This case asks us to decide whether garment manufacturers who hired contractors to stitch and finish pieces of clothing were “joint employers” within the meaning of the Fair Labor Standards Act of 1938 (“FLSA”), 29 U.S.C. § 201 et seq., and New York law. Plaintiffs, garment workers in New York City who were directly employed by the contractors, claim that the manufacturers were their joint employers because they worked predominantly on the manufacturers’ garments, they performed a line-job that was integral to the production of the manufacturer’s product, and their work was frequently and directly supervised by the manufacturers’ agents. The manufacturers respond that the con*64tractors, who, among other things, hired and paid plaintiffs to assemble clothing for numerous manufacturers, were plaintiffs’ sole employers. Both plaintiffs and the manufacturers moved for summary judgment on the issue of joint employment.
The United States District Court for the Southern District of New York (Richard Conway Casey, Judge), applying the four-factor test set forth in Carter v. Dutchess Community College, 735 F.2d 8 (2d Cir. 1984), granted the manufacturers’ motion, and held that the manufacturers could not be held liable for violations of the FLSA or its New York statutory analogues. The District Court also declined to exercise supplemental jurisdiction over a surviving New York claim.
We conclude that the District Court erred when it limited its analysis to the four factors identified in Carter. Accordingly, we vacate the judgment of the District Court and remand the cause to the District Court with instructions concerning further proceedings.
BACKGROUND
The relevant facts are laid out in Judge Casey’s opinion, Zheng v. Liberty Apparel Co., 2002 WL 398663, at *1-2 (S.D.N.Y. Mar.13, 2002), and we recount only those facts necessary to resolve the issues on appeal. Unless otherwise noted, the facts are undisputed.
Plaintiffs-Appellants are 26 non-English-speaking adult garment workers who worked in a factory at 103 Broadway in New York’s Chinatown. They brought this action against both (1) their immediate employers, six contractors doing business at 103 Broadway (“Contractor Corporations”) and their principals (collectively, “Contractor Defendants”), and (2) Liberty Apparel Company, Inc. (“Liberty”) and its principals, Albert Ni-gri and Hagai Laniado (collectively, “Liberty Defendants”). Because the Contractor Defendants either could not be located or have ceased doing business, plaintiffs have voluntarily dismissed their claims against those defendants with prejudice. Accordingly, plaintiffs now seek damages only from the Liberty Defendants.
Liberty, a “jobber” in the parlance of the garment industry, is a manufacturing company that contracts out the last phase of its production process. That process, in broad terms, worked as follows: First, Liberty employees developed a pattern for a garment, cut a sample from the pattern, and sent the sample to a customer for approval. Once the customer approved the pattern, Liberty purchased the necessary fabric from a vendor, and the vendor delivered the fabric to Liberty’s warehouse. There, the fabric was graded and marked, spread out on tables, and, finally, cut by Liberty employees.
After the fabric was cut, Liberty did not complete the production process on its own premises. Instead, Liberty delivered the cut fabric, along with other essential materials, to various contractors for assembly. The assemblers, in turn, employed workers to stitch and finish the pieces, a process that included sewing the fabrics, buttons, and labels into the garments, cuffing and hemming the garments, and, finally, hanging the garments. The workers, including plaintiffs, were paid at a piece rate for their labor.
From March 1997 through April 1999, Liberty entered into agreements with the Contractor Corporations under which the Contractor Corporations would assemble garments to meet Liberty’s specifications. During that time period, Liberty utilized as many as thirty to forty assemblers, including the Contractor Corporations. Liberty did not seek out assemblers; in*65stead, assemblers came to Liberty’s warehouse looking for assembly work. In order to obtain such work, a prospective assembler was required by Liberty to sign a form agreement.
Plaintiffs claim that approximately 70-75% of their work during the time period at issue was for Liberty. They explain that they knew they were working for Liberty based on both the labels that were sewn into the garments and the specific lot numbers that came with the garments. Liberty’s co-owner, Albert Nigri, asserts that the percentage of the Contractor Corporations’ work performed for Liberty was closer to 10-15%. He derives that figure from individual plaintiffs’ handwritten notes and records.
The parties do not dispute that Liberty employed people to monitor Liberty’s garments while they were being assembled. However, the parties dispute the extent to which Liberty oversaw the assembly process. Various plaintiffs presented affidavits to the District Court stating that two Liberty representatives — a man named Ah Sen and “a Taiwanese woman” — visited the factory approximately two to four times a week for up to three hours a day, and exhorted the plaintiffs to work harder and faster. In their affidavits, these plaintiffs claim further that, when they finished working on garments, Liberty representatives — -as opposed to employees of the Contractor Corporations — inspected their work and gave instructions directly to the workers if corrections needed to be made. One of the plaintiffs also asserts that she informed the “Taiwanese woman” that the workers were not being paid for then-work at the factory.
Albert Nigri, on the other hand, avers that Liberty’s quality control person made brief visits to assemblers’ factories and was instructed to speak only with Lai Huen Yam, a co-owner of the Contractor Corporations, or with his wife. Furthermore, Nigri asserts in his affidavit that Liberty representatives were expected to spend just thirty minutes at each of the assemblers’ work sites. Finally, Nigri states that Liberty did not employ two quality control persons simultaneously; did not employ a quality control person during some of the relevant time period; and did not employ a man as a quality control person.
In their complaint, plaintiffs alleged that both the Liberty Defendants and the Contractor Defendants violated 29 U.S.C. § 206 and N.Y. Lab. Law § 652(1) (“§ 652(1)”), which require an employer to pay employees a legally mandated minimum wage. Plaintiffs alleged further that all of the defendants, including Liberty and its principals, violated 29 U.S.C. § 207 and N.Y. Comp.Codes R. & Regs. tit. 12, § 142-2.2 (“§ 142-2”), which require employers to compensate employees at one- and-one-half times the regular rate when an employee works in excess of 40 hours per week. Additionally, plaintiffs alleged that defendants violated N.Y. Lab. Law § 191, which requires employers to pay manual workers on a weekly basis, and N.Y. Lab. Law § 193, which prevents employers from making unauthorized deductions from employees’ wages. Finally, plaintiffs alleged that, in violation of N.Y. Lab. Law § 345-a (“§ 345-a”) — a statutory provision that applies to the apparel industry only — Liberty Defendants entered into contracts with the Contractor Corporations even though they knew, or should have known, that the Contractor Corporations failed to comply with the provisions of New York law that govern the payment of wages.
