Ragnar E. Danneskjold, an inmate at Attica state prison, appeals from Judge Larimer’s grant of summary judgment for appel-lees, the Commissioner and the Deputy Commissioner of the New York State Department of Correctional Services and the Director of the Consortium of Niagara Frontier. Judge Larimer held that appellant’s claim for unpaid minimum wages under the Fair Labor Standards Act (“FLSA” or “the Act”), 29 U.S.C. §§ 201-219, failed as a matter of law because Danneskjold was not an “employee” within the meaning of the statute. In reaching that conclusion, he applied the “economic reality” test set out in our decision in Carter v. Dutchess Community College, 735 F.2d 8 (2d Cir.1984). Judge Larimer also dismissed Danneskjold’s claim under 42 U.S.C. § 1983 because that claim depended, in part, on the success of the FLSA claim.
We modify the test established in Carter.1 We hold that the FLSA does not apply to prison. inmates in circumstances in which their labor provides services to the prison, whether or not the work is voluntary, whether it is performed inside or outside the prison, and whether or not a private contractor is involved. Because Danneskjold worked in an education program that provided rehabilitative services only to inmates, he was not an employee for purposes of the FLSA. We therefore affirm.
BACKGROUND
On appeal from a grant of summary judgment, we view the facts in the light most favorable to the nonmoving party, in this ease Danneskjold. See Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255, 106 S.Ct. 2505, 2513-14, 91 L.Ed.2d 202 (1986); Henry v. Daytop Village, Inc., 42 F.3d 89, 92 (2d Cir.1994).
Since 1975, the Consortium of the Niagara Frontier (the “Consortium”), an association of Canisius College, Daemen College, and Niagara University, has offered inmates at Attica Correctional Facility an opportunity to earn college degrees by taking courses while serving their sentences. In addition to a professional teaching staff, the Consortium uses trained student inmates to assist in the administration of the program and to tutor other student inmates. In March 1987, Danneskjold responded to a notice posted by the Consortium seeking a clerk-tutor. After reviewing Danneskjold’s application and interviewing him, Consortium personnel requested permission from the Department of Corrections inmate program committee to hire Danneskjold. The program committee approved the Consortium’s request.
From March 16, 1987 until October 10, 1988, Danneskjold worked for the Consor*40tium as a clerk-tutor. His duties included assisting and tutoring student inmates, assisting professors with academic and curricular matters, and correcting papers. For this work, the Department paid him between $.95 and $1.45 per day in accordance with the Department’s inmate wage system.
On October 12,1988, Danneskjold filed this suit, pro se, alleging that as a clerk-tutor he had been an “employee” of the Consortium and, as such, was entitled to receive the minimum wage for any and all hours worked. Based on our decision in Carter, the district court granted appellees’ motion for summary judgment on the ground that Danneskjold and the Consortium were not in an employment relationship.
DISCUSSION
On appeal from a grant of summary judgment, we review the ruling de novo. Litton Indus., Inc. v. Lehman Bros. Kuhn Loeb Inc., 967 F.2d 742, 746 (2d Cir.1992). Summary judgment is proper only if, viewing all evidence in the light most favorable to the nonmoving party, there is no genuine issue of material fact, and the movant is entitled to judgment as matter of law. Id.
The minimum wage provisions of the FLSA apply only to workers who are “employees” within the meaning of the Act. 29 U.S.C. § 206(a)(1). The Act defines “employee” as “any individual employed by an employer,” id. § 203(e)(1), with certain exceptions not relevant to this case. An “employer” is one who acts “directly or indirectly in the interest of an employer in relation to an employee.” Id. § 203(d). The Supreme Court has held that these terms are to be applied in light of the “economic reality” of the relationship between the parties. Goldberg v. Whitaker House Coop., Inc., 366 U.S. 28, 33, 81 S.Ct. 933, 936-37, 6 L.Ed.2d 100 (1961). Interpreting this standard, the Ninth Circuit, in non-prisoner cases, has fashioned a four-factor test to determine the nature of that relationship. The four factors ask whether the alleged employer: (i) had the power to hire and fire; (ii) supervised workers and controlled the conditions of employment; (in) determined the rate and method of payment; and (iv) maintained employment records. Bonnette v. California Health & Welfare Agency, 704 F.2d 1465, 1470 (9th Cir.1983).
