefits and holiday, overtime, and promotion policies. Id., at 13; Illinois Dept. of Central Management Servs. v. AFSCME, Council 31, No. S–CB–16–17 etc., 33 PERI ¶67 (ILRB Dec. 13, 2016) (ALJ Decision), pp. 26–28, 63–66, 224. But when the State offered cost-saving proposals on these issues, the Union countered with very different suggestions. Among other things, it advocated wage and tax increases, cutting spending “to Wall Street financial institutions,” and reforms to Illinois’ pension and tax systems (such as closing “corporate tax loopholes,” “[e]xpanding the base of the state sales tax,” and “allowing an income tax that is adjusted in accordance with ability to pay”). Id., at 27–28. To suggest that speech on such matters is not of great public concern—or that it is not directed at the “public square,”&�Բ��;post, at 16 (Kagan, J., dissenting)—is to deny reality.

In addition to affecting how public money is spent, union speech in collective bargaining addresses many other important matters. As the examples offered by respondents’ own amici show, unions express views on a wide range of subjects—education, child welfare, healthcare, and minority rights, to name a few. See, e.g., Brief for American Federation of Teachers as Amicus Curiae 15–27; Brief for Child Protective Service Workers et al. as Amici Curiae 5–13; Brief for Human Rights Campaign et al. as Amici Curiae 10–17; Brief for National Women’s Law Center et al. as Amici Curiae 14–30. What unions have to say on these matters in the context of collective bargaining is of great public importance.

Take the example of education, which was the focus of briefing and argument in Friedrichs. The public importance of subsidized union speech is especially apparent in this field, since educators make up by far the largest category of state and local government employees, and education is typically the largest component of state and local government expenditures.[15]

Speech in this area also touches on fundamental questions of education policy. Should teacher pay be based on seniority, the better to retain experienced teachers? Or should schools adopt merit-pay systems to encourage teachers to get the best results out of their students?[16] Should districts transfer more experienced teachers to the lower performing schools that may have the greatest need for their skills, or should those teachers be allowed to stay where they have put down roots?[17] Should teachers be given tenure protection and, if so, under what conditions? On what grounds and pursuant to what procedures should teachers be subject to discipline or dismissal? How should teacher performance and student progress be measured—by standardized tests or other means?

Unions can also speak out in collective bargaining on controversial subjects such as climate change,[18] the Confederacy,[19] sexual orientation and gender identity,[20] evolution,[21] and minority religions.[22] These are sensitive political topics, and they are undoubtedly matters of profound “ ‘value and concern to the public.’ ”&�Բ��;Snyder v. Phelps,562 U. S. 443, 453 (2011). We have often recognized that such speech “ ‘occupies the highest rung of the hierarchy of First Amendment values’ ” and merits “ ‘special protection.’ ”&�Բ��;Id., at 452.

What does the dissent say about the prevalence of such issues? The most that it is willing to admit is that “some” issues that arise in collective bargaining “raise important non-budgetary disputes.”&�Բ��;Post, at 17. Here again, the dissent refuses to recognize what actually occurs in public-sector collective bargaining.

Even union speech in the handling of grievances may be of substantial public importance and may be directed at the “public square.”&�Բ��;Post, at 16. For instance, the Union respondent in this case recently filed a grievance seeking to compel Illinois to appropriate $75 million to fund a 2% wage increase. State v. AFSCME Council 31, 2016 IL 118422, 51 N. E. 3d 738, 740–742, and n. 4. In short, the union speech at issue in this case is overwhelmingly of substantial public concern.

C

The only remaining question under Pickering is whether the State’s proffered interests justify the heavy burden that agency fees inflict on nonmembers’ First Amendment interests. We have already addressed the state interests asserted in Abood—promoting “labor peace” and avoiding free riders, see supra, at 11–18—and we will not repeat that analysis.

In Harris and this case, defenders of Abood have as- serted a different state interest—in the words of the Harris dissent, the State’s “interest in bargaining with an adequately funded exclusive bargaining agent.” 573 U. S., at ___ (Kagan, J., dissenting) (slip op., at 7); see also post, at 6–7 (Kagan, J., dissenting). This was not “the interest Abood recognized and protected,”&�Բ��;Harris, supra, at ___ (slip op., at 7) (Kagan, J., dissenting), and, in any event, it is insufficient.

Although the dissent would accept without any serious independent evaluation the State’s assertion that the absence of agency fees would cripple public-sector unions and thus impair the efficiency of government operations, see post, at 8–9, 11, ample experience, as we have noted, supra, at 12, shows that this is questionable.

Especially in light of the more rigorous form of Pickering analysis that would apply in this context, see supra, at 23–25, the balance tips decisively in favor of the employees’ free speech rights.[23]

We readily acknowledge, as Pickering did, that “the State has interests as an employer in regulating the speech of its employees that differ significantly from those it possesses in connection with regulation of the speech of the citizenry in general.” 391 U. S., at 568. Our analysis is consistent with that principle. The exacting scrutiny standard we apply in this case was developed in the context of commercial speech, another area where the government has traditionally enjoyed greater-than-usual power to regulate speech. See supra, at 10. It is also not disputed that the State may require that a union serve as exclusive bargaining agent for its employees—itself a significant impingement on associational freedoms that would not be tolerated in other contexts. We simply draw the line at allowing the government to go further still and require all employees to support the union irrespective of whether they share its views. Nothing in the Pickering line of cases requires us to uphold every speech restriction the government imposes as an employer. See Pickering, supra, at 564–566 (holding teacher’s dismissal for criticizing school board unconstitutional); Rankin v. McPherson,483 U. S. 378, 392 (1987) (holding clerical employ- ee’s dismissal for supporting assassination attempt on President unconstitutional); Treasury Employees, 513 U. S., at 477 (holding federal-employee honoraria ban unconstitutional).

VI

For the reasons given above, we conclude that public-sector agency-shop arrangements violate the First Amendment, and Abood erred in concluding otherwise. There remains the question whether stare decisisnonetheless counsels against overruling Abood. It does not.

“Stare decisis is the preferred course because it promotes the evenhanded, predictable, and consistent development of legal principles, fosters reliance on judicial decisions, and contributes to the actual and perceived integrity of the judicial process.”&�Բ��;Payne v. Tennessee,501 U. S. 808, 827 (1991). We will not overturn a past decision unless there are strong grounds for doing so. United States v. International Business Machines Corp.,517 U. S. 843, 855–856 (1996); Citizens United, 558 U. S., at 377 (Roberts, C. J., concurring). But as we have often recognized, stare decisis is “ ‘not an inexorable command.’ ”&�Բ��;Pearson v. Callahan,555 U. S. 223, 233 (2009); see also Lawrence v. Texas,539 U. S. 558, 577 (2003); State Oil Co. v. Khan,522 U. S. 3, 20 (1997); Agostiniv. Felton,521 U. S. 203, 235 (1997); Seminole Tribe of Fla. v. Florida,517 U. S. 44, 63 (1996); Payne, supra, at 828.

The doctrine “is at its weakest when we interpret the Constitution because our interpretation can be altered only by constitutional amendment or by overruling our prior decisions.”&�Բ��;Agostini, supra, at 235. And stare decisis applies with perhaps least force of all to decisions that wrongly denied First Amendment rights: “This Court has not hesitated to overrule decisions offensive to the First Amendment (a fixed star in our constitutional constellation, if there is one).”&�Բ��;Federal Election Comm’n v. Wisconsin Right to Life, Inc.,551 U. S. 449, 500 (2007) (Scalia, J., concurring in part and concurring in judgment) (internal quotation marks omitted); see also Citizens United, supra, at 362–365 (overruling Austin,494 U. S. 652); Barnette, 319 U. S., at 642 (overruling Minersville School Dist. v. Gobitis,310 U. S. 586 (1940)).

Our cases identify factors that should be taken into account in deciding whether to overrule a past decision. Five of these are most important here: the quality of Abood’s reasoning, the workability of the rule it established, its consistency with other related decisions, developments since the decision was handed down, and reliance on the decision. After analyzing these factors, we conclude that stare decisis does not require us to retain Abood.

A

An important factor in determining whether a precedent should be overruled is the quality of its reasoning, see Citizens United, 558 U. S., at 363–364; id., at 382–385 (Roberts, C. J., concurring); Lawrence, 539 U. S., at 577–578, and as we explained in Harris, Abood was poorly reasoned, see 573 U. S., at ___–___ (slip op., at 17–20). We will summarize, but not repeat, Harris’s lengthy discussion of the issue.

Abood went wrong at the start when it concluded that two prior decisions, Railway Employes v. Hanson,351 U. S. 225 (1956), and Machinists v. Street,367 U. S. 740 (1961), “appear[ed] to require validation of the agency-shop agreement before [the Court].” 431 U. S., at 226. Properly understood, those decisions did no such thing. Both cases involved Congress’s “bare authorization” of private-sector union shops under the Railway Labor Act. Street, supra, at 749 (emphasis added).[24] Abood failed to appreciate that a very different First Amendment question arises when a State requires its employees to pay agency fees. See Harris, supra, at ___ (slip op., at 17).

Moreover, neither Hanson nor Street gave careful consideration to the First Amendment. In Hanson, the primary questions were whether Congress exceeded its power under the Commerce Clause or violated substantive due process by authorizing private union-shop arrangements under the Commerce and Due Process Clauses. 351 U. S., at 233–235. After deciding those questions, the Court summarily dismissed what was essentially a facial First Amendment challenge, noting that the record did not substantiate the challengers’ claim. Id., at 238; see Harris, supra, at ___ (slip op., at 17). For its part, Street was decided as a matter of statutory construction, and so did not reach any constitutional issue. 367 U. S., at 749–750, 768–769. Abood nevertheless took the view that Hanson and Street “all but decided” the important free speech issue that was before the Court. Harris, 573 U. S., at ___ (slip op., at 17). As we said in Harris, “[s]urely a First Amendment issue of this importance deserved better treatment.”&�Բ��;Ibid.