Liberty Defendants moved for summary judgment on all claims against them on the ground that plaintiffs were not their employees. Plaintiffs cross-moved for sum*66mary judgment, seeking a determination that they were employed by Liberty. In a March 13, 2002 Opinion and Order, the District Court granted Liberty Defendants’ motion for summary judgment and dismissed every federal and state claim in the complaint on the merits except the claim arising under N.Y. Lab. Law § 345-a, which does not require an employment relationship. See Zheng, 2002 WL 398663, at *7-9. The District Court determined that Liberty Defendants were not joint employers under the FLSA because, based on the plaintiffs’ own admissions, these defendants did not (1) hire and fire the plaintiffs, (2) supervise and control their work schedules or conditions of employment, (3) determine the rate and method of payment, or (4) maintain employment records. See id. (applying the factors identified in Carter, 735 F.2d at 12). The District Court then declined to exercise pendent jurisdiction over the surviving claim under N.Y. Lab. Law § 345, and dismissed the complaint. Id. at *9.
Plaintiffs appeal (1) the District Court’s grant of summary judgment dismissing the FLSA claims; (2) the dismissal on the merits of plaintiffs’ claims under New York’s statutory analogues to the FLSA, ie., the § 652(a) and § 142-2 claims; and (3) the District Court’s refusal to exercise pendent jurisdiction over the § 345-a claim.1
DISCUSSION
I. FLSA Claims
A. Competing Economic Reality Tests
The FLSA defines “employee” as “any individual employed by an employer.” 29 U.S.C. § 203(e)(1). An entity “employs” an individual under the FLSA if it “suffer[s] or permit[s]” that individual to work. 29 U.S.C. § 203(g). This definition is necessarily a broad one, in accordance with the remedial purpose of the FLSA. See United States v. Rosenwasser, 323 U.S. 360, 363 & n. 3, 65 S.Ct. 295, 89 L.Ed. 301 (1945); Brock v. Superior Care, Inc., 840 F.2d 1054, 1058 (2d Cir.1988).
An entity “suffers or permits” an individual to work if, as a matter of “economic reality,” the entity functions as the individual’s employer. See Goldberg v. Whitaker House Coop., Inc., 366 U.S. 28, 33, 81 S.Ct. 933, 6 L.Ed.2d 100 (1961) (explaining that “ ‘economic reality’ rather than ‘technical concepts’ is ... the test of employment” under the statute); see also Herman v. RSR Security Services Ltd., 172 F.3d 132, 139 (2d Cir.1999) (“[T]he overarching concern is whether the alleged employer possessed the power to control the workers in question.”). The regulations promulgated under the FLSA expressly recognize that a worker may be employed by more than one entity at the same time. See 29 C.F.R. § 791.2 (2003); see also Torres-Lopez v. May, 111 F.3d 633, 639-45 (9th Cir.1997) (permitting claims against joint employers under FLSA); Antenor v. D & S Farms, 88 F.3d 925, 929-38 (11th Cir.1996) (same). In the present case, it is undisputed that the Contractor Defendants, who are no longer parties to this suit, employed plaintiffs. The issue is whether the Liberty Defendants also employed them.
In previous cases, we have applied two different tests to determine whether an employment relationship exists in light of the Supreme Court’s admonition that “economic reality” govern our application of the FLSA. In Carter v. Dutchess Community College, 735 F.2d 8 (2d Cir.1984), we *67held that an inmate conducting tutorial classes in a program managed by a community college had raised genuine issues of material fact as to whether the college was an employer under the FLSA, where, among other things, the college sent compensation directly to the inmate and set the inmate’s tutoring schedule. See Carter, 735 F.2d at 13-15. To reach that conclusion, we evaluated whether the college
(1) had the power to hire and fire the employees, (2) supervised and controlled employee work schedules or conditions of employment, (3) determined the rate and method of payment, and (4) maintained employment records.
See Carter, 735 F.2d at 12 (borrowing factors from Bonnette v. Cal. Health & Welfare Agency, 704 F.2d 1465, 1470 (9th Cir. 1983)).
More recently, we applied the same four-factor test in Herman v. RSR Security Services Ltd., 172 F.3d 132, 139 (2d Cir.1999), and affirmed the trial court’s judgment, after a bench trial, that a company’s chairman and co-owner qualified as a joint employer under the FLSA. We ratified the District Court’s judgment because “[ejvidence in the record supported] at least three of the four [Carter ] factors.” Id. at 140.
Four years after deciding Carter but more than ten years before deciding Herman, we decided Brock v. Superior Care, Inc., 840 F.2d 1054 (2d Cir.1988). There, en route to affirming the District Court’s determination after a bench trial that nurses engaged by a health-care service were employees of the service rather than independent contractors, we applied a different, more expansive test. That test, drawn from cases distinguishing employees from independent contractors, examined
(1) the degree of control exercised by the employer over the workers, (2) the workers’ opportunity for profit or loss and their investment in the business, (3) the degree of skill and independent initiative required to perform the work, (4) the permanence or duration of the working relationship, and (5) the extent to which the work is an integral part of the employer’s business.
See Superior Care, 840 F.2d at 1058-59 (citing, inter alia, United States v. Silk, 331 U.S. 704, 716, 67 S.Ct. 1463, 91 L.Ed. 1757 (1947)).
We have stated that “[t]he [four-factor] Bonnette test,” borrowed by us from the Ninth Circuit in Carter, “is useful largely in cases involving claims of joint employment,]” Danneskjold v. Hausrath, 82 F.3d 37, 43 (2d Cir.1996),2 because, when an entity exercises those four prerogatives, that entity, in addition to any primary employer, must be considered a joint employer. The Superior Care factors, on the other hand, and particularly factors two and three — the workers’ investment in the business, and the degree of skill and independent initiative required of workers— have been used primarily to distinguish independent contractors from employees, because, unlike the Carter factors, they do *68not bear directly on whether workers who are already employed by a primary employer are also employed by a second employer. Instead, they help courts determine if particular workers are independent of all employers.
