Before POSNER and COFFEY, Circuit Judges, and KELLAM, Senior District Judge.*fn*
POSNER, Circuit Judge. The Continental Web Press, a commercial printer, asks us to set aside an order by the National Labors Relations Board, and the Board asks us to enforce it. The Board found that Continental had committed an unfair labor practice by refusing to bargain with a union that had won an election to represent the company's pressman, and the Board ordered Continental to bargain. The principal basis for Continental's refusal was its contention that pressmen are not an appropriate unit for collective bargaining, so that the election should never have been held. We must decide whether the Board's unit determination can be upheld.
Section 9(b) of the National Labor Relations Act, as amended, 29 U.S.C. § 159(b), tells the Board to pick in each case "the unit appropriate for the purposes of collective bargaining," but (with immaterial exceptions) offers the Board no guidance in determining appropriateness beyond stating that the unit can be the firm, the craft, the plant, or a subdivision of the plant and that in deciding which one it is to be the Board shall try "to assure to employees the fullest freedom in exercising the rights" guaranteed by the Act. All this, as we shall see in a moment, comes to very little; and the absence of statutory criteria has reinforced the natural tendency of reviewing courts to defer to specialized administrative bodies, with the result that the Board's unit determinations are rarely disturbed on appeal. The court's obeisant attitude is epitomized by the frequent statement that the Board need only choose an appropriate unit -- its choice need not be the most appropriate unit. See, e.g., Magic Pan, Inc. v. NLRB, 627 F.2d 105, 107 (7th Cir. 1980) (per curiam); NLRB v. Chicago Health & Tennis Clubs, Inc., 567 F.2d 331, 334-35 (7th Cir. 1977). As this is not what the statute, says, the real point must be that the reviewing court is in no position to decide which among several alternative units is the most appropriate.
Of course the court cannot defer to the Board completely; if convinced that a unit determination is arbitrary -- which is to say unreasonable -- the court must reverse. But because the statute does not define an "appropriate" bargaining unit, it is very hard for a reviewing court to tell whether a unit determination is arbitrary. See, e.g., NLRB v. Res-Care, Inc., 705 F.2d 1461, 1469 (7th Cir. 1983). About all the court can do -- recognizing that section 9(b) is a broad delegation of power to the Board -- is to insist that the Board apply with reasonable consistency whatever standard it adopts to guide the exercise of its delegated power.
The Board has a standard: it will approve a unit if but only if the members have a "community of interest." But the words provide little direction. There is a sense in which all people employed in the same firm, the same craft, or even the same industry share a community of interest, and there is a sense in which every worker is a separate "community of interest" because workers differ in age, in the value they place on leisure, in their preferred trade-off between present and future consumption, in their attitude toward risk, and in other dimensions, which, when all are added together, make every worker unique. It would have been helpful, therefore if the Board had tried to give "community of interest" a precise meaning, or at least had explained the purpose behind the formula. The Board has done neither of these things. All one can say about the practical application of the Board's standard is that, in deciding how much "community of interest" is required for a proposed unit to be deemed appropriate, the Board seems to weigh, though not explicitly, the following considerations. See NLRB v. Res-Care, Inc., supra, 705 F.2d at 1468-69; Gorman, Basic Text on Labor Law 67-68 (1976).
On the one hand, the more homogeneous the group of workers composing a proposed unit is with respect to the terms and conditions of employment -- the greater, in other words, the community of interest among those workers is -- the more likely they are to be well served by a single, exclusive bargaining representative, and the less likely there is to be a dissatisfied minority. This consideration argues for requiring a high degree of homogeneity as a condition for approving a separate unit. On the other hand, an extremely homogeneous group is likely to be very small, which in turn implies that there may be many bargaining units -- each perhaps represented by a different union -- in the same plant. It is costly for an employer to have to negotiate separately with a number of different unions, and the costs are not borne by the employer alone. The different unions may have inconsistent goals, yet any one of the unions may be able to shut down the plant (or curtail its operations) by a strike, thus imposing costs on other workers as well as on the employer's shareholders, creditors, suppliers, and customers. (Congress expressed concern about the effects of "proliferation of bargaining units" when it made nonprofit health-care institutions subject to the National Labor Relations Act in 1974. See Res-Care, supra, 705 F.2d at 1469-70; S. Rep. No. 766, 93d Cong., 2d Sess. 5 (1974); H.R. Rep. No. 1051, 93d Cong., 2d Sess. 7 (1974). Apart from concern with the impact on the employer's operations, breaking up a work force into many small units creates a danger that some of them will be so small and powerless that it will be worth no one's while to organize them, in which event the members of these units will be left out of the collective bargaining process. But the other side of this coin is that small, homogeneous units usually are cheaper to organize than larger ones, and often are more effective. Fewer workers have to be contacted, and there is less danger that conflicts of interest will prevent the formation of an effective majority coalition. Also, the smaller a group is (other things being equal), the cheaper it will be for the employer to meet the group's demands (provided there are not a lot of other small units, each with its own stiff demands); thus the benefits of organization will be concentrated but the costs spread over the whole output of the plant.
