MIKVA, Circuit Judge:

This case involves an inter sese dispute between the present and former shareholders of a closely held corporation as to which of them is personally liable to the Small Business Administration (SBA) for the corporation's default on a loan. We hold that the dispute must be resolved by reference to District of Columbia law and that, because the parties have not provided any guidance as to what that law might be and the district court did not apply District law, the case must be remanded.

FACTS

Practa, Inc. was a District of Columbia corporation formed in 1969 to own and manage a gourmet grocery store named "What in the World." In 1974, the corporation entered into loan negotiations with The National Bank of Washington, which culminated with the Bank extending Practa credit up to $50,000. The loan was guaranteed by the SBA. In November 1974, Practa secured a $20,000 disbursement on the loan. At the time of this disbursement, appellee Pena and his wife owned 45% of the corporation's stock, appellee Callahan and his wife owned 10%, and appellant Trotter owned 45%. The SBA required all of these stockholders to execute personal guarantees in favor of the SBA on the loan.

In January 1975, the corporation changed hands, with appellant Eghinis, an employee, purchasing the Pena and Callahan shares. The parties agree that, at the time of the stock purchase, nothing was said about the personal loan guarantees of Pena and Callahan, nor was any express assignment of that liability to Eghinis entered into when the shares were transferred. From January 1975 on, Eghinis and Trotter were the exclusive owners and managers of Practa's grocery business, while Pena and Callahan ceased to have any financial interest, apart from their personal guarantee of the SBA loan, in the corporation. At the request of Eghinis and Trotter, the rest of the loan, totalling $30,000, was distributed to Practa in four separate payments during 1975 and 1976.

There is no dispute that the loan money was applied to legitimate business needs. Nonetheless, by 1978 the corporation was in financial extremis, and in September of that year the corporation defaulted on its loan to the Bank. Shortly after, the corporation was liquidated, with a principal of $15,845 remaining owing to the Bank. After purchasing its guaranteed portion of the loan from the Bank, the SBA, seeking to recover on the shareholders' personal guarantees of the loan, sued Pena, Callahan, and Trotter in the United States District Court for the District of Columbia.

Pena and Callahan then cross-claimed against Trotter and filed a third-party complaint against Eghinis in an effort to have Trotter and Eghinis indemnify them for any liability they were found to have to the SBA. The trial court, per Judge Johnson, granted summary judgment for the SBA against the loan guarantors on January 18, 1982, and, following further extensive discovery between the parties, then held that Trotter and Eghinis were required to indemnify Pena and Callahan for the judgment that the SBA had received against the latter. From this decision, Trotter and Eghinis now appeal.

DISCUSSION

There is no dispute over the judgment in favor of the SBA. The argument on appeal instead centers on the trial court's decision to require the present stockholders to indemnify the former ones for the liability which the latter were found to have to the SBA. The trial court's decision to do so was based on two principles: first, that an implied promise to indemnify runs from every principal to its sureties, and thus that the corporation was liable to Pena and Callahan for their personal guarantee of the loan; second, that the corporate veil of Practa should be pierced so that Trotter and Eghinis would be liable for the corporation's implied promise to indemnify Pena and Callahan. In support of this second principle, the court drew on four cases to find that the corporate fiction of Practa, Inc. should be disregarded.

We are of the opinion, however, that none of the decisions upon which the district court relied is controlling in the context of this case. The proper resolution of this litigation hinges on the absence of any federal interest of any sort in this particular inter sese shareholder dispute over who is ultimately liable for the corporation's loan default. The SBA's recovery on the personal guarantees of Pena, Callahan, and Trotter satisfied the only federal interest in the entire case. Put another way, no federal law gave or denied Pena and Callahan the right to be indemnified by Trotter and Eghinis; the existence and contours of such a right were to be defined solely by reference to local law.

The same point can be elucidated by considering the jurisdictional bases upon which the various claims in this case were brought into federal court. Jurisdiction for the SBA's action against the loan guarantors rested on 28 U.S.C. § 1345 (1976), which applies to suits in which the United States is a plaintiff; when the United States sues, federal interests are sufficiently implicated that federal law, including federal common law, may be invoked to define the rights and liabilities of the parties. See, e.g., Clearfield Trust Co. v. United States, 318 U.S. 363, 63 S.Ct. 573, 87 L.Ed. 838 (1943). In contrast, Pena's and Callahan's cross-claim against Trotter and third-party claim against Eghinis were dependent upon the outcome of the federal claim, but were not based on any federal right.