Abood’s unwarranted reliance on Hanson and Street appears to have contributed to another mistake: Abood judged the constitutionality of public-sector agency fees under a deferential standard that finds no support in our free speech cases. (As noted, supra, at 10–11, today’s dissent makes the same fundamental mistake.)Abood did not independently evaluate the strength of the government interests that were said to support the challenged agency-fee provision; nor did it ask how well that provision actually promoted those interests or whether they could have been adequately served without impinging so heavily on the free speech rights of nonmembers. Rather, Abood followed Hanson and Street, which it interpreted as having deferred to “the legislative assessment of the important contribution of the union shop to the system of labor relations established by Congress.” 431 U. S., at 222 (emphasis added). But Hanson deferred to that judgment in deciding the Commerce Clause and substantive due process questions that were the focus of the case. Such deference to legislative judgments is inappropriate in deciding free speech issues.

If Abood had considered whether agency fees were actually needed to serve the asserted state interests, it might not have made the serious mistake of assuming that one of those interests—“labor peace”—demanded, not only that a single union be designated as the exclusive representative of all the employees in the relevant unit, but also that nonmembers be required to pay agency fees. Deferring to a perceived legislative judgment, Abood failed to see that the designation of a union as exclusive representative and the imposition of agency fees are not inextricably linked. See supra, at 11–12; Harris, supra, at ___ (slip op., at 31).

Abood also did not sufficiently take into account the difference between the effects of agency fees in public- and private-sector collective bargaining. The challengers in Abood argued that collective bargaining with a government employer, unlike collective bargaining in the private sector, involves “inherently ‘political’ ” speech. 431 U. S., at 226. The Court did not dispute that characterization, and in fact conceded that “decisionmaking by a public employer is above all a political process” driven more by policy concerns than economic ones. Id., at 228; see id., at 228–231. But (again invoking Hanson), the Abood Court asserted that public employees do not have “weightier First Amendment interest[s]” against compelled speech than do private employees. Id., at 229. That missed the point. Assuming for the sake of argument that the First Amendment applies at all to private-sector agency-shop arrangements, the individual interests at stake still differ. “In the public sector, core issues such as wages, pensions, and benefits are important political issues, but that is generally not so in the private sector.”&�Բ��;Harris, 573 U. S., at ___ (slip op., at 17).

Overlooking the importance of this distinction, “Abood failed to appreciate the conceptual difficulty of distinguishing in public-sector cases between union expenditures that are made for collective-bargaining purposes and those that are made to achieve political ends.”&�Բ��;Id., at ___ (slip op., at 18). Likewise, “Abood does not seem to have anticipated the magnitude of the practical administrative problems that would result in attempting to classify public-sector union expenditures as either ‘chargeable’ . . . or nonchargeable.”&�Բ��;Ibid.Nor did Abood “foresee the practical problems that would face objecting nonmembers.”&�Բ��;Id., at ___ (slip op., at 19).

In sum, as detailed in Harris, Abood was not well reasoned.[25]

B

Another relevant consideration in the stare decisis calculus is the workability of the precedent in question, Montejo v. Louisiana,556 U. S. 778, 792 (2009), and that factor also weighs against Abood.

Abood’s line between chargeable and nonchargeable union expenditures has proved to be impossible to draw with precision. We tried to give the line some definition in Lehnert. There, a majority of the Court adopted a three-part test requiring that chargeable expenses (1) be “ ‘germane’ ” to collective bargaining, (2) be “justified” by the government’s labor-peace and free-rider interests, and (3) not add “significantly” to the burden on free speech, 500 U. S., at 519, but the Court splintered over the application of this test, see id., at 519–522 (plurality opinion); id., at 533–534 (Marshall, J., concurring in part and dissenting in part). That division was not surprising. As the Lehnert dissenters aptly observed, each part of the majority’s test “involves a substantial judgment call,”&�Բ��;id., at 551 (opinion of Scalia, J.), rendering the test “altogether malleable” and “no[t] principled,”&�Բ��;id., at 563 (Kennedy, J., concurring in judgment in part and dissenting in part).

Justice Scalia presciently warned that Lehnert’s amorphous standard would invite “perpetua[l] give-it-a-try litigation,”&�Բ��;id., at 551, and the Court’s experience with union lobbying expenses illustrates the point. The Lehnert plurality held that money spent on lobbying for increased education funding was not chargeable. Id., at 519–522. But Justice Marshall—applying the same three-prong test—reached precisely the opposite conclusion. Id., at 533–542. And Lehnert failed to settle the matter; States and unions have continued to “give it a try” ever since.

In Knox, for example, we confronted a union’s claim that the costs of lobbying the legislature and the electorate about a ballot measure were chargeable expenses under Lehnert. See Brief for Respondent in Knox v. Service Employees, O. T. 2011, No. 10–1121, pp. 48–53. The Court rejected this claim out of hand, 567 U. S., at 320–321, but the dissent refused to do so, id., at 336 (opinion of Breyer, J.). And in the present case, nonmembers are required to pay for unspecified “[l]obbying” expenses and for “[s]ervices” that “may ultimately inure to the benefit of the members of the local bargaining unit.” App. to Pet. for Cert. 31a–32a. That formulation is broad enough to encompass just about anything that the union might choose to do.

Respondents agree that Abood’s chargeable-nonchargeable line suffers from “a vagueness problem,” that it sometimes “allows what it shouldn’t allow,” and that “a firm[er] line c[ould] be drawn.” Tr. of Oral Arg. 47–48. They therefore argue that we should “consider revisiting” this part of Abood. Tr. of Oral Arg. 66; see Brief for Union Respondent 46–47; Brief for State Respondents 30. This concession only underscores the real- ity that Abood has proved unworkable: Not even the parties defending agency fees support the line that it has taken this Court over 40 years to draw.

Objecting employees also face a daunting and expensive task if they wish to challenge union chargeability determinations. While Hudson requires a union to provide nonmembers with “sufficient information to gauge the propriety of the union’s fee,” 475 U. S., at 306, the Hudson notice in the present case and in others that have come before us do not begin to permit a nonmember to make such a determination.

In this case, the notice lists categories of expenses and sets out the amount in each category that is said to be attributable to chargeable and nonchargeable expenses. Here are some examples regarding the Union respondent’s expenditures:

See App. to Pet. for Cert. 35a–36a.

How could any nonmember determine whether these numbers are even close to the mark without launching a legal challenge and retaining the services of attorneys and accountants? Indeed, even with such services, it would be a laborious and difficult task to check these figures.[26]

The Union respondent argues that challenging its chargeability determinations is not burdensome because the Union pays for the costs of arbitration, see Brief for Union Respondent 10–11, but objectors must still pay for the attorneys and experts needed to mount a serious challenge. And the attorney’s fees incurred in such a proceeding can be substantial. See, e.g., Knox v. Chiang, 2013 WL 2434606, *15 (ED Cal., June 5, 2013) (attorney’s fees in Knox exceeded $1 million). The Union respondent’s suggestion that an objector could obtain adequate review without even showing up at an arbitration, see App. to Pet. for Cert. 40a–41a, is therefore farfetched.

C

Developments since Abood, both factual and legal, have also “eroded” the decision’s “underpinnings” and left it an outlier among our First Amendment cases. United States v. Gaudin,515 U. S. 506, 521 (1995).

Abood pinned its result on the “unsupported empirical assumption” that “the principle of exclusive representation in the public sector is dependent on a union or agency shop.”&�Բ��;Harris, 573 U. S., at ___ (slip op., at 20); Abood, 431 U. S., at 220–222. But, as already noted, experience has shown otherwise. See supra, at 11–12.

It is also significant that the Court decided Abood against a very different legal and economic backdrop. Public-sector unionism was a relatively new phenomenon in 1977. The first State to permit collective bargaining by government employees was Wisconsin in 1959, R. Kearney & P. Mareschal, Labor Relations in the Public Sector 64 (5th ed. 2014), and public-sector union membership remained relatively low until a “spurt” in the late 1960’s and early 1970’s, shortly before Abood was decided, Freeman, Unionism Comes to the Public Sector, 24 J. Econ. Lit. 41, 45 (1986). Since then, public-sector union membership has come to surpass private-sector union membership, even though there are nearly four times as many total private-sector employees as public-sector employees. B. Hirsch & D. Macpherson, Union Membership and Earnings Data Book 9–10, 12, 16 (2013 ed.).

This ascendance of public-sector unions has been marked by a parallel increase in public spending. In 1970, total state and local government expenditures amounted to $646 per capita in nominal terms, or about $4,000 per capita in 2014 dollars. See Dept. of Commerce, Statistical Abstract of the United States: 1972, p. 419; CPI Inflation Calculator, BLS, http://data.bls.gov/cgi-bin/cpicalc.pl. By 2014, that figure had ballooned to approximately $10,238 per capita. ProQuest, Statistical Abstract of the United States: 2018, pp. 17, Table 14, 300, Table 469. Not all that increase can be attributed to public-sector unions, of course, but the mounting costs of public-employee wages, benefits, and pensions undoubtedly played a substantial role. We are told, for example, that Illinois’ pension funds are underfunded by $129 billion as a result of generous public-employee retirement packages. Brief for Jason R. Barclay et al. as Amici Curiae 9, 14. Unsustainable collective-bargaining agreements have also been blamed for multiple municipal bankruptcies. See Brief for State of Michigan et al. as Amici Curiae 10–19. These developments, and the political debate over public spending and debt they have spurred, have given collective-bargaining issues a political valence that Abooddid not fully appreciate.

Abood is also an “anomaly” in our First Amendment jurisprudence, as we recognized in Harris and Knox.Harris, supra, at ___ (slip op., at 8); Knox, 567 U. S., at 311. This is not an altogether new observation. In Abood itself, Justice Powell faulted the Court for failing to perform the “ ‘exacting scrutiny’ ” applied in other cases involving significant impingements on First Amendment rights. 431 U. S., at 259; see id., at 259–260, and n. 14. Our later cases involving compelled speech and association have also employed exacting scrutiny, if not a more demanding standard. See, e.g., Roberts, 468 U. S., at 623; United Foods, 533 U. S., at 414. And we have more recently refused, even in agency-fee cases, to extend Abood beyond circumstances where it directly controls. See Knox, supra, at 314; Harris, supra, at ___–___ (slip op., at 28–29).

