Browning-Ferris Indus. of Cal., Inc. v. Nat'l Labor Relations Bd.

Millett, Circuit Judge:

*1199 Browning-Ferris Industries of California, Inc. operates one of the largest recycling plants in the world. To operate its plant, Browning-Ferris contracts with Leadpoint Business Services to provide workers to sort through the incoming material, clear jams that occur in the sorting process, and keep the sorting areas clean. In 2013, a local union petitioned to represent those workers as a bargaining unit under the National Labor Relations Act, see 29 U.S.C. § 159 (a), designating Browning-Ferris and Leadpoint as "joint employers" of the workers.

In concluding that Browning-Ferris and Leadpoint were joint employers of the workers in the petitioned-for unit, the National Labor Relations Board ruled that it would consider a putative joint employer's reserved right to control the workers at issue, as well as any indirect control exercised over the workers, as among a number of factors relevant to determining joint-employer status. Browning-Ferris challenges both of those aspects of the Board's test.

We hold that the right-to-control element of the Board's joint-employer standard has deep roots in the common law. The common law also permits consideration *1200 of those forms of indirect control that play a relevant part in determining the essential terms and conditions of employment. Accordingly, we affirm the Board's articulation of the joint-employer test as including consideration of both an employer's reserved right to control and its indirect control over employees' terms and conditions of employment. But because the Board did not confine its consideration of indirect control consistently with common-law limitations, we grant the petition for review in part, deny the cross-application for enforcement, dismiss without prejudice the application for enforcement as to Leadpoint, and remand for further proceedings consistent with this opinion.

I

A

Congress enacted the National Labor Relations Act of 1935, 29 U.S.C. § 151 et seq. , to "protect the right of workers to act together to better their working conditions," NLRB v. Washington Aluminum Co. , 370 U.S. 9 , 14, 82 S.Ct. 1099 , 8 L.Ed.2d 298 (1962), and to "promot[e] stable collective-bargaining relationships," Auciello Iron Works, Inc. v. NLRB , 517 U.S. 781 , 790, 116 S.Ct. 1754 , 135 L.Ed.2d 64 (1996). To that end, the Act mediates the relationship between "employees" and "employers" by, among other things, conferring upon employees a right to unionize, 29 U.S.C. § 157 , prohibiting employers from engaging in specified unfair labor practices, id. § 158(a), and imposing obligations on employers to collectively bargain with representatives of employees, id. § 158(d). The National Labor Relations Board is charged with administering the Act. Id. § 153; NLRB v. SW General, Inc., --- U.S. ----, 137 S.Ct. 929 , 937, 197 L.Ed.2d 263 (2017).

But how do those statutory obligations on employers work when an employee has more than one putative employer? After all, a Board order that an employer bargain with a union over the terms and conditions of employment may well be futile if another entity, not subject to an order to bargain, exercises the final say over a working condition or has the power to override a choice negotiated in a collective-bargaining agreement. See Herbert Harvey, Inc. v. NLRB , 385 F.2d 684 , 686 (D.C. Cir. 1967) (discussing such a situation). To address that not-uncommon scenario, the Board has long recognized that two entities may be joint employers in the eyes of the National Labor Relations Act. See, e.g. , Franklin Simon & Co. , 94 N.L.R.B. 576 , 579 (1951). This case involves the standard that the Board applies in making that joint-employer determination.

On this point, the National Labor Relations Act gives no direct guidance. The Act provides no relevant definition of "employer," let alone of "joint employer." See 29 U.S.C. § 152 (2) (providing only that the term "employer" "includes any person acting as an agent of an employer, directly or indirectly" and excluding listed entities not relevant here).

The Supreme Court, meanwhile, has addressed the question of joint-employer status under the Act only once. In Boire v. Greyhound Corp. , 376 U.S. 473 , 84 S.Ct. 894 , 11 L.Ed.2d 849 (1964), the Court held that a putative joint employer must "possess[ ] sufficient control over the work of the employees to qualify as a joint employer," id. at 481 , 84 S.Ct. 894 . That inquiry, the Court stressed, is essentially "factual," and is not controlled by the fact that one putative employer is an independent contractor of another. See id.

In the years that followed, the test that courts and the National Labor Relations *1201 Board applied to determine joint-employer status resisted consistency or reliable delineation. Compare, e.g. , Springfield Ret. Residence , 235 N.L.R.B. 884 , 891 (1978) (finding joint-employer status where employer had the power to hire and fire), with, e.g. , Mobil Oil Corp. , 219 N.L.R.B. 511 , 515-516 (1975) (finding joint-employer status where employer had the power to set working conditions and make personnel decisions).

Almost twenty years later, the Third Circuit articulated a standard around which both the Board and courts began to coalesce. In NLRB v. Browning-Ferris Industries of Pennsylvania, Inc. , 691 F.2d 1117 (3d Cir. 1982), the Third Circuit ruled that separate business entities are joint employers if they each "exert significant control over the same employees" in that they "share or co-determine those matters governing essential terms and conditions of employment," id. at 1124 ; see also id. at 1123 . The Board soon adopted that same articulation of the test. See TLI, Inc. , 271 N.L.R.B. 798 , 798 (1984) ; Laerco Transp. & Warehouse , 269 N.L.R.B. 324 , 325 (1984).

This court's test for joint-employer status, like that of a number of other circuits, echoes the Third Circuit's standard, holding that "[t]wo separate entities may be joint employers of 'a single * * * [work force] if they share or co-determine those matters governing [the] essential terms and conditions of employment,' " Dunkin' Donuts Mid-Atlantic Distrib. Ctr., Inc. v. NLRB , 363 F.3d 437 , 440 (D.C. Cir. 2004) (quoting Aldworth Co. , 338 N.L.R.B. 137 , 139 (2002) ). See also 3750 Orange Place Ltd. P'ship v. NLRB , 333 F.3d 646 , 660 (6th Cir. 2003) ; Holyoke Visiting Nurses Ass'n v. NLRB , 11 F.3d 302 , 306 (1st Cir. 1993).

Following Laerco and TLI , however, the Board added additional requirements that constricted the joint-employer test. For one thing, the Board said that a joint-employer relationship depends on evidence of the actual exercise of control by each employer, not merely a reserved right to control. See AM Property Holding Corp. , 350 N.L.R.B. 998 , 1000 (2007) (Board "does not rely merely on the existence of * * * contractual provisions" to determine whether a joint-employer relationship exists, but "rather looks to the actual practice of the parties"). In addition, the Board held that "[t]he essential element in [the] analysis is whether a putative joint employer's control over employment matters is direct and immediate." In re Airborne Freight Co. , 338 N.L.R.B. 597 , 597 n.1 (2002). For several years, then, the Board would rely in analyzing joint-employer claims only on evidence of (i) actual control, as opposed to the right to control, and (ii) direct and immediate control, not indirect control. See NLRB v. CNN America, Inc. , 865 F.3d 740 , 748-749 (D.C. Cir. 2017).

The Board's decision in this case changed both of those factors by making the right to control and indirect control relevant considerations in determining joint employer status.

B

Browning-Ferris operates the Newby Island Recyclery in Milpitas, California. As one of the largest recycling facilities in the world, Newby Island receives approximately 1,200 tons of mixed materials, waste, and recyclables every day. Inside the facility, four conveyor belts-called "sort lines" or "material streams"-carry different categories of materials that must be sorted so that the remaining portion may be recycled.

This case involves three groups of Newby Island workers: sorters, screen cleaners, *1202 and housekeepers. Sorters, as the title suggests, remove and sort non-recyclable materials from the stream lines coming into the facility. Screen cleaners clear jams in the sort lines. Housekeepers clean the areas around the sort lines.

Browning-Ferris, by itself, employs approximately sixty workers at Newby Island. Most of those individuals work outside of the facility as loader operators, equipment operators, forklift operators, and sort-line equipment operators. One of those Browning-Ferris employees, however, is a sorter. Browning-Ferris also has supervisors who oversee and manage the operations of its employees. While Browning-Ferris employs the one sorter, it does not by itself employ the other sorters, or any screen cleaners or housekeepers. Instead, Browning-Ferris contracts with a staffing agency to provide those workers.

In 2009, Browning-Ferris entered into an exclusive service contract with Leadpoint, known as the Temporary Labor Services Agreement ("Agreement"), to staff Newby Island's sorting, screen cleaning, and housekeeping positions. Leadpoint provides approximately 240 workers for Browning-Ferris's Newby Island plant, most of whom fill the sorting, screen cleaning, and housekeeping positions. In addition, Leadpoint employs its own onsite managers and supervisors who oversee the sorters, screen cleaners, and housekeepers.

Under the Agreement, Leadpoint handles the hiring of workers from start to finish, but must ensure that the sorters, screen cleaners, and housekeepers at Newby Island meet certain conditions and qualifications required by Browning-Ferris in the Agreement. Those conditions include passing a "five-panel urinalysis drug screen" or equivalent drug test, and "hav[ing] the appropriate qualifications * * *, consistent with all applicable laws and instructions from [Browning-Ferris], to perform the general duties of the assigned position." J.A. 19. The Agreement further provides that Leadpoint workers cannot be assigned to Newby Island for more than six months at a time. But evidence in the record indicates that the time limit is not consistently enforced and some Leadpoint workers have continued working for more than six months.

Leadpoint "has the sole responsibility to counsel, discipline, review, evaluate, determine pay rates, and terminate" the workers that it provides to Browning-Ferris. J.A. 20. Browning-Ferris "reserves the right to ensure that" personnel from Leadpoint work "free from the effects of alcohol and illegal drug use." Id. Browning-Ferris also "may reject" or "discontinue the use of" a worker at its facility "for any or no reason." J.A. 21.

Leadpoint is responsible for paying the workers, as well as providing their benefits and unemployment insurance. Leadpoint determines the wages the workers will be paid, and it sends Browning-Ferris weekly invoices documenting the services performed and the total hours clocked by the workers. While Browning-Ferris generally has no direct input on the wages that Leadpoint pays, a March 2013 increase in the local minimum wage prompted Leadpoint and Browning-Ferris to amend the Agreement's wage schedule to comply with the new law. In addition, the Agreement provides that Leadpoint workers may not, without approval from Browning-Ferris, earn a higher wage than that earned by any Browning-Ferris worker performing similar tasks. The lone Browning-Ferris sorter earns approximately five dollars more per hour than all of the Leadpoint sorters.

For all workers at Newby Island, Browning-Ferris has determined that there will be three shifts per day, and it *1203 sets the hours for those shifts. Each shift lasts approximately eight hours, but may occasionally run into overtime. In addition, Browning-Ferris supervisors decide daily which of the four sort lines will run and provide Leadpoint supervisors with a target headcount of how many workers will be needed to operate those lines. Browning-Ferris does not decide which workers will work on which sort lines or during which shifts; Leadpoint makes that call. If Browning-Ferris supervisors determine that a sort line will run overtime, they convey that information to Leadpoint supervisors, who then make the necessary staffing arrangements.

The Board found inconsistencies in the frequency and nature of Browning-Ferris supervisors' communications with the workers. Some Browning-Ferris supervisors testified that their only direct communication with the workers involved referring the workers and any problems they raised to Leadpoint supervisors. According to those Browning-Ferris supervisors, they did not directly or specifically instruct those workers on how to perform their jobs. Instead, if they spotted something untoward, they would just tell Leadpoint supervisors "that there's a problem." J.A. 141. For example, the sorting lines are designed with an emergency stop switch to halt the flow of materials. One Browning-Ferris supervisor explained that he and his colleagues generally instruct Leadpoint supervisors, not the workers, on when the emergency stop switch can be used. They left it up to the Leadpoint supervisors to convey that information to the sorters.

