RESILIENT FLOOR COVERING PENSION TRUST FUND BOARD OF TRUSTEES; Resilient Floor Covering Pension Trust Fund, Plaintiffs-Appellants, v. MICHAEL'S FLOOR COVERING, INC., Defendant-Appellee, 801 F.3d 1079

OPINION BERZON, Circuit Judge: We decide in this case two related issues: (1) whether a successor employer, both generally and in the construction industry in particular, can be subject to withdrawal liability under the Multiem-ployer Pension Plan Amendments Act (“MPPAA”), 29 U.S.C. § 1381-1453, amendments to the Employee Retirement Income Security Act (“ERISA”), 29 U.S.C. § 1001 et seq.; and (2) if so, what factors are most relevant to determining whether a construction industry employer is a successor for purposes of imposing MPPAA withdrawal liability.

We conclude that a construction industry successor employer can be subject to MPPAA withdrawal liability, so long as the successor took over the business with notice of the liability. We also hold that the most important factor in assessing whether an employer is a successor for purposes of imposing MPPAA withdrawal liability is whether there is substantial continuity in the business operations between the predecessor and the successor, as determined in large part by whether the new employer has taken over the economically critical bulk of the prior employer’s customer base.

The district court, after a bench trial, held Defendant-Appellee Michael’s Floor Covering, Inc. (“Michael’s”) not liable as a successor employer. In doing so, the district court weighed continuity of the workforce as the most important factor, and, moreover, applied an incorrect test to determine whether there was continuity of the workforce. We therefore reverse and remand for further proceedings applying the correct standards. I. A. Studer’s Floor Covering, Inc. (“Stu-der’s”) was a construction industry employer that sold and installed floor covering materials to commercial and residential customers.

From the 1960s until it ceased doing business on December 31, 2009, Stu-der’s operated out of a storefront and warehouse on Anderson Avenue in Vancouver, Washington. At the time of its closing, Studer’s was a party to a collective bargaining agreement with the Linoleum, Carpet and Soft Tile Applicators Local Union No. 1236, pursuant to which Stu-der’s made contributions to the Resilient Floor Covering Pension Trust Fund (“the Fund”), a multiemployer defined benefit pension plan covered by the MPPAA amendments to ERISA. See 29 U.S.C. § 1002(37)(A).

Toward the end of 2009, the president and chairman of Studer’s, Scott Studer, informed his sales staff that Studer’s would close at the end of the year. Shortly after that announcement, one of those staff members, Michael Haasl, told Studer “that he intended to bid for projects for the sale and installation of floor covering materials for his own company,” Michael’s Floor Covering, LLC (“Michael’s”). Haasl incorporated Michael’s in October 2009. On November 30, 2009, while Studer’s was still in operation, Michael’s obtained a lease on the same storefront and warehouse Studer’s had long occupied.

That lease’s term began on January 1, 2010, the day immediately after Studer’s’ lease terminated. Around the same time, Haasl purchased signs for the Michael’s location very similar to those that Studer’s used. Both spelled out the name “Michael’s”/“Studer’s” in red cursive, and “Floor Coverings” in black block capitals, on a white background. Additionally, at Michael’s’ request, Studer’s gave its authorization to Quest, Studer’s’ telephone service provider, for Michael’s to take over Studer’s’ business telephone numbers at the end of 2009.

Studer’s sold most, though not all, of its tools, equipment, and inventory at a publicly advertised liquidation sale in the fall of 2009. At that sale, Michael’s purchased about 30% of Studer’s’ tools, equipment and inventory.

According to Scott Studer, although “Studer’s did not sell, give[,] or otherwise assign its customer lists or any portion of its customer information to Michael’s[,] Mike Haasl knew the identity of many of Studer’sf] customers and suppliers through his work over the course of 19 years as a salesman for Studer’s.” Michael’s used those existing business relationships in developing its business. The district court found that “Michael’s performs much the same work as Stu-der’s,” though Michael’s added product lines to its showroom that Studer’s had not carried.

For example, the purchasing manager for one major business customer of both Studer’s and Michael’s, New Tradition Homes, testified that Michael’s was asked to “pick up where [Studer’s] left off” and did; that “the type of work done” by Michael’s and Studer’s was “[t]he same”; and that there were no “differences in the type of work done by Michael’s Floor Covering as opposed to what was done by Studer’s Floor Covering.” That same purchasing manager also reported that New Tradition Homes did not “put out a request for bids to replace Studer’s.” Although New Tradition Homes’ “usual bid process” did involve competitive bidding from “a broader number of potential suppliers,” it did not require bidding in this instance, because a “sales rep that [they] were very comfortable with was starting his business,” referring to Haasl and Michael’s.

He also noted that there was only “[v]ery minimal” “disruption caused by the transition from Studer’s to Michael’s”: “[m]ostly it was internal with our systems. We had to make sure that our purchase orders went out on one day to Studer’s and then on the next day to Michael’s Floor Coverings.” In Michael’s’ first two years of operation, it employed eight installers; otherwise, Michael’s outsourced installation work to independent contractors. Of the eight employee installers, five had previously worked for Studer’s at one time or another.

Several of those installers 'stated that the range of work they did for Michael’s was substantially similar to, although slightly broader than, the work they had previously done for Studer’s. The proportion of Studer’s customers retained by Michael’s depends on the mode of calculation used. The district court found that “many of Studer’s[’] customers became Michael’s!’] customers.” The Fund asserts that Michael’s obtained the bulk of its business during its start-up phase from Studer’s’ customers, largely business customers.

For example, all but seven of Michael’s’ business customers in its first three months of operation had been Studer’s’ customers during Studer’s last year of business. Michael’s counters that only 80 or so of the 868 customers Michael’s served in its first two years were former Studer’s clients; this head count includes both large commercial customers with repeat contracts for housing developments and apartment buildings, and individual homeowners, who are more likely to contract on a one-time basis and for fairly small jobs. B.