Navigating through such a welter of competing considerations is obviously a difficult tasks and one for which (as we have said) neither the statute nor the standard that the Board has adopted to carry out its statutory duty provides guidance. Maybe, as suggested in Res-Care, supra, 705 F.2d at 1469, the provision in section 9(b) that tells the Board to choose a unit that will best protect the workers' statutory rights suggests a tilt in favor of whatever unit a union petitioning for recognition proposes. But section 9(c)(5) of the Act, 29 U.S.C. § 159(c)(5), is against this suggestion ("In determining whether a unit is appropriate for the purposes specified in subsection (b) of this section the extent to which the employees have organized shall not be controlling"); and, as we just noted, a unit preferred by one group of workers might be inimical to another group. This is a particularly significant consideration because section 7 of the Act, 29 U.S.C. § 157, guarantees not only the right of employees "to self-organization, to form, join, or assist labor organizations, [and] to bargain collectively through representatives of their own choosing," but also "the right to refrain from any or all of such activities. . . ." Maybe, then a neutral stance, giving no weight to the fact that a union may prefer a particular unit, is the appropriate one. But that makes the task for the Board even harder. It also means that a reviewing court is unlikely to set aside the Board's determination unless the unit seems not to make any sense, as would be true if the Board directed that a homogeneous work force be broken up into separate units, see, e.g., NLRB v. Meyer Label Co., 597 F.2d 18, 22 (2d Cir. 1979); Amalgamated Clothing Workers of America v. NLRB, 491 F.2d 595, 598-99 (5th Cir. 1974); NLRB v. Pinkerton's Inc., 428 F.2d 479 (6th Cir. 1970) -- a course most unlikely to be in the best interests of the work force as a whole.
But that, Continental Web says, is precisely what the Board did in this case: it broke up a homogeneous work force into separate units. According to the facts as found by the Board, Continental's lithographic (i.e., photo-offset) production process consists of two stages. In the first, "preparatory employees" as they are called make plates by photographing the customer's art work and then mounting the negatives. They work in rooms off the main area of the plant (where the printing presses are), presumably to control lighting and assure cleanliness while the customer's art work is being photographed. The pressmen (the workers who run the printing presses) come into the preparatory department to pick up the plates, which are then run in the presses. Sometimes the pressmen bring the plates back to the preparatory department to be checked or corrected, and sometimes preparatory employees make corrections on the plates in the presses. The pressmen have a different supervisor and a different apprenticeship program from the preparatory employees"; transfers between the two departments are rate; and the pressmen start work an hour or two earlier in the morning. But pressmen and preparatory employees are paid about the same (for pressmen the range is $6 to $13 an hour, for preparatory employees it is $6 to $12), have the same fringe benefits, and, so far as appears, work the same total number of hours. At the time of the election the company employed 36 pressmen and 7 preparatory employees.
The union won the election to represent the pressmen by a vote of 19 to 11, with 6 not voting. Although the union's margin of victory exceeded the number of preparatory employees, the Board does not argue that this makes the unit-determination question academic. As should be apparent from our earlier discussion, one cannot assume that the union would have attracted as many votes from the pressmen if the unit had been more inclusive.
Like Continental Web we find it difficult to understand why there should be separate units for pressmen and preparatory employees. The preparatory employees are lithographers, too -- members of the same skilled craft -- doing similar work, for similar pay, in an adjacent area of the same plant; and sometimes their work actually takes them into the pressmen's work area. The Board does not suggest that the preparatory employees have as great a "community of interest" with any other workers in the plant as they do with the pressmen. Apparently they are to have their only tiny unit of seven members, or be left out in the collective bargaining process. It is as if the Board had said that the workers at the beginning of an assembly line belong in a different bargaining unit from the ones at the end; for preparation and printing are successive stages in a single lithographic production process.