Some workers at the Newby facility had different experiences. They testified that Browning-Ferris supervisors would occasionally direct the workers' removal of materials from the sort lines or their cleaning of certain areas, and would also warn them against pressing the emergency stop switch too frequently. In addition, a Browning-Ferris supervisor admitted that he had at times held informal meetings with sorters to teach them how to differentiate between organic and inorganic material on the sort lines.

Although the Agreement makes Leadpoint ultimately responsible for disciplining the workers it provides, Browning-Ferris has, on occasion, alerted Leadpoint supervisors to incidents that Browning-Ferris believed warranted disciplinary action. For example, in June 2013, a Browning-Ferris supervisor, Paul Keck, sent an email "request[ing] the[ ] immediate dismissal" of a worker seen passing a bottle of alcohol and the worker to whom it was passed. J.A. 34. A Leadpoint supervisor questioned both workers and sent them to a clinic for drug and alcohol testing. Based on the results of the testing, one of the workers was terminated from Leadpoint's employ, and the other continued to work for Leadpoint, but was reassigned to another company's facility. Keck later testified that he did not know what action Leadpoint had taken with respect to those two workers, although he noticed that one was no longer at Newby and was unsure about the other.

In that same e-mail, Keck informed the Leadpoint supervisor that he had reviewed video surveillance tapes showing a Leadpoint worker damaging a wall mount. Keck closed the e-mail by stating: "I hope you'll agree [that] this Leadpoint employee should be immediately dismissed." J.A. 34. Following the e-mail, Leadpoint supervisors first suspended and then terminated the worker involved for destroying or defacing property. Keck again testified that he did not follow up to learn what happened to that employee.

On another occasion, Keck advised a Leadpoint supervisor that the size of a pre-sort line should be reduced by two *1204 workers per shift, and that two other workers on the pre-sort line should be repositioned. The e-mail closed with: "This staffing change is effective Monday, August 5, 2013." J.A. 32.

C

1

In July 2013, the International Brotherhood of Teamsters Local 350 ("Union") filed a petition with the Board seeking to represent a new bargaining unit consisting of "full time and regular part-time employees" that were "employed by [Leadpoint] and [Browning-Ferris], joint employers," at Newby Island. J.A. 344. As relevant here, the petitioned-for unit included Leadpoint sorters, housekeepers, and screen cleaners, but not Leadpoint supervisors. At the time, the Union already represented a separate bargaining unit consisting of the sixty workers at Newby Island directly employed by Browning-Ferris, including the sole Browning-Ferris sorter.

After an evidentiary hearing, the Acting Regional Director concluded that Browning-Ferris and Leadpoint were not *1205 joint employers of the workers in the petitioned-for bargaining unit. Instead, the Director concluded that employees of Leadpoint alone composed the appropriate bargaining unit, and directed that an election be held for that unit. In the Director's view, the evidence was insufficient to establish that Browning-Ferris controlled or co-determined those matters governing the essential terms and conditions of the workers' employment, such as wages, benefits, hiring, discipline, termination, daily work responsibilities, and shift schedules.

The Union filed a petition for review, and the Board solicited briefing from the parties and any interested amici on whether the joint-employer test should be updated and how it should apply in this case. On August 27, 2015, the Board issued a decision concluding that Browning-Ferris and Leadpoint are joint employers of the workers in the petitioned-for bargaining unit. Browning-Ferris Indus. of Cal., Inc. , 362 N.L.R.B. No. 186, at 2 (Aug. 27, 2015). In so ruling, the Board "restate[d]" and "reaffirm[ed]" its longstanding joint-employer standard, adopted from the Third Circuit's Browning-Ferris decision, under which "two or more statutory employers are joint employers of the same statutory employees if they 'share or codetermine those matters governing the essential terms and conditions of employment.' " Id. (citation omitted).

In applying that test, the Board announced for the first time that it would subdivide the inquiry, asking first "whether there is a common-law employment relationship with the employees in question." Browning-Ferris , 362 N.L.R.B. No. 186, at 2. If so, the Board would ask secondly "whether the putative joint employer possesses sufficient control over employees' essential terms and conditions of employment to permit meaningful collective bargaining." Id. In applying both prongs of that test, the Board announced that it would "no longer require that a joint employer not only possess the authority to control employees' terms and conditions of employment, but also exercise that authority." Id. Nor would the Board anymore demand that "a statutory employer's control * * * be exercised directly and immediately" "to be relevant to the joint-employer inquiry." Id. Instead, the Board would consider both reserved control and indirect control as potentially "probative" in the joint-employer analysis. See id. at 2, 13, 16, 17 n.94.

Applying that test, the Board concluded that Browning-Ferris and Leadpoint were joint employers of the workers in the petitioned-for bargaining unit. Browning-Ferris , 362 N.L.R.B. No. 186, at 20. Among the evidence the Board viewed as demonstrating Browning-Ferris's control were Keck's reports of misconduct by workers and requests for their discipline and removal; Browning-Ferris's control over the speed of the sort lines, including direct admonitions to workers to sort faster, work smarter, and not stop the sort lines; and the contractual condition that workers earn no more than Browning-Ferris employees performing similar work. Id. at 18-20.

Two members of the Board dissented. In their view, the requirements that control actually be exercised and that it be direct and immediate were required by the common law of agency. See Browning-Ferris , 362 N.L.R.B. No. 186, at 28-32 (Members Miscimarra & Johnson, dissenting). The dissent also expressed concern that retroactive application of the new aspects of the test would disrupt the longstanding expectations of parties who had structured their labor relationships based on the Board's previous joint-employer standard. See id. at 22-23.

Browning-Ferris timely petitioned for review of the Board's order, while the Board cross-applied for enforcement of the order against Browning-Ferris and separately applied for enforcement of the order against Leadpoint. 1

2

While this case was pending, the Board again changed course on the joint-employer question. In Hy-Brand Industrial Contractors, Ltd. , 365 N.L.R.B. No. 156 (Dec. 14, 2017) (later overruled by Hy-Brand Industrial Contractors, Ltd. , 366 N.L.R.B. No. 26 (Feb. 26 2018) ), the Board expressly overruled its Browning-Ferris decision and announced that "a finding of joint-employer status" would require (1) "proof that the alleged joint-employer entities have actually exercised joint control over essential employment terms (rather than merely having 'reserved' the right to exercise control)," (2) the control exercised "must be 'direct and immediate' (rather than indirect)," and (3) "joint-employer status will not result from control that is 'limited and routine.' " Id. at 35.

Following the Hy-Brand decision, the Board moved this court to remand Browning-Ferris's case to the agency for further consideration. We granted that motion on December 22, 2017.

While that remand was still pending before the Board, an investigation conducted by the Board's Inspector General uncovered that one of the Board members that decided the Hy-Brand case was a shareholder in the law firm that represented Leadpoint before the Board in Browning-Ferris . On that basis, the Inspector General concluded that the Member's participation in the Hy-Brand decision amounted to "a serious and flagrant problem and/or deficiency in the Board's administration of its deliberative process." Memorandum of NLRB Inspector General David P. Berry (Feb. 9, 2018), available at https://www.nlrb.gov/who-we-are/inspector-general. The Inspector General explained that "the practical effect of the Hy-Brand deliberative process was a 'do *1206 over' for the Browning-Ferris parties," and so that Member should have recused himself. Id. at 2, 5.

In light of the Inspector General's report, the Board unanimously vacated its Hy-Brand decision and announced that "the overruling of the Browning-Ferris decision is of no force or effect." Hy-Brand Industrial Contractors, Ltd. , 366 N.L.R.B. No. 26 (Feb. 26, 2018). The Board then moved this court to recall its remand mandate and asked this court to proceed with resolving Browning-Ferris's petition for review and the Board's cross-application for enforcement. We granted that motion on April 6, 2018, and recalled our mandate, but held the case in abeyance pending the Board's disposition of Hy-Brand's motion for reconsideration. The Board denied reconsideration two months later. Hy-Brand Industrial Contractors, Ltd. , 366 N.L.R.B. No. 93 (June 6, 2018).

On May 9, 2018, the Board announced its plan to undertake a rulemaking on the standard for joint-employer status. The Board was explicit that any new rule that might result from that process would be prospective only. Browning-Ferris Mot. to Remand at 9, 12 (June 13, 2018).

In June 2018, the Board specifically requested that this court proceed to decide the case, notwithstanding the pending rulemaking. See Board Opp. to Mot. to Remand (June 15, 2018); see also Board Mot. to Govern Future Proceedings (June 13, 2018); Tr. of Oral Argument at 13 (July 3, 2018).

On September 14, 2018, the Board published its notice of proposed rulemaking that suggested reinstating its prior "direct and immediate control" test for joint-employer status. "[T]o be deemed a joint employer under the proposed regulation, an employer must possess and actually exercise substantial direct and immediate control over the employees' essential terms and conditions of employment of another employer's employees in a manner that is not limited and routine." 83 Fed. Reg. 46681 , 46686 (Sept. 14, 2018).

Since issuing its proposed rule, the Board has reiterated its request that this court resolve the pending petitions for review in this case. See Letter from Linda Dreeben, Deputy Associate General Counsel, National Labor Relations Board to Mark J. Langer, Clerk of Court, U.S. Court of Appeals for the District of Columbia Circuit (September 19, 2018).

II

We start with the question of what, if any, deference is owed to the Board's adjustments to the joint-employer standard. The Board claims that its "reasonable" judgment merits "considerable deference." See Board Br. 16 (citations omitted); cf. Chevron, U.S.A., Inc. v. Natural Resources Def. Council, Inc. , 467 U.S. 837 , 842-844, 104 S.Ct. 2778 , 81 L.Ed.2d 694 (1984) (courts defer to an agency's "reasonable interpretation" of ambiguous terms in a statute administered by the agency). Browning-Ferris says that the Board gets no deference. We hold that, to the extent that the Board's joint-employer standard is predicated on interpreting the common law, Browning-Ferris is correct. The content and meaning of the common law is a pure question of law that we review de novo without deference to the Board.

Under Supreme Court and circuit precedent, the National Labor Relations Act's test for joint-employer status is determined by the common law of agency. The Supreme Court has often held that, when Congress leaves undefined statutory terms like "employee" and "employer" that have longstanding common-law meanings, courts should presume that Congress intended *1207 to incorporate those meanings, unless the statute, directs otherwise. See Microsoft Corp. v. i4i Ltd. P'ship , 564 U.S. 91 , 103, 131 S.Ct. 2238 , 180 L.Ed.2d 131 (2011) ("Where Congress uses terms that have accumulated settled meaning under * * * the common law, [we] must infer, unless the statute otherwise dictates, that Congress means to incorporate the established meaning of those terms.") (alterations in original) (quoting Neder v. United States , 527 U.S. 1 , 21, 119 S.Ct. 1827 , 144 L.Ed.2d 35 (1999) ); Community for Creative Non-Violence v. Reid , 490 U.S. 730 , 739-740, 109 S.Ct. 2166 , 104 L.Ed.2d 811 (1989) ("[W]hen Congress has used the term 'employee' without defining it, * * * Congress intended to describe the conventional master-servant relationship as understood by common-law agency doctrine."); id. (citing additional cases holding that "employee," "employer," and "scope of employment" must be interpreted in light of agency law).