In Carter, 735 F.2d at 10, we addressed facts similar to those in the instant matter. The prisoner-plaintiff in Carter sought a ruling that his service as a tutor in an educational program for inmates conducted by Dutchess County Community College rendered him an employee of the College for FLSA purposes. The district court held that because the prison maintained “ultimate control” over the inmates, they were not employees within the meaning of the FLSA. Id. at 11.
In reversing the district court, we rejected a per se rule that prisoners may never be considered employees for purposes of the FLSA. Id. at 13. We based that conclusion on three grounds. First, we reasoned that exempting an entire class of workers from the Act’s coverage on account of their status could, in some instances, undermine one of the primary purposes of the FLSA: preventing unfair competition among employers if some were able to pay less than the minimum wage. Id. at 13; see also 29 U.S.C. § 202(a). Second, we noted that the statute does not include prison inmates in its specific exemptions from coverage. Carter, 735 F.2d at 13. Third, we reasoned that existing case-law required a “particularized inquiry into the facts of each case” and that a blanket exclusion of prisoners was inappropriate. Id. Accordingly, we reversed the district court and held that it should apply the four-factor inquiry articulated by the Ninth Circuit in Bonnette as the applicable economic reality test. Id. at 13-14. We implied, but did not hold, that the facts alleged by Carter might be sufficient to warrant FLSA coverage. Carter was remanded for further proceedings but appears never to have been decided on the merits.
Although we do not disturb Carter’s rejection of a rule that a prisoner’s labor is at all times and in all circumstances exempt from the FLSA, or discard use of an economic reality test to determine whether such labor is subject to the FLSA, we do reexam*41ine and reject its use of the four-part Bon-nette test to make that determination. We do so for three reasons.
First, Bonnette did not involve prison labor and is not well suited to determining the status of that labor under the FLSA. It involved the question of whether the state was the employer of persons, styled “chore workers,” who provided state-mandated in-home services to the aged, blind, and disabled. Bonnette held that the state was a joint employer of the chore workers, along with the recipients of their services. In the prison context, however, application of Bon-nette leads to a radical result. Literally applied, the Bonnette factors would render all prison labor, including involuntary labor inside the penal institution, such as in a prison laundry, subject to minimum wage laws. No court has ever suggested, much less held, that the FLSA applies to such labor.
Second, because of a quirk of legal history, Carter was decided at a time (1984) when application of the FLSA to common state prison labor was barred by Supreme Court caselaw. In 1974,2 Congress extended the FLSA to cover state and municipal employees. However, in National League of Cities v. Usery, 426 U.S. 833, 852, 96 S.Ct. 2465, 2474, 49 L.Ed.2d 245 (1976), the Supreme Court erected a constitutional bar to the application of the federal minimum wage to state employees. See Wentworth v. Solem, 548 F.2d 773, 775 (8th Cir.1977) (per curiam); see also Garcia v. San Antonio Metro. Transit Auth., 469 U.S. 528, 105 S.Ct. 1005, 83 L.Ed.2d 1016 (1985) (overruling National League of Cities). It was in this legal context that Carter arose. Presumably because of National League of Cities, the FLSA portion of Carter involved only the question whether Dutchess Community College was Carter’s employer. (A constitutional claim appears to have been asserted against the state defendants.) Carter’s adoption of the Bonnette test was thus in the context of a Supreme Court decision barring application of the FLSA to state employees, and consequently to inmates of state prisons employed by the prison. Indeed, a reading of Carter leaves the firm impression that the panel in no way believed that the FLSA applied to ordinary prison labor.