Despite this distinction between joint employment cases and independent contractor cases, we have never suggested that, in analyzing joint employment, the four Carter factors alone are relevant, and that other factors that bear on the relationship between workers and potential joint employers should be ignored. Thus, in Lopez v. Silverman, 14 F.Supp.2d 405 (S.D.N.Y.1998), which arose out of a factual backdrop somewhat similar to ours, Judge Denise Cote attempted to combine the tests used in Carter and Superior Care. Drawing on those two cases, along with the Supreme Court’s decision in Rutherford Food Corp. v. McComb, 331 U.S. 722, 67 S.Ct. 1473, 91 L.Ed. 1772 (1947),3 Judge Cote concluded that the following seven factors should be considered in determining whether garment workers are jointly employed by a “jobber”:
(1) the extent to which the workers perform a discrete line-job forming an integral part of the putative joint employer’s integrated process of production or overall business objective;
(2) whether the putative joint employer’s premises and equipment were used for the work;
(3) the extent of the putative employees’ work for the putative joint employer;
(4) the permanence or duration of the working relationship between the workers and the putative joint employer;
(5) the degree of control exercised by the putative joint employer over the workers;
(6) whether responsibility under the contract with the putative joint employer passed “without material changes” from one group of potential joint employees to another; and
(7) whether the workers had a “business organization” that could or did shift as a unit from one putative joint employer to another.
See Lopez 14 F.Supp.2d at 419-20.4 In choosing these factors, Judge Cote concluded that the four factors identified in Carter could not, on their own, constitute the “economic reality” test, because they rarely permit a finding of joint employment “outside of situations involving direct corporate subsidiaries or managing administrators.” Id. at 415. However, Judge Core concluded also that the five Superior Care factors, on their own, likewise could not make up the “economic reality” test, because those factors, when applied to joint employment cases, are “equally skewed in the opposite direction.” Id.
The District Court in this case expressly declined to follow Lopez. Citing Herman, 172 F.3d at 139, Judge Casey reasoned that, in order to be held liable under the FLSA’s economic reality test, “an alleged employer must, at a minimum, possess the power to control the workers in question.” Zheng, 2002 WL 398663, at *6. The Court then declared that the Carter test, which was reaffirmed in Herman, remains the law of the Circuit for determining whether an entity possesses the requisite control over workers, and that Lopez cannot be reconciled with Carter. Id. Applying the *69four Carter factors only, the District Court concluded that the Liberty Defendants lacked sufficient control over plaintiffs to qualify as joint employers, and thus granted their motion for summary judgment. Id. at *1.
B. Need for Remand
We conclude, for the reasons set forth below, that the District Court erred when, based exclusively on the four factors mentioned in Carter, it determined that the Liberty Defendants were not, as a matter of law, joint employers under the FLSA. In our view, the broad language of the FLSA, as interpreted by the Supreme Court in Rutherford, demands that a district court look beyond an entity’s formal right to control the physical performance of another’s work before declaring that the entity is not an employer under the FLSA. Moreover, as explained below, neither Carter nor Herman supports the application of a rigid four-part test in the instant case. In those cases, we held only that the four factors applied by the District Court in this case can be sufficient to establish employer status. We did not hold, nor under Rutherford could we have held, that a positive finding on those four factors is necessary to establish an employment relationship. Accordingly, the District Court’s judgment in favor of the Liberty Defendants must be vacated.
1. The Language of the Statute
As noted above, the relevant provision of the FLSA, 29 U.S.C. § 203(g), defines “employ” as including “to suffer or permit to work.” This is “ ‘the broadest definition [of ‘employ’] that has ever been included in any one act,’ ” United States v. Rosenwas-ser, 323 U.S. 360, 363 n. 3, 65 S.Ct. 295, 89 L.Ed. 301 (1945) (quoting 81 Cong. Rec. 7657 (1937) (statement of Sen. Hugo L. Black)),5 and it encompasses “working relationships, which prior to [the FLSA], were not deemed to fall within an employer-employee category,” Walling v. Portland Terminal Co., 330 U.S. 148, 150-51, 67 S.Ct. 639, 91 L.Ed. 809 (1947).
Measured against the expansive language of the FLSA, the four-part test employed by the District Court is unduly narrow, as it focuses solely on the formal right to control the physical performance of another’s work. That right is central to the common-law employment relationship, see Restatement of Agency § 220(1) (1933) (“A servant is a person employed to perform service for another in his affairs and who, with respect to his physical conduct in the performance of the service, is subject to the other’s control or right to control.”), and, therefore, the four-factor test may approximate the common-law test for identifying joint employers. However, the four-factor test cannot be reconciled with the “suffer or permit” language in the statute, which necessarily reaches beyond traditional agency law. See Nationwide Mut. Ins. Co. v. Darden, 503 U.S. 318, 326, 112 S.Ct. 1344, 117 L.Ed.2d 581 (1992) (noting that the “suffer or permit to work” formulation “stretches the meaning of ‘employee’ to cover some parties who might not qualify as such under a strict application of traditional agency law principles”); Bartels v. Birmingham, 332 U.S. 126, 130, 67 S.Ct. 1547, 91 L.Ed. 1947 (1947) (explaining that, under the economic reality test, “employees” are not limited to those who are subject to the physical control of an employer). Indeed, the narrow approach used by the District Court may be *70more rigid than the common-law approach.6
2. Rutherford
Rutherford confirmed that the definition of “employ” in the FLSA cannot be reduced to formal control over the physical performance of another’s work. In Rutherford, the Supreme Court held that a slaughterhouse jointly employed workers who de-boned meat on its premises, despite the fact that a boning supervisor— with whom the slaughterhouse had entered into a contract — directly controlled the terms and conditions of the meat boners’ employment. Specifically, the supervisor, rather than the slaughterhouse, (i) hired and fired the boners, (ii) set their hours, and, (iii) after being paid a set amount by the slaughterhouse for each one hundred pounds of de-boned meat, paid the boners for their work. Rutherford, 331 U.S. at 726, 730, 67 S.Ct. 1473.
In determining that the meat boners were employees of the slaughterhouse notwithstanding the role played by the boning supervisor, the Court examined the “circumstances of the whole activity,” id. at 730, 67 S.Ct. 1473, but also isolated specific relevant factors that help distinguish a legitimate contractor from an entity that “suffers or permit[s]” its subcontractor’s employees to work. First, the Court noted that the boners “did a specialty job on the production line”; that is, their work was “a part of the integrated unit of production” at the slaughterhouse. Id. at 729-30, 67 S.Ct. 1473. The Court noted also that- responsibility under the boning contracts passed from one boning supervisor to another “without material changes” in the work performed at the slaughterhouse; that the slaughterhouse’s premises and equipment were used for the boners’ work; that the group of boners “had no business organization that could or did shift as a unit from one slaughterhouse to another”; and that the managing official of the slaughterhouse, in addition to the boners’ purported employer, closely monitored the boners’ performance and productivity. Id. Based on its analysis of these factors, the Court imposed FLSA liability on the slaughterhouse.