That presumption applies with full force to the employer-employee relationship under the National Labor Relations Act. In NLRB v. Hearst Publications, Inc. , 322 U.S. 111 , 64 S.Ct. 851 , 88 L.Ed. 1170 (1944), the Supreme Court bypassed the common-law meaning of "employee" in favor of a definition that potentially swept in independent contractors, reasoning that the latter definition better advanced the policies underlying the National Labor Relations Act, see id. at 131-132 , 64 S.Ct. 851 . Congress promptly and emphatically rejected that approach, amending the Act to specifically exclude "independent contractors" from the Act's definition of "employees." See Labor Management Relations Act of 1947, Pub. L. 80-101, 61 Stat. 136 (codified as amended at 29 U.S.C. §§ 141 - 197 ) ("Taft-Hartley Amendments"). "The obvious purpose" of the Taft-Hartley Amendments, the Supreme Court later ruled, "was to have the Board and the courts apply general [common-law] agency principles in distinguishing between employees and independent contractors under the Act." NLRB v. United Insurance Co. of America , 390 U.S. 254 , 256, 88 S.Ct. 988 , 19 L.Ed.2d 1083 (1968) ; see also Nationwide Mut. Ins. Co. v. Darden , 503 U.S. 318 , 324-325, 112 S.Ct. 1344 , 117 L.Ed.2d 581 (1992) (explaining the congressional reaction to Hearst ).

For purposes of determining our standard of review, the lesson from the Taft-Hartley Amendments and United Insurance is that Congress delegated to the Board the authority to make tough calls on matters concerning labor relations, but not the power to recast traditional common-law principles of agency in identifying covered employees and employers. Instead, the inquiry into the content and meaning of the common law is a "pure" question of law, and its resolution requires "no special administrative expertise that a court does not possess." United Insurance , 390 U.S. at 260 , 88 S.Ct. 988 .

For that reason, we review the Board's interpretation of the common law de novo . See FedEx Home Delivery v. NLRB , 849 F.3d 1123 , 1128 (D.C. Cir. 2017) ("[T]his particular question [regarding who is an employee or independent contractor] under the Act is not one to which we grant the Board Chevron deference[.]"); cf. International Longshoremen's Ass'n v. NLRB , 56 F.3d 205 , 212 (D.C. Cir. 1995) (because the term "agent" in the Act "incorporat[es] common law agency principles," courts do not "defer to the agency's judgment as we normally might under [ Chevron ]").

That no-deference rule applies just as much to the common-law meaning of "employer" under the Act as it does to that of "employee." That is because both inquiries turn on pure questions of law about *1208 the scope of traditional common-law agency principles. Cf. Community for Creative Non-Violence , 490 U.S. at 739-740 , 109 S.Ct. 2166 . 2

The Board argues that, even if its articulation of the common law does not get full-fledged Chevron deference, the proper standard of review is still not de novo . Citing language in Atrium of Princeton, LLC v. NLRB , 684 F.3d 1310 (D.C. Cir. 2012), and International Longshoremen's Association , 56 F.3d at 212 , the Board argues that we must accept its understanding of the common law so long as it reflects a choice between "two fairly conflicting views." Board Br. 16 (citation omitted).

That is not correct. The "two fairly conflicting views" standard applies to the Board's application of the common law to the facts of a particular case-which is a mixed question of law and fact. It does not extend to the Board's articulation of the common law, which is a pure question of law. See FedEx , 849 F.3d at 1128 ; Aurora Packing Co. v. NLRB , 904 F.2d 73 , 75 (D.C. Cir. 1990) ("[D]eference would only be extended to the Board's determination of employee status-an 'application of law to fact'-insofar as [the Board] made a 'choice between two fairly conflicting views' in a particular case.") (quoting United Insurance , 390 U.S. at 260 , 88 S.Ct. 988 ). Our decisions in Atrium of Princeton and International Longshoremen's Association are of the same mind. See Atrium of Princeton , 684 F.3d at 1315-1316 (rejecting the Board's formulation of the relevant common-law agency standard and effectively applying de novo analysis of the common law); International Longshoremen's Ass'n , 56 F.3d at 213 (finding no dispute as to the "fundamental principle of hornbook agency law" that governed, and applying the "two fairly conflicting views" standard only to the Board's application of the law to the facts). We also note that the Board's decision in Hy-Brand agreed that courts owe its interpretation of the common law no deference. Hy-Brand , 365 N.L.R.B. No. 156 at 4.

For those reasons, we review de novo whether the Board's joint-employer test comports with traditional common-law principles of agency.

Finally, it is precisely because Congress has tasked the courts, and not the Board, with defining the common-law scope of "employer" that this court accepts the Board's repeated request that we resolve this case notwithstanding the pending rulemaking. The policy expertise that the Board brings to bear on applying the National Labor Relations Act to joint employers is bounded by the common-law's definition of a joint employer. The Board's rulemaking, in other words, must color within the common-law lines identified by the judiciary. That presumably is why the Board has thrice asked this court to dispose of the petitions in this case during its *1209 rulemaking process. Like the Board, and unlike the dissenting opinion (at pp. 4-8), we see no point to waiting for the Board to take the first bite of an apple that is outside of its orchard.

III

The Board was certainly correct that, for roughly the last 25 years, the governing framework for the joint-employer inquiry has been whether both employers "exert significant control over the same employees" in that they "share or co-determine those matters governing the essential terms and conditions of employment." Browning-Ferris , 691 F.2d at 1124 . This court so held in Dunkin' Donuts , 363 F.3d at 440 .

The question in this case is whether the common-law analysis of joint-employer status can factor in both (i) an employer's authorized but unexercised forms of control, and (ii) an employer's indirect control over employees' terms and conditions of employment. See Browning-Ferris , 362 N.L.R.B. No. 186, at 2. In answering that question, we look first and foremost to the "established" common-law definitions at the time Congress enacted the National Labor Relations Act in 1935 and the Taft-Hartley Amendments in 1947, Microsoft , 564 U.S. at 103 , 131 S.Ct. 2238 (citation omitted). See Field v. Mans , 516 U.S. 59 , 70, 116 S.Ct. 437 , 133 L.Ed.2d 351 (1995) ("look[ing] to the [common-law] concept of 'actual fraud' as it was understood in 1978 when that language was added to [the statute]").

We conclude that the Board's right-to-control standard is an established aspect of the common law of agency. The Board also correctly determined that the common-law inquiry is not woodenly confined to indicia of direct and immediate control; an employer's indirect control over employees can be a relevant consideration. The Board in Hy-Brand , in fact, agreed that both reserved and indirect control are relevant considerations recognized in the common law. See Hy-Brand, 365 N.L.R.B. No 156 at 4. In applying the indirect-control factor in this case, however, the Board failed to confine it to indirect control over the essential terms and conditions of the workers' employment. We accordingly remand that aspect of the decision to the Board for it to explain and apply its test in a manner that hews to the common law of agency.

A

1

The Board's conclusion that joint-employer status considers not only the control an employer actually exercises over workers, but also the employer's reserved but unexercised right to control the workers and their essential terms and conditions of employment, finds extensive support in the common law of agency.

First , this court has already squarely addressed that common-law question. In International Chemical Workers Union Local 483 v. NLRB , 561 F.2d 253 (D.C. Cir. 1977), this court was explicit that "[w]hether [two entities are] joint employers" under the National Labor Relations Act "depends upon the amount of actual and potential control that" the putative joint employer "ha[s] over the * * * employees," id. at 255 (emphasis added). That inquiry, we added, "depend[s] upon the amount of and nature of control that [the putative employer] exercise[s] and [ is ] authorized to exercise under the contract." Id. (emphasis added). This court's decision in International Chemical Workers is, of course, binding on this panel. See LaShawn A. v. Barry , 87 F.3d 1389 , 1393 (D.C. Cir. 1996) (en banc).

*1210 The rule established in International Chemical Workers also makes great sense. Retained but unexercised control has long been a relevant factor in assessing the common-law master-servant relationship. The Supreme Court has held that the reserved right to control certain aspects of the work underpins the common-law master-servant dynamic. See Chicago, Rock Island & Pac. Ry. Co. v. Bond , 240 U.S. 449 , 456, 36 S.Ct. 403 , 60 L.Ed. 735 (1916) (worker held not to be an employee because the company "did not retain the right to direct the manner in which the business should be done, as well as the results to be accomplished, or, in other words, did not retain control not only of what should be done, but how it should be done") (emphases added); Singer Mfg. Co. v. Rahn , 132 U.S. 518 , 523, 10 S.Ct. 175 , 33 L.Ed. 440 (1889) ("[T]he relation of master and servant exists whenever the employer retains the right to direct the manner in which the business shall be done, as well as the result to be accomplished[.]") (emphasis added). 3

State-court decisions applying the common law of agency are equally clear that unexercised control bears on employer status. That was the common-law rule at the time of the National Labor Relations Act's passage in 1935. 4 That was also the common-law rule at the time of the Taft-Hartley Amendments in 1947. 5 And, for what it is worth, it is still the common-law rule today. 6

*1211 In addition, the "right to control" runs like a leitmotif through the Restatement (Second) of Agency. It starts right out of the box with the definitional provision of the master-servant relationship: a "master" "controls or has the right to control the physical conduct of [another] in the performance of [a] service," RESTATEMENT (SECOND) OF AGENCY § 2(1), at 12 ( AM. LAW INST. 1958) (emphasis added), while a "servant" likewise "is controlled or is subject to the right to control by the master," id. § 2(2), at 12 (emphasis added). And that refrain keeps repeating. See id. § 14 cmt. a, at 60 ("The extent of the right to control the physical acts of the agent is an important factor in determining whether or not a master-servant relationship between them exists."); id. § 220(1), at 485; id. § 250 cmt. a, at 550 (identifying the "right to control physical details as to the manner of performance" as "characteristic of the relation of master and servant").

In short, "[a]t common law the relevant factors defining the master-servant relationship focus on the master's control over the servant," whether that means the servant " 'is controlled or is subject to the right to control by the master,' " and so that "common-law element of control is the principal guidepost" in determining whether an entity is an employer of another. Clackamas Gastroenterology Associates, P. C. v. Wells , 538 U.S. 440 , 448, 123 S.Ct. 1673 , 155 L.Ed.2d 615 (2003) (emphases added) (quoting RESTATEMENT (SECOND) OF AGENCY § 2(2) ).

Indeed, precedent is so clear on this point that Browning-Ferris admitted at oral argument that the Board "can consider" unexercised control as a relevant factor in the joint-employer determination. Oral Arg. Tr. at 11:2. The Board's subsequent decision in Hy-Brand agreed as well that reserved control may be one "indicia" that is "probative of joint-employer status" under the common law. Hy-Brand , 365 N.L.R.B. No. 156 at 4.

Second , consideration of unexercised control accords with the common law's analogous "dual master doctrine": the concept that "[a] person may," under certain circumstances, "be the servant of two masters * * * at one time as to one act," as long as "the service to one [master] does not involve abandonment of the service to the other," RESTATEMENT (SECOND) OF AGENCY § 226, at 498, and "the act is within the scope of his employment for both," id. § 226 cmt. a, at 499. In the comments to Section 226, the Restatement (Second) specifically notes that the " right of the [putative] master[s] to control the conduct of the servant" is determinative of whether the servant has two masters at the same time. Id. § 226 cmt. a, at 498 (emphasis added).

