JUSTICE STEVENS discounts these systemic effects when he charac- terizes patronage as fostering partisan, rather than public, in- terests. Ante, at 9. But taking JUSTICE STEVENS at his word, ---- one wonders why patronage can ever be an ``appropriate require- ---- ment for the position involved,'' ante, at 1. ---- Patronage, moreover, has been a powerful means of achieving the social and political integration of excluded groups. See, e. g., - - Elrod, supra, at 379 (Powell, J., dissenting); Cornwell, Bosses, ----- ----- Machines and Ethnic Politics, in Ethnic Group Politics 190, 195-197 (H. Bailey, Jr., & E. Katz eds. 1969). By supporting and ultimately dominating a particular party ``machine,'' racial and ethnic minorities have--on the basis of their politics rather than their race or ethnicity--acquired the patronage awards the machine had power to confer. No one disputes the historical ac- curacy of this observation, and there is no reason to think that patronage can no longer serve that function. The abolition of patronage, however, prevents groups that have only recently ob- tained political power, especially blacks, from following this path to economic and social advancement. `Every ethnic group that has achieved political power in American cities has used the bureaucracy to provide jobs in re- turn for political support. It's only when Blacks begin to play the same game that the rules get changed. Now the use of such jobs to build political bases becomes an ``evil'' activi- ty, and the city insists on taking the control back ``down- town.'' ' '' New York Amsterdam News, Apr. 1, 1978, p. A-4, quoted in Hamilton, The Patron-Recipient Relationship and Minority Politics in New York City, 94 Pol. Sci. Q. 211, 212 (1979). While the patronage system has the benefits argued for above, it also has undoubted disadvantages. It facilitates financial corr- uption, such as salary kickbacks and partisan political activity on government-paid time. It reduces the efficiency of govern- ment, because it creates incentives to hire more and less- qualified workers and because highly qualified workers are reluc- tant to accept jobs that may only last until the next election. And, of course, it applies some greater or lesser inducement for individuals to join and work for the party in power. To hear the Court tell it, this last is the greatest evil. That is not my view, and it has not historically been the view of the American people. Corruption and inefficiency, rather than abridgement of liberty, have been the major criticisms leading to enactment of the civil-service laws--for the very good reason that the patronage system does not have as harsh an effect upon conscience, expression, and association as the Court suggests. As described above, it is the nature of the pragmatic, patronage-based, two-party system to build alliances and to suppress rather than foster ideological tests for participation in the division of political ``spoils.'' What the patronage sys- tem ordinarily demands of the party worker is loyalty to, and ac- tivity on behalf of, the organization itself rather than a set of political beliefs. He is generally free to urge within the or- ------ --- --- ganization the adoption of any political position; but if that ---------- position is rejected he must vote and work for the party nonethe- less. The diversity of political expression (other than expres- sion of party loyalty) is channeled, in other words, to a dif- ferent stage--to the contests for party endorsement rather than the partisan elections. It is undeniable, of course, that the patronage system entails some constraint upon the expression of views, particularly at the partisan-election stage, and consider- able constraint upon the employee's right to associate with the other party. It greatly exaggerates these, however, to describe them as a general `` `coercion of belief,' '' ante, at 9, quoting ---- Branti, 445 U. S., at 516; see also ante, at 11-12; Elrod, 427 ------ ---- ----- U. S., at 355 (plurality opinion). Indeed, it greatly exag- gerates them to call them ``coercion'' at all, since we generally make a distinction between inducement and compulsion. The public official offered a bribe is not ``coerced'' to violate the law, and the private citizen offered a patronage job is not ``coerced'' to work for the party. In sum, I do not deny that the patronage system influences or redirects, perhaps to a sub- stantial degree, individual political expression and political association. But like the many generations of Americans that have preceded us, I do not consider that a significant impairment of free speech or free association. In emphasizing the advantages and minimizing the disadvantages (or at least minimizing one of the disadvantages) of the pa- tronage system, I do not mean to suggest that that system is best. It may not always be; it may never be. To oppose our Elrod-Branti jurisprudence, one