JERRY E. SMITH, Circuit Judge:

On behalf of itself and the other interest-holders in 523 West Texas oil wells, Parker Parsley Petroleum Company ("Parker Parsley") filed suit in federal district court against Dresser Industries, Inc., Titan Services, Inc., BJ Services U.S.A., Inc., BJ-Hughes Holding Company, Baker Hughes Production Tools, Inc., and Baker Hughes Incorporated (hereinafter collectively "Dresser"), charging that Dresser defrauded Parker Parsley by shorting it on materials used in oil well stimulation procedures. Parker Parsley based federal jurisdiction upon violations of the Racketeer Influenced and Corrupt Organizations Act ("RICO"), 18 U.S.C. § 1961 et seq., and appended Texas state claims for fraud, breach of contract, breach of implied warranty, negligence, and gross negligence.

The district court dismissed the RICO claims but retained pendent jurisdiction over the state claims. After a jury trial on the state claims, the district court entered judgment awarding $85 million actual and $100 million punitive damages. After a separate proceeding, the court awarded the plaintiffs attorneys' fees of approximately $1.8 million. We vacate the judgment and dismiss for lack of federal jurisdiction.

I.

Parker Parsley operated a large number of oil wells in West Texas. Some of the wells were not as productive as the company wished, so it contracted with Dresser in 1983 and 1984 to "fracture" the wells to stimulate them. Apparently through the efforts of Dresser's Odessa division manager, Gary "Zeke" Lancaster, Dresser shorted Parker Parsley, using less sand and gel than it had agreed to use for the fracturing, which, Parker asserted, reduced the amount of oil that eventually could be extracted.

In 1985, Dresser's Titan subdivision entered into a partnership with a BJ-Hughes Holding Co. subsidiary and remained in the business as BJ-Titan. In 1986 and 1987, Parker Parsley awarded its fracturing contracts to BJ-Titan, and the shorting apparently continued. In 1987, Baker Hughes Incorporated acquired BJ Holding Co. and later became the corporate parent of all the BJ-Titan partners. The company fired Lancaster for embezzling, and it seems that his attorney informed Dresser of the shorting, which he said had been approved by high executives of his former employers.

II.

The RICO claim was dismissed about nine months after the suit was filed and a month before trial was scheduled to begin. The district court retained jurisdiction over the state law fraud, contract, and tort claims, but then continued the case for three months. Dresser appeals the court's retention of pendent jurisdiction and challenges the award of punitive damages, the measure of actual damages, and the exclusion of evidence relating to a witness's alleged bias and, in a separate appeal now consolidated, attorneys' fees.

III.

Parker Parsley grounded its RICO claims on 18 U.S.C. § 1962(a) and (c). The district court held that Parker Parsley had failed to allege a proper RICO enterprise or a cognizable RICO injury, that the BJ-Titan partners were not "persons" for purposes of the statute, and that, because Parker Parsley's substantive claims had failed, its conspiracy claims should be dismissed as well. Parker Parsley crossappeals, arguing that its RICO claim should have survived the dismissal motion. We affirm the dismissal.

As stated in Sedima, S.P.R.L. v. Imrex Co., 473 U.S. 479, 496, 105 S.Ct. 3275, 3285, 87 L.Ed.2d 346 (1985), a viable claim under section 1962(c) "requires (1) conduct (2) of an enterprise (3) through a pattern (4) of racketeering activity." Parker Parsley averred three potential enterprises. First, it alleged an association-in-fact composed of the "servicing entity's" field employees who carried out the short-changing. Alternatively, it pleaded that each respective corporate defendant, as the servicing entity, was the enterprise. Third, it alleged that the BJ-Titan partnership, as the servicing entity, was the enterprise. The district court held first that the only bases for the association-in-fact were the employees' relationship with the defendant companies and the alleged wrongful conduct. The court noted that such an association must be "an entity separate and apart from the pattern of activity in which it engages," see Atkinson v. Anadarko Bank Trust Co., 808 F.2d 438, 441 (5th Cir.) ( quoting United States v. Turkette, 452 U.S. 576, 583, 101 S.Ct. 2524, 2529, 69 L.Ed.2d 246 (1981)), cert. denied, 483 U.S. 1032, 107 S.Ct. 3276, 97 L.Ed.2d 780 (1987), and that the acts of the members of the alleged association took place within the course of their conduct as employees, which basis this court disallowed in Elliot v. Foufas, 867 F.2d 877, 881 (5th Cir. 1989). The district court rejected the other possible enterprises because the alleged acts were "committed" by the "enterprise" in the course of its regular business and because the RICO "persons" that were alternatively alleged were not claimed to have committed the predicate acts.