The greatest conflicts of interest among workers are over wages, fringe benefits, and working conditions. But there do not seem to be any significant differences between the preparatory employees and the pressmen in Continental Web's plant along any of these dimensions. Of course they do different jobs; but if the workers on a single assembly line do many different tasks but are all paid the same and work under the same conditions, the Board cannot divide the work force into as many bargaining units as there are differentiable tasks. The only differences in working conditions at Continental Web's printing plant that the Board (either in its own opinion, or in that of the Regional Director, which the Board affirmed) mentioned were the different hours of starting work (the presses have to be warmed up before they can be run), and the facts that (1) most of the pressmen wear smocks, while none of the preparatory employees do; (2) the groups have different employee handbooks; and (3) pressroom time cards are designated by the number 103 but preparatory-department time cards by the number 102. We can also infer that (4) the pressman's work is dirtier (because they alone wear smocks) -- but the Board did not mention this point. All these differences seem small an unimportant, and the third seems utterly trivial.
The fact that we are unable to find a significant potential for conflicts of interest that might justify putting the preparatory employees in a different unit from the pressmen would not be important if the Board or the Regional Director could find one; but if they can, they are not telling us about it. Their opinions recite the differences in working conditions that we have mentioned (along with the separate supervisors, separate apprenticeship programs, and infrequency of interdepartmental transfers -- facts that do not necessarily imply a difference in terms of employment or in actual working conditions), and then tack on a conclusion that therefore the pressmen have a sufficient community of interest to be a separate unit. We understand everything but the "therefore."
Of course, this is not the first case in which the Board has had to designate a bargaining unit in the printing trade, which is heavily unionized; and if all the Board had done in this case was to apply established principles recognizing pressmen as a separate unit, a citation to a prior case would have been all the reasoning required. The Administrative Procedure Act does not require loquacity. But far from applying settled principles, the Board apparently has reversed a long-established presumption in favor of combining pressmen and preparatory employees into a single unit of lithographic production workers. For many years, through many changes of Administration, the Board has been saying that pressmen and preparatory employees compose one unit in lithographic printing plants. See, e.g., NLRB v. Lord Baltimore Press, Inc., 370 F.2d 397, 400 and n. 4 (8th Cir. 1966); Meyer Label Co., 232 N.L.R.B. 933 (1977), rev'd, 597 F.2d 18 (2d Cir. 1979); Moore Business Forms, Inc., 216 N.L.R.B. 833, 834 (1975); George Rice & Sons, 212 N.L.R.B. 947 (1974); Paramount Press, Inc., 187 N.L.R.B. 586 (1970); Sherwin Williams Co., 173 N.L.R.B. 316 (1968); A. B. Hirschfeld Press, Inc., 140 N.L.R.B. 212, 216 (1962); Allen, Lane & Scott, 137 N.L.R.B. 223, 225 (1962) ("long-standing rules"); Earl Litho Printing Co., 116 N.L.R.B. 1538, 1539 (1956) ("The Board has consistently held that all employees engaged in the lithographic production process form a cohesive unit for collective-bargaining purposes and has established units which included all such employees without regard to the specific skills of the employees or the departmental nature of the unit sought"); A. B. Hirschfeld Press, Inc., 96 N.L.R.B. 1068, 1069 (1951) ("absent unusual circumstances, all employees engaged in the lithographic process form an indivisible entity for the purposes of collective bargaining"); Johnson City Publishing Co., 81 N.L.R.B. 1341, 1342 (1949); Manz Corp., 79 N.L.R.B. 211, 212 (1948); cf. NLRB v. Krieger-Ragsdale & Co., 379 F.2d 517, 520 (7th Cir. 1967);National Blank Book Co., 246 N.L.R.B. 921 (1979); Olin Corp., 229 N.L.R.B. 793 (1977). The earlier cases are more emphatic on this score than the later ones, but, so far as we can tell, this is only because the policy had become so well established that allusion was all that was required to invoke it. See, e.g., Comfort & Co., 267 N.L.R.B. No. 47, at p. 5 (Aug. 25, 1983) (citing Allen, Lane & Scott, supra).