need not believe that the pa- ----- ------ tronage system is necessarily desirable; nor even that it is al- ----------- ways and everywhere arguably desirable; but merely that it is a -------- political arrangement that may sometimes be a reasonable choice, and should therefore be left to the judgment of the people's elected representatives. The choice in question, I emphasize, is not just between patronage and a merit-based civil service, but rather among various combinations of the two that may suit dif- ferent political units and different eras: permitting patronage hiring, for example, but prohibiting patronage dismissal; permit- ting patronage in most municipal agencies but prohibiting it in the police department; or permitting it in the mayor's office but prohibiting it everywhere else. I find it impossible to say that, always and everywhere, all of these choices fail our ``balancing'' test. C The last point explains why Elrod and Branti should be over- ----- ------ ruled, rather than merely not extended. Even in the field of constitutional adjudication, where the pull of stare decisis is ----- ------- at its weakest, see Glidden Co. v. Zdanok, 370 U. S. 530, 543 ------- -- ------ (1962) (opinion of Harlan, J.), one is reluctant to depart from precedent. But when that precedent is not only wrong, not only recent, not only contradicted by a long prior tradition, but also has proved unworkable in practice, then all reluctance ought to disappear. In my view that is the situation here. Though unwil- ling to leave it to the political process to draw the line between desirable and undesirable patronage, the Court has nei- ther been prepared to rule that no such line exists (i. e., that - - all patronage is unconstitutional) nor able to design the line --- itself in a manner that judges, lawyers, and public employees can understand. Elrod allowed patronage dismissals of persons in ----- ``policymaking'' or ``confidential'' positions. 427 U. S., at 367 (plurality opinion); id., at 375 (Stewart, J., concurring). -- Branti retreated from that formulation, asking instead ``whether ------ the hiring authority can demonstrate that party affiliation is an appropriate requirement for the effective performance of the pub- lic office involved.'' 445 U. S., at 518. What that means is anybody's guess. The Courts of Appeals have devised various tests for determining when ``affiliation is an appropriate re- quirement.'' See generally Martin, A Decade of Branti Decisions: ------ A Government Officials' Guide to Patronage Dismissals, 39 Am. U. L. Rev. 11, 23-42 (1989). These interpretations of Branti are ------ not only significantly at variance with each other; they are still so general that for most positions it is impossible to know whether party affiliation is a permissible requirement until a court renders its decision. A few examples will illustrate the shambles Branti has produced. ------ A city cannot fire a deputy sheriff because of his political af- filiation, but then again perhaps it can, especially if he is called the ``police captain.'' A county cannot fire on that basis its attorney for the department of social services, nor its assistant attorney for family court, but a city can fire its solicitor and his assistants, or its assistant city attorney, or its assistant state's attorney, or its corporation counsel. A city cannot discharge its deputy court clerk for his political affiliation, but it can fire its legal assistant to the clerk on that basis. Firing a juvenile court bailiff seems impermissible, but it may be permissible if he is assigned permanently to a single judge. A city cannot fire on partisan grounds its director of roads, but it can fire the second in command of the water department. A government cannot discharge for political reasons the senior vice president of its development bank, Standefer and O'Brien do not allege that their political affiliation was the reason they were laid off, but only that it was the reason they were not recalled. Complaint PP 9, 21-22, App. to Respondent's Brief in Opposition; 641 F. Supp. 249, 256, 257 (CDIll. 1986). Those claims are essentially identical to the claims of persons wishing to be hired; neither fall within the narrow rule of Elrod and Branti against patronage firing. ----- ------ The examples could be multiplied, but this summary should make obvious that the ``tests'' devised to implement Branti have pro- ------ duced inconsistent and unpredictable results. That uncertainty undermines the purpose of both the nonpatronage rule and the ex- ception. The rule achieves its objective of preventing the ``coercion'' of political affiliation, see supra, at ----, only ----- if the employee is confident that he can engage in (or refrain from) political activities without risking dismissal. Since the current doctrine leaves many employees utterly in the dark about whether their jobs are protected, they are likely to play