Third, caselaw after Carter has been quite unkind to the use of the four-factor Bonnette test in determining the status of prison labor under the FLSA. In 1985, the Supreme Court overruled National League of Cities in Garcia v. San Antonio Metro. Transit Auth., 469 U.S. 528, 546-47, 105 S.Ct. 1005, 1015-16, 83 L.Ed.2d 1016 (1985), opening the possibility that prison authorities might be deemed FLSA employers if Bonnette were literally applied. Although the first post-Gama Court of Appeals decision addressing the relationship between prison labor and the FLSA held that the labor in question was covered by the Act, Watson v. Graves, 909 F.2d 1549 (5th Cir.1990), it applied an economic reality test quite different from that required by Bonnette. In Watson, inmates in a sheriffs custody were assigned to work for a construction company owned by the sheriffs daughter and son-in-law. The construction company had no employees other than the owners and used only prison labor and subcontractors. The inmates were paid twenty dollars a day without regard to the nature or amount of work performed.
Applying the four factors articulated in Bonnette, the district court held that, although the inmate-laborers were supervised exclusively by the owners of the construction company without interference from jail personnel, the sheriff made the decision as to whether or not an inmate was entitled to participate in the work release program and determined the rate of pay. Neither the sheriff nor the construction company maintained any employment records on the inmates. In light of these facts, the district court held that the inmates were not employees under the Act and granted summary judgment for the defendants.
*42The Fifth Circuit reversed, holding that the fact that the sheriff maintained the power to “hire and fore” and set the daily wage was not dispositive. Examining the economic reality of the relationship in light of the policies underlying the FLSA, the Court held that the workers were being used for the economic advantage of the company. Id. at 1555. Furthermore, other construction contractors in the area could not compete on a level playing ground with a company facing minimal labor and overhead costs. Given these economic realities, the Fifth Circuit held that the inmates were “employees” of the company and were therefore entitled to the federal minimum wage. Id. at 1554-56.
Although Watson cited Carter and set out the four-part Bonnette test, the decision actually applied an economic reality test at a higher level of generality than Bonnette. Instead of concentrating on details concerning power to hire, fire, supervise, and set wages (which in the case of prisoners always rests ultimately with prison officials) and the location of employment records (an almost irrelevant factor in this context) the court looked to the underlying policies of the FLSA. It noted that the construction company got the benefit of labor in the private economy without paying FLSA wages. Id. at 1555. The labor provided services on the cheap that afforded the construction company a competitive advantage over similar employers who had to pay FLSA wages. Such competition, of course, tended to undermine compliance with the FLSA. The court was at pains, moreover, to note that labor was not part of the sentences of the prisoners in question. Id. at 1556.
Decisions subsequent to Watson have universally denied FLSA wages to prisoners, although the factual contexts of those decisions differ radically from that in Watson. The subsequent cases also have rather uniformly declined to use the four-part Bonnette test. For example, Vanskike v. Peters, 974 F.2d 806, 809-10 (7th Cir.1992), cert. denied, 507 U.S. 928, 113 S.Ct. 1303, 122 L.Ed.2d 692 (1993), declined to apply Bonnette to forced prison labor providing services within a prison. It noted that in “economic reality,” the prisoner was not an employee in any conventional sense, although under Bonnette there was virtually an open and shut case for application of the FLSA to the prison as an employer. Id.
We agree with, and adopt, the reasoning of Vanskike that forced prison labor for the prison is not subject to the FLSA. The relationship is not one of employment; prisoners are taken out of the national economy; prison work is often designed to train and rehabilitate; prisoners’ living standards are determined by what the prison provides; and most such labor does not compete with private employers. Id. at 810-11. Moreover, for reasons stated in Vanskike, Congress most certainly did not intend the FLSA to apply to forced prison labor. Id. at 810-12. Indeed, the continued existence of the Ashurst-Summers Act, 18 U.S.C. §§ 1761-62, which regulates the interstate transportation of prison-made goods to avoid competition between low-cost prison labor and free labor, reveals a congressional assumption that prison labor will not be paid at FLSA minimum wage levels.