Like the case at bar, Rutherford was a joint employment case, as it is apparent from the Supreme Court’s opinion that the boners were, first and foremost, employed by the boning supervisor who had entered into a contract with the slaughterhouse. See id. at 724-25 (explaining that the boning supervisor exercised the prerogatives of an employer, including hiring workers, managing their work, and paying them). Rutherford thus held that, in certain circumstances, an entity can be a joint employer under the FLSA even when it does not hire and fire its joint employees, directly dictate then-hours, or pay them.7
*713. Carter and Herman
Carter and Herman are consistent with Rutherford, and neither case supports the application of the test used by the District Court in the circumstances presented in the instant case to negate employer liability. The question before us in Carter was whether prisoners who performed work for an educational institution could be considered employees of that institution under the FLSA. Carter, 735 F.2d at 12. In rejecting the district court’s conclusion that they could not be considered employees as a matter of law, we did not purport to identify all factors that could bear on the employer status question. Instead, we stated that the economic reality test “include[s]” an inquiry into the four factors. Id. (emphasis added). We then determined that summary judgment in favor of the defendant was inappropriate considering that the defendant exercised most of the “typical employer prerogatives” encompassed by the four factors. Id. at 14-15. Carter thus stands solely for the proposition that the four factors applied by the District Court in the instant case can be sufficient to establish employer status. Carter did not hold, nor could it have held in light of Rutherford, that those factors are necessary to establish an employment relationship.8
In our more recent decision in Herman, moreover, we affirmed the District Court’s determination after a bench trial that a company chairman jointly employed the company’s employees where the chairman exercised three of the four employer prerogatives identified in Carter. In doing so, we reiterated that “economic reality is determined based upon all the circumstances, [and] any relevant evidence may be examined so as to avoid having the test confined to a narrow legalistic definition.” Herman, 172 F.3d at 139 (emphasis in original). Thus, as in Carter, we indicated in Herman that where the four factors weigh in favor of a district court’s finding of joint employment, that finding will not be disturbed on appeal. We did not suggest — indeed, we expressly denied — that the four factors borrowed from the Ninth Circuit in Carter are the exclusive touchstone of the joint employment inquiry under the FLSA.
Because the District Court in the instant case interpreted our precedents to demand an exclusive four-factor test, we vacate its judgment and remand for further proceedings.
C. Instructions on Remand
1. Factors to Be Applied
On remand, the District Court must determine whether the Liberty Defendants should be deemed to have been the plaintiffs’ joint employer. This determination is to be based on “the circumstances of the whole activity,” Rutherford, 331 U.S. at 730, 67 S.Ct. 1473, viewed in light of “economic reality,” Goldberg, 366 U.S. at 33, 81 S.Ct. 933. We discuss below factors, drawn from Rutherford, which we think the court will find illuminating in these circumstances. The court is also free to *72consider any other factors it deems relevant to its assessment of the economic realities.
The factors we find pertinent in these circumstances, listed in no particular order, are (1) whether Liberty’s premises and equipment were used for the plaintiffs’ work; (2) whether the Contractor Corporations had a business that could or did shift as a unit from one putative joint employer to another; (3) the extent to which plaintiffs performed a discrete line-job that was integral to Liberty’s process of production; (4) whether responsibility under the contracts could pass from one subcontractor to another without material changes; (5) the degree to which the Liberty Defendants or their agents supervised plaintiffs’ work; and (6) whether plaintiffs worked exclusively or predominantly for the Liberty Defendants. See Rutherford, 331 U.S. at 724-25, 730, 67 S.Ct. 1473; see also Lopez, 14 F.Supp.2d at 416-18 (summarizing the factors considered in Rutherford ).
These particular factors are relevant because, when they weigh in plaintiffs’ favor, they indicate that an entity has functional control over workers even in the absence of the formal control measured by the Carter factors. Thus, in Rutherford, by looking beyond the boning supervisor’s formal prerogatives, the Supreme Court determined, based principally on the factors listed above, that the slaughterhouse dictated the terms and conditions of the boners’ employment. First, although it did not literally pay the workers, the slaughterhouse de facto set the workers’ wages, because the boners did no meat boning for any other firm and shared equally in the funds paid to the boning supervisor. See Rutherford, 331 U.S. at 726, 730, 67 S.Ct. 1473. The slaughterhouse also controlled employee work schedules, both because the boners’ hours were dependent on the number of cattle slaughtered, and also because the slaughterhouse manager was constantly “after” the boners about their work. See Rutherford, 331 U.S. at 726, 67 S.Ct. 1473. Finally, the slaughterhouse effectively “controlled the [boners’] ... conditions of employment,” Carter, 735 F.2d at 12, because the boners worked for the slaughterhouse as an in-house boning unit on the slaughterhouse’s premises, see id. at 730, 67 S.Ct. 1473. In sum, the relationship between the slaughterhouse and the successive boning supervisors who managed the boners had no substantial, independent economic purpose; instead, it was most likely a subterfuge meant to evade the FLSA or other labor laws.
The first two factors derived from Rutherford require minimal discussion. The first factor — namely, whether a putative joint employer’s premises and equipment are used by its putative joint employees— is relevant because the shared use of premises and equipment may support the inference that a putative joint employer has functional control over the plaintiffs’ work. Similarly, the second factor' — namely, whether the putative joint employees are part of a business organization that shifts as a unit from one putative joint employer to another' — is relevant because a subcontractor that seeks business from a variety of contractors is less likely to be part of a subterfuge arrangement than a subcontractor that serves a single client. Although neither shared premises nor the absence of a broad client base is anything close to a perfect proxy for joint employment (because they are both perfectly consistent with a legitimate subcontracting relationship), the factfinder can use these readily verifiable facts as a starting point in uncovering the economic realities of a business relationship.