To be sure, Section 226 addresses situations in which an individual is a "servant of two masters, not joint employers ." RESTATEMENT (SECOND) OF AGENCY § 226, at 498 (emphasis added). But if unexercised control is relevant to identifying two distinct employers, that consideration logically applies to identifying simultaneous joint employers as well. Indeed, the Supreme Court has, in the context of the Federal Employers' Liability Act, 45 U.S.C. § 51 et seq. , identified the dual master doctrine as a "common-law" "method[ ] by which [an *1212 individual] can establish his 'employment' with [one entity] even while he is nominally employed by another." See Kelley v. Southern Pac. Co. , 419 U.S. 318 , 324, 95 S.Ct. 472 , 42 L.Ed.2d 498 (1974).

2

Browning-Ferris argues that the "most important" component of the employee-or-independent-contractor inquiry is the "extent of the actual supervision exercised." Browning-Ferris Br. 27 (emphases omitted) (quoting Aurora Packing , 904 F.2d at 76 ). Considering the independent-contractor inquiry to be "essentially the same" as the joint-employer inquiry, id. 31, Browning-Ferris tells us that we should import the same focus here. Both steps of that argument fail.

a

For starters, the common law's analysis of independent contractor status, if anything, has long agreed that "the right of control and not [merely] the exercise of that right * * * is relevant" to establishing that a worker is an employee rather than an independent contractor. Local 814, Int'l Bhd. of Teamsters v. NLRB , 512 F.2d 564 , 571 n.13 (D.C. Cir. 1975) (emphasis added); see, e.g. , Construction, Bldg. Material, Ice & Coal Drivers, Helpers & Inside Employees Union, Local No. 221 v. NLRB , 899 F.2d 1238 , 1242 (D.C. Cir. 1990) (R.B. Ginsburg, J.) ("The right to control the 'means and manner' of job performance * * * is * * * recurrent in the cases in point" addressing employee versus independent-contractor status) (emphasis added); Dovell v. Arundel Supply Corp. , 361 F.2d 543 , 544 (D.C. Cir. 1966) ("The decisive test in determining whether the relation of master and servant exists is whether the employer has the right to control and direct the servant in the performance of his work and in the manner in which the work is to be done. It will be noted from the above, it is not the manner in which the alleged master actually exercised his authority to control and direct the action of the servant which controls, but it is his right to do so that is important."); Gracev. Magruder , 148 F.2d 679 , 681 (D.C. Cir. 1945) ("The vital element which negatives such independence, in the relation between employer and employee, is the right to control the employee, not only as to the final result, but in the performance of the task itself. And, it is the right to control, not control or supervision itself, which is most important."); RESTATEMENT (SECOND) OF AGENCY § 220 (1958) (defining an independent contractor as "a person who contracts with another to do something for him but who is not controlled by the other nor subject to the other's right to control with respect to his physical conduct in the performance of the undertaking.") (emphasis added); cf. Logue v. United States , 412 U.S. 521 , 527, 93 S.Ct. 2215 , 37 L.Ed.2d 121 (1973) ("[T]he modern common law as reflected in the Restatement of Agency * * * make[s] the distinction between the servant or agent relationship and that of independent contractor turn on the absence of authority in the principal to control the physical conduct of the contractor in performance of the contract.") (emphasis added). 7

*1213 Lastly, the parties and amici dispute the appropriateness of relying on the Restatement (Second) of Agency as a relevant source of common law. Some amici argue that the Restatement (Second)'s primary focus is on assigning liability for specific tortious conduct or breaches of contracts, not on determining the relationship between a putative employer and employee. Chamber of Commerce et al. Br. 22-23. Nevertheless, the Supreme Court has repeatedly relied on the Restatement (Second) to answer questions of employment under the common law of agency. See, e.g. , Community for Creative Non-Violence , 490 U.S. at 752 & n.31, 109 S.Ct. 2166 ("In determining whether a hired party is an employee under the general common law of agency, we have traditionally looked for guidance to the Restatement [ (Second) ] of Agency."); Town & Country , 516 U.S. at 94-95 , 116 S.Ct. 450 ; Darden , 503 U.S. at 324 , 112 S.Ct. 1344 .

This court too has relied specifically on Section 220 of the Restatement (Second) of Agency to determine whether a worker is an employee or independent contractor under traditional common-law principles in National Labor Relations Act cases. E.g. , FedEx , 849 F.3d at 1125 ; Lancaster Symphony Orchestra v. NLRB , 822 F.3d 563 , 565-566 (D.C. Cir. 2016) ; North American Van Lines v. NLRB , 869 F.2d 596 , 599-600 (D.C. Cir. 1989). Accordingly, controlling precedent makes the Restatement (Second) of Agency a relevant source of traditional common-law agency standards in the National Labor Relations Act context.

In any event, both the first Restatement of Agency and the Restatement (Third) of Agency also identify the "right to control" as a relevant factor in establishing a master-servant or employment relationship. RESTATEMENT OF AGENCY § 2(1)-(2), at 11 ( AM. LAW INST. 1933) (A "master" "controls or has the right to control the physical conduct of the other in the performance of [a] service," while a "servant" "is controlled or is subject to the right to control by the master[.]"); 2 RESTATEMENT (THIRD) OF AGENCY § 7.07(3)(a), at 198 ( AM. LAW INST. 2006) ("For purposes of this section, * * * an employee is an agent whose principal controls or has the right to control the manner and means of the agent's performance of work[.]").

In sum, the Board's conclusion that an employer's authorized or reserved right to control is relevant evidence of a joint-employer relationship wholly accords with traditional common-law principles of agency. And because the Board relied on evidence that Browning-Ferris both had a "right to control" and had "exercised that control," Browning-Ferris , 362 N.L.R.B. No. 186, at 18, this case does not present the question whether the reserved right to control, divorced from any actual exercise of that authority, could alone establish a joint-employer relationship.

b

Beyond all that, Browning-Ferris's contention that the joint-employer and independent-contractor tests are virtually identical lacks any precedential grounding. Browning-Ferris cites no case in which we have applied an employee-or-independent-contractor test to resolve a question of joint employment, and we have found none. Cf.

*1214 Redd v. Summers , 232 F.3d 933 , 938 (D.C. Cir. 2000) (noting in the Title VII context that "[t]his court has never invoked" the independent-contractor test "to resolve an issue of joint employment," but avoiding the issue). 8

That lack of precedent is understandable because, at bottom, the independent-contractor and joint-employer tests ask different questions. The independent-contractor test considers who, if anyone, controls the worker other than the worker herself. See Lancaster Symphony Orchestra , 822 F.3d at 566 . The joint-employer test, by contrast, asks how many employers control individuals who are unquestionably superintendent.

In this case, there is no question that the workers Leadpoint provides are employees of (at least) Leadpoint, not independent contractors. See Browning-Ferris Br. 31 n.14 ("It is undisputed that the persons in the petitioned-for bargaining unit are employees, not independent contractors."). Indeed, there is nothing independent at all about those employees' work lives.

In addition, an important aspect of the independent-contractor inquiry is whether the workers in question are operating their own independent businesses. See United Insurance , 390 U.S. at 258-259 , 88 S.Ct. 988 (listing whether workers "operate their own independent businesses" as a "decisive factor[ ]" in the employee-or-independent-contractor inquiry); RESTATEMENT (SECOND) OF AGENCY § 220(2)(b), at 485 (listing "whether or not the [worker] is engaged in a distinct occupation or business" as a factor in the employee-or-independent-contractor inquiry). That consideration is of no help to the joint-employer inquiry.

Similarly, under the Restatement (Second) of Agency, several of the factors that guide the employee-or-independent-contractor determination are aimed at characterizing the nature of the work performed. See , e.g. , RESTATEMENT (SECOND) OF AGENCY § 220(2)(c), at 485 (considering "the kind of occupation, with reference to whether, in the locality, the work is usually done under the direction of the employer or by a specialist without supervision"); id. § 220(2)(d), at 485 (considering "the skill required in the particular occupation"). Those factors shed no meaningful light on the question of Browning-Ferris's status here.

To be sure, as Browning-Ferris notes, both tests ultimately probe the existence of a common-law master-servant relationship. 9 And central to establishing a master-servant relationship-whether for purposes of the independent-contractor inquiry or the joint-employer inquiry-is the nature and extent of a putative master's control. 10 Accordingly, employee-or-independent-contractor *1215 cases can still be instructive in the joint-employer inquiry to the extent that they elaborate on the nature and extent of control necessary to establish a common-law employment relationship. Beyond that, a rigid focus on independent-contractor analysis omits the vital second step in joint-employer cases, which asks, once control over the workers is found, who is exercising that control, when , and how .

In short, using the independent-contractor test exclusively to answer the joint-employer question would be rather like using a hammer to drive in a screw: it only roughly assists the task because the hammer is designed for a different purpose.

c

The dissenting opinion is of the view that Leadpoint's purported status as an independent contractor per se resolves the issue before us, reasoning that employees of an independent contractor cannot be employees of the company that hired the contractor. See Dissent Op. 9. Controlling precedent says otherwise.

In Boire v. Greyhound Corp. , the only Supreme Court case to address the question of joint employer status, the Court was explicit that the joint employer inquiry is " unaffected by any possible determination" that one employer is an independent contractor of another employer. 376 U.S. at 481 , 84 S.Ct. 894 (emphasis added); id . ("Greyhound has never suggested that the employees [of the independent contractor] themselves occupy an independent contractor status.").

This court's precedent is of the same view. In Herbert Harvey v. NLRB , the World Bank hired Herbert Harvey Inc.-an independent contractor providing building repair services. 385 F.2d at 684-685 ; see Herbert Harvey, Inc. v. NLRB , 424 F.2d 770 , 775 (D.C. Cir. 1969) (noting that it was "plain" to the Board that the World Bank and Herbert Harvey contracted for "a completely independent relationship"). We nevertheless held that, as to Herbert Harvey's employees, the "record clearly shows a basis for finding that Harvey and the Bank are joint employers[.]" Id. ; see also International Chem. Workers Union Local 483 v. NLRB , 561 F.2d 253 , 256 (D.C. Cir. 1977) (explaining that an employer' status as an independent contractor is "not determinative" of the other putative employer's control over the employees at issue).

The dissenting opinion dismisses Boire as a decision about "jurisdiction." Dissenting Op. 1228 n.3. True. But in resolving the question of jurisdiction in Boire , the Supreme Court was explicit that the statutory carve-out from the National Labor Relations Act for independent contractors-and, importantly, a related jurisdictional exception-did not apply because the Board's jurisdiction was "unaffected" by Floors' independent-contractor status. Boire , 376 U.S. at 481 , 84 S.Ct. 894 . The Supreme Court's analysis of why the independent contractor's status did not solve Greyhound's jurisdictional problem, accordingly, was necessary to the decision. "When an opinion issues for the [Supreme] Court, it is not only the result but also those portions of the opinion necessary to that result by which we are bound." Seminole Tribe of Florida v. Florida , 517 U.S. 44 , 67, 116 S.Ct. 1114 , 134 L.Ed.2d 252 (1996).