The MPPAA amendments to ERISA provide, in part, that “[i]f an employer withdraws from a multiemployer [pension] plan in a complete withdrawal ..., then the employer is liable to the plan” for “withdrawal liability.” 29 U.S.C. § 1381(a). Withdrawal liability “is the amount determined [under the statutory calculation method] ... to be the allocable amount of unfunded vested benefits” accrued at the time of the employer’s withdrawal. § 1381(b); see also § 1391.

For “employees] that ha[ve] an obligation to contribute under a plan for work performed in the building and construction industry,” however, there is no withdrawal liability if they cease operations entirely for at least five years. § 1383(b)(1). The dispute in this case concerns whether this construction industry exception applies here because Studer’s permanently ceased performing work covered by the Fund, or whether, instead, it does not apply, because Michael’s essentially took over the work Studer’s would have done, yet did not make contributions to the Fund.

Taking the latter position, the Fund, believing Michael’s to be Studer’s’ successor, assessed withdrawal liability in the amount of $2,291,014.00 against Studer’s and Michael’s and sued Michael’s to recover that amount. After discovery, the Fund and Michael’s filed cross-motions for

summary judgment. Michael’s moved for

summary judgment on the grounds that the Fund could not establish that Michael’s was a successor of Studer’s, and, even if Michael’s were Studer’s’ successor, the Fund could not show Michael’s was subject to its predecessor’s withdrawal liability, for two reasons: first, Studer’s had not itself continued business in the area; and second, Michael’s did not have adequate notice of Studer’s’ liability. The Fund moved for partial

summary judgment on the ground that Michael’s was a successor to Studer’s, so a statutory withdrawal triggering liability occurred when Michael’s continued Studer’s’ business but failed to make contributions to the Fund. At the hearing on the parties’ cross-motions for

summary judgment, the district court suggested that the parties consent to converting the motion to a bench trial on the successorship question only (that is, not on the question whether, if a successor, Michael’s had sufficient notice of the liability). The parties orally agreed to a bench trial “on the record.” About two weeks after the

summary judgment hearing, the Fund filed a motion for leave to supplement the record with additional invoices from Michael’s and Stu-der’s. The Fund noted that the possibility of a bench trial on the record was first raised at the

summary judgment hearing, and explained that, “[hjaving given the matter consideration after the hearing,” the Fund wished to supplement the record with these additional invoices. The Fund had previously included Studer’s invoices from the last three months of 2009 and Michael’s invoices from the first three months of 2010. “[F]or purposes of creating a more complete trial record,” the Fund explained, it was seeking to submit Michael’s’ invoices for the remainder of 2010 and the entirety of 2011, and some additional Studer’s invoices as well.

Michael’s opposed the motion on the grounds that (1) it was “premature” and (2) the evidence “lack[ed] relevance, materiality or probity” because there was no “basis for imposing withdrawal liability on Studer’s,” when Studer’s did not continue work as Studer’s after 2009. The district court issued findings of fact and conclusions of law on November 1, 2012. It determined that Michael’s was not a successor to Studer’s and therefore not subject to withdrawal liability. Applying the multi-factor successorship test set forth in NLRB v.

Jeffries Lithograph Co., 752 F.2d 459, 463 (9th Cir.1985), the district court concluded that, although Michael’s used the same plant that Studer’s had, the other factors either weighed against a finding of successorship (continuity of the workforce; whether the same jobs exist under the same working conditions; whether the same supervisors were employed) or were neutral (whether the same machinery, equipment, and methods of production are used; whether the same service is offered; and whether there was substantial continuity of the business).

The district court characterized the inquiry as concerning whether the successor has “ ‘basically the same owners and operators as ... the predecessor employer,’ ” and that the “changes between predecessor and successor were technical in nature rather than a substantive change in the management.’ ” (quoting New England Mech., Inc. v. Laborers Local Union 294, 909 F.2d 1339, 1343 (9th Cir.1990)). According to the district court, “[t]he question here is whether Michael’s is ‘essentially the same’ as Studer’s....

It is not.” The district court also denied the Fund’s motion to supplement the record, because the Fund had not shown “good cause for the late filing,” and because the “customer issue [wa]s not dispositive of the successor employer determination.” II. “We review the district court’s findings of fact after a bench trial for clear error.” OneBeacon Ins. Co. v. Haas Indus., Inc., 634 F.3d 1092, 1096 (9th Cir.2011). “Questions of law and mixed questions of fact and law are reviewed de novo.” M.M. v. Lafayette Sch. Dist., 767 F.3d 842, 851 (9th Cir.2014) (as amended).

Additionally, “[w]e review for abuse of discretion a district court’s denial of a motion to supplement the record.” E.E.O.C. v. Peabody W. Coal Co., 773 F.3d 977, 982 (9th Cir.2014). A district court abuses its discretion where it applies the wrong legal standard or where its “application of the correct legal standard was (1) ‘illogical,’ (2) ‘implausible,’ or (3) without ‘support in inferences that may be drawn from the facts in the record.’ ” United States v. Hinkson, 585 F.3d 1247, 1262 (9th Cir.2009) (en banc) (quoting Anderson v.

City of Bessemer City, 470 U.S. 564, 577, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985)). Additionally, “ ‘[i]f an exercise of discretion is based on an erroneous

interpretation of the law, the ruling should be overturned.’ ” Estate of Darulis v. Garate, 401 F.3d 1060, 1063 (9th Cir.2005) (quoting Miles v. California, 320 F.3d 986, 988 (9th Cir.2003)); see also Conservation N.W. v. Sherman, 715 F.3d 1181, 1185 (9th Cir.2013). III. A. ERISA, the federal comprehensive private employee benefits statute, includes provisions designed “to ensure that employees and their beneficiaries would not be deprived of anticipated retirement benefits by the termination of pension plans before sufficient funds have been accumulated in the plans.” Pension Benefit Guar. Corp. v. R.A.