*73The other factors we have pointed out are less straightforward. Rutherford considered the extent to which plaintiffs performed a line-job that is integral to the putative joint employer’s process of production. Interpreted broadly, this factor could be said to be implicated in every subcontracting relationship, because all subcontractors perform a function that a general contractor deems “integral” to a product or a service. However, we do not interpret the factor quite so broadly. The factor is derived from the Rutherford Court’s statement that the boners at the slaughterhouse should be considered joint employees because, inter alia, “[they] did a specialty job on the production line.” Rutherford, 331 U.S. at 730, 67 S.Ct. 1473. Based on this statement in Rutherford,9 along with similar language in decisions interpreting Rutherford, see, e.g., Anterior, 88 F.3d at 937 (noting that farmworkers were “analogous to employees working at a particular position on a larger production line” (citing Rutherford, 331 U.S. at 729-30, 67 S.Ct. 1473)), we construe Rutherford to mean that work on a production line occupies a special status under the FLSA, at least when it lies on “the usual path of an employee,” id. at 729, 67 S.Ct. 1473.
Rutherford, however, offers no firm guidance as to how to distinguish work that “in its essence, follows the usual path of an employee,” id., from work that can be outsourced without attracting increased scrutiny under the FLSA. In our view, there is no bright-line distinction between these two categories of work. On one end of the spectrum lies the type of work performed by the boners in Rutherford— i.e., piecework on a producer’s premises that requires minimal training or equipment, and which constitutes an essential step in the producer’s integrated manufacturing process. On the other end of the spectrum lies work that is not part of an integrated production unit, that is not performed on a predictable schedule, and that requires specialized skills or expensive technology. In classifying business relationships that fall in between these two poles, we are mindful of the substantial and valuable place that outsourcing, along with the subcontracting relationships that follow from outsourcing, have come to occupy in the American economy. See, e.g., The Outing of Outsourcing, The Economist, Nov. 25, 1995, at 57, 57 (noting that outsourcing “is part and parcel of the way American companies of all sizes do business”). We are also mindful that manufacturers, and especially manufacturers of relatively sophisticated products that require multiple components, may choose to outsource the production of some of those components in order to increase efficiency. See, e.g., Ravi Venkatesan, Strategic Sourcing: To Make or Not to Make, Harv. Bus. Rev., Nov./Dec.l992, at 98 (arguing that manufacturers should outsource the production of components to maximize efficiency). Accordingly, we resist the temptation to say that any work on a so-called production line — no matter what product is being manufactured' — -should attract heightened scrutiny. Instead, in determining the weight and degree of factor (3), we believe that both industry custom and historical practice should be consulted. Industry custom may be relevant because, insofar as the practice of using subcontractors to complete a particular task is widespread, it is unlikely to be a mere subterfuge to avoid complying with labor laws. At the same time, historical practice may *74also be relevant, because, if plaintiffs can prove that, as a historical matter, a contracting device has developed in response to and as a means to avoid applicable labor laws, the prevalence of that device may, in particular circumstances, be attributable to widespread evasion of labor laws. Ultimately, this factor, like the other factors derived from Rutherford, is not independently determinative of a defendant’s status, because the mere fact that a manufacturing job is not typically outsourced does not necessarily mean that there is no substantial economic reason to outsource it in a particular case. However, as Rutherford indicates, the type of work performed by plaintiffs can bear on the overall determination as to whether a defendant may be held liable for an FLSA violation.10
The fourth factor the Court considered in Rutherford is whether responsibility under the contracts could pass from one subcontractor to another without material changes. That factor is derived from the Rutherford Court’s observation that “[t]he responsibility under the boning contracts without material changes passed from one boner to another.” Rutherford, 331 U.S. at 730, 67 S.Ct. 1473. In the quoted passage, the Supreme Court was referring to the fact that, even when the boning supervisor abandoned his position and another supervisor took his place (as occurred several times, see id. at 725, 67 S.Ct. 1473), the same employees would continue to do the same work in the same place. Under Rutherford, therefore, this factor weighs in favor of a determination of joint employment when employees are tied to an entity such as the slaughterhouse rather than to an ostensible direct employer such as the boning supervisor. In such circumstances, it is difficult not to draw the inference that a subterfuge arrangement exists. Where, on the other hand, employees work for an entity (the purported joint employer) only to the extent that their direct employer is hired by that entity, this factor does not in any way support the determination that a joint employment relationship exists.11
The fifth factor listed above — namely, the degree to which the defendants supervise the plaintiffs’ work — also requires some comment, as it too can be misinterpreted to encompass run-of-the-mill subcontracting relationships. Although Rutherford indicates that a defendant’s extensive supervision of a plaintiffs work is indicative of an employment relationship, see Rutherford, 331 U.S. at 730, 67 S.Ct. 1473 (noting that “[t]he managing *75official of the plant kept close touch on the operation”), Rutherford indicates also that such extensive supervision weighs in favor of joint employment only if it demonstrates effective control of the terms and conditions of the plaintiffs employment, see Rutherford, 331 U.S. at 726, 67 S.Ct. 1473 (suggesting the slaughterhouse owner’s close scrutiny of the boners’ work played a role in setting the boners’ schedule); see also Anterior, 88 F.3d at 934 (growers exercised control over farm-workers when their supervision of the workers affected the workers’ schedule). By contrast, supervision with respect to contractual warranties of quality and time of delivery has no bearing on the joint employment inquiry, as such supervision is perfectly consistent with a typical, legitimate subcontracting arrangement. See Moreau v. Air France, 343 F.3d 1179, 1188 (9th Cir.2003) (supervision of workers not indicative of joint employment where principal merely gave “specific instructions to a service provider” concerning performance under a service contract); cf. James Brian Quinn and Frederick G. Hilmer, Strategic Outsourcing, Sloan Mgmt. Rev., Summer 1994, at 43, 53 (explaining that “[t]he most successful outsourcers find it absolutely essential to have both close personal contact and rapport at the floor level and political clout and understanding with the supplier’s top management”).
Finally, the Rutherford Court considered whether the purported joint employees worked exclusively or predominantly for the putative joint employer. In describing that factor, we use the words “exclusively or predominantly” on purpose. As noted in Lopez, the extent of work performed for a putative joint employer is “not described in any decision ... as a separate factor for consideration.” Id. at 417. However, it has “implicitly [been] a factor,” id., in cases in which the purported joint employees worked exclusively or predominantly for the purported joint employer. See Rutherford, 331 U.S. at 724-25, 67 S.Ct. 1473 (meat boners worked full-time on slaughterhouse’s premises); Ante-nor, 88 F.3d at 927 (harvesters worked solely on growers’ land even though they were hired by contractor). In those situations, the joint employer may de facto become responsible, among other things, for the amount workers are paid and for their schedules, which are traditional indicia of employment.12 On the other hand, where a subcontractor performs merely a majority of its work for a single customer, there is no sound basis on which to infer that the customer has assumed the prerogatives of an employer.