So we take the Supreme Court at its word, as did the Fifth Circuit on remand in Boire , *1216 NLRB v. Greyhound Corp. , 368 F.2d 778 , 780-781 (5th Cir. 1966) (applying the Supreme Court's standard to hold that Greyhound and the independent contractor were joint employers), and the Third Circuit in its watershed joint-employer decision, Browning-Ferris , 691 F.2d at 1122-1123 . See also id. at 1123 (noting that, under Boire, Greyhound's status as a joint employer "is unaffected by any possible determination as to Floors' status as an independent contractor") (quoting Boire , 376 U.S. at 481 , 84 S.Ct. 894 ).

Lastly, the dissenting opinion cites to the 1925 edition of Corpus Juris for the proposition that:

An independent contractor is not the servant of his employer. The relation of master and servant does not exist between an employer and the servants of an independent contractor, nor between an independent contractor and the servant of a subcontractor, and he is not responsible as a master, either to or for them.

Dissenting Op. 1228 (quoting 39 C.J. Master and Servant § 8, at 37-38 (1925) ) (emphasis omitted).

As between Supreme Court precedent and Corpus Juris , we hew to the former. But as it turns out, we need not make that choice here because the cited passage does not stop where the dissenting opinion does. Corpus Juris adds in the very next sentence:

If, however, the employer retains or assumes control over the means and method by which the work of a contractor is to be done, the relation of master and servant exists between him and servants of such a contractor, and the mere fact of nominal employment by an independent contractor will not relieve the master of liability where the servant is in fact in his employ.

39 C.J. Master and Servant § 8, at 38 (emphasis added).

B

The Board also ruled that an employer's control need not "be exercised directly and immediately" "to be relevant to the joint-employer inquiry"; indicia of "indirect[ ]" control can also be considered. Browning-Ferris , 362 N.L.R.B. No. 186, at 2. The Board again correctly discerned the content of the common law-indirect control can be a relevant factor in the joint-employer inquiry. But in failing to distinguish evidence of indirect control that bears on workers' essential terms and conditions from evidence that simply documents the routine parameters of company-to-company contracting, the Board overshot the common-law mark.

1

a

Traditional common-law principles of agency do not require that "control * * * be exercised directly and immediately" to be " relevant to the joint-employer inquiry." Browning-Ferris , 362 N.L.R.B. No. 186, at 2 (emphasis added). In fact, the National Labor Relations Act itself expressly recognizes that agents acting "indirectly" on behalf of an employer could also count as employers. 29 U.S.C. § 152 (2) (the term "employer" "includes any person acting as an agent of an employer, directly or indirectly"). The Act thus textually indicates that the statute looks at all probative indicia of employer status, whether exercised "directly or indirectly." Id .

Browning-Ferris's proposed rigid distinction between direct and indirect control has no anchor in the common law. Neither Browning-Ferris nor the dissenting opinion cites any case holding that consideration of indirect control is forbidden. Nor have we found any. To the contrary, common-law decisions have repeatedly recognized *1217 that indirect control over matters commonly determined by an employer can, at a minimum, be weighed in determining one's status as an employer or joint employer, especially insofar as indirect control means control exercised "through an intermediary," id .

To begin with, courts applying the traditional common law of agency have explicitly considered indirect control as relevant to the existence of a master-servant relationship. See White v. Morris, 114 Ga.App. 618 , 152 S.E.2d 417 , 419 (1966) ("[E]vidence and inferences therefrom indicating [a putative employer's] indirect control * * * are relevant for consideration" of "the existence of a master-servant relationship," "because the alleged relationship can exist by virtue of indirect control of the servant's performance as well as by direct control."); Wallowa Valley Stages, Inc. v. Oregonian Pub. Co., 235 Or. 594 , 386 P.2d 430 , 433 (1963) (en banc) (finding sufficient evidence "that the [putative master] indirectly exercised some control over the detail of [the putative servant's] operations"), repudiated on other grounds by Woody v. Waibel, 276 Or. 189 , 554 P.2d 492 (1976) (en banc). 11

In particular, the common law has never countenanced the use of intermediaries or controlled third parties to avoid the creation of a master-servant relationship. See, e.g. , Nicholson v. Atchison, T. & S. F. Ry. Co. , 147 P. 1123 , 1126 (Kan. 1915) (putative master's use of "branch company" as a "mere instrumentality" "did not break the relation of master and servant existing between the plaintiff and the [putative master]"); 39 C.J. Master and Servant § 8, at 38 ("Where an independent contractor is created or is operating as a subterfuge, an employee will be regarded as the servant of the principal employer.").

Our cases too have considered indirect control relevant to employer status. See, e.g. , Dunkin' Donuts , 363 F.3d at 440 (in addition to direct control, joint employer's warehouse supervisor "reported his opinion about [warehouse applicants'] qualifications, which [contractor] generally followed," and joint employer's transportation manager "prevented hiring of [driver] applicants he did not approve"); Al-Saffy v. Vilsack , 827 F.3d 85 , 97 (D.C. Cir. 2016) (in Title VII context, this court cited as relevant evidence supporting reversal of summary judgment the fact that officials working for putative employer had recommended plaintiff's dismissal).

In addition, control that is exercised through an intermediary is a defining feature of the subservant doctrine. 12 Much as *1218 the joint-employer inquiry arises in situations in which an employee has multiple masters at the same time, the subservant doctrine analogously governs arrangements in which an employee has, as simultaneous masters, both "his immediate employer and [his immediate employer's] master." RESTATEMENT (SECOND) OF AGENCY § 226 cmt. a, at 499. Given the central role that indirect control plays in the subservant doctrine, there is no sound reason that the related joint-employer inquiry would give that factor a cold shoulder.

Even the now-vacated Board decision in Hy-Brand acknowledged that indirect control can be relevant to the joint employer question. Hy-Brand , 365 N.L.R.B. No. 156, at 4 ("Our fundamental disagreement with the Browning-Ferris test is not that it treats indicia of indirect, and even potential, control to be probative of joint-employer status, but that it makes such indicia potentially dispositive without any evidence of direct control in even a single area."). There is thus broad agreement that the common law factors indirect control into the analysis of employer status.

Accordingly, the Board's conclusion that it need not avert its eyes from indicia of indirect control-including control that is filtered through an intermediary-is consonant with established common law. And that is the only question before this court. Hy-Brand 's concern about whether indirect control can be "dispositive" is not at issue in this case because the Board's decision turned on its finding that Browning-Ferris exercised control "both directly and indirectly." Browning-Ferris , 362 N.L.R.B. No. 186, at 18 ; see also id. at 19 ("We find that all of these forms of control-both direct and indirect-are indicative of an employer-employee relationship.").

Browning-Ferris's argument that the common law of agency closes its mind to evidence of indirect control is unsupported by law or logic. First, Browning-Ferris points to a passage in the comments to Section 220 of the RESTATEMENT (SECOND) , which distinguishes employees from independent contractors, and argues that the relevant factors do "not look[ ] to indirect control." Browning-Ferris Br. 24 (quoting Browning-Ferris , 362 N.L.R.B. No. 186, at 29 (Members Miscimarra & Johnson, dissenting) ). In fact, the comments say nothing one way or the other about direct versus indirect control. All they demonstrate is the entirely uncontroversial proposition that the stronger the indicia of control, the clearer the indication of employee rather than independent-contractor status. See, e.g. , RESTATEMENT (SECOND) OF AGENCY § 220 cmt. j, at 490 (short period of employment makes worker "less apt" to be subject to sufficient control and "more likely" to be considered an independent contractor); id. § 220 cmt. k, at 490 ("fact that a worker supplies his own tools is some evidence that he is not a servant") (emphasis added). And, once again, Browning-Ferris's exclusive focus on the independent-contractor test ill fits the joint-employer inquiry into who is pulling the strings when it comes to managing and supervising workers who are admittedly employees.

Second, Browning-Ferris points to our decision in Local 777, Democratic Union Organizing Committee, Seafarers International Union of North America v. NLRB , 603 F.2d 862 (D.C. Cir. 1978), which contrasted "economic controls" that are insufficient to establish a common-law employment relationship with "the more usual forms of direct control typical of an employer/employee relationship," id . at 873. See Browning-Ferris Br. 29. But that statement indicates only that "direct control" is "typical[ly]" or "usual[ly]" present in employment relationships. It does not hold either that indirect control is categorically *1219 excluded from the matrix of relevant factors, or that direct control of all the essential terms and conditions of employment is the sine qua non of employer status under the traditional common-law principles of agency. 13

We should also hesitate to find the common law at war with common sense. A categorical rule against even considering indirect control-no matter how extensively the would-be employer exercises determinative or heavily influential pressure and control over all of a worker's working conditions-would allow manipulated form to flout reality. If, for example, a company entered into a contract with Leadpoint under which that company made all of the decisions about work and working conditions, day in and day out, with Leadpoint supervisors reduced to ferrying orders from the company's supervisors to the workers, the Board could sensibly conclude that the company is a joint employer. This is especially so if that company retains the authority to step in and exercise direct authority any time the company's indirect mandates are not followed. After all, as Justice Scalia commented, "the soul of the law * * * is logic and reason." Hein v. Freedom from Religion Found., Inc. , 551 U.S. 587 , 633, 127 S.Ct. 2553 , 168 L.Ed.2d 424 (2007) (Scalia, J., concurring in the judgment); cf. United States v. Bradley , 35 U.S. (10 Pet.) 343, 364, 9 L.Ed. 448 (1836) (applying rule because "[t]his is not only the dictate of the common law, but of common sense").

2

The problem with the Board's decision is not its recognition that indirect control (and certainly control exercised through an intermediary) can be a relevant consideration in the joint-employer analysis. It is the Board's failure when applying that factor in this case to hew to the relevant common-law boundaries that prevent the Board from trenching on the common and routine decisions that employers make when hiring third-party contractors and defining the terms of those contracts. To inform the joint-employer analysis, the relevant forms of indirect control must be those that "share or co-determine those *1220 matters governing essential terms and conditions of employment." Dunkin' Donuts , 363 F.3d at 440 (citation omitted); see also Browning-Ferris , 691 F.2d at 1123 ; Laerco , 269 N.L.R.B. at 325 . By contrast, those types of employer decisions that set the objectives, basic ground rules, and expectations for a third-party contractor cast no meaningful light on joint-employer status.

The Board's analysis of the factual record in this case failed to differentiate between those aspects of indirect control relevant to status as an employer, and those quotidian aspects of common-law third-party contract relationships. For example, the Board treated as equally relevant to employer status (i) evidence that Browning-Ferris supervisors "communicated detailed work directions to employees on the stream," which may well have dictated a term or condition of employment, and (ii) Browning-Ferris's and Leadpoint's use of a "cost-plus contract," a frequent feature of third-party contracting and sub-contracting relationships. See Browning-Ferris , 362 N.L.R.B. No. 186, at 18-20.

In addition, the Board provided no blueprint for what counts as "indirect" control. At some points, the Board indicated that indirect control means control that is conveyed "through an intermediary." Browning-Ferris , 362 N.L.R.B. No. 186, at 2. Such use of an intermediary either to transmit Browning-Ferris directions to a Leadpoint sorter, see Oral Arg. Tr. 39, 41-42, or to implement Browning-Ferris-influenced disciplinary measures, J.A. 32, may well be found to implicate the essential terms and conditions of work. On the other hand, routine contractual terms, such as a very generalized cap on contract costs, or an advance description of the tasks to be performed under the contract, would seem far too close to the routine aspects of company-to-company contracting to carry weight in the joint-employer analysis. Cf. NLRB v. Denver Bldg. & Const. Trades Council , 341 U.S. 675 , 689-690, 71 S.Ct. 943 , 95 L.Ed. 1284 (1951) ("[T]hat the contractor had some supervision over the subcontractor's work, did not eliminate the status of each as an independent contractor or make the employees of one the employees of the other.").