Gray & Co., 467 U.S. 717, 720, 104 S.Ct. 2709, 81 L.Ed.2d 601 (1984). ERISA originally sought to accomplish this purpose by creating an insurance program for pension plans, administered by the Pension Benefit Guaranty Corporation (“PBGC”); the insurance program initially covered only single-employer plans, but was later extended to multiem-ployer plans.

See id. at 720-22, 104 S.Ct. 2709 (noting that the provision obligating the PBGC to pay benefits for single employer plans took effect immediately when ERISA was enacted in 1974 and that mandatory coverage of multiemployer pension plans was to take effect in 1978). The MPPAA amendments to ERISA were prompted by Congress’s realization that in some instances, ERISA as it stood did “not adequately protect [multiemployer .pension] plans from the adverse consequences that resulted when individual employers terminatefd] their participation in, or withdr[e]w from, multiemployer plans.” Id. at 722, 104 S.Ct. 2709.

The concern was that “a significant number of [mul-tiemployer] plans were experiencing extreme financial hardship” as a result of individual employer withdrawals from the plans, which saddled the remaining employers with increased funding obligations. Id. at 721, 104 S.Ct. 2709. These withdrawals caused a domino effect of cascading additional withdrawals that eventually “could have resulted in the termination of numerous plans.” Id. Large numbers of plan terminations, in turn, could have jeopardized the entire PBGC insurance program once the provision extending coverage to multiemployer plans became effective.

See id. To address this dilemma, Congress enacted the MPPAA, which imposed “new rules under which a withdrawing employer would be required to pay whatever share of the plan’s unfunded vested liabilities was attributable to that employer’s participation,” thereby protecting the financial health of the plan and safeguarding the PBGC insurance program. Id. at 723, 104 S.Ct. 2709. The MPPAA amendments to ERISA make employers liable for unfunded vested benefits if they withdraw from a multiemployer plan. § 1381; see also § 1391.

In general, a complete withdrawal triggers withdrawal liability where an employer “permanently ceases to have an obligation to contribute under the plan” or “permanently ceases all covered operations under the plan.” § 1383(a). But that general standard for withdrawal, and so for withdrawal liability, does not always apply. Central to this case is the special MPPAA rule for “employer[s] that ha[ve] an obligation to contribute under a plan for work performed in the building and construction industry.” § 1383(b)(1). Under that rule, a complete withdrawal occurs only if: (

A) an employer ceases to have an obligation to contribute under the plán, and (

B) the employer— (

i) continues to perform work in the jurisdiction of the collective bargaining agreement of the type for which contributions were previously required, or (ii) resumes such work within 5 years after the date on which the obligation to contribute • under the plan ceases, and does not renew the obligation at the time of the resumption. § 1383(b)(2).

In other words, under § 1383(b), known as “the MPPAA construction industry exception,” employers in that industry who entirely cease operations are not subject to the withdrawal liability that § 1381 would otherwise impose, unless they resume construction work within five years without also renewing their obligation to contribute to the plan. See Carpenters Pension Trust Fund for N. Cal. v. Underground Constr. Co., 31 F.3d 776, 779 (9th Cir.1994). In enacting the MPPAA, Congress “recognized the transitory nature of contracts and employment in the building and construction industry.” Id. at 778.

The exception is rooted in the understanding that “[construction industry] employers [will] come and go[,] [but] as long as the base of construction projects in the area covered by the plan [continues] funding the plan’s obligations, the plan, is not threatened” by an individual employer’s departure. Id. It is on this premise that § 1383(b) “aims to extract withdrawal contributions only from those employers who may threaten the plan by reducing the plan’s contribution base,” that is, those employers who continue to do work in the area covered by the plan without contributing to it. Id.

The “contribution base” concept is thus at the core of the MPPAA construction industry withdrawal liability concept. We have previously recognized the centrality of the contribution base in applying the construction industry exception to MPPAA withdrawal liability. In H.C. Elliott, Inc. v. Carpenters Pension Trust Fund for Northern California, 859 F.2d 808 (9th Cir.1988), we observed that “[i]n the construction industry, the funding base of the plan is the construction projects in the area” where the plan is administered. Id. at 812 (quoting H.R.Rep. No. 96-869, 96th Cong., 2d Sess., pt. 1, at 75 (1980)).

We noted further that “as long as contributions are made for whatever work is done in the area,” there is no threat to the plan’s future funding viability; if an individual employer withdraws and goes out of business, other employers who contribute to the pension plan on behalf of their employees will perform that work. Id. (quoting H.R.Rep. No. 96-869, at 75). As we have also explained, “[t]he withdrawal of an[] employer from the plan does decrease the [funding] base ... if the employer stays in the industry but goes non-union and ceases making payments to the plan.” Id. (emphasis added).

In that case, employers continue to undertake construction work without contributing to the plan. So, assuming a constant number of construction projects in a locale, the number of employee hours for which contributions are made will go down. In short, because of concern about shrinking contribution bases, the § 1383(

b) construction industry exception imposes withdrawal liability on employers who cease making payments to the plan while continuing to do business in the area. B. In the fields of labor and employment law, federal courts have developed a common-law doctrine of successorship liability that “provides an exception from the general rule that a purchaser of assets does not acquire a seller’s liabilities.” Chi. Truck Drivers, Helpers & Warehouse Workers Union (Indep.) Pension Fund v. Tasemkin, Inc., 59 F.3d 48, 49 (7th Cir.1995).

The successorship doctrine extends to legal obligations arising under the National Labor Relations Act (“NLRA”), the Fair Labor Standards Act (“FLSA”), Title VII of the Civil Rights Act of 1964 (“Title VII”), and the Family and Medical Leave Act (“FMLA”), among others. See, e.g., Fall River Dyeing & Finishing Corp. v. NLRB, 482 U.S. 27, 107 S.Ct. 2225, 96 L.Ed.2d 22 (1987) (NLRA); Steinbach v. Hubbard, 51 F.3d 843 (9th Cir.1995) (FLSA); Bates v. Pac. Maritime Ass’n, 744 F.2d 705 (9th Cir.1984) (Title VII); Sullivan v.