In sum, by looking beyond a defendant’s formal control over the physical performance of a plaintiffs work, the “economic reality” test — which has been distilled into a nonexclusive and overlapping set of fac*76tors — gives content to the broad “suffer or permit” language in the statute. See 29 U.S.C. § 203(g) (stating that an entity “employs” an individual for purposes of the FLSA if it' “suffer[s] or permit[s]” that individual to work). However, by limiting FLSA liability to cases in which defendants, based on the totality of the circumstances, function as employers of the plaintiffs rather than mere business partners of plaintiffs’ direct employer, the test also ensures that the statute is not interpreted to subsume typical outsourcing relationships. The “economic reality” test, therefore, is intended to expose outsourcing relationships that lack a substantial economic purpose, but it is manifestly not intended to bring normal, strategically-oriented contracting schemes within the ambit of the FLSA.
2. Application of the Factors on Remand
We intimate no view as to whether plaintiffs, under a proper application of the economic reality test derived from Rutherford, will have presented sufficient evidence to survive a renewed motion for summary judgment on remand. We note, however, that under our existing precedents, the inquiry as to whether an entity is an employer for purposes of the FLSA involves three types of determinations. First, there are historical findings of fact that underlie each of the relevant factors. Second, there are findings as to the existence and degree of each factor. Finally, there is the conclusion of law to be drawn from applying the factors, ie., whether an entity is a joint employer. See Superior Care, 840 F.2d at 1059 (citing, inter alia, Brock v. Mr. W Fireworks, Inc., 814 F.2d 1042, 1043-*5 (5th Cir.1987)).
The first two determinations — the findings of historical fact and the findings as to the existence and degree of each factor — are findings of fact that must be accepted on appeal unless clearly erroneous.13 See Superior Care, 840 F.2d at 1059; cf. Walling v. General Indus. Co., 330 U.S. 545, 550, 67 S.Ct. 883, 91 L.Ed. 1088 (1947) (whether each of the individual factors exist that render an employee an “executive” under the FLSA is a factual question subject to the “clearly erroneous” standard set forth in Fed.R.Civ.P. 52(a)). Only the last determination — the ultimate decision as to whether a party is an employer- — -is a legal conclusion that is reviewed de novo. See Superior Care, 840 F.2d at 1059; see also Icicle Seafoods, Inc. v. Worthington, 475 U.S. 709, 714, 106 S.Ct. 1527, 89 L.Ed.2d 739 (1986) (“The question whether [the plaintiffs’] particular activities exclude them from the overtime benefits of the FLSA is a question of law.”).
In order to grant summary judgment for defendants, the District Court would have to conclude that, even where both the historical facts and the relevant factors are interpreted in the light most favorable to plaintiffs, defendants are still entitled to judgment as a matter of law.14 To reach *77this conclusion, the Court need not decide that every factor weighs against joint employment. The Ninth Circuit, for example, recently concluded — in affirming a grant of summary judgment — that Air France did not jointly employ ground maintenance workers at the airline’s San Francisco terminal for the purposes of the Family and Medical Leave Act (“FMLA”),15 despite the fact that Air France representatives directly supervised many of the workers, and even trained some of them. See Mor-eau, 343 F.3d at 1189-90. The Court reasoned that, even if all reasonable inferences were drawn in the plaintiffs favor,16 there was insufficient evidence in the record on which to base the conclusion that the plaintiff was an employee of the airline. Id. at 1190-91.
Although summary judgment might also be granted to plaintiffs even when isolated factors point against imposing joint liability, see, e.g., Anterior, 88 F.3d at 937-38 (growers are employers as a matter of law even though middleman rather than growers exercised some employer prerogatives), the District Court’s conclusion that, in the present circumstances, the record cannot support summary judgment in plaintiffs’ favor, remains undisturbed. This case is quite different from Rutherford, in which the Supreme Court concluded that the slaughterhouse was a joint employer as a matter of law. In Rutherford, unlike in this case, every relevant factor described above weighed in favor of a joint employment relationship, and the record as a whole compelled the conclusion that the slaughterhouse exercised functional control over the boners. See Rutherford, 331 U.S. at 730, 67 S.Ct. 1473. Should the District Court, on remand, deny summary judgment in favor of defendants, it will be incumbent upon the Court to conduct a trial.17
*78II. State Law Claims
The District Court specifically addressed just two of plaintiffs’ five state law claims. First, the District Court dismissed plaintiffs’ overtime compensation claim under N.Y. Comp.Codes R. & Regs, tit. 12, § 142-2.2 — which is analogous to the overtime compensation claim brought under 29 U.S.C. § 207 — based on “the same analysis” it applied to the FLSA claims. Zheng, 2002 WL 398663, at *6 n. 3. Next, the Court refused to dismiss plaintiffs’ N.Y. Lab. Law § 345-a claim— which forbids a garment manufacturer from contracting out work to an entity knowing that the entity failed to comply with New York law governing the payment of wages — on the ground that a disputed issue of fact exists as to whether Liberty Defendants knew or should have known how much plaintiffs were being paid. However, the Court declined to exercise pendent jurisdiction over the § 345-a claim.
The District Court did not refer to plaintiffs’ claims under N.Y. Labor Law §§ 191, 193, and 652(1). Section 652(1) is analogous to 19 U.S.C. § 206, and requires an employer to pay the minimum wage. Section 191 requires employers to pay manual workers on a weekly basis, and section 193 forbids unauthorized deductions from employees’ wages. Like the § 142-2 claim and unlike the § 345-a claim, each of these claims is dependent on a finding of joint employment. Compare N.Y. Labor Law § 191 (“Every employer shall pay wages.”), N.Y. Labor Law § 193 (“No employer shall make any deduction .... ”), and § 652(1) (“Every employer shall pay .... ”), with § 345-a (imposing liability on any “manufacturer ... who contracts ... with ... [a] contractor ... who knew or should have known ... of [the] contractor’s failure to comply with article six or nineteen of this chapter .... ”). Thus, the District Court’s analysis of the § 142-2 claim applies to these three claims, and we assume the Court intended to dismiss these claims on the merits.