Abood particularly sticks out when viewed against our

cases holding that public employees generally may not be required to support a political party. See Elrod,427 U. S. 347; Branti,445 U. S. 507; Rutan,497 U. S. 62; O’Hare Truck Service, Inc. v. City of Northlake,518 U. S. 712 (1996). The Court reached that conclusion despite a “long tradition” of political patronage in government.Rutan, supra, at 95 (Scalia, J., dissenting); see also Elrod, 427 U. S., at 353 (plurality opinion); id., at 377–378 (Powell, J., dissenting). It is an odd feature of our First Amendment cases that political patronage has been deemed largely unconstitutional, while forced subsidization of union speech (which has no such pedigree) has been largely permitted. As Justice Powell observed: “I am at a loss to understand why the State’s decision to adopt the agency shop in the public sector should be worthy of greater deference, when challenged on First Amendment grounds, than its decision to adhere to the tradition of political patronage.”Abood, supra, at 260, n. 14 (opinion concurring in judgment) (citing Elrod, supra, at 376–380, 382–387 (Powell, J., dissenting); emphasis added). We have no occasion here to reconsider our political patronage decisions, but Justice Powell’s observation is sound as far as it goes. By overruling Abood, we end the oddity of privileging compelled union support over compelled party support and bring a measure of greater coherence to our First Amendment law.

D

In some cases, reliance provides a strong reason for adhering to established law, see, e.g., Hilton v. South Carolina Public Railways Comm’n,502 U. S. 197, 202–203 (1991), and this is the factor that is stressed most strongly by respondents, their amici, and the dissent. They contend that collective-bargaining agreements now in effect were negotiated with agency fees in mind and that unions may have given up other benefits in exchange for provisions granting them such fees. Tr. of Oral Arg. 67–68; see Brief for State Respondents 54; Brief for Union Respondent 50; post, at 22–26 (Kagan, J., dissenting). In this case, however, reliance does not carry decisive weight.

For one thing, it would be unconscionable to permit free speech rights to be abridged in perpetuity in order to preserve contract provisions that will expire on their own in a few years’ time. “The fact that [public-sector unions] may view [agency fees] as an entitlement does not establish the sort of reliance interest that could outweigh the countervailing interest that [nonmembers] share in having their constitutional rights fully protected.”&�Բ��;Arizona v. Gant,556 U. S. 332, 349 (2009).

For another, Abood does not provide “a clear or easily applicable standard, so arguments for reliance based on its clarity are misplaced.”&�Բ��;South Dakota v. Wayfair, Inc., ante, at 20; see supra, at 38–41.

This is especially so because public-sector unions have been on notice for years regarding this Court’s misgivings about Abood. In Knox, decided in 2012, we described Abood as a First Amendment “anomaly.” 567 U. S., at 311. Two years later in Harris, we were asked to overrule Abood, and while we found it unnecessary to take that step, we cataloged Abood’s many weaknesses. In 2015, we granted a petition for certiorari asking us to review a decision that sustained an agency-fee arrangement under Abood. Friedrichs v. California Teachers Assn., 576 U. S. ___. After exhaustive briefing and argument on the question whether Abood should be overruled, we affirmed the decision below by an equally divided vote. 578 U. S. ___ (2016) (per curiam). During this period of time, any public-sector union seeking an agency-fee provision in a collective-bargaining agreement must have understood that the constitutionality of such a provision was uncertain.

That is certainly true with respect to the collective-bargaining agreement in the present case. That agreement initially ran from July 1, 2012, until June 30, 2015. App. 331. Since then, the agreement has been extended pursuant to a provision providing for automatic renewal for an additional year unless either party gives timely notice that it desires to amend or terminate the contract. Ibid. Thus, for the past three years, the Union could not have been confident about the continuation of the agency-fee arrangement for more than a year at a time.

Because public-sector collective-bargaining agreements are generally of rather short duration, a great many of those now in effect probably began or were renewed since Knox (2012) or Harris (2014). But even if an agreement antedates those decisions, the union was able to protect itself if an agency-fee provision was essential to the overall bargain. A union’s attorneys undoubtedly understand that if one provision of a collective-bargaining agreement is found to be unlawful, the remaining provisions are likely to remain in effect. See NLRB v. Rockaway News Supply Co.,345 U. S. 71, 76–79 (1953); see also 8 R. Lord, Williston on Contracts §19:70 (4th ed. 2010). Any union believing that an agency-fee provision was essential to its bargain could have insisted on a provision giving it greater protection. The agreement in the present case, by contrast, provides expressly that the invalidation of any part of the agreement “shall not invalidate the remaining portions,” which “shall remain in full force and effect.” App. 328. Such severability clauses ensure that “entire contracts” are not “br[ought] down” by today’s ruling. Post, at 23, n. 5 (Kagan, J., dissenting).

In short, the uncertain status of Abood, the lack of clarity it provides, the short-term nature of collective-bargaining agreements, and the ability of unions to protect themselves if an agency-fee provision was crucial to its bargain all work to undermine the force of reliance as a factor supporting Abood.[27]

* * *

We recognize that the loss of payments from nonmembers may cause unions to experience unpleasant transition costs in the short term, and may require unions to make adjustments in order to attract and retain members. But we must weigh these disadvantages against the considerable windfall that unions have received under Abood for the past 41 years. It is hard to estimate how many billions of dollars have been taken from nonmembers and transferred to public-sector unions in violation of the First Amendment. Those unconstitutional exactions cannot be allowed to continue indefinitely.

All these reasons—that Abood’s proponents have abandoned its reasoning, that the precedent has proved unworkable, that it conflicts with other First Amendment decisions, and that subsequent developments have eroded its underpinnings—provide the “ ‘special justification[s]’ ” for overruling Abood. Post, at 19 (Kagan, J., dissenting) (quoting Kimble v. Marvel Entertainment, LLC, 576 U. S. ___, ___ (2015) (slip op., at 8)).[28]

VII

For these reasons, States and public-sector unions may no longer extract agency fees from nonconsenting employees. Under Illinois law, if a public-sector collective-bargaining agreement includes an agency-fee provision and the union certifies to the employer the amount of the fee, that amount is automatically deducted from the nonmember’s wages. §315/6(e). No form of employee consent is required.

This procedure violates the First Amendment and cannot continue. Neither an agency fee nor any other payment to the union may be deducted from a nonmember’s wages, nor may any other attempt be made to collect such a payment, unless the employee affirmatively consents to pay. By agreeing to pay, nonmembers are waiving their First Amendment rights, and such a waiver cannot be presumed. Johnson v. Zerbst,304 U. S. 458, 464 (1938); see also Knox, 567 U. S., at 312–313. Rather, to be effective, the waiver must be freely given and shown by “clear and compelling” evidence. Curtis Publishing Co. v. Butts,388 U. S. 130, 145 (1967) (plurality opinion); see also College Savings Bank v. Florida Prepaid Postsecondary Ed. Expense Bd.,527 U. S. 666, 680–682 (1999). Unless employees clearly and affirmatively consent before any money is taken from them, this standard cannot be met.

* * *

Abood was wrongly decided and is now overruled. The judgment of the United States Court of Appeals for the Seventh Circuit is reversed, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

Notes

1 See Bureau of Labor Statistics (BLS), Labor Force Statistics From the Current Population Survey (Table 42) (2017), https://www.bls.gov/cps/tables.htm (all Internet materials as visited June 26, 2018).

2 See Union Membership and Coverage Database From the Current Population Survey (Jan. 21, 2018), unionstats.com.

3 See National Conference of State Legislatures, Right-to-Work States (2018), http://www.ncsl.org/research/labor-and-employment/right-to-work-laws-and-bills.aspx#chart; see also, e.g., Brief for Mackinac Center for Public Policy as Amicus Curiae 27–28, 34–36.

4 The collective-action problem cited by the dissent, post, at 6, is not specific to the agency-fee context. And contrary to the dissent’s suggestion, it is often not practical for an entity that lobbies or advocates on behalf of the members of a group to tailor its message so that only its members benefit from its efforts. Consider how effective it would be for a group that advocates on behalf of, say, seniors, to argue that a new measure should apply only to its dues-paying members.

5 In order to obtain that status, a union must petition to be recognized and campaign to win majority approval. Ill. Comp. Stat., ch. 5, §315/9(a) (2016); see, e.g., County of Du Page v. Illinois Labor Relations Bd., 231 Ill. 2d 593, 597–600, 900 N. E. 2d 1095, 1098–1099 (2008). And unions eagerly seek this support. See, e.g., Brief for Employees of the State of Minnesota Court System as Amici Curiae 9–17.

6 There is precedent for such arrangements. Some States have laws providing that, if an employee with a religious objection to paying an agency fee “requests the [union] to use the grievance procedure or arbitration procedure on the employee’s behalf, the [union] is authorized to charge the employee for the reasonable cost of using such procedure.”&�Բ��;E.g.,Cal. Govt. Code Ann. §3546.3 (West 2010); cf. Ill. Comp. Stat., ch. 5, §315/6(g) (2016). This more tailored alternative, if applied to other objectors, would prevent free ridership while imposing a lesser burden on First Amendment rights.

7 Indeed, under common law, “collective bargaining was unlawful,”&�Բ��;Teamsters v. Terry,494 U. S. 558, 565–566 (1990) (plurality opinion); see N. Citrine, Trade Union Law 4–7, 9–10 (2d ed. 1960); Notes, Legal-ity of Trade Unions at Common Law, 25 Harv. L. Rev. 465, 466 (1912), and into the 20th century, every individual employee had the “liberty of contract” to “sell his labor upon such terms as he deem[ed] proper,”&�Բ��;Adair v. United States,208 U. S. 161, 174–175 (1908); see R. Morris, Government and Labor in Early America 208, 529 (1946). So even the concept of a private third-party entity with the power to bind employees on the terms of their employment likely would have been foreign to the Founders. We note this only to show the problems inherent in the Union respondent’s argument; we are not in any way questioning the foundations of modern labor law.

8 See, e.g., Ellsworth, The Landholder, VII (1787), in Essays on the Constitution of the United States 167–171 (P. Ford ed. 1892); Webster, On Test Laws, Oaths of Allegiance and Abjuration, and Partial Exclusions from Office, in A Collection of Essays and Fugitiv[e] Writings 151–153 (1790).

9 Justice Powell’s separate opinion did invoke Pickering in a relevant sense, but he did so only to acknowledge the State’s relatively greater interest in regulating speech when it acts as employer than when it acts as sovereign. Abood v. Detroit Bd. of Ed.,431 U. S. 209, 259 (1977) (concurring in judgment). In the very next sentence, he explained that “even in public employment, a significant impairment of First Amendment rights must survive exacting scrutiny.”&�Բ��;Ibid. (internal quotation marks omitted). That is the test we apply today.