The Board's employment of the indirect-control factor, in other words, requires it to erect some legal scaffolding that keeps the inquiry within traditional common-law bounds and recognizes that "[s]ome such supervision is inherent in any joint undertaking, and does not make the contributing contractors employees." Radio City Music Hall Corp. v. United States , 135 F.2d 715 , 718 (2d Cir. 1943) (L. Hand, J.). After all, "global oversight" is a routine feature of independent contracts. See North American Van Lines , 869 F.2d at 599 ("[G]lobal oversight * * * is fully compatible with the relationship between a company and an independent contractor."). 14 Wielding direct and indirect control over the "essential terms and conditions" of employees' work lives is not. Dunkin' Donuts , 363 F.3d at 440 (citation omitted). The Board's decision obscures that line.

The Board's assurance that " 'influence' is not enough * * * if it does not amount to control" misses the point that not every aspect of control counts. Browning-Ferris , 362 N.L.R.B. No. 186, at 13 n.68. The critical question is what is being controlled. Whether Browning-Ferris influences *1221 or controls the basic contours of a contracted-for service-such as requiring four lines' worth of sorters plus supporting screen cleaners and housekeepers-would not count under the common law.

Counsel for the Board assured the court at oral argument that the Board will determine the boundaries of the indirect-control element as it proceeds, on a case-by-case basis. See Oral Arg. Tr. 61-62. In principle, there is nothing wrong with the Board fleshing out the operation of a legal test that Congress has delegated to the Board to administer through case-by-case adjudication. See Eastex, Inc. v. NLRB , 437 U.S. 556 , 574-575 (1978) ("[T]he 'nature of the problem, as revealed by unfolding variant situations,' requires 'an evolutionary process for its rational response, not a quick, definitive formula as a comprehensive answer.' ") (quoting Local 761, Int'l Union of Electric, Radio & Machine Workers v. NLRB , 366 U.S. 667 , 674, 81 S.Ct. 1285 , 6 L.Ed.2d 592 (1961) ).

But the Board's decision here is one of those cases-the one in which the Board first applied that indirect-control factor, and did so at times in a manner that appears to have pushed beyond the common-law's bounds. Because the Board has no administrative expertise when it comes to discerning the traditional common-law meaning of "employer," see United Insurance , 390 U.S. at 260 , 88 S.Ct. 988 , that step-by-step approach depends on the Board starting with a correct articulation of the governing common-law test. Here, that legal standard is the common-law principle that a joint employer's control-whether direct or indirect, exercised or reserved-must bear on the "essential terms and conditions of employment," Dunkin' Donuts , 363 F.3d at 440 (citation omitted), and not on the routine components of a company-to-company contract.

Because we cannot tell from this record what facts proved dispositive in the Board's determination that Browning-Ferris is a joint employer, and we are concerned that some of them veered beyond the orbit of the common law, we remand for further proceedings consistent with this opinion. 15

C

There is a second half to the Board's new test that bears mention. The Board held that, even if it finds that the common law would deem a business to be a joint employer, the Board will also ask "whether the putative joint employer possesses sufficient control over employees' essential terms and conditions of employment to permit meaningful collective bargaining" before finding joint-employer status under the Act. See Browning-Ferris , 362 N.L.R.B. No. 186, at 2. "In other words," according to the Board, "the existence of a common-law employment relationship is necessary, but not sufficient, to find joint-employer status [under the Act]." Id . at 12.

The Board, however, did not meaningfully apply the second step of its test here. In concluding that Browning-Ferris and Leadpoint were joint employers of the workers in the petitioned-for unit, the Board simply noted that Browning-Ferris's collective-bargaining obligation applies "only with respect to those terms and conditions over which it possesses sufficient control for bargaining to be meaningful." Browning-Ferris , 362 N.L.R.B. No. 186, at 2 n.7. But the Board never delineated *1222 what terms and conditions are "essential" to make collective bargaining "meaningful," id . at 2, instead declaring that it would adhere to an "inclusive" and "non-exhaustive" approach to the meaning of "essential terms and conditions of employment," id. at 15. Nor did the Board clarify what "meaningful collective bargaining" might require in an arrangement like this.

We trust that, if the Board were again to find that Browning-Ferris is a joint employer of the Leadpoint workers under the common law, it would not neglect to (i) apply the second half of its announced test, (ii) explain which terms and conditions are "essential" to permit "meaningful collective bargaining," and (iii) clarify what "meaningful collective bargaining" entails and how it works in this setting.

IV

In this case the Board both refined its joint-employer standard and immediately applied it retroactively to conclude that Browning-Ferris and Leadpoint were joint employers of the workers in the petitioned-for unit. Browning-Ferris challenges that retroactive application as manifestly unjust. Because we conclude that the Board insufficiently explained the scope of the indirect-control element's operation and how a properly limited test would apply in this case, it would be premature for us to decide Browning-Ferris's challenge to the Board's retroactive application of its test. We do not know whether, under a properly articulated and cabined test of indirect control, Browning-Ferris will still be found to be a joint employer. In addition, the lawfulness of the retroactive application of a new decision cannot be evaluated reliably without knowing with more precision what that new test is and how far it departs (or does not) from reasonable, settled expectations.

Nevertheless, we note that the Board in this case "carefully examined three decades of its precedents," "concluded that the joint-employer standard they reflected required 'direct and immediate' control," and "[t]hereafter * * * forthrightly overruled those cases and set forth * * * 'a new rule.' " CNN America , 865 F.3d at 749-750 (quoting Browning-Ferris , 362 N.L.R.B. No. 186, at 3 ). In rearticulating its joint-employer test on remand, then, the Board should keep in mind that while retroactive application may be "appropriate for new applications of [existing] law," it may be unwarranted or unjust "when there is a substitution of new law for old law that was reasonably clear," and on which employers may have relied in organizing their business relationships. Epilepsy Found. of Ne. Ohio v. NLRB , 268 F.3d 1095 , 1102 (D.C. Cir. 2001) (alteration in original; internal quotation marks omitted) (quoting Public Serv. Co. of Colo. v. FERC , 91 F.3d 1478 , 1488 (D.C. Cir. 1996) ); cf. American Tel. & Tel. Co. v. FCC , 454 F.3d 329 , 333-334 (D.C. Cir. 2016) (finding retroactive application "not manifestly unjust" where the agency's previous rulings "reflect[ed] a highly fact-specific, case-by-case style of adjudication" that did not establish "a clear rule of law exempting" certain conduct).

* * * * *

In sum, we uphold as fully consistent with the common law the Board's determination that both reserved authority to control and indirect control can be relevant factors in the joint-employer analysis. We reverse, however, the Board's articulation and application of the indirect-control element in this case to the extent that it failed to distinguish between indirect control that the common law of agency considers intrinsic to ordinary third-party contracting relationships, and indirect control over the essential terms and conditions of *1223 employment. We accordingly grant Browning-Ferris's petition in part, deny the Board's cross-application, dismiss without prejudice the Board's application for enforcement as to Leadpoint, and remand for further proceedings consistent with this opinion.

So ordered.

This case presents the question whether, under the National Labor Relations Act, Browning-Ferris is the joint employer of Leadpoint's employees. While the case was pending before our court, the Board's Chairman announced that the Board will conduct a rulemaking to establish standards for determining joint employer status. The Board then published its Notice of Proposed Rulemaking. The Standard for Determining Joint-Employer Status, 83 Fed. Reg. 46,681 (Sept. 14, 2018).

In response to the Chairman's announcement, Browning-Ferris moved to remand the case to the Board pending the outcome of the rulemaking. I voted to grant the motion. My colleagues denied it and now release their opinion on the questions the Board is considering in its rulemaking.

I dissent because the majority should not have issued any merits opinion in light of the pending rulemaking proceedings. I dissent as well because the majority opinion misstates the common law, misframes the questions in the case, and adds to the uncertainty the Board's Browning-Ferris decision has generated.

I.

The unusual twists and turns in this case need to be recounted in order to appreciate where matters now stand.

In 2015 the Board, with a full complement of 5 Members, issued its 3 to 2 "representation" decision that Leadpoint and Browning-Ferris jointly employed Leadpoint's employees at the Browning-Ferris facility in California, and thus constituted a single bargaining unit. Browning-Ferris Indus. of Cal. , 362 N.L.R.B. No. 186 (Aug. 27, 2015).

This intermediate Board decision overturned decades of settled law. Direct and immediate control of employees, not just indirect control or potential control, had been required before a company could be deemed a joint employer of another company's employees for the purposes of collective bargaining. See, e.g. , Int'l Chem. Workers Union Local 483 v. NLRB , 561 F.2d 253 , 255-57 (D.C. Cir. 1977) ; AM Prop. Holding Corp. , 350 N.L.R.B. 998 , 999-1002 (2007), enforced in relevant part sub nom. Serv. Emps. Int'l Union, Local 32BJ v. NLRB , 647 F.3d 435 , 442-45 (2d Cir. 2011) ; Airborne Freight Co. , 338 N.L.R.B. 597 , 597 n.1 (2002) ; TLI, Inc. , 271 N.L.R.B. 798 , 798-99 (1984), aff'd sub nom. Gen. Teamsters Local Union No. 326 v. NLRB , 772 F.2d 894 (3d Cir. 1985) (unpublished table mem.); Laerco Transp. & Warehouse , 269 N.L.R.B. 324 , 325-26 (1984).

The implications of the Board's decision were profound and attracted much attention. Statements in its 3-2 opinion affected countless business relationships across the country and, according to a Committee of the House of Representatives, did so almost always in a negative way. See H.R. Rep. No. 115-379, at 9-17 (2017); see also Browning-Ferris , 362 N.L.R.B. No. 186, at 35-47 (dissenting op. of Members Miscimarra & Johnson). The House Committee held hearings and reported a bill that would overrule the Board's decision and restore the joint employer test the Board had been following for decades. The bill *1224 passed the House, but at the time of this writing the Senate had not acted. Save Local Business Act, H.R. 3441, 115th Cong. (as passed by House, July 11, 2017).

In the meantime, the Board in this case ordered an election to implement its representation decision. At the time, Browning-Ferris had 60 employees at the California facility who were represented by a union. That union sought to represent the collective BFI and Leadpoint employees under a single bargaining unit. In the Board-ordered election, the employees of the combined bargaining unit voted in favor of the union representing them in joint bargaining with Browning-Ferris and Leadpoint. When Browning-Ferris refused to come to the table, the Board issued a bargaining order. Browning-Ferris Indus. of Cal. , 363 N.L.R.B. No. 95 (Jan. 12, 2016). Browning-Ferris responded with its petition for judicial review in this court, and the Board cross-petitioned for enforcement.

We heard oral argument in March of 2017. Thereafter the composition of the Board changed. In December 2017, in another 3 to 2 decision, the Board overruled its decision in this case. Hy-Brand Indus. Contractors, Ltd. , 365 N.L.R.B. No. 156 (Dec. 14, 2017). At the urging of the Board's General Counsel, we sent the case back to the Board for reconsideration in light of Hy-Brand .