We agree that Parker Parsley alleged no RICO enterprise under section 1962(c). The initial averred association-in-fact, consisting of the shortchanging field employees, either has no existence as an entity separate and apart from the actual pattern of racketeering, see, e.g., Old Time Enters. v. International Coffee Corp., 862 F.2d 1213, 1217 (5th Cir. 1989); Delta Truck Tractor v. J.I. Case Co., 855 F.2d 241, 243 (5th Cir. 1988), cert. denied, 489 U.S. 1079, 109 S.Ct. 1531, 103 L.Ed.2d 836 (1989), or is the defendant corporate entity functioning through its employees in the course of their employment. See Old Time Enters., 862 F.2d at 1217; see also Atkinson, 808 F.2d at 441. Because neither of these can constitute a RICO enterprise, see Elliot, 867 F.2d at 881, and because a corporation cannot be both the enterprise and the RICO perpetrator, Bishop v. Corbitt Marine Ways, 802 F.2d 122, 123 (5th Cir. 1986), this association cannot be a RICO enterprise.

The alternative RICO enterprises also fail. The corporate partners in the servicing entity, or alternatively, BJ-Titan, committed the predicate acts, if such acts may be attributed to them, in the course of their regular business. Additionally, as the district court noted, if the corporations or partnership are to be held liable as RICO "persons," they must have committed the predicate acts, but Parker Parsley, despite the claim in its brief, has not alleged that the partners did so. See United States v. Cauble, 706 F.2d 1322, 1332-33 (5th Cir. 1983), cert. denied, 465 U.S. 1005, 104 S.Ct. 996, 79 L.Ed.2d 229 (1984).

The acts of the servicing entity, or the partnership, cannot, for RICO purposes, be attributed vicariously to the individual partners. See Schofield v. First Commodity Corp., 793 F.2d 28, 32 (1st Cir. 1986). Having determined that the claims were properly dismissed for failure to state a RICO enterprise, we need not address Parker Parsley's other arguments regarding the section 1962(c) claims.

Heretofore we have not explicitly applied the foregoing analysis to a section 1962(a) claim, but we need not do so now in order to affirm, for the district court also dismissed Parker Parsley's section 1962(a) claims for failure to allege a RICO injury. We see no reason to disturb this ruling. Section 1964(c) states, "Any person injured in his business or property by reason of a violation of section 1962 of this chapter may sue therefor ... and shall recover threefold the damages he sustains." Section 1962(a) provides,

It shall be unlawful for any person who has received any income derived, directly or indirectly, from a pattern of racketeering activity or through collection of an unlawful debt . . . to use or invest, directly or indirectly, any part of such income, or the proceeds of such income, in acquisition of any interest in, or the establishment or operation of, any enterprise which is engaged in, or the activities of which affect, interstate or foreign commerce.

As the district court noted, it is obvious from the complaint and the RICO case statement that the only damages Parker Parsley is attempting to recover are those caused by inadequate fracturing jobs, not from any investment of income derived from the alleged shorting. As all but one of the circuits that have considered the issue have held, the causal language of section 1964(c) requires that the compensable injury stem from the violation of the RICO section in question, so any injury under section 1962(a) must flow from the use or investment of racketeering income. Parker Parsley's injury does not stem from the investment of the income from racketeering activity; therefore, it has pleaded no cause of action under section 1962(a), and the district court properly dismissed the RICO claims.