10 See Brief for State of Michigan et al. as Amici Curiae 9–24. Nationwide, the cost of state and local employees’ wages and benefits, for example, is nearly $1.5 trillion—more than half of those jurisdictions’ total expenditures. See Dept. of Commerce, Bureau of Economic Analysis, National Data, GDP & Personal Income, Table 6.2D, line 92 (Aug. 3, 2017), and Table 3.3, line 37 (May 30, 2018), https://www.bea.gov/iTable/iTable.cfm?reqid=19&step=2#reqid=19&step=2&isuri=1&1921=survey. And many States and cities struggle with unfunded pension and retiree healthcare liabilities and other budget issues.

11 PEW Charitable Trusts, Fiscal 50: State Trends and Analysis (updated May 17, 2016), http://www.pewtrusts.org/en/research-and-analysis/data-visualizations/2014/fiscal-50#ind4.

12 See Brief for Jason R. Barclay et al. as Amici Curiae 9; M. Egan, How Illinois Became America’s Most Messed-Up State, CNN Money (July 1, 2017), https://cnnmon.ie/2tp9NX5.

13 Brief for Jason R. Barclay et al. as Amici Curiae 9.

14 E. Campbell, S&P, Moody’s Downgrade Illinois to Near Junk, Lowest Ever for a U. S. State, Bloomberg (June 1, 2017), https://bloom.bg/2roEJUc.

15 See National Association of State Budget Officers, Summary: Spring 2018 Fiscal Survey of States 2 (June 14, 2018), http://www.nasbo.org; ProQuest Statistical Abstract of the United States: 2018, pp. 306, Table 476, 321, Table 489.

16 See Rogers, School Districts ‘Race to the Top’ Despite Teacher Dispute, Marin Independent J., June 19, 2010.

17 See Sawchuk, Transferring Top Teachers Has Benefits: Study Probes Moving Talent to Low-Performing Schools, Education Week, Nov. 13, 2013, pp. 1, 13.

18 See Tucker, Textbooks Equivocate on Global Warming: Stanford Study Finds Portrayal ‘Dishonest,’ San Francisco Chronicle, Nov. 24, 2015, p. C1.

19 See Reagan, Anti-Confederacy Movement Rekindles Texas Textbook Controversy, San Antonio Current, Aug. 4, 2015.

20 See Watanabe, How To Teach Gay Issues in 1st Grade? A New Law Requiring California Schools To Have Lessons About LGBT Americans Raises Tough Questions, L. A. Times, Oct. 16, 2011, p. A1.

21 See Goodstein, A Web of Faith, Law and Science in Evolution Suit, N. Y. Times, Sept. 26, 2005, p. A1.

22 See Golden, Defending the Faith: New Battleground in Textbook Wars: Religion in History, Wall St. J., Jan. 25, 2006, p. A1.

23 Claiming that our decision will hobble government operations, the dissent asserts that it would prevent a government employer from taking action against disruptive non-unionized employees in two carefully constructed hypothetical situations. See post, at 17–18. Both hypotheticals are short on potentially important details, but in any event, neither would be affected by our decision in this case. Rather, both would simply call for the application of the standard Pickering test. In one of the hypotheticals, teachers “protest merit pay in the school cafeteria.”&�Բ��;Post, at 17. If such a case actually arose, it would be important to know, among other things, whether the teachers involved were supposed to be teaching in their classrooms at the time in question and whether the protest occurred in the presence of students during the student lunch period. If both those conditions were met, the teachers would presumably be violating content-neutral rules regarding their duty to teach at specified times and places, and their conduct might well have a disruptive effect on the educational process. Thus, in the dissent’s hypothetical, the school’s interests might well outweigh those of the teachers, but in this hypothetical case, as in all Pickering cases, the particular facts would be very important. In the other hypothetical, employees agitate for a better health plan “at various inopportune times and places.”&�Բ��;Post, at 17. Here, the lack of factual detail makes it impossible to evaluate how the Pickering balance would come out. The term “agitat[ion]” can encompass a wide range of conduct, as well as speech. Post, at 17. And the time and place of the agitation would also be important.

24 No First Amendment issue could have properly arisen in those cases unless Congress’s enactment of a provision allowing, but not requiring, private parties to enter into union-shop arrangements was sufficient to establish governmental action. That proposition was debatable when Abood was decided, and is even more questionable today. See American Mfrs. Mut. Ins. Co. v. Sullivan,526 U. S. 40, 53 (1999); Jackson v. Metropolitan Edison Co.,419 U. S. 345, 357 (1974). Compare, e.g., White v. Communications Workers of Am., AFL–CIO, Local 13000, 370 F. 3d 346, 350 (CA3 2004) (no state action), and Kolinske v. Lubbers, 712 F. 2d 471, 477–478 (CADC 1983) (same), with Beck v. Communications Workers of Am., 776 F. 2d 1187, 1207 (CA4 1985) (state action), and Linscott v. Millers Falls Co., 440 F. 2d 14, 16, and n. 2 (CA1 1971) (same). We reserved decision on this question in Communications Workers v. Beck,487 U. S. 735, 761 (1988), and do not resolve it here.

25 Contrary to the dissent’s claim, see post, at 19, and n. 4, the fact that “[t]he rationale of [Abood] does not withstand careful analysis”&�Բ��;is a reason to overrule it, e.g., Lawrence v. Texas,539 U. S. 558, 577 (2003). And that is even truer when, as here, the defenders of the precedent do not attempt to “defend [its actual] reasoning.”&�Բ��;Citizens United v. Federal Election Comm’n,558 U. S. 310, 363 (2010); id., at 382–385 (Roberts, C. J., concurring).

26 For this reason, it is hardly surprising that chargeability issues have not arisen in many Court of Appeals cases. See post, at 22 (Kagan, J., dissenting).

27 The dissent emphasizes another type of reliance, namely, that “[o]ver 20 States have by now enacted statutes authorizing [agency-fee] provisions.”&�Բ��;Post, at 23. But as we explained in Citizens United,“[t]his is not a compelling interest for stare decisis. If it were, legislative acts could prevent us from overruling our own precedents, thereby interfering with our duty ‘to say what the law is.’ ” 558 U. S., at 365 (quoting Marbury v. Madison, 1 Cranch 137, 177 (1803)). Nor does our decision “ ‘require an extensive legislative response.’ ”&�Բ��;Post, at 23. States can keep their labor-relations systems exactly as they are—only they cannot force nonmembers to subsidize public-sector unions. In this way, these States can follow the model of the federal government and 28 other States.

28 Unfortunately, the dissent sees the need to resort to accusations that we are acting like “black-robed rulers” who have shut down an “energetic policy debate.”&�Բ��;Post, at 27–28. We certainly agree that judges should not “overrid[e] citizens’ choices” or “pick the winning side,”&�Բ��;ibid.—unless the Constitution commands that they do so. But when a federal or state law violates the Constitution, the American doctrine of judicial review requires us to enforce the Constitution. Here, States with agency-fee laws have abridged fundamental free speech rights. In holding that these laws violate the Constitution, we are simply enforcing the First Amendment as properly understood, “[t]he very purpose of [which] was to withdraw certain subjects from the vicissitudes of political controversy, to place them beyond the reach of majorities and officials and to establish them as legal principles to be applied by the courts.”&�Բ��;West Virginia Bd. of Ed. v. Barnette,319 U. S. 624, 638 (1943).

Justice Kagan, with whom Justice Ginsburg, Justice Breyer, and Justice Sotomayor join, dissenting.

For over 40 years, Abood v. Detroit Bd. of Ed., 431 U. S. 209 (1977), struck a stable balance between public employees’ First Amendment rights and government entities’ interests in running their workforces as they thought proper. Under that decision, a government entity could require public employees to pay a fair share of the cost that a union incurs when negotiating on their behalf over terms of employment. But no part of that fair-share payment could go to any of the union’s political or ideological activities.

That holding fit comfortably with this Court’s general framework for evaluating claims that a condition of public employment violates the First Amendment. The Court’s decisions have long made plain that government entities have substantial latitude to regulate their employees’ speech—especially about terms of employment—in the interest of operating their workplaces effectively. Abood allowed governments to do just that. While protecting public employees’ expression about non-workplace matters, the decision enabled a government to advance important managerial interests—by ensuring the presence of an exclusive employee representative to bargain with. Far from an “anomaly,”&�Բ��;ante, at 7, the Abood regime was a paradigmatic example of how the government can regulate speech in its capacity as an employer.

Not any longer. Today, the Court succeeds in its 6-year campaign to reverse Abood. See Friedrichs v. California Teachers Assn., 578 U. S. ___ (2016) (per curiam); Harris v. Quinn, 573 U. S. ___ (2014); Knox v. Service Employees, 567 U. S. 298 (2012). Its decision will have large-scale consequences. Public employee unions will lose a secure source of financial support. State and local governments that thought fair-share provisions furthered their interests will need to find new ways of managing their workforces. Across the country, the relationships of public employees and employers will alter in both predictable and wholly unexpected ways.

Rarely if ever has the Court overruled a decision—let alone one of this import—with so little regard for the usual principles of stare decisis. There are no special justifications for reversing Abood. It has proved workable. No recent developments have eroded its underpinnings. And it is deeply entrenched, in both the law and the real world. More than 20 States have statutory schemes built on the decision. Those laws underpin thousands of ongoing contracts involving millions of employees. Reliance interests do not come any stronger than those surrounding Abood. And likewise, judicial disruption does not get any greater than what the Court does today. I respectfully dissent.

I

I begin with Abood, the 41-year-old precedent the majority overrules. That case involved a union that had been certified as the exclusive representative of Detroit’s public school teachers. The union’s collective-bargaining agreement with the city included an “agency shop” clause, which required teachers who had not joined the union to pay it “a service charge equal to the regular dues required of [u]nion members.”&�Բ��;Abood, 431 U. S., at 212. A group of non-union members sued over that clause, arguing that it violated the First Amendment.