Dollar Tree Stores, Inc., 623 F.3d 770, 780-81 (9th Cir.2010) (recognizing regulations that incorporate common law successorship principles in defining successors-in-interest for purposes of FMLA liability). Striking a “balance between the need to effectuate federal labor and employment ... policies and the need ... to facilitate the fluid transfer of corporate assets,” the successorship doctrine, when applicable, holds legally responsible for obligations arising under federal labor and employment statutes businesses that are substantial continuations of entities with such obligations.

Upholsterers’ Int’l Union Pension Fund v. Artistic Furniture of Pontiac, 920 F.2d 1323, 1326 (7th Cir.1990). “The inquiry [in these successor-ship cases] is [therefore] not merely whether the new employer is a ‘successor’ in the strict corporate-law sense of the term. The successorship inquiry in the labor-law context is much broader.” Sullivan, 623 F.3d at 781. “The primary question in [labor and employment] successorship cases is whether, under the totality of the circumstances, there is ‘substantial continuity’ between the old and new enterprise.” Haw. Carpenters Trust Funds v.

Waiola Carpenter Shop, Inc., 823 F.2d 289, 294 (9th Cir.1987); see also New England Mech., Inc. v. Laborers Local Union 294, 909 F.2d 1339, 1342 (9th Cir.1990); Steinbach, 51 F.3d at 846.

To address whether the new business is the successor of an old business, we consider the following factors, which are “not ... exhaustive”: [Whether] there has been a substantial continuity of the same business operations^] [whether] the new employer uses the same plant; [whether] the same or substantially the same work force is employed; [whether] the same jobs exist under the same working conditions; [whether] the same supervisors are employed; [whether] the same machinery, equipment, and methods of production are used; and [whether] the same product is manufactured or the same service [is] offered.

Jeffries Lithograph, 752 F.2d at 463 (quoting Premium Foods, Inc., 260 N.L.R.B. 708, 714 (1982), enforced 709 F.2d 623 (9th Cir.1983)) (last alteration in original); see also Haw. Carpenters, 823 F.2d at 294. Other cases have considered whether the body of customers is the same. See, e.g., Fall River Dyeing, 482 U.S. at 43, 107 S.Ct. 2225. “There is, and can be, no single definition of ‘successor’ which is applicable in every legal context. A new employer ... may be a successor for some purposes and not for others.” Howard Johnson Co. v. Detroit Local Joint Exec. Bd., Hotel & Rest.

Emps. & Bartenders Int’l Union, AFL-CIO, 417 U.S. 249, 262 n. 9, 94 S.Ct. 2236, 41 L.Ed.2d 46 (1974). “[Djecisions on successorship must balance, inter alia, the national policies underlying the statute at issue and the interests of the affected parties,” Sullivan, 623 F.3d at 782 (quoting Steinbach, 51 F.3d at 846) (alteration in original). “Because the origins of successor liability are equitable, fairness is a prime consideration in its application.” Id. (Quoting Criswell v. Delta Air Lines, Inc., 868 F.2d 1093, 1094 (9th Cir.1989)).

Thus, these decisions requiref ] analysis of the interests of the new employer and the employees and of the policies of the labor laws in light of the facts of each case and the particular legal obligation which is at issue, whether it be the duty to recognize and bargain with the union, the duty to remedy unfair labor practices, the duty to arbitrate, etc. Id. (quoting Howard Johnson, 417 U.S. at 262 n. 9, 94 S.Ct. 2236). The individual successorship factors outlined in Jeffries are, accordingly, given greater or lesser weight depending on the statutory context.

Moreover, “in light of ... the myriad factual circumstances and legal contexts in which [the employment law successorship issue] can arise,-and the absence of congressional guidance as to its resolution, emphasis on the facts of each case as it arises is especially appropriate.” Howard Johnson, 417 U.S. at 256, 94 S.Ct. 2236. Finally, as the successorship test is “more functional than formal,” “the absence of one ... factor” does not compel a particular conclusion. Hawaii Carpenters, 823 F.2d at 293, 294.

Depending on the statutory context and the type of claim, certain factors may warrant greater or lesser emphasis. For example, under § 8(a)(5) of the NLRA, which imposes on employers a duty to bargain in good faith with the chosen representative of their employees, the NLRB has determined “substantial continuity” with an emphasis on “the employees’ perspective.” Fall River Dyeing, 482 U.S. at 43, 107 S.Ct. 2225.

The reason for this emphasis is that a successor’s § 8(a)(5) duty to bargain ,in good faith derives from the rebuttable presumption of majority support a union obtains once it has been certified as the unit’s bargaining representative. Id. at 37-38, 107 S.Ct. 2225.

The majority presumption generally furthers the NLRA’s “overriding policy” of “ ‘industrial peace’ ” by “promoting] stability in collective-bargaining relationships.” Id. at 38, 107 S.Ct. 2225 (quoting Terrell Machine Co., 173 N.L.R.B. 1480 (1969), enf'd, 427 F.2d 1088 (4th Cir.), cert. denied, 398 U.S. 929, 90 S.Ct. 1821, 26 L.Ed.2d 91 (1970)) (some internal quotation marks omitted) (alteration in original).

Requiring a successor to bargain with the incumbent union even after a change in corporate structure assures employees that their choice of representative is not “subject to the vagaries of an enterprise’s transformation,” and so promotes industrial peace. Id. at 39-40, 107 S.Ct. 2225. Further, “a mere change in ownership, without an essential change in working conditions, is not likely to change employees’ attitudes toward union representation.” Jeffries Lithograph, 752 F.2d at 463.

Consequently, when determining whether a company is a successor with a duty to recognize and bargain with the incumbent union, “the touchstone remains whether there was an ‘essential change in the business that could have affected employee attitudes toward representation.’ ” Id. at 464.