Callahan's claim was in federal court only because it was ancillary to the federal action of the SBA. Ancillary jurisdiction generally refers to the exercise of federal jurisdiction over state law claims between non-diverse parties, when those claims are brought into federal court by a defendant or an intervenor in an action already properly within a federal court's jurisdiction. See generally Matasar, Rediscovering "One Constitutional Case": Procedural Rules and the Rejection of the Gibbs Test for Supplemental Jurisdiction, 71 CALIF.L.REV. 1401, 1401 n. 1 (1983). The rationale which allows federal courts to exercise jurisdiction over such state-law claims is that a sufficiently close nexus exists between those claims and the federal claim properly before the court that effective vindication of the latter requires allowing a party to bring the state and federal claims in one suit. See United Mine Workers v. Gibbs, 383 U.S. 715, 86 S.Ct. 1130, 16 L.Ed.2d 218 (1966). While the exercise of ancillary jurisdiction to bring new parties into a case has been questioned, see Aldinger v. Howard, 427 U.S. 1, 96 S.Ct. 2413, 49 L.Ed.2d 276 (1976), the Supreme Court appears to have endorsed the particular type of ancillary jurisdiction invoked in this case. See Owen Equipment Erection Co. v. Kroger, 437 U.S. 365, 376, 98 S.Ct. 2396, 2403, 57 L.Ed.2d 274 (1978) (treating impleader of third-party defendant as example of appropriate use of ancillary jurisdiction).

Pena's claim, while similarly within the ancillary jurisdiction of the federal court, may also have satisfied the requirements of the diversity statute, 28 U.S.C. § 1332 (1976). See United States v. Pena, Civ. No. 80-2739, mem. op. at 2 n.[*] (D.D.C. June 18, 1982). The district court was apparently satisfied that Eghinis was a citizen of a different state than Pena, although the record reveals no reason for this conclusion. As a result, no federal question basis for jurisdiction existed for either of the claims under review. Those claims thus arose under state law.

Resolution of all claims that arise under state law, whether brought in federal court or not, is controlled by the substantive law of the state that creates the cause of action, Erie R.R. Co. v. Tompkins, 304 U.S. 64, 58 S.Ct. 817, 82 L.Ed. 1188 (1938); after Erie, federal courts are not free to concoct a "general federal common law" for the resolution of all disputes which find their way into federal court. This court has held that the principle of Erie, if not its constitutional underpinning, fully applies to federal courts in the District of Columbia when they exercise jurisdiction over state-created causes of action. Anchorage-Hynning Co. v. Moringiello, 697 F.2d 356, 360-61 (D.C. Cir. 1983). In consequence, the district court could have looked only to District of Columbia precedents to determine the circumstances under which the local courts would pierce the corporate veil of District corporations.

We note that there can be no dispute that the particular state's law which must be applied is that of the District of Columbia. Practa, Inc. was incorporated under the law of the District of Columbia, the loan was made in the District, and the stock transfer that underlies this dispute occurred here. Under these circumstances, District of Columbia courts would apply the law of the District. See, e.g., Owen v. Owen, 427 A.2d 933 (D.C.App. 1981). A federal court is obligated to do likewise. See Klaxon Co. v. Stentor Electric Manufacturing Co., 313 U.S. 487, 61 S.Ct. 1020, 85 L.Ed. 1477 (1941).

