Opinion for the court filed by Senior Circuit Judge FRIEDMAN.
Opinion dissenting in part filed by Circuit Judge CLEVENGER.
FRIEDMAN, Senior Circuit Judge.
In this Winstar-related appeal, the acquirer of financially-distressed savings and loans associations (also known as “thrifts”) seeks damages from the United States for the latter’s breach of agreements permitting the acquired thrifts to treat as capital, for regulatory purposes, an item known as “supervisory goodwill,” and to amortize that “asset” over a substantial period. The Court of Federal Claims found that the parties had entered into agreements regarding such treatment of “supervisory goodwill” and that the government had breached those agreements. Granite Mgmt Corp. v. United States, 53 Fed.Cl. 228, 241 (2002) (“Liab.Op.”). The court then granted summary judgment for the government on damages, rejecting all the theories upon which the acquirer based its damage claims. Granite Mgmt. Corp. v. United States, 58 Fed.Cl. 766 (2003) (“Damages Op.").
With one exception, we affirm the trial court’s rejection of those damage claims. We hold that the Court of Federal Claims improperly granted summary judgment for the government on the claim that the acquirer could have sold the acquired thrifts for a higher price if the thrifts had been allowed to continue treating their “supervisory goodwill” as regulatory capital. We conclude that further development of the facts on that issue is necessary, and therefore we remand to the Court of Federal Claims for that purpose.
I
A. The history and circumstances surrounding the savings and loan industry crisis during the late 1970s and early 1980s have been extensively discussed in opinions of the Supreme Court and this court. See United States v. Winstar Corp., 518 U.S. 839, 116 S.Ct. 2432, 135 L.Ed.2d 964 (1996); Winstar Corp. v. United States, 64 F.3d 1531 (Fed.Cir.1995) (en banc); see also, e.g., Fifth Third Bank of W. Ohio v. United States, 402 F.3d 1221 (Fed.Cir.2005); Home Sav. of Am., FSB v. United States, 399 F.3d 1341 (Fed.Cir.2005); Glendale Fed. Bank, FSB v. United States, 239 F.3d 1374 (Fed.Cir.2001).
During those years, more than 400 thrifts failed and many others reached the brink of insolvency. To deal with the crisis, federal regulators encouraged healthy financial institutions and outside investors to purchase troubled thrifts. The regulators offered various financial incentives to thrift acquirers, including cash contributions from the government and favorable accounting treatment of an intangible asset known as “supervisory goodwill,” which reflected the amount by which the assumed liabilities of the acquired thrifts exceeded the value of the acquired assets. Typically, the acquirers were permitted to include “supervisory goodwill” in the thrift’s reserve capital requirements and to amortize that “asset” over many years.
In response to continued financial problems in the savings and loan industry, [1377]*1377Congress in 1989 enacted the Financial Institutions Reform, Recovery, and Enforcement Act of 1989, Pub.L. No. 101-73, 101 Stat. 183 (“FIRREA”). That Act, among other things, barred the thrifts’ use of “supervisory goodwill” as regulatory capital. As a result, many thrifts no longer complied with federal regulatory capital requirements; a number of them became insolvent and were seized by regulatory authorities.
Thrift acquirers filed lawsuits alleging that by enacting and enforcing FIRREA, and thus eliminating use of “supervisory goodwill” as a regulatory asset, the government had breached its contracts with the acquirers. In United States v. Winstar Corp., 518 U.S. 839, 116 S.Ct. 2432, 135 L.Ed.2d 964 (1996), the Supreme Court affirmed this court’s en banc determination that FIRREA had that effect, and that the government was liable in damages for breach of contract. Since Winstar, this court has decided numerous Winstar-relat-ed cases.
B. The following underlying facts in this case are undisputed. In 1986, the appellant Granite Management Corporation (“Granite”) (then known as “First Nationwide Financial Corporation”) acquired four financially-troubled thrifts in three separate transactions, one transaction each in Ohio (State Savings & Loan Company and Citizens Home Savings Company), Missouri (St. Louis Federal Savings & Loan Association), and Kentucky (Lincoln Federal Savings & Loan). The institutions were combined into a single thrift called “First Nationwide Bank” that was run by Granite and its parent company, Ford Motor Company (Ford). As a result of the acquisitions, Granite also obtained “branching rights” that permitted that bank to expand from its California base into ten new states where the acquired thrifts were operating, some years before deregulation allowed competitors to enter interstate banking.
The government made a direct cash contribution of $168 million to First Nationwide’s capital. Under so-called Assistance Agreements by which the government authorized the three transactions, Granite was permitted to treat nearly $150 million as “supervisory goodwill” and to amortize that “asset” over 25 years. The parties dispute the amount of the unamortized balance remaining in 1989, when FIRREA was enacted. According to Granite, the unamortized balance was approximately $275 million. Damages Op. at 769.