At the same time, in the collective bargaining context, a successor is only obligated.to bargain when “the new employer makes a conscious decision to maintain generally the same business and to hire a majority of its employees from the predecessor ... [and indeed] intends to take advantage of the trained work force of its predecessor.” Fall River Dyeing, 482 U.S. at 41, 107 S.Ct. 2225. Thus limited, the doctrine safeguards employers’ interest in being able to rearrange or sell their business for legitimate purposes. Id.

Balancing these pertinent considerations, courts determine successorship in the context of the NLRA duty to bargain by examining, among other factors, “whether the business of both employers is essentially the same; whether the employees of the new company are doing the same jobs in the same working conditions under the same supervisors; and whether the new entity has the same production process, produces the same products, and basically has the same body of customers,” Fall River Dyeing, 482 U.S. at 43, 107 S.Ct. 2225, all while “keep[ing] in mind the question whether those employees who have been retained will understandably view their job situations as essentially unaltered.” Id. (quoting Golden State Bottling Co. v.

NLRB, 414 U.S. 168, 184, 94 S.Ct. 414, 38 L.Ed.2d 388 (1973)). By contrast, in a different NLRA context — deciding whether to impose successor liability for a predecessor’s unfair labor practices — the Supreme Court placed the emphasis on the employers’ economic considerations, while continuing to take the employees’ perspective into account.

Golden State Bottling determined that a successor could be required to remedy its predecessor’s unlawful discharge of an employee under §§ 8(a)(3) and (1) of the NLRA so long as (1) the successor had obtained substantial assets of the predecessor; (2) there were sufficient indicia of substantial continuity of business operations; and (3) the successor took over with notice of the unfair labor practice liability. 414 U.S. at 184-85, 94 S.Ct. 414.

Golden State Bottling explained that the policies that allow employees to engage in protected concerted activity without incurring retribution support this approach where the predecessor entity engaged in unfair labor practices. “Avoidance of labor strife, prevention of a deterrent effect on the exercise of rights guaranteed by § 7 of the [NLRA], ... and protection for the victimized employee” were all “important policies” that would be undermined absent the imposition of successor liability for unfair labor practices. Id. at 185, 94 S.Ct. 414.

Taking those policies into account, Golden State Bottling held that a successor employer is liable for remedying a predecessor’s violation of its employees’ organizational rights “[w]hen a new employer ... has acquired substantial assets of its predecessor and continued, without interruption or substantial change, the predecessor’s business operations.” Id. at 184, 94 S.Ct. 414. If successor liability were not imposed under those circumstances, “the successor may benefit from the unfair labor practices due to a continuing deterrent effect on union activities.” Id.

Turning to fairness to employers, Golden State Bottling held that successor employers would be held liable only when they took over the business with notice of the liability. Id. at 185, 94 S.Ct. 414. With that protection, the liability could “be reflected in the price [it] pays for the [predecessor’s] business” assets. Id. By focusing on the economic realities of the business transition, Golden State Bottling adapted the successorship doctrine to address a successor’s liability for a predecessor’s unfair labor practice.

The Title VII employment discrimination context provides another example of tailoring successorship factors. There, the “three principal factors [that] bear[] on the appropriateness of successor liability for employment discrimination [are]: (1) the continuity in operations and work force of the successor and predecessor employers, (2) the notice to the successor employer of its predecessor’s legal obligation, and (3) the ability of the predecessor to provide adequate relief directly.” Bates, 744 F.2d at 709-10.

Imposing successor liability under those circumstances is fair, Bates held, even where the successor did not purchase or merge with the predecessor, because a successor “well aware” of its predecessor’s liability is able to consider that information before deciding to continue the predecessor’s business. See id. at 710. Where such notice is provided, the successor’s “choice to take over [its predecessor’s] operations informally through the hiring of its former employees and the purchase of some of its equipment, rather than through a more formal acquisition, [does] not shield it from successorship liability.” Id.

In sum, the cases that have considered in various labor and employment law eon-texts whether an employer is a successor have tailored their analyses to the particular policy concerns underlying the applicable statute and to the particular claim. The successorship standards are flexible and must' be tailored to the circumstances at hand. C. We have not previously decided whether a successor employer can be subject to MPPAA withdrawal liability. We have, however, held, in a closely related context, that a successor can be liable for' its predecessor’s delinquent ERISA contributions.

See Trs. for Alaska LaborersConstr. Indus. Health & Sec. Fund v. Ferrell, 812 F.2d 512, 516 (9th Cir.1987); Hawaii Carpenters, 823 F.2d at 293. Other circuits agree with that result. See Einhorn v. M.L. Ruberton Constr. Co., 632 F.3d 89, 98-99 (3d Cir.2011); Stotter Div. of Graduate Plastics Co. v. Dist. 65, UAW, AFL-CIO, 991 F.2d 997, 1002 (2d Cir.1993); Artistic Furniture, 920 F.2d at 1327-29. We see no reason why the successorship doctrine should not apply to MPPAA withdrawal liability just as it does to the obligation to make delinquent ERISA contributions.

The primary reason for making a successor responsible for its predecessor’s delinquent ERISA contributions is that, “[a]bsent the imposition of successor liability, present and future employer participants in .the union pension plan will bear the burden of [the predecessor’s] failure to pay its share,” which will threaten the health of the plan while the successor reaps a windfall. Artistic Fur niture, 920 F.2d at 1328. That rationale applies with equal, if not greater, force to a predecessor’s MPPAA withdrawal liability.

A primary purpose of ERISA is “to ensure that employees and their beneficiaries [a]re not ... deprived of anticipated retirement benefits by the termination of pension plans before sufficient funds have been accumulated in the plans.” R.A. Gray & Co., 467 U.S. at 722, 104 S.Ct. 2709. The MPPAA’s purpose is better to effectuate ERISA’s purposes. By assessing proportional liability to individual employers who withdraw from a plan, the MPPAA avoids overburdening the remaining participating employers and increases the likelihood that multiemployer plans remain fully funded. See id. at 722-25, 104 S.Ct. 2709.