The cases upon which the district court relied, however, involve situations in which some federal interest was implicated by the decision whether to pierce the corporate veil; as a result, the courts had both a jurisdictional and substantive basis for resorting to a federal common law of veilpiercing. See generally Note, Piercing the Corporate Veil: The Alter Ego Doctrine Under Federal Common Law, 95 Harv.L. Rev. 853 (1982) (outlining circumstances under which federal law controls question whether corporate veil ought to be pierced). For example, in Capital Telephone Co. v. FCC, 498 F.2d 734 (D.C. Cir. 1974), the FCC denied an application for a radio license on the theory that the corporate applicant was wholly owned and controlled by an individual who exercised similar control over another corporation which already had a license to broadcast in the area. As the court noted, id. at 738, the question whether granting a license under these circumstances would be consistent with the Communications Act of 1934 was inherently a federal question: "Where the statutory purpose could be easily frustrated through the use of separate corporate entities a regulatory commission is entitled to look through corporate entities and treat the separate entities as one for purposes of regulation." Id. at n. 10. A similar situation was presented in Quinn v. Butz, 510 F.2d 743 (D.C. Cir. 1975), in which the Secretary of Agriculture had ascribed to a corporation's vice-president the corporation's violations of a federal statute; our consideration in that case of whether the corporate veil had properly been pierced is precedent only for cases in which some federal interest is implicated by the veil-piercing inquiry. See also Anderson v. Abbott, 321 U.S. 349, 64 S.Ct. 531, 88 L.Ed. 793 (1944) (piercing corporate veil of bank-stock holding company for purposes of Federal Reserve Act and National Bank Act). The analysis in these cases is not controlling for situations in which there is no substantive federal interest in whether the corporate fiction should be disregarded. The question whether a corporate veil ought to be pierced for purposes of applying some federal statute is distinct from whether a corporate veil ought to be pierced for purposes of allocating state tort or contract liabilities.

The parties on appeal appear to have abandoned the cases cited by the district court. Instead, the argument from precedent is focused on our recent decision in Labadie Coal Co. v. Black, 672 F.2d 92 (D.C. Cir. 1982). Close analysis, however, reveals that that case, too, is not of authoritative aid in the effort to resolve the present dispute. In Labadie, a Kentucky corporation had contracted to sell coal through a Virginia corporation (FAI) of which Black was the director, president, and sole employee. When FAI defaulted on over $100,000 of obligations to Labadie Coal Co., the coal company sought to make Black personally liable for FAI's debts. Our opinion in that case laid out several factors to guide analysis of when a shareholder, director, or manager may be made personally liable for a corporation's debts. Labadie, however, did not address the choice-of-law problem with which we are concerned today, for Labadie, which appears to have been a diversity action, simply assumed that a general federal common law of veil-piercing could be constructed to settle the dispute in that case. Nowhere in Labadie is any reference made to the law of either of the two states — Kentucky or Virginia — which appear to have had a substantive interest in the circumstances under which its corporations' veils would be pierced. We therefore read Labadie as not dispositive of the threshold question of whether state or federal law ought to govern analysis of a veil-piercing case. Labadie applies only after it has been determined that some federal interest is implicated by the veil-piercing inquiry. When, as in this case, the sole basis upon which a claim is in federal court is ancillary jurisdiction — so that the claim by definition has no federal content — Labadie is of no precedential value.

The district court also noted Francis O. Day Co. v. Shapiro, 267 F.2d 669 (D.C. Cir. 1959), which, like Labadie, appears to have been a diversity action. To the extent that Francis O. Day Co. applied local law, however, it would almost certainly have been Maryland, rather than District of Columbia, law that was applied. In that case, appellant — a Maryland corporation — had recovered in Maryland state court for the failure of another corporation to perform its contractual obligation regarding property located in Maryland. Moreover, Francis O. Day Co. did not address the choice-of-law issue and relied centrally on two pre- Erie cases, United States v. Milwaukee Refrigerator Transit Co., 142 F. 247 (E.D. Wis. 1905) and Callas v. Independent Taxi Owners' Ass'n, 66 F.2d 192 (D.C. Cir.), cert. denied, 290 U.S. 669, 54 S.Ct. 89, 78 L.Ed. 578 (1933). We therefore read Francis O. Day Co., like Labadie, as not relevant to the threshold question of whether state or federal veil-piercing law ought to be applied.

We are thus faced with a situation in which the parties have not cited a single decision that sheds any light on the key issue in the case — whether, under the law of the District of Columbia, Trotter and Eghinis would be liable to Pena and Callahan for the personal guarantees that the latter signed in favor of the SBA. Were the correct way of resolving that issue the veil-piercing approach taken by the district court, we might nonetheless reach out to decide the merits of the issue despite the failure of the parties to brief it properly. As a general matter, veil-piercing analysis depends more on the facts than on any sharply defined underlying legal standard, for the essential legal test in all jurisdictions is whether it would be equitable to allow the stockholders to invoke the corporate mantle to shield themselves from personal liability, 1 Fletcher, Cyclopedia of Corporations, § 41.30 at 431 (1983); as a result, the damage done by relying on federal rather than local veil-piercing cases may in itself have been minimal. But we believe that the district court's decision contains a second error which, in conjunction with the court's failure to follow the command of Erie, militates in favor of a remand.