IV. A.

The district court refused to surrender jurisdiction over the pendent state claims, noting that the suit had been filed more than nine months earlier, since which time it had survived a "serious attack upon the propriety of venue," "rigorous deposition schedules," "ungodly amounts of discovery documents," and a hearing on discovery disputes. The court stated that dismissal would be a tremendous financial drain to all the parties as well as a waste of judicial resources; it thus concluded that "the equities weigh heavily in favor of maintenance of the case." Dresser argues that the court erred in exercising its pendent jurisdiction.

B.

We review the decision to retain jurisdiction over the pendent state claims for abuse of discretion. Rosado v. Wyman, 397 U.S. 397, 401, 90 S.Ct. 1207, 1212, 25 L.Ed.2d 442 (1970); La Porte Constr. Co. v. Bayshore Nat'l Bank, 805 F.2d 1254, 1257 (5th Cir. 1986). The Supreme Court explained the extent of pendent jurisdiction in United Mine Workers v. Gibbs, 383 U.S. 715, 726, 86 S.Ct. 1130, 1139, 16 L.Ed.2d 218 (1966), noting that the justification for pendent jurisdiction

lies in considerations of judicial economy, convenience and fairness to litigants; if these are not present a federal court should hesitate to exercise jurisdiction over state claims, even though bound to apply state law to them. Needless decisions of state law should be avoided both as a matter of comity and to promote justice between the parties, by procuring for them a surer-footed reading of applicable law. Certainly, if the federal claims are dismissed before trial, even though not insubstantial in a jurisdictional sense, the state claims should be dismissed as well. [Footnotes and citations omitted.]

The Court has not treated Gibbs as establishing a bright-line rule for pendent jurisdiction but has called for a more flexible analysis, balancing the values of economy, convenience, fairness, federalism, and comity. See, e.g., Carnegie-Mellon Univ. v. Cohill, 484 U.S. 343, 350 n. 7, 108 S.Ct. 614, 619 n. 7, 98 L.Ed.2d 720 (1988) (citing Rosado, 397 U.S. at 403-05, 90 S.Ct. at 1213-14). The Carnegie-Mellon Court did state, though, that when the single federal-law claim is eliminated at an "early stage" of the litigation, the district court has "a powerful reason to choose not to continue to exercise jurisdiction." Id. 484 U.S. at 351, 108 S.Ct. at 619. Our general rule is to dismiss state claims when the federal claims to which they are pendent are dismissed. Wong v. Stripling, 881 F.2d 200, 204 (5th Cir. 1989).

C.

Dresser argues that the federal claim was dismissed at a "preliminary stage" in the proceedings and that the district court failed to articulate specific considerations of judicial economy, convenience, and fairness that would support pendent jurisdiction, perhaps because there were none. Thus, Dresser contends that when the RICO claims were dismissed, there had been no "substantial commitment of judicial resources to the nonfederal claims," W.R. Grace Co. v. Continental Casualty Co., 896 F.2d 865, 872 (5th Cir. 1990), as the court had conducted only one hearing, other proceedings having been held before a magistrate judge.

Parker Parsley responds that the court did not abuse its discretion in maintaining jurisdiction over the state-law claims. It stresses our deference to the discretion of the district court, see id. at 870, and argues that weighing the equities and factors set out in Carnegie-Mellon should lead us to affirm.

D.

The instant case is distinguishable from the cases each party finds dispositive. Parker Parsley argues that we should follow Hudak v. Economic Research Analysts, 499 F.2d 996, 1001 (5th Cir. 1974), cert. denied, 419 U.S. 1122, 95 S.Ct. 805, 42 L.Ed.2d 821 (1975), in which the district court tried the state claims with the federal claims in a seemingly appropriate use of pendent jurisdiction. When we found that the limitation period had run on the federal claim, so that the claim should have been dismissed at the beginning of the litigation, we held that to dismiss the state claim would not serve the interests of judicial economy, convenience, and fairness. Id.