In considering their challenge, the Court canvassed the purposes of the “agency shop” clause. It was rooted, the Court understood, in the “principle of exclusive union representation”—a “central element” in “industrial relations” since the New Deal. Id., at 220. Significant benefits, the Court explained, could derive from the “designation of a single [union] representative” for all similarly situated employees in a workplace. Ibid. In particular, such arrangements: “avoid[ ] the confusion that would result from attempting to enforce two or more agreements specifying different terms and conditions of employment”; “prevent[ ] inter-union rivalries from creating dissension within the work force”; “free[ ] the employer from the possibility of facing conflicting demands from different unions”; and “permit[ ] the employer and a single union to reach agreements and settlements that are not subject to attack from rival labor organizations.”&�Բ��;Id., at 220–221. As proof, the Court pointed to the example of exclusive-representation arrangements in the private-employment sphere: There, Congress had long thought that such schemes would promote “peaceful labor relations” and “labor stability.”&�Բ��;Id., at 219, 229. A public employer like Detroit, the Court believed, could reasonably make the same calculation.

But for an exclusive-bargaining arrangement to work, such an employer often thought, the union needed adequate funding. Because the “designation of a union as exclusive representative carries with it great responsibilities,” the Court reasoned, it inevitably also entails substantial costs. Id., at 221. “The tasks of negotiating and administering a collective-bargaining agreement and representing the interests of employees in settling disputes and processing grievances are continuing and difficult ones.”&�Բ��;Ibid. Those activities, the Court noted, require the “expenditure of much time and money”—for example, payment for the “services of lawyers, expert negotiators, economists, and a research staff.”&�Բ��;Ibid. And there is no way to confine the union’s services to union members alone (and thus to trim costs) because unions must by law fairly represent all employees in a given bargaining unit—union members and non-members alike. See ibid.

With all that in mind, the Court recognized why both a government entity and its union bargaining partner would gravitate toward an agency-fee clause. Those fees, the Court reasoned, “distribute fairly the cost” of collective bargaining “among those who benefit”—that is, all employees in the work unit. Id., at 222. And they “counteract[ ] the incentive that employees might otherwise have to become ‘free riders.’ ”&�Բ��;Ibid. In other words, an agency-fee provision prevents employees from reaping all the “benefits of union representation”—higher pay, a better retirement plan, and so forth—while leaving it to others to bear the costs. Ibid. To the Court, the upshot was clear: A government entity could reasonably conclude that such a clause was needed to maintain the kind of exclusive bargaining arrangement that would facilitate peaceful and stable labor relations.

But the Court acknowledged as well the “ First Amendment interests” of dissenting employees. Ibid. It recognized that some workers might oppose positions the union takes in collective bargaining, or even “unionism itself.”&�Բ��;Ibid. And still more, it understood that unions often advance “political and ideological” views outside the collective-bargaining context—as when they “contribute to political candidates.”&�Բ��;Id., at 232, 234. Employees might well object to the use of their money to support such “ideological causes.”&�Բ��;Id., at 235.

So the Court struck a balance, which has governed this area ever since. On the one hand, employees could be required to pay fees to support the union in “collective bargaining, contract administration, and grievance adjustment.”&�Բ��;Id., at 225–226. There, the Court held, the “important government interests” in having a stably funded bargaining partner justify “the impingement upon” public employees’ expression. Id., at 225. But on the other hand, employees could not be compelled to fund the union’s political and ideological activities. Outside the collective-bargaining sphere, the Court determined, an employee’s First Amendment rights defeated any conflicting government interest. See id., at 234–235.

II

Unlike the majority, I see nothing “questionable” about Abood’s analysis. Ante, at 7 (quoting Harris, 573 U. S., at ___ (slip op., at 17)). The decision’s account of why some government entities have a strong interest in agency fees (now often called fair-share fees) is fundamentally sound. And the balance Abood struck between public employers’ interests and public employees’ expression is right at home in First Amendment doctrine.

A

Abood’s reasoning about governmental interests has three connected parts. First, exclusive representation arrangements benefit some government entities because they can facilitate stable labor relations. In particular, such arrangements eliminate the potential for inter-union conflict and streamline the process of negotiating terms of employment. See 431 U. S., at 220–221. Second, the government may be unable to avail itself of those benefits unless the single union has a secure source of funding. The various tasks involved in representing employees cost money; if the union doesn’t have enough, it can’t be an effective employee representative and bargaining partner. See id., at 221. And third, agency fees are often needed to ensure such stable funding. That is because without those fees, employees have every incentive to free ride on the union dues paid by others. See id., at 222.

The majority does not take issue with the first point. See ante, at 33 (It is “not disputed that the State may require that a union serve as exclusive bargaining agent for its employees” in order to advance the State’s “interests as an employer”). The majority claims that the second point never appears in Abood, but is willing to assume it for the sake of argument. See ante, at 31–32; but see Abood, 431 U. S., at 221 (The tasks of an exclusive representative “often entail expenditure of much time and money”). So the majority stakes everything on the third point—the conclusion that maintaining an effective system of exclusive representation often entails agency fees. Ante, at 12 (It “is simply not true” that exclusive representation and agency fees are “inextricably linked”); see ante, at 14.

But basic economic theory shows why a government would think that agency fees are necessary for exclusive representation to work. What ties the two together, as Abood recognized, is the likelihood of free-riding when fees are absent. Remember that once a union achieves exclusive-representation status, the law compels it to fairly represent all workers in the bargaining unit, whether or not they join or contribute to the union. See supra, at 4. Because of that legal duty, the union cannot give special advantages to its own members. And that in turn creates a collective action problem of nightmarish proportions. Everyone—not just those who oppose the union, but also those who back it—has an economic incentive to withhold dues; only altruism or loyalty—as against financial self-interest—can explain why an employee would pay the union for its services. And so emerged Abood’s rule allowing fair-share agreements: That rule ensured that a union would receive sufficient funds, despite its legally imposed disability, to effectively carry out its duties as exclusive representative of the government’s employees.

The majority’s initial response to this reasoning is simply to dismiss it. “[F]ree rider arguments,” the majority pronounces, “are generally insufficient to overcome First Amendment objections.”&�Բ��;Ante, at 13 (quoting Knox, 567 U. S., at 311). “To hold otherwise,” it continues, “would have startling consequences” because “[m]any private groups speak out” in ways that will “benefit[ ] nonmembers.”&�Բ��;Ante, at 13. But that disregards the defining characteristic of this free-rider argument—that unions, unlike those many other private groups, must serve members and non-members alike. Groups advocating for “senior citizens or veterans” (to use the majority’s examples) have no legal duty to provide benefits to all those individuals: They can spur people to pay dues by conferring all kinds of special advantages on their dues-paying members. Unions are—by law—in a different position, as this Court has long recognized. See, e.g., Machinists v. Street, 367 U. S. 740, 762 (1961). Justice Scalia, responding to the same argument as the majority’s, may have put the point best. In a way that is true of no other private group, the “law requires the union to carry” non-members—“indeed, requires the union to go out of its way to benefit [them], even at the expense of its other interests.”&�Բ��;Lehnert v. Ferris Faculty Assn., 500 U. S. 507, 556 (1991) (opinion concurring in part and dissenting in part). That special feature was what justified Abood: “Where the state imposes upon the union a duty to deliver services, it may permit the union to demand reimbursement for them.” 500 U. S., at 556.

The majority’s fallback argument purports to respond to the distinctive position of unions, but still misses Abood’s economic insight. Here, the majority delivers a four-page exegesis on why unions will seek to serve as an exclusive bargaining representative even “if they are not given agency fees.”&�Բ��;Ante, at 14; see ante, at 14–17. The gist of the account is that “designation as the exclusive representative confers many benefits,” which outweigh the costs of providing services to non-members. Ante, at 15. But that response avoids the key question, which is whether unions without agency fees will be able to (not whether they will want to)carry on as an effective exclusive representative. And as to that question, the majority again fails to reckon with how economically rational actors behave—in public as well as private workplaces. Without a fair-share agreement, the class of union non-members spirals upward. Employees (including those who love the union) realize that they can get the same benefits even if they let their memberships expire. And as more and more stop paying dues, those left must take up the financial slack (and anyway, begin to feel like suckers)—so they too quit the union. See Ichniowski & Zax, Right-to-Work Laws, Free Riders, and Unionization in the Local Public Sector, 9 J. Labor Economics 255, 257 (1991).[1] And when the vicious cycle finally ends, chances are that the union will lack the resources to effectively perform the responsibilities of an exclusive representative—or, in the worst case, to perform them at all. The result is to frustrate the interests of every government entity that thinks a strong exclusive-representation scheme will promote stable labor relations.

Of course, not all public employers will share that view. Some would rather not bargain with an exclusive representative. Others would prefer that representative to be poorly funded—to serve more as a front than an effectual bargaining partner. But as reflected in the number of fair-share statutes and contracts across the Nation, see supra, at 2, many government entities think that effective exclusive representation makes for good labor relations—and recognize, just as Abood did, that representation of that kind often depends on agency fees. See, e.g., Harris, 573 U. S., at ___ (slip op., at 24) (Kagan, J., dissenting) (describing why Illinois thought that bargaining with an adequately funded exclusive representative of in-home caregivers would enable the State to better serve its dis- abled citizens). Abood respected that state interest; today’s majority fails even to understand it. Little wonder that the majority’s First Amendment analysis, which involves assessing the government’s reasons for imposing agency fees, also comes up short.

B

In many cases over many decades, this Court has addressed how the First Amendment applies when the government, acting not as sovereign but as employer, limits its workers’ speech. Those decisions have granted substantial latitude to the government, in recognition of its significant interests in managing its workforce so as to best serve the public. Abood fit neatly with that caselaw, in both reasoning and result. Indeed, its reversal today creates a significant anomaly—an exception, applying to union fees alone, from the usual rules governing public employees’ speech.

“Time and again our cases have recognized that the Government has a much freer hand” in dealing with its employees than with “citizens at large.”&�Բ��;NASA v. Nelson, 562 U. S. 134, 148 (2011) (internal quotation marks omitted). The government, we have stated, needs to run “as effectively and efficiently as possible.”&�Բ��;Engquist v. Oregon Dept. of Agriculture, 553 U. S. 591, 598 (2008) (internal quotation marks omitted). That means it must be able, much as a private employer is, to manage its workforce as it thinks fit. A public employee thus must submit to “certain limitations on his or her freedom.”&�Բ��;Garcetti v. Ceballos, 547 U. S. 410, 418 (2006). Government workers, of course, do not wholly “lose their constitutional rights when they accept their positions.”&�Բ��;Engquist, 553 U. S., at 600. But under our precedent, their rights often yield when weighed “against the realities of the employment context.”&�Բ��;Ibid. If it were otherwise—if every employment decision were to “bec[o]me a constitutional matter”—“the Government could not function.”&�Բ��;NASA, 562 U. S., at 149 (internal quotation marks omitted).