For the same reasons that we vacate so much of the District Court’s judgment as dismissed the FLSA claims, we also vacate so much of the District Court’s judgment as dismissed the analogous New York claims — that is, those claims brought under § 142-2 and § 652(1). The overtime compensation claim under § 142-2 is governed by the definitions in 12 NYCRR § 142-2.16, and the minimum wage claim under § 652(1) is governed by the definitions in New York Labor Law § 651. Like the FLSA, the former covers “any individual employed, suffered or permitted to work by an employer,” 12 NYCRR § 142-2.16(a), and the latter covers “any individual employed or permitted to work by an employer in any occupation,” N.Y. Lab. Law § 651(5).
Just as federal courts interpreting the “suffer or permit” language have looked beyond common-law agency principles in analyzing joint employment relationships, so too have New York courts. See, e.g., People ex rel. Price v. Sheffield Farms-Slawsort-Decker Co., 225 N.Y. 25, 121 N.E. 474, 476 (1918) (Cardozo, J.) (applying “sufferance” or “permission” standard to hold milk delivery business liable for delivery man’s practice of employing a minor even though the delivery man directly set the terms and conditions of the minor’s employment). The District Court therefore erred when it relied exclusively on the *79four Carter factors to dismiss both the § 142-2 claim and the § 652(1) claim, and these claims are reinstated.
Plaintiffs’ § 345-a claim, over which the District Court declined to exercise pendent jurisdiction, is likewise reinstated. It appears that the District Court did not exercise jurisdiction over the § 345-a claim principally because the FLSA claims were dismissed before trial. See Zheng, 2002 WL 398663, at *9 (apparently relying on 28 U.S.C. § 1367(c)(3), which permits a District Court not to exercise pendent jurisdiction when it “has dismissed all claims over which it had original jurisdiction”).18 Because the FLSA claims are now reinstated, 28 U.S.C. § 1367(c)(3) no longer provides a basis for declining to exercise jurisdiction over the § 345-a claim.
As a final matter, we reiterate that plaintiffs have not challenged the dismissal of their claims under N.Y. Labor Law §§ 191 and 193, which are governed by a narrower definition of employment applicable to Article 6 of New York’s labor statute. See N.Y. Lab. Law § 190 (defining “employer” and “employee” without using the “suffer or permit” language). Accordingly, we need not address those claims, the disposition of which remains undisturbed.
CONCLUSION
To summarize, we hold that the District Court erred when, based exclusively on the four factors identified in Carter v. Dutch-ess Community College, 735 F.2d 8 (2d Cir.1984), it concluded, as a matter of law, that Liberty and its principals were not joint employers within the meaning of the FLSA. Under the broad language of 29 U.S.C. § 203(g), as interpreted in Rutherford Food Corp. v. McComb, 331 U.S. 722, 67 S.Ct. 1473, 91 L.Ed. 1772 (1947), a district court must look beyond traditional agency principles before declaring that the entity is not an employer under the FLSA. Accordingly, although an entity’s exercise of an employer’s formal prerogatives — hiring and firing, supervising schedules, determining rate and method of payment, and maintaining records — may be sufficient to establish joint employment under the FLSA, it is not necessary to establish joint employment.
For the same reasons, we also hold that the District Court erred when it concluded, based exclusively on the four Carter factors, that Liberty and its principals were not joint employers within the meaning of New York’s statutory analogues to the FLSA.
The District Court’s judgment dismissing the FLSA claims, the New York statu*80tory analogues to those claims, and the N.Y. Labor Law § 345-a claim is therefore vacated, and the cause is remanded to the District Court for further proceedings consistent with this opinion and our instructions.
. Plaintiffs do not appeal the dismissal of the claims brought under N.Y. Lab. Law § 191 (payment on a weekly basis) and N.Y. Lab. Law § 193 (unauthorized deductions).
. Other courts of appeals have made the same observation. See Morgan v. MacDonald, 41 F.3d 1291, 1293 (9th Cir.1994) ("The Bon-nette factors are properly applied when an individual is clearly employed by one of several entities and the only question is which one. They are of no help, however, in deciding the more fundamental question present here: whether the inmates are 'employed' in the relevant sense at all.''); Vanskike v. Peters, 974 F.2d 806, 809 (7th Cir. 1992) ("The Bonnette factors, with their emphasis on control over the terms and structure of the employment relationship, are particularly appropriate where (as in Bonnette itself) it is clear that some entity is an 'employer' and the question is which one.”).
. Rutherford, is analyzed in detail at pp. 70-71, post.
. Judge Cote also considered factors that were specific to the factual record before her, namely the family ties between the manufacturer and the contractor and the manufacturers’ acceptance of substandard work from the contractor. Id. at 422-23.
. Senator Black introduced the bill that would become the Fair Labor Standards Act of 1938. See John S. Forsythe, Legislative History of the Fair Labor Standards Act, 6 Law & Contemp. Probs. 464, 466 (1939).
. At common law, factors other than those identified by the District Court can be considered in determining whether the requisite control exists to establish an employment relationship. See Restatement of Agency § 220(2) (explaining that, inter alia, the skill required in a particular occupation and the length of time that an individual is employed are relevant in determining if a worker is an employee or an independent contractor); cf. Restatement of Agency § 220, cmt. b ("The relationship of master and servant is one not capable of exact definition.”).
. There is no discussion in the Supreme Court’s opinion in Rutherford about whether the slaughterhouse kept employment records for the workers in the de-boning operation, but the Tenth Circuit’s earlier decision in the case indicates that it did not. See Walling v. Rutherford Food Corp., 156 F.2d 513, 515-16 (10th Cir.1946) (noting that the boners were not mentioned in any union contracts, that the slaughterhouse did not collect their union dues, and that "[t]he plant operator never included the boners in its Workmen’s Com*71pensation liability policy and never made any deductions for unemployment compensation or withholding taxes”).
. Although it dealt with the distinction between an independent contractor and an employee rather than with joint employment, Superior Care likewise counsels against a rigid approach. In Superior Care, we stated that ”[t]he factors that have been identified by various courts in applying the economic reality test are not exclusive.” Superior Care, 840 F.2d at 1059. Rather, "[s]ince the test concerns the totality of the circumstances, any relevant evidence may be considered, and mechanical application of the test is to be avoided.” Id.
. In the course of its opinion, the Rutherford Court noted also that the boners' work constituted “part of the integrated unit of production.” id. at 729. 67 S.Ct. 1473. and that "[t]he slaughterhouse operations, of which the boning is a part, are carried on in a series of interdependent steps,” id. at 725, 67 S.Ct. 1473.