By 1990, First Nationwide had become severely undercapitalized. In order to bring the bank into compliance with regulatory requirements, its parent, Ford, provided it with $250 million in additional capital. In 1994, Ford and Granite sold First Nationwide Bank.
In 1995, Granite filed suit in the Court of Federal Claims seeking damages for the government’s alleged breach of contract and related claims. In 2002, the trial court granted Granite’s motion for partial summary judgment on liability. The court ruled that the government had entered into contracts with Granite covering the three transactions by which Granite acquired the failing thrifts; that the contracts authorized Granite to treat “supervisory goodwill” as part of its regulatory capital and to amortize it over twenty-five years; and that the government, by enacting FIRREA, breached these contracts. Liab. Op. at 234, 241.
Granite then presented several alternative theories for determining damages, under which it would recover between $104.3 [1378]*1378million and $421.9 million. Damages Op. at 770. First, Granite sought its costs of administering and operating the acquired thrifts — its “cost of performance” — under both reliance and restitution theories. Second, Granite contended that by acquiring the thrifts, it conferred a benefit on the government, and claimed the value of that benefit. Third, Granite sought the “lost value” of the “supervisory goodwill” it lost under FIRREA. Granite proposed three bases for those “lost value” damages: (1) Granite’s estimated cost of replacing the lost regulatory capital; (2) the cost of Ford’s $250 million capital infusion, which Granite contends was required to replace the lost goodwill; and (3) the difference between the price at which First Nationwide Bank was sold and the higher price Granite contends it could have obtained if First Nationwide had included the “supervisory goodwill.” Damages Op. at 769-70, 777.
The Court of Federal Claims granted the government’s motion for summary judgment on damages. It rejected all of Granite’s damages theories, several of which; it held, this court had previously denied in other cases.
II
In its brief as appellee, the government challenges the Court of Federal Claims’ ruling that the government entered into a contract with Granite regarding the Ohio transaction under which Granite acquired two thrifts, State Savings and Citizens Home. Citing two of our prior decisions, the government argues that “as in” those cases, it “here made no contractual promise concerning a 25-year amortization period in connection [with] the State Savings/Citizens transaction and therefore committed no breach with respect to that transaction.” Because the government did not cross-appeal on that issue, it cannot now raise it.
An appellee may rely upon any ground supported by the record for affirmance of the judgment, whether or not the lower court relied on that ground. If, however, the appellee seeks to change or modify the judgment rather than just affirm it, the appellee must file a cross-appeal. Radio Steel & Mfg. Co. v. MTD Prods., Inc., 731 F.2d 840, 844 (Fed.Cir.1984) (stating that “a party will not be permitted to argue before us an issue on which it has lost and on which it has not appealed, where the result of acceptance of its argument would be a reversal or modification of the judgment rather than an affirmance”); see also El Paso Natural Gas Co. v. Neztsosie, 526 U.S. 473, 479-82, 119 S.Ct. 1430, 143 L.Ed.2d 635 (1999).
In the present case, in which the government did not file a cross-appeal, the government’s argument is not made as an alternative ground for affirmance of the judgment. To the contrary, the government is seeking to modify or change the judgment of the trial court. This is shown by the conclusion of the government’s brief, which states:
For the foregoing reasons, the judgment of the trial court should be reversed with respect to the Statewide Savings/Citizens Home agreement and affirmed upon all other grounds concerning damages.
Ordinarily, a ruling that the appellee cannot seek a change in the judgment is made in circumstances where it is clear that the appellee could have cross-appealed. The present case presents a more complex situation in that the government effectively lost on the issue of liability for [1379]*1379breach of contract, but ultimately prevailed on the issue of contract damages, making it less clear that the government could have cross-appealed from the final judgment. However, the facts do not warrant a different conclusion on this issue.
The trial court’s earlier liability order, which the government as appellee seeks to change, was not a final judgment and the government could not have directly appealed it. If the government had attempted to appeal from the final judgment insofar as that judgment incorporated the earlier liability order involving the Ohio transaction, the government would have been faced with the problem that ordinarily a litigant cannot appeal from a judgment in his favor. Under the peculiar circumstances of this case, where the government challenges only part of the earlier order granting partial summary judgment, it might have been able to appeal. See Glaros v. H.H. Robertson Co., 797 F.2d 1564, 1573 (Fed.Cir.1986) (stating that grant of partial summary judgment “from which no immediate appeal lies” is “merged” into the final judgment and reviewable on appeal from that judgment), cited with approval in Fifth Third Bank, 402 F.3d at 1236. The government also could have made its argument as an alternative ground for affirmance of the final judgment without seeking a modification of it.