Those principles apply with full force when public employees’ expressive rights are at issue. As we have explained: “Government employers, like private employers, need a significant degree of control over their employees’ words” in order to “efficient[ly] provi[de] public services.”&�Բ��;Garcetti, 547 U. S., at 418. Again, significant control does not mean absolute authority. In particular, the Court has guarded against government efforts to “leverage the employment relationship” to shut down its employees’ speech as private citizens. Id., at 419. But when the government imposes speech restrictions relating to workplace operations, of the kind a private employer also would, the Court reliably upholds them. See, e.g., id., at 426; Connick v. Myers, 461 U. S. 138, 154 (1983).

In striking the proper balance between employee speech rights and managerial interests, the Court has long applied a test originating in Pickering v. Board of Ed. of Township High School Dist. 205, Will Cty., 391 U. S. 563 (1968). That case arose out of an individual employment action: the firing of a public school teacher. As we later described the Pickering inquiry, the Court first asks whether the employee “spoke as a citizen on a matter of public concern.”&�Բ��;Garcetti, 547 U. S., at 418. If she did not—but rather spoke as an employee on a workplace matter—she has no “possibility of a First Amendment claim”: A public employer can curtail her speech just as a private one could. Ibid. But if she did speak as a citizen on a public matter, the public employer must demonstrate “an adequate justification for treating the employee differently from any other member of the general public.”&�Բ��;Ibid. The government, that is, needs to show that legitimate workplace interests lay behind the speech regulation.

Abood coheres with that framework. The point here is not, as the majority suggests, that Abood is an overt, one-to-one “application of Pickering.”&�Բ��;Ante, at 26. It is not. Abood related to a municipality’s labor policy, and so the Court looked to prior cases about unions, not to Pickering’s analysis of an employee’s dismissal. (And truth be told, Pickering was not at that time much to look at: What the Court now thinks of as the two-step Pickering test, as the majority’s own citations show, really emerged from Garcetti and Connick—two cases post-dating Abood. See ante, at 22.)[] But Abood and Pickering raised variants of the same basic issue: the extent of the government’s authority to make employment decisions affecting expression. And in both, the Court struck the same basic balance, enabling the government to curb speech when—but only when—the regulation was designed to protect its managerial interests. Consider the parallels:

Like Pickering, Abood drew the constitutional line by analyzing the connection between the government’s managerial interests and different kinds of expression. The Court first discussed the use of agency fees to subsidize the speech involved in “collective bargaining, contract administration, and grievance adjustment.” 431 U. S., at 225–226. It understood that expression (really, who would not?) as intimately tied to the workplace and employment relationship. The speech was about “working conditions, pay, discipline, promotions, leave, vacations, and terminations,”&�Բ��;Borough of Duryea v. Guarnieri, 564 U. S. 379, 391 (2011); the speech occurred (almost always) in the workplace; and the speech was directed (at least mainly) to the employer. As noted earlier, Abood described the managerial interests of employers in channeling all that speech through a single union. See 431 U. S., at 220–222, 224–226; supra, at 3. And so Abood allowed the government to mandate fees for collective bargaining—just as Pickering permits the government to regulate employees’ speech on similar workplace matters. But still, Abood realized that compulsion could go too far. The Court barred the use of fees for union speech supporting political candidates or “ideological causes.” 431 U. S., at 235. That speech, it understood, was “unrelated to [the union’s] duties as exclusive bargaining representative,” but instead was directed at the broader public sphere. Id., at 234. And for that reason, the Court saw no legitimate managerial interests in compelling its subsidization. The employees’ First Amendment claims would thus prevail—as, again, they would have under Pickering.

Abood thus dovetailed with the Court’s usual attitude in First Amendment cases toward the regulation of public employees’ speech. That attitude is one of respect—even solicitude—for the government’s prerogatives as an employer. So long as the government is acting as an employer—rather than exploiting the employment relationship for other ends—it has a wide berth, comparable to that of a private employer. And when the regulated expression concerns the terms and conditions of employment—the very stuff of the employment relationship—the government really cannot lose. There, managerial interests are obvious and strong. And so government employees are . . . just employees, even though they work for the government. Except that today the government does lose, in a first for the law. Now, the government can constitutionally adopt all policies regulating core workplace speech in pursuit of managerial goals—save this single one.

The majority claims it is not making a special and unjustified exception. It offers two main reasons for declining to apply here our usual deferential approach, as exemplified in Pickering, to the regulation of public employee speech. First, the majority says, this case involves a “blanket” policy rather than an individualized employment decision, so Pickering is a “painful fit.”&�Բ��;Ante, at 23. Second, the majority asserts, the regulation here involves compelling rather than restricting speech, so the pain gets sharper still. See ante, at 24–25.And finally, the majority claims that even under the solicitous Pickering standard, the government should lose, because the speech here involves a matter of public concern and the government’s managerial interests do not justify its regulation. See ante, at 27–31. The majority goes wrong at every turn.

First, this Court has applied the same basic approach whether a public employee challenges a general policy or an individualized decision. Even the majority must concede that “we have sometimes looked to Pickering in considering general rules that affect broad categories of employees.”&�Բ��;Ante, at 23. In fact, the majority cannot come up with any case in which we have not done so. All it can muster is one case in which while applying the Pickering testto a broad rule—barring any federal employee from accepting any payment for any speech or article on any topic—the Court noted that the policy’s breadth would count against the government at the �ٱ��’s second step. See United States v. Treasury Employees, 513 U. S. 454 (1995). Which is completely predictable. The inquiry at that stage, after all, is whether the government has an employment-related interest in going however far it has gone—and in Treasury Employees, the government had indeed gone far. (The Court ultimately struck down the rule because it applied to speech in which the government had no identifiable managerial interest. See id., at 470, 477.) Nothing in Treasury Employees suggests that the Court defers only to ad hoc actions, and not to general rules, about public employee speech. That would be a perverse regime, given the greater regularity of rulemaking and the lesser danger of its abuse. So I would wager a small fortune that the next time a general rule governing public employee speech comes before us, we will dust off Pickering.

Second, the majority’s distinction between compelling and restricting speech also lacks force. The majority posits that compelling speech always works a greater injury, and so always requires a greater justification. See ante, at 8. But the only case the majority cites for that reading of our precedent is possibly (thankfully) the most exceptional in our First Amendment annals: It involved the state forcing children to swear an oath contrary to their religious beliefs. See ibid. (quoting West Virginia Bd. of Ed. v. Barnette, 319 U. S. 624 (1943)). Regulations challenged as compelling expression do not usually look anything like that—and for that reason, the standard First Amendment rule is that the “difference between compelled speech and compelled silence” is “without constitutional significance.”&�Բ��;Riley v. National Federation of Blind of N. C., Inc., 487 U. S. 781, 796 (1988); see Wooley v. Maynard, 430 U. S. 705, 714 (1977) (referring to “[t]he right to speak and the right to refrain from speaking” as “complementary components” of the First Amendment). And if anything, the First Amendment scales tip the opposite way when (as here) the government is not compelling actual speech, but instead compelling a subsidy that others will use for expression. See Brief for Eugene Volokh et al. as Amici Curiae 4–5 (offering many examples to show that the First Amendment “simply do[es] not guarantee that one’s hard-earned dollars will never be spent on speech one disapproves of”).[3] So when a government mandates a speech subsidy from a public employee—here, we might think of it as levying a tax to support collective bargaining—it should get at least as much deference as when it restricts the employee’s speech. As this case shows, the former may advance a managerial interest as well as the latter—in which case the government’s “freer hand” in dealing with its employees should apply with equal (if not greater) force. NASA, 562 U. S., at 148.

Third and finally, the majority errs in thinking that under the usual deferential approach, the government should lose this case. The majority mainly argues here that, at Pickering’s first step,“union speech in collective bargaining” is a “matter of great public concern” because it “affect[s] how public money is spent” and addresses “other important matters” like teacher merit pay or tenure. Ante, at 27, 29 (internal quotation marks omitted). But to start, the majority misunderstands the threshold inquiry set out in Pickering and later cases. The question is not, as the majority seems to think, whether the public is, or should be, interested in a government employee’s speech. Instead, the question is whether that speech is about and directed to the workplace—as contrasted with the broader public square. Treasury Employees offers the Court’s fullest explanation. The Court held there that the government’s policy prevented employees from speaking as “citizen[s]” on “matters of public concern.” 513 U. S., at 466 (quoting Pickering, 391 U. S., at 568). Why? Because the speeches and articles “were addressed to a public audience, were made outside the workplace, and involved content largely unrelated to their Government employment.” 513 U. S., at 466; see id., at 465, 470 (repeating that analysis twice more). The Court could not have cared less whether the speech at issue was “important.”&�Բ��;Ante, at 29. It instead asked whether the speech was truly of the workplace—addressed toit, made in it, and (most of all) about it.

Consistent with that focus, speech about the terms and conditions of employment—the essential stuff of collective bargaining—has never survived Pickering’s first step. This Court has rejected all attempts by employees to make a “federal constitutional issue” out of basic “employment matters, including working conditions, pay, discipline, promotions, leave, vacations, and terminations.”&�Բ��;Guarnieri, 564 U. S., at 391; see Board of Comm’rs, Wabaunsee Cty. v. Umbehr, 518 U. S. 668, 675 (1996) (stating that public employees’ “speech on merely private employment matters is unprotected”). For that reason, even the Justices who originally objected to Abood conceded that the use of agency fees for bargaining on “economic issues” like “salaries and pension benefits” would not raise significant First Amendment questions. 431 U. S., at 263, n. 16 (Powell, J., concurring in judgment). Of course, most of those issues have budgetary consequences: They “affect[ ] how public money is spent.”&�Բ��;Ante, at 29. And some raise important non-budgetary disputes; teacher merit pay is a good example, see ante, at 30. But arguing about the terms of employment is still arguing about the terms of employment: The workplace remains both the context and the subject matter of the expression. If all that speech really counted as “of public concern,” as the majority suggests, the mass of public employees’ complaints (about pay and benefits and workplace policy and such) would become “federal constitutional issue[s].”&�Բ��;Guarnieri, 564 U. S., at 391. And contrary to decades’ worth of precedent, government employers would then have far less control over their workforces than private employers do. See supra, at 9–11.