Hudak is not controlling here, however. In the instant case the court knew that it had no federal question before it before trial began, while in Hudak the court conducted a trial as though the federal claim were before it. It makes no sense, for purposes of judicial economy, convenience, or comity, to "punish" a district court for abusing discretion it does not know it has; given the apparent vitality of the federal claim, the Hudak court properly retained jurisdiction.

Dresser asserts that this case is indistinguishable from La Porte Construction, where we stated that when a litigant's federal cause of action has been dismissed at a preliminary stage, we must remand with instructions to dismiss the state claims for lack of subject matter jurisdiction. 805 F.2d at 1257. La Porte Construction is persuasive but does not control our decision, either; it predated Carnegie-Mellon and did not establish a rule that where federal claims are dismissed before trial, the pendent state claims must be treated similarly.

Rather, we stated that where the federal claim was dismissed at an "early stage" of the proceedings (which we did not define), so that there had been no commitment of federal judicial resources, dismissal would not prejudice the litigants (namely, the nonmoving defendants), who had not devoted much effort to defending the claim. Thus, dismissal was required, particularly in light of the Gibbs Court's caution against unnecessary decisions of state law. Accordingly, we must weigh the equities as laid out in Gibbs and Carnegie-Mellon.

V.

As the Supreme Court noted in Carnegie-Mellon, under Gibbs a court "should consider and weigh in each case, and at every stage of the litigation, the values of judicial economy, convenience, fairness, and comity" to decide whether to exercise pendent jurisdiction. 484 U.S. at 350, 108 S.Ct. at 619. The Court further stated that "in the usual case in which all federal-law claims are eliminated before trial, the balance of factors to be considered under the pendent jurisdiction doctrine — judicial economy, convenience, fairness, and comity — will point toward declining to exercise jurisdiction over the remaining state-law claims." Id. at 350 n. 7, 108 S.Ct. at 619 n. 7.

No single factor — such as whether the case is in an "early stage" or involves novel issues of state law — is dispositive. Rather, we look to all the factors under the specific circumstances of a given case. Having done so here, we conclude that this matter justified no departure from the usual rule that dismissal was required.

A.

Obviously, to retry in state court a matter that already has been tried in the federal courts would not in itself serve judicial economy. But that is not the issue; a court cannot obtain jurisdiction over a case merely by trying it; otherwise, its decision to retain jurisdiction would be, effectively, unreviewable. Thus we must look at the case as of the filing of the motion to dismiss and not with the benefit of hindsight.

At the stage of the proceedings when the motion was filed, judicial economy would have been better served by dismissal. It is true that some substantial development had occurred in this case, which was pending before a district judge with a reputation for moving cases promptly to trial. For example, a number of discovery matters had been presented to, and decided by, the magistrate judge. Nonetheless, the proceedings were at a relatively early stage compared to those in, e.g., Newport. The case had been pending for only nine months, not four years; trial was scheduled soon but was still a few weeks away; and discovery had not been completed. In short, the parties were not ready for trial.

Second, only one week before Dresser filed its motion to dismiss, Parker Parsley filed a second amended complaint that markedly revised its theories of recovery. That complaint added new theories as to damages, including an assertion that they should be measured by the reduction in the market value of the damaged wells; added claims involving acidizing treatments for wells; and requested that the district court "pierce the corporate veils" of several of the defendants. The second amended complaint also appended an allegation that a defendant shorted other customers besides Parker Parsley, added two bases for liability stemming from the relationship of the defendants and Lancaster, and changed the number of wells included in the underlying dispute by roughly ten percent.

The filing of a pleading that so substantially changed important aspects of the case meant that the case was at an earlier stage than the parties and the court previously might have thought, for the new theories needed development before trial. We also note that by filing the second amended complaint, Parker Parsley brought this problem upon itself.