In EF Operating Corp. v. Am. Bldgs., 993 F.2d 1046 (3d Cir.1993), the Third Circuit dealt with a similar situation. In that damages suit, the defendant, Flaherty, moved for summary judgment, and also moved to dismiss the plaintiffs complaint for lack of personal jurisdiction. The district court granted Flaherty’s summary judgment motion, and, without explicitly ruling on Flaherty’s personal jurisdiction motion, it ruled in favor of another defendant that had raised the same jurisdictional issue. The “logical conclusion!’ was that “the district court decided against Flaherty on the jurisdictional issue[.]” Id. at 1048.
The plaintiff appealed from the final order granting summary judgment. The defendant, Flaherty, attempted to contest the personal jurisdiction ruling without filing a cross-appeal. Id. The Third Circuit held that because the defendant, “[b]y seeking dismissal of the complaint for lack of personal jurisdiction,” was “not seeking to support the summary judgment on different grounds” but rather was “seeking] to vacate the summary judgment,” the court would not consider the personal jurisdiction issue in the absence of a cross-appeal. Id. at 1049.
In this case, the government states that although it made the Ohio transaction argument to the Court of Federal Claims in its “motion for summary judgment upon liability, the trial court never specifically addressed it,” just as the district court in EF Operating did not explicitly rule on the defendant’s motion for personal jurisdiction. Like the defendant in EF Operating, the first time the government raised this issue during this appeal was in its brief responding to Granite’s appeal. The government made no attempt to appeal the trial court’s ruling on liability at any time. Because the government does not simply seek to support the grant of summary judgment on damages, but instead explicitly asks that “the judgment of the trial court ... be reversed with respect to the Statewide Savings/Citizens Home agreement,” we rule, as the Third Circuit did in EF Operating, that for lack of a cross-appeal, the government cannot now raise this issue. The circumstances here do not [1380]*1380warrant an exception to the general rule that without itself appealing, an appellee cannot seek a change in the judgment under review.
Ill
In its appeal Granite argues the same theories of damages that it presented to the Court of Federal Claims. Except for the claim discussed in Part IV, below, we find its arguments no more persuasive than that court did. Indeed, our precedent fully supports the trial court’s rejection of these claims.
A. Granite seeks to recover its “cost of performance” for operating the acquired thrifts, under either a reliance or a restitution theory. “Reliance” damages cover the amount a non-breaching party expends in performing the contract in reliance on the other party’s anticipated performance. “Restitution” damages are designed to restore the non-breaching party to the situation that would have existed had there been no contract and no breach. These two theories may be analyzed together because, as this court stated in a previous Winstar-related case, a restitution claim “based on recovery of the expenditures of the non-breaching party in performance of the contract” “can be viewed as a form of reliance damages.” LaSalle Talman Bank, F.S.B. v. United States, 317 F.3d 1363, 1376 (Fed.Cir.2003). Granite conceded at trial that “there was no substantive difference between the two theories” and sought to recover an “identical amount” of damages under each. Damages Op. at 773.
Granite’s claim is that it was damaged by its acquisition and long-term operation of thrifts whose liabilities exceeded their assets, and that this deficit constituted its cost of performing the contract. In several Winstar-related cases, however, this court has rejected damage claims based upon the excess of liabilities assumed over assets acquired. Cal. Fed. Bank, FSB v. United States, 245 F.3d 1342, 1351-52 (Fed.Cir.2001); Glendale, 239 F.3d at 1381-82. As the Court of Federal Claims here stated, “assumed liabilities ... are not a usable measure of either cost to the thrift or benefit to the government” when calculating damages. Damages Op. at 773-77.
Furthermore, as the government points out in its brief, Granite’s alleged liability stemmed from the poor quality of the thrifts’ loans it acquired, rather than from the government’s breach of the contract permitting the use of “supervisory goodwill.” Roger Orders, Granite’s damages expert who performed the accounting upon which Granite based this theory of damages, stated that Granite’s costs were “incurred as a result of the acquisitions, not as a result of the breach.” Since Granite’s “cost of performance” thus did not result from the government’s breach, it is not an appropriate measure of damages. See, e.g., Glendale, 239 F.3d at 1382 (noting non-breaching party’s entitlement to “damages for any losses actually sustained as a result of the breach”).
B. Granite contends that it may recover the value of the benefit it conferred on the government by acquiring the four troubled thrifts. Granite relies on the government’s own analyses of the savings it realized by contracting with Granite rather than liquidating the four thrifts.