Consider an analogy, not involving union fees: Suppose a government entity disciplines a group of (non-unionized) employees for agitating for a better health plan at various inopportune times and places. The better health plan will of course drive up public spending; so according to the majority’s analysis, the employees’ speech satisfies Pickering’s “public concern” test. Or similarly, suppose a public employer penalizes a group of (non-unionized) teachers who protest merit pay in the school cafeteria. Once again, the majority’s logic runs, the speech is of “public concern,” so the employees have a plausible First Amendment claim. (And indeed, the majority appears to concede as much, by asserting that the results in these hypotheticals should turn on various “factual detail[s]” relevant to the interest balancing that occurs at the Pickering �ٱ��’s second step. Ante, at 32, n. 23.) But in fact, this Court has always understood such cases to end at Pickering’s&�Բ��;first step: If an employee’s speech is about, in, and directed to the workplace, she has no “possibility of a First Amendment claim.”&�Բ��;Garcetti, 547 U. S., at 418; see supra, at 11. So take your pick. Either the majority is exposing government entities across the country to increased First Amendment litigation and liability—and thus preventing them from regulating their workforces as private employers could. Or else, when actual cases of this kind come around, we will discover that today’s majority has crafted a “unions only” carve-out to our employee-speech law.

What’s more, the government should prevail even if the speech involved in collective bargaining satisfies Pickering’s first part. Recall that the next question is whether the government has shown “an adequate justification for treating the employee differently from any other member of the general public.”&�Բ��;Garcetti, 547 U. S., at 418; supra, at 11. That inquiry is itself famously respectful of government interests. This Court has reversed the government only when it has tried to “leverage the employment relationship” to achieve an outcome unrelated to the workplace’s “effective functioning.”&�Բ��;Garcetti, 547 U. S., at 419; Rankin v. McPherson, 483 U. S. 378, 388 (1987). Nothing like that is true here. As Abood described, many government entities have found agency fees the best way to ensure a stable and productive relationship with an exclusive bargaining agent. See 431 U. S., at 220–221, 224–226; supra, at 3–4. And here, Illinois and many governmental amicihave explained again how agency fees advance their workplace goals. See Brief for State Respondents 12, 36; Brief for Governor Tom Wolf et al. as Amici Curiae 21–33. In no other employee-speech case has this Court dismissed such work-related interests, as the majority does here. See supra, at 6–9 (discussing the majority’s refusal to engage with the logic of the State’s position). Time and again, the Court has instead respected and acceded to those interests—just as Abood did.

The key point about Abood is that it fit naturally with this Court’s consistent teaching about the permissibility of regulating public employees’ speech. The Court allows a government entity to regulate that expression in aid of managing its workforce to effectively provide public services. That is just what a government aims to do when it enforces a fair-share agreement. And so, the key point about today’s decision is that it creates an unjustified hole in the law, applicable to union fees alone. This case is sui generis among those addressing public employee speech—and will almost surely remain so.

III

But the worse part of today’s opinion is where the majority subverts all known principles of stare decisis. The majority makes plain, in the first 33 pages of its decision, that it believes Abood was wrong.[4] But even if that were true (which it is not), it is not enough. “Respecting stare decisis means sticking to some wrong decisions.”&�Բ��;Kimble v. Marvel Entertainment, LLC, 576 U. S. ___, ___ (2015) (slip op., at 7). Any departure from settled precedent (so the Court has often stated) demands a “special justification—over and above the belief that the precedent was wrongly decided.”&�Բ��;Id., at ___ (slip op., at 8) (internal quotation marks omitted); see, e.g., Arizona v. Rumsey, 467 U. S. 203, 212 (1984). And the majority does not have anything close. To the contrary: all that is “special” in this case—especially the massive reliance interests at stake—demands retaining Abood, beyond even the normal precedent.

Consider first why these principles about precedent are so important. Stare decisis—“the idea that today’s Court should stand by yesterday’s decisions”—is “a foundation stone of the rule of law.”&�Բ��;Kimble, 576 U. S., at ___ (slip op., at 7) (quoting Michigan v. Bay Mills Indian Community, 572 U. S. ___, ___ (2014) (slip op., at 15)). It “promotes the evenhanded, predictable, and consistent development” of legal doctrine. Payne v. Tennessee, 501 U. S. 808, 827 (1991). It fosters respect for and reliance on judicial decisions. See ibid. And it “contributes to the actual and perceived integrity of the judicial process,”&�Բ��;ibid., by ensuring that decisions are “founded in the law rather than in the proclivities of individuals,”&�Բ��;Vasquez v. Hillery, 474 U. S. 254, 265 (1986).

And Abood is not just any precedent: It is embedded in the law (not to mention, as I’ll later address, in the world) in a way not many decisions are. Over four decades, this Court has cited Abood favorably many times, and has affirmed and applied its central distinction between the costs of collective bargaining (which the government can charge to all employees) and those of political activities (which it cannot). See, e.g., Locke v. Karass, 555 U. S. 207, 213–214 (2009); Lehnert, 500 U. S., at 519; Teachers v. Hudson, 475 U. S. 292, 301–302 (1986); Ellis v. Railway Clerks, 466 U. S. 435, 455–457 (1984). Reviewing those decisions not a decade ago, this Court—unanimously—called the Abood rule “a general First Amendment principle.”&�Բ��;Locke, 555 U. S., at 213. And indeed, the Court has relied on that rule when deciding cases involving compelled speech subsidies outside the labor sphere—cases today’s decision does not question. See, e.g., Keller v. State Bar of Cal., 496 U. S. 1, 9–17 (1990) (state bar fees); Board of Regents of Univ. of Wis. System v. Southworth, 529 U. S. 217, 230–232 (2000) (public university student fees); Glickman v. Wileman Brothers & Elliott, Inc., 521 U. S. 457, 471–473 (1997) (commercial advertising assessments); see also n. 3, supra.

Ignoring our repeated validation of Abood, the majority claims it has become “an outlier among our First Amendment cases.”&�Բ��;Ante, at 42. That claim fails most spectacularly for reasons already discussed: Abood coheres with the Pickering approach to reviewing regulation of public employees’ speech. See supra, at 11–13. Needing to stretch further, the majority suggests that Abood conflicts with “our political patronage decisions.”&�Բ��;Ante, at 44. But in fact those decisions strike a balance much like Abood’s. On the one hand, the Court has enabled governments to compel policymakers to support a political party, because that requirement (like fees for collective bargaining) can reasonably be thought to advance the interest in workplace effectiveness. See Elrod v. Burns, 427 U. S. 347, 366–367 (1976); Branti v. Finkel, 445 U. S. 507, 517 (1980). On the other hand, the Court has barred governments from extending that rule to non-policymaking employees because that application (like fees for political campaigns) can’t be thought to promote that interest, see Elrod, 427 U. S., at 366; the government is instead trying to “leverage the employment relationship” to achieve other goals, Garcetti, 547 U. S., at 419. So all that the majority has left is Knox and Harris. See ante, at 43. Dicta in those recent decisions indeed began the assault on Abood that has culminated today. But neither actually addressed the extent to which a public employer may regulate its own employees’ speech. Relying on them is bootstrapping—and mocking stare decisis. Don’t like a decision? Just throw some gratuitous criticisms into a couple of opinions and a few years later point to them as “special justifications.”

The majority is likewise wrong to invoke “workability” as a reason for overruling Abood. Ante, at 38. Does Abood require drawing a line? Yes, between a union’s collective-bargaining activities and its political activities. Is that line perfectly and pristinely “precis[e],” as the majority demands? Ante, at 38. Well, not quite that—but as exercises of constitutional linedrawing go, Abood stands well above average. In the 40 years since Abood, this Court has had to resolve only a handful of cases raising questions about the distinction. To my knowledge, the circuit courts are not divided on any classification issue; neither are they issuing distress signals of the kind that sometimes prompt the Court to reverse a decision. See, e.g., Johnson v. United States, 576 U. S. ___ (2015) (overruling precedent because of frequent splits and mass confusion). And that tranquility is unsurprising: There may be some gray areas (there always are), but in the mine run of cases, everyone knows the difference between politicking and collective bargaining. The majority cites some disagreement in two of the classification cases this Court decided—as if non-unanimity among Justices were something startling. And it notes that a dissenter in one of those cases called the Court’s approach “malleable” and “not principled,”&�Բ��;ante, at 39—as though those weren’t stock terms in dissenting vocabulary. See, e.g., Murr v. Wisconsin, 582 U. S. ___, ___ (2017) (Roberts, C. J., dissenting) (slip op., at 2); Dietz v. Bouldin, 579 U. S. ___, ___ (2016) (Thomas, J., dissenting) (slip op., at 1); Alabama Legislative Black Caucus v. Alabama, 575 U. S. ___, ___ (2015) (slip op., at 13) (Scalia, J., dissenting). As I wrote in Harris a few Terms ago: “If the kind of hand-wringing about blurry lines that the majority offers were enough to justify breaking with precedent, we might have to discard whole volumes of the U. S. Reports.” 573 U. S., at ___ (slip op., at 15).

And in any event, one stare decisis factor—reliance—dominates all others here and demands keeping Abood.Stare decisis, this Court has held, “has added force when the legislature, in the public sphere, and citizens, in the private realm, have acted in reliance on a previous decision.”&�Բ��;Hilton v. South Carolina Public Railways Comm’n, 502 U. S. 197, 202 (1991). That is because overruling a decision would then “require an extensive legislative response” or “dislodge settled rights and expectations.”&�Բ��;Ibid. Both will happen here: The Court today wreaks havoc on entrenched legislative and contractual arrangements.

Over 20 States have by now enacted statutes authorizing fair-share provisions. To be precise, 22 States, the District of Columbia, and Puerto Rico—plus another two States for police and firefighter unions. Many of those States have multiple statutory provisions, with variations for different categories of public employees. See, e.g., Brief for State of California as Amicus Curiae 24–25. Every one of them will now need to come up with new ways—elaborated in new statutes—to structure relations between government employers and their workers. The majority responds, in a footnote no less, that this is of no proper concern to the Court. See ante, at 47, n. 27. But in fact, we have weighed heavily against “abandon[ing] our settled jurisprudence” that “[s]tate legislatures have relied upon” it and would have to “reexamine [and amend] their statutes” if it were overruled. Allied-Signal, Inc. v. Director, Div. of Taxation, 504 U. S. 768, 785 (1992); Hilton, 502 U. S., at 203.