. It bears noting that this factor does not simply measure the amount of initiative or judgment required for a job. We agree with the view of Judge Cote in Lopez that the amount of initiative or judgment required for a job, which we considered in Superior Care, 840 F.2d at 1058-61, is of "minimal value for deciding whether workers already acknowledged to be the employees of one employer should likewise be deemed the employees of the second employer,” Lopez, 14 F.Supp.2d at 415, as it tests principally for a worker's independence from all employers rather than for a worker's dependence on a particular employer, see id. Although there may be an overlap between jobs that require little initiative or judgment and line-jobs, there are undoubtedly jobs that require little initiative or judgment that are not line-jobs, and that are typically performed by subcontractors. Likewise, there are undoubtedly jobs that require substantial initiative and judgment that are rarely outsourced.
. The district court in Lopez appears to have concluded that this factor weighs in favor of joint employment when a general contractor uses numerous subcontractors who compete for work and have different employees. See Lopez, 14 F.Supp.2d at 422. We reject this view. Indeed, if this view were adopted, factor (4) would classify nearly all subcontracting relationships as joint employment relationships — a result that finds no support either in the law or in our country's commercial practices.
. Factor (2) listed above — i.e., whether the Contractor Corporations had a business that could or did shift as a unit from one putative joint employer to another — overlaps substantially with the factor discussed here — i.e., whether all or nearly all of the Contractor Corporations’ work was performed for the Liberty Defendants. The factors are not identical, however, and capture different aspects of a business relationship's "economic reality." For example, factor (2), but not factor (6), would weigh in favor of joint employment if a subcontractor were to work exclusively tor two or three general contractors on those contractors’ premises without the resources to work for any other contractor. By contrast, factor (6), but not factor (2), would weigh in favor of joint employment if a subcontractor worked solely for a single client but had the ability to seek out other clients at any time. Together, these two factors help the factfinder determine if a subcontractor's apparent dependence on particular contractors translates into functional control by those contractors over the subcontractor's employees.
. The fact-intensive character of the joint employment inquiry is highlighted by the fact .that two of the three leading cases in this circuit were appeals from judgments following bench trials. See Herman, 172 F.3d at 135; Superior Care, 840, F.2d at 1057. In the third case, we decided that genuine issues of material fact precluded summary judgment on the ultimate issue of FLSA coverage. See Carter, 735 F.2d at 15.
. The Supreme Court has specifically cautioned courts of appeals not to engage in independent fact-finding after concluding that a district court has applied the wrong factors in classifying an individual or an entity under the FLSA. See Icicle, 475 U.S. at 714, 106 S.Ct. 1527 (concluding that the Ninth Circuit erred when it made factual findings to support its determination that individuals were not "seamen” within the meaning of the *77FLSA). Here, to determine that defendants are entitled to judgment as a matter of law even if we interpret the facts in the light most favorable to plaintiffs, we would not need to engage in formal factfinding. Nevertheless, Icicle remains relevant, because in Icicle, the Supreme Court vacated the Ninth Circuit's judgment even though the Ninth Circuit's factual findings appeared to be undisputed. See Icicle, 475 U.S. at 716-17, 106 S.Ct. 1527 (Stevens, J., dissenting). Thus, the Court suggested that, with respect to the fact-intensive classifications required by the FLSA, courts of appeals should avoid even straightforward factual determinations.
Because the process of isolating and then applying the relevant factors, even at the summary judgment stage, is closely akin to the District Court’s fact-finding function under Fed.R.Civ.P. 52(a), we do not think it is advisable to apply the Rutherford factors in the first instance on appeal.
.The definition of "employ” in the FMLA is the same as the definition of "employ” in the FLSA. See 29 U.S.C. § 2611(3) ("The terms 'employ', employee', and 'State' have the same meanings given such terms in subsections (c), (e), and (g) of section 203 of this title.”). Thus, in determining whether a joint employment relationship existed under the FMLA, the Moreau Court borrowed directly from the FLSA's joint employment case law. See Moreau, 343 F.3d at 1183-85 (relying primarily on the Ninth Circuit’s own decisions in Bonnette and Torres-Lopez).
. Notably, although the Ninth Circuit doubted whether "many of the functions, such as food service or cargo transport, are actually 'integral' to a passenger airline,” and also whether "this factor translates well outside of the production line employment situation,” id. at *27, the Court assumed, for the purposes of Air France’s summary judgment motion, that the various services are "integral” to the airline, id. Nevertheless, the Court concluded that "this factor (even if coupled with Air France’s ‘indirect’ supervision of employees) does not outweigh the numerous significant factors discussed above.” Id.
. To the extent it conflicts with this analysis, we disagree with the District Court’s decision in Lopez to grant summary judgment in plaintiffs’ favor. However, we note that Lopez was different from this case in several ways: First, the "historical facts” in Lopez were largely undisputed, see Lopez 14 F.Supp.2d at 407-*7808; second, the Court characterized the relationship between the manufacturer and one of the contractors as "becoming so extensive and regular as to approach one of exclusive agency,” id. at 421; and third, the Court considered several factors that were "wholly specific to [its] record,” id. at 422.
. The District Court’s exact rationale for declining jurisdiction over the § 345-a claim is somewhat unclear. After noting that a district court may retain jurisdiction over a state law claim even after all federal claims have been dismissed, the Court quoted United Mine Workers of America v. Gibbs, 383 U.S. 715, 726, 86 S.Ct. 1130, 16 L.Ed.2d 218 (1966), for the proposition that "if the federal claims are dismissed before trial ... the state claims should be dismissed as well.” Instead of resting on Gibbs alone, however, the Court went on to note that "it appears that, as of yet, no state court has interpreted section 345-a of New York Labor Law,” Zheng, 2002 WL 398663, at *9, thus invoking a separate ground for declining jurisdiction, see 28 U.S.C. § 1367(c)(1) ("The district courts may decline to exercise supplemental jurisdiction over a claim ... if the claim raises a novel or complex issue of State law[.]”). Based on its quotation from Gibbs, we interpret the District Court to have declined to exercise jurisdiction over the § 345-a claim on the ground that the FLSA claims were dismissed. We intimate no view as to whether, on remand, the District Court should exercise any discretion it may have to dismiss the § 345-a claim on the separate and independent ground that it presents a novel or complex issue of state law.