Contrary to Michael’s’ submissions, “there is no underlying congressional policy here militating against the imposition of [successor] liability.” Golden State Bottling, 414 U.S. at 181, 94 S.Ct. 414. Although Michael’s argues that ERISA § 1384, is in tension with application of the traditional employment law successorship doctrine to impose withdrawal liability on successors, that is not so.

First, 29 U.S.C. § 1384 allows a contributing employer to avoid withdrawal liability where it sells its assets in “a bona fide, arm’s-length sale” and the purchaser both takes on “an obligation to contribute to the plan ... for substantially the same number of contribution base units for which the seller had an obligation to contribute to the plan,” § 1384(a)(1), and provides a bond or other financial assurance sufficient to cover five years of contributions. If the purchaser withdraws from the plan within five years, the seller is subject to withdrawal liability along with the purchaser. § 1384(a)(1)(C).

Although § 1384 establishes one circumstance in which an employer who might — ■ but would not necessarily — otherwise fit into the successor category is not liable for withdrawal payments, it does not address whether the broader employment and labor law successorship doctrine applies where those stringent conditions are not met.

Nor does § 1392, which imposes withdrawal liability on an employer who engages in “any transaction” for which the “principal purpose ... is to evade or avoid liability under [the MPPAA],” suggest any basis for holding the employment and labor law successor liability doctrine inapplicable to MPPAA withdrawal liability.

Section 1392 is essentially punitive. It imposes withdrawal liability for “any” purposely evasive or devious transaction, regardless of the potential impact on the contribution base or on the employees covered by the pension plan. Given its punitive focus, § 1392 does not suggest any intention to displace the usual employment and labor law successorship doctrine, which is remedial rather than punitive and so focuses on objective factors, not on the employer’s purpose in engaging in the transaction.

Finally, the narrow construction industry exception to MPPAA withdrawal liability is fully consistent with the generally applicable successorship doctrine. As explained above, the exception recognizes that, so long as a previously contributing construction employer ceases doing business at the time it withdraws, the funding will remain relatively constant. Where that occurs, other contributing employers are likely to pick up the construction projects that would previously have gone to the withdrawing employer, a. C. Elliott, 859 F.2d at 812.

But “[t]he withdrawal of a[ ] [construction] employer from the plan does decrease the [funding] base ... if the employer stays in the industry but goes non-union and ceases making payments to the plan.” Id. (emphasis added). Then, contributions are not made for the construction jobs the employer is continuing to do in the area. Id. The same detrimental impact occurs where a successor business picks up the work the predecessor would have performed.

Like § 1383(b), which imposes withdrawal liability on employers who cease contributing but continue working in the area, imposing traditional employment and labor law successor liability on employers who substantially continue the business of a construction industry predecessor without contributing to the plan protects the viability of pension funds in the face of a shrinking contribution base.

For all these reasons, we hold that a bona fide successor can be liable for its predecessor’s MPPAA withdrawal liability, both in general and with regard to the special building and construction trade provisions in particular, so long as the successor had notice of the liability. D. We now consider how the established successorship factors are to be weighed in the context of MPPAA withdrawal liability in the construction industry context.

Keeping in mind the flexible successorship inquiry discussed above, “substantial continuity” is “the primary question,” and so the most important consideration, in assessing whether an employer is a successor for purposes of imposing other labor law liabilities. Id.

Fall River Dyeing determined “substantial continuity” by examining, inter alia, “whether the business of both employers is essentially the same; whether the employees of the new company are doing the same jobs in the same working conditions under the same supervisors; and whether the new entity has the same production process, produces the same products, and basically has the same body of customers.” 482 U.S. at 43, 107 S.Ct. 2225.

This definition of “substantial continuity” contains an element Jeffries did not expressly enumerate—whether the successor has “basically the same body of customers” as the predecessor. Id. But Jeffries was decided before Fall River Dyeing, and Jeffries’ list of the pertinent factors was expressly “not ... exhaustive.” Jeffries Lithograph, 752 F.2d at 463. In the current context, we conclude, the “same body of customers” factor is of special significance when determining successorship for purposes of withdrawal liability under the MPPAA construction industry exception.

The consideration whether the successor deliberately takes over “basically the same body of customers,” Fall River Dyeing, 482 U.S. at 43, 107 S.Ct. 2225, dovetails more precisely than any other Fall River Dyeing or Jeffries factors with the underlying rationale for the construction industry exception to MPPAA withdrawal liability—that an employer’s complete withdrawal and cessation of work usually does not harm the plan because other contributing employers will pick up the construction jobs (i.e. the customers) that would have gone to the withdrawing company.

If, instead, an employer uses its insider knowledge to draw a great many of the predecessor’s customers, and so can “pick up where [the predecessor] left off,” doing “[t]he same” “type of work” as the predecessor, yet neither contributes to the pension plan nor pays withdrawal liability, the assumption that animates the construction industry exception collapses. Instead, the plan’s contribution base is compromised, and the plan’s financial stability threatened.

For that reason, focusing the successorship inquiry on business retention through exploitation of the predecessor’s contacts, public presentation, and good will effectuates the purposes of the MPPAA construction industry- withdrawal provisions. It is possible, of course, for a new employer to inherit a substantial portion of a prior employer’s customer base without making any deliberate attempt to do so.

Where that is the case, the entrepreneurial interests of putative successor employers predominate, just as they do in the NLRA successorship context when there is no intention “to take advantage of the trained work force of [their] predecessor[s].” Fall River Dyeing, 482 U.S. at 41, 107 S.Ct. 2225. Where, however, the objective factors indicate that the new employer “ma[de] a conscious decision,” id., to take over the predecessor’s customer base, the equitable origins of the successor liability doctrine support the conclusion that the successor must pay withdrawal liability.