A number of cases explicitly refuse to sever preparatory employees from pressmen -- the severance performed in this case. Among the cases we have cited, severance was refused in Moore, Paramount, the two Hirschfeld cases, Earl Litho, Johnson, and Manz; to which can be added Con P. Curran Printing Co., 57 N.L.R.B. 185, 188-89 (1944). And the fact that here the union wanted a smaller unit (as in some of the previous cases too) could not justify the Board's certifying such a unit if it were otherwise inappropriate. See section 9(c)(5); supra; Gorman, supra, at 73-74. Otherwise unions could gerrymander the bargaining units (corresponding to elector districts in political elections) to their hearts' content. The "unusual circumstances" exception mentioned in a few cases cannot help the Board either. Continental Web put in evidence (not controverted) that there is nothing unusual about the layout or operation or use of labor in its lithographic production process, and the Board did not suggest otherwise.
The Board is allowed to reverse course. But it has got to give reasons, or else the reversal is arbitrary. This general principle of administrative law, see, e.g., Motor Vehicle Mfrs. Ass'n v. State Farm Mutual Automobile Ins. Co., 463 U.S. 29, 103 S. Ct. 2856, 2866, 77 L. Ed. 2d 443 (1983); Illinois Bell Tel. Co. v. FCC, os. 84-1145 et al., slip op. at 7-8 (7th Cir. June 29, 1984); Distrigas of Massachusetts Corp. v. FERC, 737 F.2d 1208, slip op. at 28 (1st Cir. 1984), is fully applicable to unit determinations. See, e.g., Consolidated Papers, Inc. v. NLRB, 670 F.2d 754, 757 (7th Cir. 1982); NLRB v. Meyer Label Co., supra, 597 F.2d at 20-21; NLRB v. Saint Francis College, 562 F.2d 246, 252 (3d Cir. 1977); NLRB v. Pioneer Natural Gas Co., 397 F.2d 573, 576 (5th Cir. 1968); but cf. Friendly Ice Cream Corp. v. NLRB, 705 F.2d 570, 576 (1st Cir. 1983). The Board gave no reasons for changing course in this case -- and in fact denied doing so. Without referring to the presumption in favor of a lithographic-production unit, the Board cited two of its cases that had relied on the presumption (Moore and Earl Litho), and it pointed out that there were factual differences between those cases and the present case. In particular, pressmen and preparatory employees spent more time in each other's work areas in those cases and had a common supervisor. The Board cited no other cases. The Regional Director in his opinion in this case had gone so far as to say that the Board has a long-established policy of recognizing pressmen as a separate unit. For this surprising proposition he cited only two cases, and in neither is there any reference to preparatory employees. See Kellogg Co., 104 N.L.R.B. 302 (1953); Metropolitan Imprinters Corp., 133 N.L.R.B. 1433 (1961). In the second case the employees excluded from the unit are described as "binders, finishers, and shippers." Id. at 1434.
We are sympathetic, though, to the technique used by the Board. It is, after all, the common law technique. The policy of combining pressmen and preparatory employees in a single unit emerged in a series of cases, and there is always a peril of over-generalization when policy is made this way. A court focusing in common law manner on the facts of particular cases is quite likely not to foresee a future case with different facts that might falsify the generalization. If such a case arises, the court may quite properly be moved to refine its statement of general policy. This is how law grows, not only in courts but in agencies like the Labor Board that rely heavily on the common law method. Finding distinctions is not reversing course; it is not like first deciding that cars must be equipped with airbags and then that they need not be; it calls for no special explanation. See Illinois Bell Tel. Co. v. FCC, supra, slip op. at 8-9. Here, however, by a course of adjudications extending over many years, the Board succeeded in developing a clear policy that seems applicable to this case; and to discard the policy without explanation would be arbitrary, especially when done with retroactive effect. Continental Web had no notice that the Board was going to depart from its long-standing policy on lithographic units. The Board could have prevented surprise by using its dormant rulemaking powers. See NLRB v. Majestic Weaving Co., 355 F.2d 854, 859-61 (2d Cir. 1966) (Friendly, J.).
The Board did not admit it was abandoning a policy. It did not mention the policy. All it said was that the facts of this case were different from those in Moore and Earl Litho. It did not explain why these factual differences should lead to a difference in legal result or even why, as an original matter, the facts here justified a unit limited to pressmen. The Board's earlier cases had not suggested that the policy of combining pressmen and preparatory employees into one unit rested on the specific production processes of the employers in those cases. The Board had seemed to think that the particulars were not important. It was free to change its mind about this; and it was not required to confess error in doing so. All it had to do was give a plausible reason why differences that had seemed unimportant for many years actually had determinative significance -- why a unit that it had again and again found to be homogeneous should be broken into subunits. But it failed to do this. Maybe an implicit rule or policy can be modified implicitly, as we suggested in Rockford League of Women Voters v. NRC, 679 F.2d 1218, 1222 (7th Cir. 1982), but an explicit rule requires more.