Then in February 2018 still another reconstituted Board vacated Hy-Brand on the ground that one Member of the three-Member majority should not have participated in the case. Hy-Brand Indus. Contractors, Ltd. , 366 N.L.R.B. No. 26 (Feb. 26, 2018). Hy-Brand thus reverted to a 2 to 2 tie about whether Browning-Ferris should be overruled.

Several months later, the newly-appointed Board Chairman announced that a majority of the Board's Members had decided that "notice-and-comment rulemaking offers the best vehicle to fully consider all views on what the [joint-employer] standard ought to be." Letter from Chairman John F. Ring, NLRB, to Sens. Elizabeth Warren, Kirsten Gillibrand & Bernard Sanders 1 (June 5, 2018) (alteration in original).

In the meantime we had restored to our docket the Browning-Ferris petition for judicial review and the Board's cross-petition for enforcement.

The Board published its notice of proposed rulemaking on September 14, 2018. The Standard for Determining Joint-Employer Status, 83 Fed. Reg. 46,681 .

II.

Apparently the majority objects to Browning-Ferris's remand request on the ground that any final Board rule would be prospective only. 1 Maj. Op. 1206. The thinking must be-why remand the case if the Board's final rule would not change the outcome? That idea is incorrect. There are at least three ways in which the rulemaking could have a significant impact on this case even though the Board's rule will not be retroactive.

First, notice and comment rulemaking can be educational. In the rulemaking on the joint employer question the Board expects many comments. Letter from Chairman John F. Ring 1. One of the advantages of rulemaking over adjudication is this: "Agencies discover [through rulemaking] that they are not always repositories of ultimate wisdom; they learn from the suggestions of outsiders and often benefit *1225 from that advice." NLRB v. Wyman-Gordon Co. , 394 U.S. 759 , 777-78, 89 S.Ct. 1426 , 22 L.Ed.2d 709 (1969) (Douglas, J., dissenting). 2

On a remand from our court without a merits opinion, the Board could take into account what it learned from the rulemaking, even though it would not directly apply its "new" rule to Browning-Ferris. A thorough historical analysis, for example, might show-contrary to the Board's opinion here-that under the common law indirect control and potential control were never enough to establish joint-employer status. If the case reached us again, either on the company's or the union's petition, the Board's revised judgment could have the "power to persuade" even though on our de novo review it lacked "the power to control." Skidmore v. Swift & Co. , 323 U.S. 134 , 140, 65 S.Ct. 161 , 89 L.Ed. 124 (1944).

Second, assume that in the rulemaking the Board retains the joint-employer standard it set forth in this case. Even so a question remains. Should the new standard be applied to Browning-Ferris? In the decision now on review the Board rejected the argument of Browning-Ferris that its new joint-employer standard should not be applied retroactively. 362 N.L.R.B. No. 186, at 1-2. On remand and in light of what the Board learned during the rulemaking, the Board might reconsider that aspect of its decision. Case law in this circuit, set forth in the margin, strongly suggests that it should. 3

The third reason is the most significant and the most probable. Suppose the final rule flatly disagrees with the Board's Browning-Ferris decision and reinstates the standard that had prevailed for decades. 4 That is what the Board's Notice of Proposed Rulemaking suggests. The proposed rule is set forth in the margin. 5

Browning-Ferris moved to remand the case to the Board pending the outcome of *1226 the rulemaking. The Board's Deputy Associate General Counsel 6 opposed the motion on the basis that the rulemaking "would not affect this case." That argument was mistaken. Board counsel so confessed in oral argument on the motion. Oral Arg. Tr. 15:9-16:8 (July 3, 2018).

The argument was mistaken for two reasons already mentioned. It was mistaken as well because the Board's application of its proposed rule to Browning-Ferris would not amount to retroactive law giving. Applying the Board's new rule would be reinstating the legal regime existing before the Board's decision in this case discarded it. The upshot is that if the Board applied its proposed "new" rule-actually the old rule-to Browning-Ferris on remand the Board would not be impermissibly attaching "new legal consequences to events completed before [the rule's] enactment." Landgraf v. USI Film Prods. , 511 U.S. 244 , 269-70, 114 S.Ct. 1483 , 128 L.Ed.2d 229 (1994). Our decision in Catholic Health Initiatives is on point. We held that a rulemaking applying a rule codifying a policy announced in an earlier adjudication did not violate the rule against retroactive rulemaking. 718 F.3d at 920-22.

Like other administrative agencies, the Board may establish standards through rulemaking or adjudication. See 29 U.S.C. § 156 . Here, after the back and forth recounted above, the Board has determined that the standards for joint employer status should be established through rulemaking. See The Standard for Determining Joint-Employer Status, 83 Fed. Reg. at 46,686. Bell Aerospace requires federal courts to respect the Board's determination to proceed by rulemaking. NLRB v. Bell Aerospace Co. , 416 U.S. 267 , 294-95, 94 S.Ct. 1757 , 40 L.Ed.2d 134 (1974). Yet the majority opinion-without any reasonable explanation-threatens to short-circuit the Board's choice, to control and confine the scope of its rulemaking, and to influence the outcome of that proceeding. 7

The majority's opinion potentially has this effect because it is rendered de novo , a standard of review the Board may not have anticipated. Board Br. 16. On de novo review it is the court, not the Board, who decides what will be the test for joint employer status. De novo review or not, our court should not be attempting to preempt the Board's forthcoming judgment in the rulemaking proceeding. The Board is not "the repository of ultimate wisdom," and neither are the judges of this court.

To sum up, the Board's attorney confessed that the rationale of the Board's General Counsel for opposing remand was in error. The Board's attorney also raised doubt that in opposing a remand, she was expressing the considered judgment of the *1227 Members of the Board. 8 Even so, the panel majority has denied the motion to remand the case pending the rulemaking. The majority's rationale is simply this: if Board counsel 9 wants the court to go ahead and decide the merits, the court should do so. In relying solely on the position of Board counsel, the majority acts as if it were dealing with some sort of "waiver," with a known right the Board itself has intentionally relinquished. See Johnson v. Zerbst , 304 U.S. 458 , 464, 58 S.Ct. 1019 , 82 L.Ed. 1461 (1938). But that is not accurate. The Board has no "right" to relinquish. To treat this controversy as the majority does is not only to ignore the substantial interests of Browning-Ferris, but also to neglect the judiciary's responsibility to avoid interfering with an agency's ongoing rulemaking proceedings.

III.

As to the merits, I rely on the comprehensive opinions of Member Miscimarra and Member Johnson dissenting in this case and of the Board majority in the now-vacated Hy-Brand case. Both opinions show how pernicious the Board's decision would be if it were implemented across the American economy. Both opinions also show that the Board majority did not accurately describe the common law of joint employer. And both opinions remain largely unanswered.

Although I cannot improve on what the Board's dissenters said in Browning-Ferris or on what the previous Board majority said in Hy-Brand , I offer a few comments about the decision of our court. I do so because the decision disregards and contradicts a strong, clear, accepted and well-founded body of common law cases. Instead of clarity it adds another layer of ambiguity. Rather than narrowing the Board's broad pronouncements, the majority opinion endorses and expands them.

A.

The majority's errors about the meaning of the common law may be traced to two sources. The first is its failure to recognize the importance of Leadpoint's clear and undisputed status as an independent contractor. 10 The majority thinks that under *1228 the common law a company's status as an independent contractor has no bearing on the joint employer question this case presents. Maj. Op. 1213. As I will explain, the opposite is true. It seems likely that the Board, in its rulemaking, will come to the same conclusion.

The other source of the majority's errors is its failure to notice that the common law of joint employer may vary according to the nature of the business arrangement between companies, or between consumers and companies. 11 The joint employer issue in franchising arrangements, for example, involves different considerations than those involved in the typical principal-independent contractor arrangement. Browning-Ferris , 362 N.L.R.B. No. 186, at 45-47 (dissenting op). Yet the majority opinion declares that "indirect control" is "relevant" across the broad spectrum of business relationships-about which neither I nor my colleagues have any experience or familiarity. 12

So I come back to the common law, which is supposed to control our decision and should have controlled the Board's. Under the common law, employees of a true independent contractor cannot be considered employees of the company who hired the contractor (the principal, or in this case Browning-Ferris). Stated in terms of the common law of agency: "An independent contractor is not the servant of his employer. The relation of master and servant does not exist between an employer and the servants of an independent contractor , nor between an independent contractor and the servant of a subcontractor, and he is not responsible as a master, either to or for them." 39 C.J. Master and Servant § 8 (1925) (emphasis added) 13 ; see also 30 C.J.S. Employer-Employee § 16 (2017) ("The relationship of employer and employee likewise does not exist between an employer or contractee and the employees of an independent contractor....").

*1229 The common law is "the dominant consensus of common-law jurisdictions." Field v. Mans , 516 U.S. 59 , 70 n.9, 116 S.Ct. 437 , 133 L.Ed.2d 351 (1995). 14 In support of the common-law rule just quoted, 60 common-law cases from across the country over the years are cited, and there are doubtless more. 15 39 C.J. Master and Servant § 8, at 38 n.53. That is indeed a "dominant consensus." In contrast, neither the Board nor the majority opinion here can cite any line of common-law cases going the other way. 16 It follows that under the common law Leadpoint's employees may not be considered employees of Browning-Ferris. As the Supreme Court held in Denver Building , a contractor's "supervision over the subcontractor's work[ ] did not eliminate the status of each as an independent contractor or make the employees of one the employees of the other ." NLRB v. Denver Bldg. & Constr. Trades Council , 341 U.S. 675 , 689-690, 71 S.Ct. 943 , 95 L.Ed. 1284 (1951) (emphasis added). "The business relationship between independent contractors is too well established in the law to be overridden without clear language doing so." Hy-Brand , 365 N.L.R.B. No. 156, at 11 (quoting id. at 690, 71 S.Ct. 943 ).

Section 5 of the Restatement (Second) of Agency, from which the majority opinion derives its so-called "indirect control" test, 17 recites the same common law rule as *1230 the 1925 treatise quoted above. "In no case are the servants of a nonservant agent the servants of the principal." Restatement (Second) of Agency § 5 ("Subagents and Subservants"), cmt. e (1958). A "servant" is an "employee." 18 An agent who is not an employee-a "non-servant agent"-is an "independent contractor." 19 Thus, "in no case" are the employees of an independent contractor employees of the company who hired the contractor. "In no case," in other words, could Leadpoint's employees also be the employees of Browning-Ferris.

The distinction between employees and independent contractors, 20 which the majority deems inconsequential, is written into the National Labor Relations Act. In NLRB v. Hearst Publications, Inc. , 322 U.S. 111 , 64 S.Ct. 851 , 88 L.Ed. 1170 (1944), the Court held that under the Act "newsboys"-adults who distributed newspapers on street corners-were "employees" of the newspaper publishers. "Congress was so incensed with the fanciful construction of its legislative intention in Hearst that in 1947 it specifically excluded 'independent contractors' from the coverage of the Act and condemned the Court's rationale in Hearst Publications as giving 'far-fetched meanings' to the words Congress has used." Local 777, Democratic Union Org. Comm. v. NLRB, 603 F.2d 862 , 905 (D.C. Cir. 1978) (on petition for rehearing); see also Labor Management Relations (Taft-Hartley) Act, 1947, Pub. L. No. 80-101, § 101, 61 Stat. 136 , 137-38 (amending § 2(3) of the National Labor Relations Act and codified at 29 U.S.C. § 152 (3) ); Harvey M. Adelstein & Harry T. Edwards, The Resurrection of NLRB v. Hearst : Independent Contractors under the National Labor Relations Act , 17 U. Kan. L. Rev. 191 (1968). In short, Congress decided that the newspaper distributors in Hearst were independent contractors, not employees of the publishers. Those distributors, those independent contractors, had employees of their own. See H.R. Rep. No. 80-245, at 18 (1947). Consistent with the common law rule set forth above, the distributors' employees could not be considered employees of the newspaper publishers. 21 If, as our court stated, Congress was "incensed" at the Supreme Court's treatment of the distributors as employees, one can only imagine Congress's reaction to treating the distributors' employees as employees of the publishers. Yet that is where the majority opinion leads. 22

*1231 B.