Third, although the magistrate judge had been active in overseeing discovery and other related matters, there is no indication that the district judge had substantial familiarity with the merits of the case when the motion to dismiss was filed. For example, the judge had conducted only one hearing on the case, a June 8 status conference. In any event, the amount of judicial resources that the case has consumed is most important for our analysis as an indication of the familiarity of the forum with the case and its ability to resolve the dispute efficiently. See Shaffer v. Board of School Directors, 730 F.2d 910, 912 (3d Cir. 1984). Here, the trial court was not so intimately involved in, and familiar with, the case that proceeding further in federal court would have prevented redundancy and would have conserved substantial judicial resources. Nor would it serve judicial economy to reward a plaintiff by allowing it into federal court when it pleads a baseless RICO suit.

B.

Dismissal would not have caused undue inconvenience to the litigants. The district court emphasized the "tremendous financial drain" to the parties, but we do not find that convincing. Little new legal research would be necessary, as the surviving claims were governed by state law, in either forum, and any additional factual research would have had to be conducted anyway. See Financial Gen. Bankshares v. Metzger, 680 F.2d 768, 774 (D.C. Cir. 1982).

Additionally, the most expensive element of the trial preparation, discovery, was largely usable in the state proceeding, as we discuss below. Moreover, while in some cases the likelihood that backlogged state courts could not resolve the dispute promptly might be a factor weighing against dismissal, the record reflects no such factor here.

C.

The fairness factor concerns the prejudice to the parties that would arise from dismissal, and it too weighs in favor of dismissal. Parker Parsley argues that it would be prejudiced by having to start over in state court, fearing that it would have to relitigate all the procedural motions that already had been ruled upon. It also argues that it might have a statute of limitations problem, because Texas Civ.Prac. Rem. Code § 16.064(b) does not block a limitations defense if federal claims were "made with intentional disregard of proper jurisdiction." Thus, Parker Parsley might have to oppose Dresser's claim that it intentionally filed suit in the wrong court and if it fails, it would lose the right to litigate its claims altogether.

We do not agree. The proscription problem is limited, as Parker Parsley's claim would not be time-barred. Section 16.064(a) provides that the statute of limitations is tolled while a case is pending in a court that lacks jurisdiction. Although section 16.064(b) says that the tolling does not apply if the plaintiff filed its initial suit "with intentional disregard of proper jurisdiction," that should not be a problem here.

Parker Parsley asserts that it will be prejudiced because it will have to prove that it was not guilty of such intentional disregard. First, however, that claim is speculative. Second, and more significantly, the plain language of the statute puts the burden of proof on the party asserting the intentional disregard, not on Parker Parsley. Third, given the widespread abuse of civil RICO, it does not seem unreasonable to require that a party risk losing its state claims if it insists upon bringing a groundless RICO claim.

Parker Parsley also asserts that it will be prejudiced because it will have wasted some of its discovery, as the Texas Rules of Civil Procedure provide only for the use of depositions in later-filed state court proceeding, not for the blanket admissibility of all discovery. See Tex.R.Civ.P. 207(b). Thus, the defendants would have the opportunity to re-contest the procedural motions and discovery rulings.

This argument is not compelling. Not all discovery is admissible, anyway. The point is that the parties would not have to repeat the effort and expense of the discovery process. See Waste Sys. v. Clean Land Air Water Corp., 683 F.2d 927, 931 (5th Cir. 1982) (fact that discovery could be used in state court proceeding weighs in favor of dismissal of case from federal court). See also Financial Gen. Bankshares, 680 F.2d at 774. In any event, the state court is a "surer-footed" arbiter of the relevance of pieces of evidence for state law claims. See Gibbs, 383 U.S. at 726, 86 S.Ct. at 1139.

Moreover, we do not expect the relitigation of other matters to pose undue hardship. The defendants can hardly contest jurisdiction, and we do not see other obstacles to resolution of the case in the state court, save those that ought to be there, as we discuss next.

D.