Certain discrete factors, including whether “the new employer uses the same plant” and whether “the same product is manufactured or the same service [is] offered” are pertinent to determining whether the successor has in fact actively and successfully captured its predecessor’s market share. See Jeffries Lithograph, 752 F.2d at 463 (alterations in original). The more closely the successor models itself on its predecessor—for example, by taking over its location and offering the same services as before—the more likely it will succeed in capturing its predecessor’s customers.

Where putative successors do not similarly rely on insider knowledge, similar public presentation, and deliberate continuity of business operations to corner their predecessor’s market share, it cannot be said that they set out to capture the predecessor’s customer base, and the successor doctrine does not apply. The other Jeffries factors are more relevant to NLRA contexts than to the MPPAA withdrawal liability context.

Although the composition of the workforce is of preeminent importance in successor-ship cases involving, for example, the duty to bargain under the NLRA, that factor is not of special relevance here. As we explained above, this factor is relevant to the duty to bargain because “a mere change of ownership, without an essential change in working conditions, is not likely to change employees’ attitudes toward union representation.” Jeffries Lithograph, 752 F.2d at 463.

In light of the presumption of continued majority support, see id., it is fair to require the successor to bargain with an incumbent union if it hires a majority of its workforce from its predecessor’s employee base. Here, by contrast, whether Michael’s hired a majority of its workforce from Studer’s’ employee base is not especially informative in determining whether the premises underlying withdrawal liability in the construction industry apply. The funding base of the Plan is not the particular individuals employed, but, rather, the construction projects in the area. See H.C. Elliott, 859 F.2d at 812.

Given the MPPAA’s primary purpose of protecting the plan’s funding base, the composition of the workforce factor may well, depending on the circumstances, deserve less weight than in the-NLRA context. Finally, whether “the same jobs exist under the same working conditions” may be quite informative' as to whether customers will continue to hire the new contractor. Still, as that consideration has a different significance than in the NLRA context, the particular job similarities that are relevant may differ as well.

Again, the focus in the MPPAA context must be on whether the successor is threatening the plan’s funding base by successfully leveraging factors pertinent to obtaining its predecessor’s market share. E. The district court did not properly identify or weigh the successorship factors as applicable to the MPPAA context. First, and most significantly, the district court did not weigh market share capture as a prime consideration, and so did not make any finding as to whether Michael’s had retained a significant portion of Stu-der’s’ business or body of customers.

Instead, the district court viewed composition of the workforce as’“perhaps the most crucial” factor. The parties disagree about the significance of the number of Studer’s customers captured by Michael’s. The Fund asserts that the bulk of Michael’s’ revenue in its first quarter came from former Studer’s clients, and, further, that by far most of Michael’s’ business customers in its first quarter had been Studer’s’ customers recently.

The spotlight, maintains the Fund, should be on the relative amount of revenue generation by Studer’s’ former customers, rather than on a simple head count of all customers, including one-time customers. Michael’s responds that only 81 of the 868 customers Michael’s served in its first two years were former Studer’s clients, and that Michael’s ability to attract large numbers of individual customers is what matters for the successorship determination. The Fund’s approach is better aligned with the policies underlying the MPPAA withdrawal liability successorship analysis than Michael’s.

The customer base inquiry is critical in this context because it is pertinent to the statutory concern with continuity of contribution rates when business changes take place. Economically, a simple headcount of the number of customers does not synchronize with that concern. Instead, a measure of the billings on the jobs worked for continuing customers by the old and new companies is more useful, as pension fund contributions are usually made based on the total employee hours worked. See, e.g., Bd. of Trustees of W. Conference of Teamsters Pension Trust Fund v. H.F.

Johnson, Inc., 830 F.2d 1009, 1011 (9th Cir.1987). The district court did find, however, that Michael’s was able to retain many of Stu-der’s’ customers, in large part because of its “personal -and business relationships” with large customers of Studer’s. New Tradition Homes did not put out bids to other contractors after Studer’s closed. Instead, New Tradition Homes gave its business to Michael’s without further inquiry, because New Tradition Homes knew Michael’s’ owner, Haasl, from his time as a salesman at Studer’s.

Michael’s may also have been able to capture other Studer’s customers, as the district court recognized, because Michael’s “use[d] ... the same location with the same telephone number and a similar looking sign,” while offering virtually the same service. As they relate to a focus on purposeful takeover of the customer base, these considerations are significant, and point toward finding Michael’s was a successor. The district court considered them, however, only as isolated, independent factors, and so did not find them weighty.

Moreover, by denying the Fund’s motion to supplement the record in part because the “customer issue [wa]s not dispositive of the successor employer determination,” the district court further undermined its consideration of customer base continuity. As we have explained, the “customer issue” could very well be “dispositive” of the successor employer determination. Substantial continuity, measured in large part by capture of Studer’s share of the construction projects in the area, is a critical factor to consider in assessing successor-ship for purposes of imposing MPPAA withdrawal liability.

Because the district court’s “exercise of discretion [in denying the motion to supplement the record was] based on an erroneous

interpretation of the law,” it cannot stand. Estate of Darulis, 401 F.3d at 1063. Further, in considering the “continuity of workforce” factor, the district court used an erroneous method of calculation.

Its conclusion that there was no “continuity of the workforce” between Studer’s and Michael’s rested on a determination that “Michael’s did not employ a majority, or even a substantial portion of Studer’s workforce.” The district court also noted that “the majority of Michael’s installation work is performed by independent contractors rather than employees,” and concluded that this factor also weighed against finding continuity of the workforce. The district court made two errors of law in its method of determining workforce continuity.