Still more, thousands of current contracts covering millions of workers provide for agency fees. Usually, this Court recognizes that “[c]onsiderations in favor of stare decisis are at their acme in cases involving property and contract rights.”&�Բ��;Payne, 501 U. S., at 828. Not today. The majority undoes bargains reached all over the country.[5] It prevents the parties from fulfilling other commitments they have made based on those agreements. It forces the parties—immediately—to renegotiate once-settled terms and create new tradeoffs. It does so knowing that many of the parties will have to revise (or redo) multiple contracts simultaneously. (New York City, for example, has agreed to agency fees in 144 contracts with 97 public-sector unions. See Brief for New York City Municipal Labor Committee as Amicus Curiae 4.) It does so knowing that those renegotiations will occur in an environment of legal uncertainty, as state governments scramble to enact new labor legislation. See supra, at 23. It does so with no real clue of what will happen next—of how its action will alter public-sector labor relations. It does so even though the government services affected—policing, firefighting, teaching, transportation, sanitation (and more)—affect the quality of life of tens of millions of Americans.

The majority asserts that no one should care much because the canceled agreements are “of rather short duration” and would “expire on their own in a few years’ time.”&�Բ��;Ante, at 45, 46. But to begin with, that response ignores the substantial time and effort that state legislatures will have to devote to revamping their statutory schemes. See supra, at 23. And anyway, it misunderstands the nature of contract negotiations when the parties have a continuing relationship. The parties, in renewing an old collective-bargaining agreement, don’t start on an empty page. Instead, various “long-settled” terms—like fair-share provisions—are taken as a given. Brief for Governor Tom Wolf et al. 11; see Brief for New York City Sergeants Benevolent Assn. as Amicus Curiae 18. So the majority’s ruling does more than advance by a few years a future renegotiation (though even that would be significant). In most cases, it commands new bargaining over how to replace a term that the parties never expected to change. And not just new bargaining; given the interests at stake, complicated and possibly contentious bargaining as well. See Brief for Governor Tom Wolf et al. 11.[6]

The majority, though, offers another reason for not worrying about reliance: The parties, it says, “have been on notice for years regarding this Court’s misgivings about Abood.”&�Բ��;Ante, at 45. Here, the majority proudly lays claim to its 6-year crusade to ban agency fees. In Knox, the majority relates, it described Abood as an “anomaly.”&�Բ��;Ante, at 45 (quoting 567 U. S., at 311). Then, in Harris, it “cataloged Abood’s many weaknesses.”Ante, at 45. Finally, in Friedrichs, “we granted a petition for certiorari asking us to” reverse Abood, but found ourselves equally divided. Ante, at 45.“During this period of time,” the majority concludes, public-sector unions “must have understood that the constitutionality of [an agency-fee] provision was uncertain.”&�Բ��;Ibid. And so, says the majority, they should have structured their affairs accordingly.

But that argument reflects a radically wrong understanding of how stare decisis operates. Justice Scalia once confronted a similar argument for “disregard[ing] reliance interests” and showed how antithetical it was to rule-of-law principles. Quill Corp. v. North Dakota, 504 U. S. 298, 320 (1992) (concurring opinion). He noted first what we always tell lower courts: “If a precedent of this Court has direct application in a case, yet appears to rest on reasons rejected in some other line of decisions, [they] should follow the case which directly controls, leaving to this Court the prerogative of overruling its own decisions.”&�Բ��;Id., at 321 (quoting Rodriguez de Quijas v. Shearson/ American Express, Inc., 490 U. S. 477, 484 (1989); some alterations omitted). That instruction, Justice Scalia explained, was “incompatible” with an expectation that “private parties anticipate our overrulings.” 406 U. S., at 320. He concluded: “[R]eliance upon a square, unabandoned holding of the Supreme Court is always justifiable reliance.”&�Բ��;Ibid. Abood’s holding was square. It was unabandoned before today. It was, in other words, the law—however much some were working overtime to make it not. Parties, both unions and governments, were thus justified in relying on it. And they did rely, to an extent rare among our decisions. To dismiss the overthrowing of their settled expectations as entailing no more than some “adjustments” and “unpleasant transition costs,”&�Բ��;ante, at 47, is to trivialize stare decisis.

IV

There is no sugarcoating today’s opinion. The majority overthrows a decision entrenched in this Nation’s law—and in its economic life—for over 40 years. As a result, it prevents the American people, acting through their state and local officials, from making important choices about workplace governance. And it does so by weaponizing the First Amendment, in a way that unleashes judges, now and in the future, to intervene in economic and regulatory policy.

Departures from stare decisis are supposed to be “exceptional action[s]” demanding “special justification,”&�Բ��;Rumsey, 467 U. S., at 212—but the majority offers nothing like that here. In contrast to the vigor of its attack on Abood, the majority’s discussion of stare decisis barely limps to the finish line. And no wonder: The standard factors this Court considers when deciding to overrule a decision all cut one way. Abood’s legal underpinnings have not eroded over time: Abood is now, as it was when issued, consistent with this Court’s First Amendment law. Abood provided a workable standard for courts to apply. And Abood has generated enormous reliance interests. The majority has overruled Abood for no exceptional or special reason, but because it never liked the decision. It has overruled Abood because it wanted to.

Because, that is, it wanted to pick the winning side in what should be—and until now, has been—an energetic policy debate. Some state and local governments (and the constituents they serve) think that stable unions promote healthy labor relations and thereby improve the provision of services to the public. Other state and local governments (and their constituents) think, to the contrary, that strong unions impose excessive costs and impair those services. Americans have debated the pros and cons for many decades—in large part, by deciding whether to use fair-share arrangements. Yesterday, 22 States were on one side, 28 on the other (ignoring a couple of in-betweeners). Today, that healthy—that democratic—debate ends. The majority has adjudged who should prevail. Indeed, the majority is bursting with pride over what it has accomplished: Now those 22 States, it crows, “can follow the model of the federal government and 28 other States.”&�Բ��;Ante, at 47, n. 27.

And maybe most alarming, the majority has chosen the winners by turning the First Amendment into a sword, and using it against workaday economic and regulatory policy. Today is not the first time the Court has wielded the First Amendment in such an aggressive way. See, e.g., National Institute of Family and Life Advocates v. Becerra, ante, p. ___ (invalidating a law requiring medical and counseling facilities to provide relevant information to users); Sorrell v. IMS Health Inc., 564 U. S. 552 (2011) (striking down a law that restricted pharmacies from selling various data).And it threatens not to be the last. Speech is everywhere—a part of every human activity (employment, health care, securities trading, you name it). For that reason, almost all economic and regulatory policy affects or touches speech. So the majority’s road runs long. And at every stop are black-robed rulers overriding citizens’ choices. The First Amendment was meant for better things. It was meant not to undermine but to protect democratic governance—including over the role of public-sector unions.

Notes

1 The majority relies on statistics from the federal workforce (where agency fees are unlawful) to suggest that public employees do not act in accord with economic logic. See ante, at 12. But first, many fewer federal employees pay dues than have voted for a union to represent them, indicating that free-riding in fact pervades the federal sector. See, e.g., R. Kearney & P. Mareschal, Labor Relations in the Public Sector 26 (5th ed. 2014). And second, that sector is not typical of other public workforces. Bargaining in the federal sphere is limited; most notably, it does not extend to wages and benefits. See Fort Stewart Schools v. FLRA, 495 U. S. 641, 649 (1990). That means union operating expenses are lower than they are elsewhere. And the gap further widens because the federal sector uses large, often national, bargaining units that provide unions with economies of scale. See Brief for International Brotherhood of Teamsters as Amicus Curiae 7. For those reasons, the federal workforce is the wrong place to look for meaningful empirical evidence on the issues here.

2 For those reasons, it is not surprising that the “categorization schemes” in Abood and Pickering are not precisely coterminous. Ante, at 25. The two casesare fraternal rather than identical twins—both standing for the proposition that the government receives great deference when it regulates speech as an employer rather than as a sovereign. See infra this page and 12–13.

3 That’s why this Court has blessed the constitutionality of compelled speech subsidies in a variety of cases beyond Abood, involving a variety of contexts beyond labor relations. The list includes mandatory fees imposed on state bar members (for professional expression); university students (for campus events); and fruit processors (for generic advertising). See Keller v. State Bar of Cal., 496 U. S. 1, 14 (1990); Board of Regents of Univ. of Wis. System v. Southworth, 529 U. S. 217, 233 (2000); Glickman v. Wileman Brothers & Elliott, Inc., 521 U. S. 457, 474 (1997); see also infra, at 20.

4 And then, after ostensibly turning to stare decisis, the majority spends another four pages insisting that Abood was “not well reasoned,” which is just more of the same. Ante, at 38; see ante, at 35–38.

5 Indeed, some agency-fee provisions, if canceled, could bring down entire contracts because they lack severability clauses. See ante, at 46 (noting that unions could have negotiated for that result); Brief for Governor Tom Wolf et al. as Amici Curiae 11.

6 In a single, cryptic sentence, the majority also claims that arguments about reliance “based on [Abood’s] clarityare misplaced” because Abood did not provide a “clear or easily applicable standard” to separate fees for collective bargaining from those for political activities. Ante, at 45. But to begin, the standard for separating those activities was clear and workable, as I have already shown. See supra, at 21–22. And in any event, the reliance Abood engendered was based not on the clarity of that line, but on the clarity of its holding that governments and unions could generally agree to fair-share arrangements.

Justice Sotomayor, dissenting.

I join Justice Kagan’s dissent in full. Although I joined the majority in Sorrell v. IMS Health Inc., 564 U. S. 552 (2011), I disagree with the way that this Court has since interpreted and applied that opinion. See, e.g.,National Institute of Family and Life Advocates v. Becerra, ante, p. ___. Having seen the troubling development in First Amendment jurisprudence over the years, both in this Court and in lower courts, I agree fully with Justice Kagan that Sorrell—in the way it has been read by this Court—has allowed courts to “wiel[d] the First Amendment in . . . an aggressive way” just as the majority does today. Post, at 27.

��ýapp�ٷ