Other circuits have similarly declined to apply the Bonnette four-point test in the context of prison labor. The Fourth Circuit has held that before a court may even consider whether the alleged employer exercises typical employer prerogatives, it must determine that the work performed by the inmate takes place outside the prison walls. Harker v. State Use Industries, 990 F.2d 131, 133 (4th Cir.) (inmate who worked at non-profit prison print shop not covered by FLSA), cert. denied, — U.S. —, 114 S.Ct. 238, 126 L.Ed.2d 192 (1993). The Tenth Circuit has rejected use of the economic reality test in its entirety with regard to work performed within a prison. Franks v. Oklahoma State Indus., 7 F.3d 971, 972 (10th Cir.1993). The D.C. Circuit has rejected the “inside/outside” distinction but has also held that unless the inmate has voluntarily contracted with a private employer, the prisoner is, as a matter of law, barred from coverage by the FLSA regardless of the four factors articulated in Bonnette. Henthorn v. Department of Navy, 29 F.3d 682, 686-87 (D.C.Cir.1994) (prisoner assigned by Bureau of Prisons to work on the grounds of the Naval Air Station not *43covered by FLSA). The Eighth Circuit has held that where forced prison labor is involved, Carter, and thus Bonnette, do not apply. See McMaster v. Minnesota, 30 F.3d 976, 980 (8th Cir.1994) (inmates assigned to work in internal prison industries not covered by FLSA), cert. denied, — U.S. —, 115 S.Ct. 1116, 130 L.Ed.2d 1080 (1995). The Fifth Circuit, which decided Watson, has since rejected use of the economic reality test to determine whether a sheriff custodian is an FLSA employer. Reimonenq v. Foti, 72 F.3d 472, 475 fn. 3 (5th Cir.1996). Finally, the court that decided Bonnette, the Ninth Circuit, has held en banc that its four factor test does not apply to work in prison programs. Hale v. Arizona, 993 F.2d 1387, 1395 (9th Cir.) (en banc) (prisoners working in prison programs pursuant to Arizona law requiring hard labor are not employees under the FLSA), cert. denied, — U.S. —, 114 S.Ct. 386, 126 L.Ed.2d 335 (1993).
We believe that the easelaw described above has essentially read Bonnette, but not necessarily the economic reality test, out of the determination of whether a particular prisoner’s labor is subject to the FLSA. We agree with those courts.
In our view, the Bonnette test is useful largely in eases involving claims of joint employment and has little relevance to the unique status of a prisoner and his or her relationship to the correctional institution. As noted, prisoners are not generally participants in the national economy. See Vanskike, 974 F.2d at 810. They have limited rights, and their conduct is closely regulated. As a result, no Court of Appeals has ever questioned the power of a correctional institution to compel inmates to perform services for the institution without paying the minimum wage. Prisoners may thus be ordered to cook, staff the library, perform janitorial services, work in the laundry, or carry our numerous other tasks that serve various institutional missions of the prison, such as recreation, care and maintenance of the facility, or rehabilitation. Such work occupies prisoners’ time that might otherwise be filled by mischief; it trains prisoners in the discipline and skills of work; and it is a method of seeing that prisoners bear a cost of their incarceration.
We also believe that so long as the labor produces goods or services for the use of the prison, voluntary labor by the prisoner — such as is involved in the instant matter — is also not subject to the FLSA. Voluntary work serves all of the penal functions of forced labor discussed above and, therefore, should not have a different legal status under the FLSA. Moreover, the performance of some services for the prison may be enhanced by making the decision to work voluntary. Where the work is such that the quality of performance is difficult to measure and may depend on motivation, volunteers may be superior to involuntary workers. In the instant case, for example, tutoring of other inmates by prisoners who volunteer may be superior to tutoring by prisoners ordered to do so. In any event, the voluntary performance of labor that serves institutional needs of the prison is not in economic reality an employment relationship. The prisoner is still a prisoner; the labor does not undermine FLSA wage structures; the opportunity is open only to prisoners; and the prison could order the labor if it chose. Indeed, to hold otherwise would lead to a perverse incentive on the part of prison officials to order the performance of labor instead of giving some choice to inmates.