Finally, although it appears that the district court did not consider the matter, dismissal would serve the important interests of federalism and comity. The federal courts are courts of limited jurisdiction, Aldinger v. Howard, 427 U.S. 1, 14-15, 96 S.Ct. 2413, 2420, 49 L.Ed.2d 276 (1976), and often are not as well equipped for determinations of state law as are state courts. Aside from the state courts' superior familiarity with their respective jurisdictions' law, the federal courts' construction of state law can be "uncertain and ephemeral." Pennhurst State School Hosp. v. Halderman, 465 U.S. 89, 122 n. 32, 104 S.Ct. 900, 920 n. 32, 79 L.Ed.2d 67 (1984). "[F]ederal courts are not the authorized expositors of state law; there is no mechanism by which their errors in such matters can be corrected on appeal by state courts." Herbert Wechsler, Federal Jurisdiction and the Revision of the Judicial Code, 13 Law Contemp.Prob. 216, 232 (1948) (cited in Gibbs, 383 U.S. at 726 n. 15, 86 S.Ct. at 1139 n. 15 and quoted in Financial Gen. Bankshares, 680 F.2d at 776). See United Gas Pipe Line Co. v. Ideal Cement Co., 369 U.S. 134, 135, 82 S.Ct. 676, 677, 7 L.Ed.2d 623 (1962) (per curiam) (state court defines authoritative meaning of state law).

In the instant case, the interests of federalism and comity point strongly toward dismissal. All of the remaining legal issues of the case, of course, are of state law, and as the district court later acknowledged, they are difficult ones.

Among the most significant issues that arose from the complaint are, first, whether the claim sounds in contract, for which punitive damages were not available in Texas, see Bellefonte Underwriters Ins. Co. v. Brown, 704 S.W.2d 742, 745 (Tex. 1986), or is a tort claim for fraud, for which punitive damages may be awarded. Although Texas courts base their analysis of the distinction upon the nature of the injury, Jim Walter Homes v. Reed, 711 S.W.2d 617, 618 (Tex. 1986), and the conduct of the defendant, Spoljaric v. Percival Tours, 708 S.W.2d 432, 434 (Tex. 1986), the state courts have not given us plain guidance. The Texas rule that where a party enters into a contract with no intention of performing, that misrepresentation may give rise to a tort action, see Crim Truck Tractor Co. v. Navistar Int'l Transp. Corp., 823 S.W.2d 591, 597 (Tex. 1992), does not tell us how to address claims that may turn out to be fraud in performance.

Second, as the defendants noted in their motion to strike the second amended complaint, Parker Parsley was proposing a measure of market value that seems nonstandard, at best. Under Texas law, the usual measure of damages is the difference in the reasonable cash market value of an oil well, "as equipped, immediately before and immediately after" the damaging event. Atex Pipe Supply v. Sesco Prod. Co., 736 S.W.2d 914, 917 (Tex.App.-Tyler 1987, writ denied).

Parker Parsley's methodology, as put forward by its expert, was a method of calculation based upon a different set of dates. Although we note that where the normal method of calculating damages would result in an unjust outcome, a court may vary the method, B.A. Mortg. Co. v. McCullough, 590 S.W.2d 955, 957 (Tex.Civ.App.-Fort Worth 1979, no writ), comity would be better served if the federal court avoided making such exceptions.

The measure-of-damages difficulty was to re-occur. As the district court recognized, the measure-of-market-value damages that it ultimately sent to the jury constituted an exception to the Texas general rule, as it compared the market value of each well (as that well would have been if properly serviced) with its value to a buyer who knew only that it had been improperly fractured.

VI.

After considering and weighing all the factors present in this case, we thus conclude that the district court, with the admirable intention of moving its docket, abused its discretion in retaining jurisdiction over the state law claims after it had dismissed the federal RICO claims. Because we find that, consequently, the court had no jurisdiction, we cannot reach any of the other issues raised on appeal. The judgment is VACATED, and a judgment of dismissal is hereby RENDERED.