The district court assumed that the touchstone for deciding this case was whether the corporate veil of Practa ought to be pierced. That is, if but only if the corporate form were disregarded, Trotter and Eghinis would be liable. Contract law principles, however, rather than veil-piercing doctrines, would seem to provide a more appropriate departure point for analysis. The parties involved in the stock transfer of this closely held corporation were all intimately familiar with each other. At least one of the new owners — Trotter — knew of the personal guarantees to the SBA that Pena and Callahan had signed. As a result, the crucial issue is whether, at the time of the stock transfer in 1975, there was any enforceable agreement, whether express or implied, that the new owners would assume liability for the personal guaranties of the former owners. If such an agreement did exist, Trotter and Eghinis must indemnify Pena and Callahan regardless of whether the corporate veil is pierced. And if no such agreement can be implied under the District's law of contracts, it is hard to imagine that Trotter and Eghinis would nonetheless be found liable under the more circuitous theory that Practa, Inc. had an obligation to indemnify Pena and Callahan and that Practa's corporate veil should be pierced. The contract issue — which, like the veil-piercing question, must be resolved in this case by reference to local law — would thus seem at least as central to the controversy as the veil-piercing inquiry.

The district court did briefly refer to the bearing of contract law principles on this case. The court said: "whether the equitable remedy sought by the movants is viewed as being grounded in contract or tort principles, the undisputed facts in these consolidated actions establish, as a matter of law, that the movants are entitled to be indemnified by the third party defendants." However, no cases — particularly local cases — were cited to support this statement. We are therefore left to speculate at the reasoning underlying the district court's statement. The statement seems to suggest that, because Pena and Callahan received no direct benefits from the corporation once they sold their stock, it would make little sense for them to have continued to act as sureties for the corporation's debt to the Bank. That may not necessarily follow, however. It may be that the price of the stock at the time of the sale already accounted for the fact that Pena and Callahan would remain personally liable should the corporation default on its SBA loan. That is, the fact that the Bank stood ready to disburse another $30,000 to Practa on the strength of the previously signed personal guarantees was obviously of some value to purchasers of Practa stock; the stock may have been purchased at a premium that incorporated this value.

There are other reasons to question whether contract doctrines would require Trotter and Eghinis to indemnify Pena and Callahan. It is clear that the parties made no express agreement as to who would be responsible for the personal guarantees that Pena, Callahan, and Trotter had already signed. That by itself may be of some significance, for District of Columbia law may require an issue of this magnitude to be addressed by the parties if the status quo is to be altered. Alternatively, the District might take the view that the implication of a promise to indemnify under these circumstances would constitute a drastic sort of judicial intervention, and that, even if the parties simply forgot about the issue, the burden of this neglect ought to be upon the party whose responsibility it was to raise the issue — in this case, the original guarantors.

By noting these countervailing considerations that weigh against the district court's conclusion that contract law would require indemnification in this case, we of course do not mean to express any view on the merits of the contract dispute. We do suggest, however, that the contract issue deserves far greater attention than it has thus far received. Most importantly, that issue must be resolved by looking to District of Columbia cases, rather than to general equitable considerations or to federal common law. To allow the district court to address this task in the first instance, the case must therefore be remanded.

Though such a remand will further delay the resolution of this dispute, several important considerations make remand the appropriate course. First, it is no mere legal technicality that requires the use of local law to settle this case; as Erie recognizes, the District of Columbia has a strong regulatory interest in determining the circumstances under which a corporation it has created will cease to provide a cloak of limited liability with regard to transactions occurring within the District. This issue is an important aspect of the District's relations with the businesses it attracts. Second, we are loathe to reach out and decide this issue without adversarial participation; not only would it unfairly surprise the parties in this case to spring a result cut wholly from judicial cloth upon them as a fait accompli, but judicial analysis would also profit from adversarial reading of the relevant precedents. Finally, this court is, for all practical purposes, the court of last resort for these parties; the Supreme Court almost never grants certiorari in a diversity action to decide a local law question. The district court should therefore pass on this issue in the first instance so that proper appellate review, should the parties desire it, can be facilitated by the presence of a prior court's analysis of the case and by the parties focused consideration of that analysis. Cf. Migra v. Warren City School District Bd. of Education, ___ U.S. ___, 104 S.Ct. 892, 899, 79 L.Ed.2d 56 (1984) ("Prudence" dictates that district court, rather than Supreme Court, be given first opportunity to interpret an issue of local law). Accordingly, we leave it to the district court on remand to decide whether, under District of Columbia contract and veil-piercing precedents, an implied promise to indemnify exists under the facts of this case.