First, the appropriate test for determining “continuity of the workforce” is whether “a majority of the new workforce once worked for the old employer,” not whether the successor employs a majority of the predecessor’s workforce. Jeffries Lithograph, 752 F.2d at 464; see also Fall River Dyeing, 482 U.S. at 46 n. 12, 107 S.Ct. 2225 (noting that the NLRB, “with the approval of the Courts of Appeals,” has adopted the

interpretation that “work force continuity ... turn[s] on whether a majority of the successor’s employees were those of the predecessor”); NLRB v. Advanced Stretchforming Int’l, Inc., 233 F.3d 1176, 1180 (9th Cir.2000); Williams Enters., Inc. v. NLRB, 956 F.2d 1226, 1232 (D.C.Cir.1992). Second, only employees in the “bargaining unit,”—that is, the installers actually employed by Michael’s who are the individuals as to whom pension fund contributions would be due— should be included in the workforce continuity test. See Small v.

Avanti Health Sys., LLC, 661 F.3d 1180, 1188 (9th Cir.2011) (stating, in context of duty to bargain, that whether there was continuity of the workforce is determined by examining employees within the relevant bargaining unit). If it had used the correct metrics, the district court might well have found there was workforce continuity here. It appears that five of Michael’s eight employee installers had previously worked for Studer’s.

That some of these employees did not “move[ ] directly from Studer’s to Michael’s,” is not dispositive; “a hiatus” between employers “is only one factor in the ‘substantial continuity’ calculus.” Fall River Dyeing, 482 U.S. at 45, 107 S.Ct. 2225.

Finally, the district court also stated that “the successor employer determination ... involvfes] [a finding that the successor has] ‘basically the same owners and operators as ... the predecessor employer,’ ” and that the “changes between predecessor and successor were technical in nature rather than a substantive change in the management.’ ” (quoting New Eng land Mech., 909 F.2d at 1348). The reliance on New England Mechanical for these propositions was mistaken.

New England Mechanical concerns the question whether a successor employer is so similar to its predecessor that it is bound by the substantive provisions of its predecessor’s collective bargaining agreement with a union. 909 F.2d at 1343. Generally, although a successor may have a duty to bargain with an incumbent union, successors are not bound by the substantive contractual terms of their predecessors’ collective bargaining agreements, to which they were not signatories. Id. at 1342; see also Fall River Dyeing, 482 U.S. at 40,107 S.Ct. 2225; NLRB v. Burns Int’l Sec.

Servs., Inc., 406 U.S. 272, 284, 92 S.Ct. 1571, 32 L.Ed.2d 61 (1972). A successor may be bound by the terms of its predecessor’s collective bargaining agreement if it has exhibited an intent to be bound, or if it is so closely related to the prior business that it is effectively an “alter ego” of that business. New England Mech., 909 F.2d at 1342. New England Mechanical thus did not disturb our general rule that “[t]he successorship inquiry in [other employment and] labor-law contexts] is much broader” than the “strict corporate-law sense of [successorship].” Sullivan, 623 F.3d at 781.

The successorship test in the MPPAA context does not require that the changes between Studer’s and Michael’s be merely “technical in nature,” nor does it require that both entities have “basically the same owners and operators.” Instead, the district court must apply the Jeffries/Fall River Dyeing successorship factors, with special emphasis on substantial continuity as measured by customer retention. IV.

The district court took an erroneously narrow view of the successorship inquiry, applied the successorship factors acontex-tually, miscalculated the continuity of the workforce factor, and imposed the unwarranted requirement that the change of ownership be merely “technical in nature.” We therefore reverse and remand for application of the labor and employment law successorship factors as appropriately weighted for MPPAA construction industry withdrawal liability purposes, and to take additional evidence as necessary to decide the relevant factual issues. REVERSED AND REMANDED. .

We note that the record in this case was sealed in the district court. Under this. Court’s rules, that sealing remains in effect on appeal unless we rule otherwise, which neither party in this case asked us to do. 9th Cir. R. 27-13. We note, however, that the sealing of several key documents, including Michael’s’ business plan, has somewhat hampered our ability fully to explain our ruling in ' this precedential opinion.

Further, we have noticed an overall tendency recently for parties to request, and district courts to grant, the sealing of records in instances in which it is hard to see any significant privacy or trade secret justification. We could, of course, request the parties to show cause as to why the record should not be unsealed in whole or in part. But that process would take time and effort away from the preparation of the opinion. We have therefore chosen instead to issue an opinion that does not contain all the facts in the record supporting it.

Our need to choose between undesirable options suggests the need to reconsider record sealing practices both in the district courts and in this court. . Hereafter, all statutory references are to

Chapter 29 of the United States Code unless otherwise indicated. . Resilient Floor Covering Pension Fund v. M & M Installation, Inc., 630 F.3d 848, 852 (9th Cir.2010) assumed without deciding that a company could be held responsible for another entity's withdrawal liability under an alter ego theory. M & M Installation also noted that it was not presented with the question whether there were other ways in which a company could be responsible for another entity’s ERISA withdrawal liability. Id. at 855. . The Seventh Circuit has so indicated as well.

See Chicago Truck Drivers, 59 F.3d at 49; see also Artistic Furniture, 920 F.2d at 1327. No circuit has held otherwise. Several district courts have reached the same conclusion. See, e.g., Cent. States, Se. & Sw. Areas Pension Fund v. Hayes, 789 F.Supp. 1430, 1436 (N.D.Ill.1992) (holding that a successor can be subject to predecessor's unpaid MPPAA withdrawal liability so long as there exists substantial continuity and notice); Auto. Indus. Pension Trust Fund v. S. City Ford, Inc., No. C 11-04590 CW, 2012 WL 1232109 (N.D.Cal. Apr. 12, 2012) (same); Trs. of Utah Carpenters’ & Cement Masons’ Pension Trust v.

Daw, Inc., No. 2:07-CV-87 TC, 2009 WL 77856, at *3 (D.Utah Jan. 7, 2009) (same). . Individual, nonrepeat customers may also reflect a functional continuity of the customer base. Word-of-mouth or professional referrals of residential customers may recommend a successor business because of the transfer of reputation and goodwill. Such factors are also not captured by a simple customer headcount, but are likely to be hard to demonstrate other than anecdotally.