it safe. On the other side, the exception was designed to permit the government to implement its electoral mandate. Elrod, supra, ----- ----- at 367 (plurality opinion). But unless the government is fairly sure that dismissal is permitted, it will leave the politically uncongenial official in place, since an incorrect decision will expose it to lengthy litigation and a large damage award, perhaps even against the responsible officials personally. This uncertainty and confusion are not the result of the fact that Elrod, and then Branti, chose the wrong ``line.'' My point ----- ------ is that there is no right line--or at least no right line that can be nationally applied and that is known by judges. Once we reject as the criterion a long political tradition showing that party-based employment is entirely permissible, yet are unwilling (as any reasonable person must be) to replace it with the princi- ple that party-based employment is entirely impermissible, we have left the realm of law and entered the domain of political science, seeking to ascertain when and where the undoubted bene- fits of political hiring and firing are worth its undoubted costs. The answer to that will vary from State to State, and indeed from city to city, even if one rejects out of hand (as the Branti line does) the benefits associated with party stability. ------ Indeed, the answer will even vary from year to year. During one period, for example, it may be desirable for the manager of a municipally owned public utility to be a career specialist, insu- lated from the political system. During another, when the effi- cient operation of that utility or even its very existence has become a burning political issue, it may be desirable that he be hired and fired on a political basis. The appropriate ``mix'' of party-based employment is a political question if there ever was one, and we should give it back to the voters of the various pol- itical units to decide, through civil-service legislation crafted to suit the time and place, which mix is best. III Even were I not convinced that Elrod and Branti were wrongly de- ----- ------ cided, I would hold that they should not be extended beyond their facts, viz., actual discharge of employees for their political affiliation. Those cases invalidated patronage firing in order to prevent the ``restraint it places on freedoms of belief and association.'' Elrod, 427 U. S., at 355 (plurality opinion); see ----- also id., at 357 (patronage ``compels or restrains'' and ``inhi- -- bits'' belief and association). The loss of one's current livel- ihood is an appreciably greater constraint than such other disap- pointments as the failure to obtain a promotion or selection for an uncongenial transfer. Even if the ``coercive'' effect of the former has been held always to outweigh the benefits of party- based employment decisions, the ``coercive'' effect of the latter should not be. We have drawn a line between firing and other em- ployment decisions in other contexts, see Wygant v. Jackson Bd. ------ ------- -- of Education, 476 U. S. 267, 282-283 (1986) (plurality opinion), -- --------- and should do so here as well. I would reject the alternative that the Seventh Circuit adopted in this case, which allows a cause of action if the employee can demonstrate that he was subjected to the ``substantial equivalent of dismissal.'' 868 F. 2d 943, 950, 954 (CA7 1989). The trouble with that seemingly reasonable standard is that it is so impre- cise that it will multiply yet again the harmful uncertainty and litigation that Branti has already created. If Elrod and Branti ------ ----- ------ are not to be reconsidered in light of their demonstrably unsa- tisfactory consequences, I would go no further than to allow a cause of action when the employee has lost his position, that is, his formal title and salary. That narrow ground alone is enough to resolve the constitutional claims in the present case. Since none of the plaintiffs has alleged loss of his position because of affiliation, I would affirm the Seventh Circuit's judgment insofar as it af- firmed the dismissal of petitioners' claims, and would reverse the Seventh Circuit's judgment insofar as it reversed the dismis- sal of cross-respondent's claims. The Court's opinion, of course, not only declines to confine El- --- rod and Branti to dismissals in the narrow sense I have proposed, --- ------ but, unlike the Seventh Circuit, even extends those opinions beyond ``constructive'' dismissals--indeed, even beyond adverse treatment of current employees--to all hiring decisions. In the long run there may be cause to rejoice in that extension. When the courts are flooded with litigation under that most unmanage- able of standards (Branti) brought by that most persistent and ------ tenacious of suitors (the disappointed office-seeker) we may be moved to reconsider our intrusion into this entire field. In the meantime, I dissent.