We note finally that the district court should also consider whether it was appropriate to continue to exercise jurisdiction over these state law claims once the United States ceased, on June 18, 1982, to be a party to the litigation — well before the local law claims were resolved. Of course, if complete diversity of citizenship existed between all the parties, the retention of jurisdiction over the state law claims would be unassailable: the outstanding obligation on the note exceeded $15,000, thus satisfying the amount in controversy requirement for diversity jurisdiction. While diversity was found to exist in 83-1599, no such finding was possible in 83-1600, since the Callahans and the Trotters are both residents of the District of Columbia.

Even if the claims do not satisfy the diversity statute, however, they were properly before the district court as ancillary claims to those claims over which the court clearly did have jurisdiction — the claims of the United States. But as the Supreme Court noted in United Mine Workers v. Gibbs, 383 U.S. 715, 86 S.Ct. 1130, 16 L.Ed.2d 218 (1966), in the interest of comity and to avoid needless interpretations of state law, "if the federal claims are dismissed before trial, even though not insubstantial in a jurisdictional sense, the state claims should be dismissed as well." Id. at 726, 86 S.Ct. at 1139. Although this and other circuits have permitted exceptions from this principle of Gibbs when, prior to dismissal of a federal claim, substantial proceedings have been conducted which would be wasted by the refusal to continue pendent jurisdiction, see Financial General Bankshares, Inc. v. Metzger, 680 F.2d 768, 773 n. 9 (D.C. Cir. 1982) (collecting cases from other circuits), the only substantial proceedings here consisted of pretrial discovery, which would not necessarily be wasted. See id. at 773. Moreover, even when Gibbs has no application, pendent jurisdiction should sometimes be declined when local law issues are unsettled, complex, or novel. See Doe v. Board on Professional Responsibility of the District of Columbia Court of Appeals, 717 F.2d 1424, 1428 (D.C. Cir. 1983). In Financial General Bankshares, supra, both the Gibbs principle and the factor of uncertainty in local law combined to produce a rejection of pendent jurisdiction. There the complaint raised federal securities law claims and a common law claim (under District of Columbia law) for breach of an attorney's fiduciary duty. The federal law claims were dismissed or settled before trial. The district court retained jurisdiction over the common law claim, held a three-day trial, and issued a lengthy opinion resolving "novel and difficult issues of local law." This court reversed, holding that, while the district court possessed the power to address the common law claim since it was initially ancillary to the federal claim, the district court had abused its discretion in doing so.

It may similarly be appropriate to decline pendent jurisdiction here. On remand, the district court should consider whether there is any reason in the present case to depart from the principle of Gibbs; and whether (on the basis of its inquiry into the substance of the District's contract and corporate law that we have described above) the local law is so unclear or unelaborated as to counsel dismissal.

There may, however, as noted above, be a basis for diversity jurisdiction over the suit in 83-1599. If such a basis exists, the fact that the district court would have to resolve the claim in 83-1599 would strongly militate in favor of retaining jurisdiction over the remaining claims in 83-1600. This may be what the district court did in its order denying summary judgment. It may also develop that either or both of these suits would now be time-barred if refiled in state or District of Columbia courts, which could also justify retention of jurisdiction under the circumstances of this case. (Dismissal could instead be conditioned upon the defendants' willingness to waive any defenses they might have under the District of Columbia statute of limitations. See, e.g., Younger v. Plunkett, 395 F.Supp. 702, 715 (E.D.Pa. 1975)). On remand, the district court will have an opportunity to consider further whether jurisdiction ought to be retained over these actions and, if so, to explain why.

It is so ordered.