A few more observations about the majority opinion are in order.

On page after page, paragraph after paragraph, the majority drags a red herring across the case. It insists that "indirect control," whatever that may encompass, and a potential right to control, are "relevant." 23 This frames the issue as if we were merely dealing with an evidentiary dispute. If only relevancy were at issue, the Federal Rules of Evidence, which the Board has adopted, 24 would control. But as everyone else recognizes, the issues before us are much more serious, and the majority opinion fails to confront them.

Consider the majority opinion on its own terms. Under Rule 401(a) of the Federal Rules of Evidence, evidence is "relevant" if it tends to make a fact of consequence "more or less probable than it would be without the evidence." 25 In any relevancy analysis there is an essential step. The majority's dozens of references to relevancy omit that step. "Relevancy is not an inherent characteristic of any item of evidence...." Fed. R. Evid. 401 advisory committee's note. As Professor James explained in a highly-regarded article, to "determine the relevancy of an offered item of evidence one must first discover to what proposition it is supposed to be relevant." 26 The "fact of consequence" made more or less probable must be identified. The majority opinion never identifies what fact it thinks evidence of indirect control makes more (or less) likely. Yet that gets to the heart of this case and is the cause of much of the controversy surrounding it.

Before its decision here, the Board's well-established, easily understood rule was that a company could not be considered *1232 a joint employer of another company's employees unless it exercised direct and immediate control or supervision over those employees. 27 Suppose that were still the law. 28 If so, evidence of indirect control would be "relevant" but not in the way the majority thinks. Such evidence would not tend to show that the company was a joint employer as the majority assumes. Just the opposite. The evidence would tend to show that the company was not a joint employer.

Take the evidence in this case. On one day a Browning-Ferris manager observed two Leadpoint employees drinking a bottle of whiskey while on duty. The Browning-Ferris manager notified Leadpoint's supervisor, and the supervisor removed the employees from the plant. The Browning-Ferris manager also sent an e-mail to Leadpoint's President requesting him to fire these two employees. (Leadpoint eventually discharged one of them.)

What, if anything, should be made of this incident on one day on one shift involving two employees in a workforce of more than two hundred employees? My colleagues think and the Board thought, Browning-Ferris , 362 N.L.R.B. No. 186, at 18, it showed that Browning-Ferris jointly employed not only the two drinking employees, but also the entire Leadpoint workforce. That is, they treat the incident as evidence that Browning-Ferris was exercising "indirect control" over Leadpoint's employees and thus was the joint employer of those employees.

The common law and any objective observer would view the majority's and Board's conclusion as nonsense. This single event was trivial in the larger picture of employer-employee-independent relations year-to-year, day-to-day, hour-to-hour at the Browning-Ferris facility. To the extent the incident had any evidentiary value, any bearing on the joint employer issue, it tended to show the opposite of what the majority seems to suppose.

The Regional Director made this point when he evaluated this evidence. He decided that the evidence tended to show that Browning-Ferris did not exercise direct control and therefore was not a joint employer. The Regional Director put it this way: "Surely if BFI had the authority to terminate Leadpoint employees, [BFI's manager] would have done this without having to email Leadpoint's President, located in Arizona, to do so." Browning-Ferris Indus. of Cal., Inc. , Case 32-RC-109684, 2013 WL 8480748 , at *9 (N.L.R.B. Aug. 16, 2013). 29

To sum up, both the Board and the Regional Director considered this example of indirect control to be relevant. To the Board the evidence made it more likely that Browning-Ferris was a joint employer. To the Regional Director the evidence made it less likely.

I have gone into detail about this one item of evidence to illustrate why the majority opinion's mere assertion that evidence of indirect control is "relevant" is not only confused and confusing, but also fails to confront one of the main issues in the case-namely, whether direct and immediate control or supervision is a necessary *1233 prerequisite to a finding of joint employer status. 30

One additional point. The Regional Director was surely correct in his assessment of this particular incident. Under the common law "the existence of the power to discharge is essential" to the right of control, and therefore to establish joint employer status. 39 C.J. Master and Servant § 4 ("Direction and Control"). Browning-Ferris did not have that power; Leadpoint did. The Board plainly erred in deciding otherwise.

C.

While endorsing "indirect control" as a common law standard for determining joint employer status, 31 the majority confesses that it does not know exactly what the Board had in mind by "indirect control" or how the common law defines those terms in the joint employer context. Maj. Op. 1220-21. This revealing admission is hardly surprising. The majority is unable to extract any "indirect control" standard from the common law 32 for an obvious reason. There is no "common law" principle as of 1947 standing for the proposition that "indirect control" could render one company a joint employer of another company's employees, especially if that other company is an independent contractor.

The majority cites the illustrations in the 1958 Restatement (Second) of Agency § 5 -the "sub-servant" doctrine-as support. Maj. Op. 1217 n.12. The Board did the same. Browning-Ferris , 362 N.L.R.B. No. 186, at 14 & n.74. But those illustrations have no bearing on the issue. In the first illustration, the miners-who hired and paid assistants-were employees of the mine operator, not independent contractors like Leadpoint. The same is true of the second illustration of a company operating "markets" (grocery stores?): unlike Leadpoint, the manager of each market was an employee of the market owner.

In other words, those illustrations would be comparable only if Leadpoint were an employee of Browning-Ferris, which it is not. The notes to this Restatement section reinforce the view stated above that under the common law employees of an independent contractor cannot be considered employees of the company that hired the independent contractor. "Except in the case of subservants, it is difficult to see how the subagent can be the principal's servant, since his employer is a nonservant agent not subject to the principal's direction." Restatement (Second) of Agency § 5 reporter's notes, at 33.

*1234 The Chamber of Commerce's amicus brief points out that the "sub-servant doctrine applies when both the servant and the sub-servant are servants of a single master." Chamber of Commerce Br. 25. In the joint-employer setting, when one of the employers is an independent contractor and not the servant of the other, the doctrine is therefore inapplicable. Id. at 26.

In the text of its opinion, the majority also seeks to fortify its view of the common law of joint employers with three state court decisions. Maj. Op. 1216. Of course three opinions over more than half a century hardly constitute some "dominant consensus of common-law jurisdictions." Field , 516 U.S. at 70 n.9, 116 S.Ct. 437 . In any event, the holdings in these cases lend no support to the majority.

The first case, White v. Morris, 114 Ga.App. 618 , 152 S.E.2d 417 (1966), was merely an intermediate appellate decision handed down 19 years after passage of the Taft-Hartley Act. To claim that the case reflects some general common law regarding joint employers in 1947 is untenable. Besides, the case presented no issue regarding joint employer status. 33

The second case the majority cites, Wallowa Valley Stages, Inc. v. Oregonian Publ'g Co., 235 Or. 594 , 386 P.2d 430 (1963) (en banc), is also inapposite. It too could not represent the dominant consensus as of 1947. The case is a weak reed anyway in light of its later repudiation by the Oregon Supreme Court. Woody v. Waibel, 276 Or. 189 , 554 P.2d 492 , 494 n.3 (1976) (en banc). Besides, no issue regarding the common law of joint employer was presented. 34

The third case, Nicholson v. Atchinson, T. & S. F. Ry., 95 Kan. 13 , 147 P. 1123 (1915), is even farther afield. The question was whether the intermediate company was an independent contractor (such as Leadpoint). The court held that it was not because the principal (the Santa Fe Company) "organized, officered, and financed [it] entirely." Id. at 1124 . It followed that the injured employee working for the intermediate company had a single employer-the Santa Fe Company. Id. at 1126 . 35

There are other common law decisions scattered throughout footnotes in the majority opinion. An analysis of these cases reveals that none of them concerned joint *1235 employment. 36 Many dealt with the question whether a tortfeasor was an employee or an independent contractor, an issue not presented in this case.

IV.

In short, the majority should not have released its opinion in the face of the Board's rulemaking. The majority has offered no reason for its rejection of Browning-Ferris's remand request. That the majority wants to preempt the rulemaking and confine it strikes me as a quite improper rationale. I dissent not only on this procedural ground, but also on the ground that the majority's analysis of the common law is inaccurate. That analysis fails to take into account the common law importance of Leadpoint's status as an independent contractor. The majority deems "indirect control" significant yet is unable to marshal any body of common law cases to support that view. And the majority, by treating this case as if it were some mere evidentiary dispute, sows confusion and ambiguity when what is needed is certainty and predictability.

ADDENDUM

July 3, 2018 Oral Argument

Transcript at 18:5-20:5

BOARD COUNSEL: ... But I want to make clear, though, that Chairman Ring's letter, although he stated clearly that the majority of the Board is committed to going to rulemaking as they're in the process of going through internal preparations *1236 to do so, the statements in his letter were his own, and that, but the one statement that is clear is that he's keeping an open mind, and I just wanted to make sure that I have that on the record given your discussion with-

JUDGE RANDOLPH: Are you suggesting that it might not be a rule-making?

COUNSEL: Well, they're committed to rule-making, and they anticipate, as his letter stated they anticipate issuing a notice of proposed rule sometime this summer.

JUDGE RANDOLPH: Okay.

COUNSEL: That statement was made in early June. But I want to emphasize, though, that, to reiterate that the Board really does believe that this Court should proceed to decision on the merits, and there's no reason other than that to even consider retroactive application.

JUDGE RANDOLPH: When you say the Board wants to, I mean, did you take a poll of the Board members?

COUNSEL: I'm standing before you, Your Honor. I'm authorized to represent the Board and the Board's position that the Board would like this case decided.

JUDGE RANDOLPH: Yes. Well, usually when you stand before us the Board has made a decision in writing, and you're defending an order and an opinion, but we don't have any order and we don't have any opinion regarding whether the Board wants to go forward with this case while the rulemaking is pending. And so, I'm asking you, you know, are, has the Board voted on that issue?

COUNSEL: Well, I'm post-decisional counsel, and the General Counsel is the one who prosecutes, and comes and defends, or seeks enforcement in this Court. I am not privy to the Board deliberations and such things as votes.

JUDGE RANDOLPH: So, you're stating the General Counsel's view?

COUNSEL: I believe if the Board consulted with the General Counsel if they had a different view we would have heard it. But the position in the papers stands. And I do want to note that when we are talking about what happens if the case were remanded, if it were remanded on the merits of course the Board would proceed with following the Court's instructions and limiting its decision position and all of its determinations in line and consistent with that decision. Here, if this Court were to remand on the basis of the news that a rule may be coming out, a rule-making may be undertaken, there's many different options the Board could potentially have, it has discretion in deciding how to handle its pending cases.