Nor did the Board say that the two cases it cited and discussed were typical cases, while this case was atypical and thus fell outside the boundaries of its rule. The Board could not have said this. Continental's lithographic production process, as we have said, appears to be standard. The facts of this case apparently are the same as the facts of many other cases that the Board has decided the opposite way.
To recapitulate: Irrespective of the Board's traditional policy of certifying lithographic production units rather than separate units of pressmen and preparatory employees, its action in separating the two groups in this case seems arbitrary because of what appears to be the virtually complete community of interest between them. The Board gave no reason why it nevertheless thought a separate pressmen's unit appropriate. And the Board did not explain why it was not applying its traditional policy towards units of lithographic production workers. We have to conclude that its decision was arbitrary.
Ordinarily the appropriate remedy in a case where the agency has failed to explain itself adequately would be to remand the case, to give the Board a chance to explain better than it has why a pressman unit is appropriate in the circumstances. We shall do this, but with one qualification. Delays in this proceeding that are the Board's responsibility rather than the employer's persuade us that if the Board on remand again orders the company to bargain with the union, and then asks us to enforce its order, we will turn the Board down because of unreasonable delay. The election was held more than four years ago; and if the Board on remand again enters a bargaining order and the company seeks judicial review of the order, it will probably be six years after that election before the order takes effect and bargaining begins, even if the proceedings on remand are conducted with all dispatch and the Board's order on remand is upheld on judicial review.
The Board impounded the ballots on the day of the election pending consideration of the company's challenge to the unit determination -- a challenge having considerable merit, as we have seen -- but did not issue its decision upholding the unit determination until more than two years later, on June 28, 1982. The Board admitted that this delay was due in part to its having misplaced certain exhibits; but whatever it was due to, it was not due to anything the company did. Meanwhile, during the two years that the ballots were sitting uncounted in the Board's regional office, the number of pressmen in Continental Web's plant doubled, to 72; and shortly after the Board declared the union the winner of the election, the company submitted a petition signed by 59 of the 72 pressmen stating that they did not want the union to represent them.
All of these things considered, the probability that the union still has the support of a majority of the pressmen is small; and if two years from now we were to uphold an order that the company bargain with the union, the probability that it would be a minority union would be great. If the delay were attributable in whole or significant part to the company, we would enforce the order nonetheless so as not to encourage delaying tactics. But the company challenged the unit determination in timely fashion, and it was a challenge made in good faith and pursued with diligence. The Board has only itself to blame for having let this proceeding stretch out to the point where it is quite likely that a bargaining order would be irrelevant to the current conditions in the plant. When the Board asks a court of appeals to enforce the Board's order, it is appealing to the court's equitable powers and is subject to some, at least, of the traditional defenses to requests for equitable relief -- and among them is the defense of laches, or unreasonable delay (on which see generally Piper Aircraft Corp. v. Wag-Aero, Inc., 741 F.2d 925 (7th Cir. 1984) (concurring opinion)). See Mosey Mfg. Co. v. NLRB, 701 F.2d 610, 613 (7th Cir. 1983) (en banc); NLRB v. Connecticut Foundry Co., 688 F.2d 871, 881 (2d Cir. 1982); NLRB v. Nixon Gear, Inc., 649 F.2d 906, 914 (2d Cir. 1981).
If, however, the Board on remand again decides that Continental Web's pressmen constitute an appropriate unit (and this time explains why), and merely orders Continental Web to post notices of its having committed an unfair labor practice and to desist from such violations in the future, the Board's order would not be subject to the objection that it might force the company to negotiate with a union not supported by a majority of the workers in the bargaining unit. Nor would it be open to that objection if, in addition to ordering the company to cease and desist and to post notices, the Board ordered a new election. The Board may decide not to do this; it may conclude that since the circumstances have changed so, it should require the union if it still wants to represent these workers to conduct a new organizing campaign. But that is for the Board to decide.
* Hon. Richard B. Kellam, of the Eastern District of Virginia, sitting by designation.