Where a prisoner’s work produces goods or services for the use of the prison, we also do not believe that the use of a private contractor as an intermediary subjects either the contractor or the institution to the FLSA unless the work undermines the FLSA in a particular labor market. For example, if a prison operated a laundry that served only inmates, clearly it might use prisoners to operate the laundry without paying FLSA wages. We perceive no distinction of legal consequence between those circumstances and the provision of similar services to the prison by a private contractor using prison labor. „ Some tasks may be more efficiently handled by private contractors, but the legal status of prison labor under the FLSA should not be altered by the fact that the boss works under a contract with the corrections authorities instead of as a prison *44employee. For the reasons stated above, we also believe that whether the labor is performed inside or outside the physical walls of the institution is irrelevant. If penal needs are better served by off-site labor, that fact does not alter the relationship for FLSA purposes.
We recognize, however, that our holding is limited to prison labor that produces goods or services for the use of the prison. We do not address the questions that arise when the prison labor is employed to produce goods or services that are sold in commerce. Compare Hale v. Arizona, 993 F.2d 1387 (9th Cir.1993). As Carter correctly noted, a principal reason for not exempting prison labor entirely from the FLSA is the possibility of “unfair competition” between prison and private labor. 735 F.2d at 13. This is not a wholly unrealistic concern. In Watson, for example, prisoners worked at sub-FLSA wages for a company that was not providing services to the prison and that competed with companies required to pay wages set by the FLSA.
In sum, we continue to follow Carter in holding that prison labor is not in all circumstances exempt from the FLSA and that an economic reality test is to be used in determining whether payment of FLSA wages is required. We discard the four-factor Bonnette test as a surrogate for an economic reality test. We hold that prison labor that produces goods or services for institutional needs of the prison, whether voluntary or involuntary, inside or outside the institution, or in connection with a private employer, is not an employment relationship within the meaning of the FLSA. Where a prisoner’s work for a private employer in the local or national economy would tend to undermine the FLSA wage scale, as in Watson, the FLSA applies. Intermediate cases will be resolved as they arise.3
Danneskjold’s work as a tutor for the consortium served only the institutional purpose of prisoner rehabilitation and in no way undermined FLSA wage scales. We therefore affirm.
. We have circulated this opinion in draft to the judges of this court, and no active judge has sought an en banc hearing of this appeal. See Kramer v. Time Warner, 937 F.2d 767, 774 (2d Cir.1991).
. The relationship between prison labor and the FLSA was subject to only sporadic litigation pri- or to 1974. The only reported decisions in this area involved prisoners performing labor for private employers. Those decisions uniformly held that the work was not subject to the FLSA. See Hudgins v. Hart, 323 F.Supp. 898 (E.D.La.1971); Sims v. Parke Davis & Co., 334 F.Supp. 774 (E.D.Mich.), aff'd, 453 F.2d 1259 (6th Cir.1971), cert. denied, 405 U.S. 978, 92 S.Ct. 1196, 31 L.Ed.2d 254 (1972); Huntley v. Gunn Furniture Co., 79 F.Supp. 110 (W.D.Mich.1948).
. In Reimonenq v. Foti, 72 F.3d 472 (5th Cir.1996), the Fifth Circuit held that the FLSA does not prohibit a sheriff custodian from compelling a prisoner working for a private employer in a work-release program to allow a deduction from his wages to contribute to an Elderly/Victim Compensation Fund. However, although the deduction caused the prisoner to receive less than FLSA wages, the private employer was paying the required minimum wage.