The Court of Federal Claims correctly rejected this argument, based on our previous holdings that similar “benefit conferred” theories in Winstar-related cases [1381]*1381“measure [damages] in terms of a liability that never came to pass” and thus are too “speculative and indeterminate” to succeed. Glendale, 239 F.3d at 1382; Cal. Fed., 245 F.3d at 1351. See LaSalle Talman, 317 F.3d at 1376-77 (affirming Court of Federal Claims’ determination that liquidation costs which the government allegedly avoided by acquirer’s assumption of thrift’s obligations were not “a meaningful measure” of damages); Admiral Fin. Corp. v. United States, 57 Fed.Cl. 418, 424 (2003) (relying upon Glendale in rejecting “liquidation cost savings” theory), affd, 378 F.3d 1336, 1343 (Fed.Cir.2004).
The trial court determined that the government had other ways to deal with the problem, such as arranging for the sale of the thrifts to another buyer. It suggested that the government might well have pursued these other options had Granite not acquired the thrifts, because “neither [the government] nor [Granite] ever liquidated the thrifts, and liquidation was an extremely rare and disfavored occurrence.” Damages Op. at 773; cf. Glendale, 239 F.3d at 1382 (noting the existence of alternatives to liquidating a failing thrift).
Granite contends that summary judgment on this issue was erroneous because material issues of fact exist regarding whether these alternatives to liquidation were truly viable. Granite’s argument that the thrifts were “dogs, plagued by credit risk problems,” which surely would have been liquidated if not acquired by Granite, is unpersuasive. The record shows that other possible acquirers beside Granite indicated interest in obtaining the thrifts; that the branching rights associated with the thrifts had significant value; and that Granite itself sold, rather than liquidated, the thrifts (combined at that point into First Nationwide Bank) after their alleged devaluation by FIRREA. Indeed, Granite itself recognizes that there were other possible acquirers of at least three of the thrifts, when it suggests, as an alternative measure of damages, that it should recover at least the difference between what it paid for the thrifts and the next lowest offer from alternative prospective acquirers. As the trial court stated, it would be “inherently speculative to try to figure out now what the Government might have done, under different circumstances, nearly a decade and a half ago.” Damages Op. at 773 (quoting Franklin Fed. Sav. Bank v. United States, 55 Fed. Cl. 108, 120 (2003)).
C. Granite seeks to recover its cost of replacing the goodwill capital it lost as a result of FIRREA. It relies on two alternative cost basés: (1) replacement of the amount of the non-amortized “supervisory goodwill” on its books that FIRREA eliminated, or (2) the $250 million cash capital infusion that Ford provided.
In support of both theories, it relies upon hypothetical studies by its financial expert, Professor Christopher James, who calculated the amount it would have cost Granite to replace the goodwill capital by assuming that Granite issued preferred stock to raise the money. Professor James “analyz[ed] [First Nationwide’s] prior capital raising activity and other comparable market transactions” and calculated “the cost of the equity capital ... by reference to the rates on issuances of preferred stock,” and then deducted “the cost of [First Nationwide’s] costing liabilities during this period in order to reflect the benefit of cash capital not provided by intangible regulatory capital.”
This court recently rejected a similar theory of capital replacement cost based [1382]*1382on hypothetical preferred stock issuance in Fifth Third Bank of Western Ohio v. United States, 402 F.3d 1221 (Fed.Cir.2005). Indeed, the plaintiff in Fifth Third Bank relied on analysis and calculations by the same financial expert, Professor James. See Fifth Third Bank of W. Ohio, 55 Fed. Cl. 223, 228-29 (2003). In neither case did the thrift actually issue any preferred stock. In Fifth Third Bank, this court ruled that the “[plaintiff’s cover damages theory [was] based entirely on hypothetical costs that were never actually incurred” and was therefore properly rejected by the trial court as “highly speculative.” 402 F.3d at 1237. Similarly, Granite did not record any actual costs in connection with obtaining the cash infusion from Ford.
This court also recently held that the Court of Federal Claims had not abused its discretion in basing a damage award for lost “supervisory goodwill” upon “hypothetical replacement costs” in Home Savings of America, FSB v. United States, 399 F.3d 1341, 1353-55 (Fed.Cir.2005). It does not follow, however, that the trial court’s refusal to accept Granite’s comparable theory constituted an abuse of discretion or otherwise involved a legal error. The two cases differ in two critical respects. First, in Home Savings, as in the present case, the source of the additional capital was the acquirer’s parent corporation. Unlike the present case, however, in Home Savings the parent corporation was also a plaintiff in the litigation, in addition to the acquiring thrift, and the parent also sought to recover its damages.
Second, there is no contention that Granite itself actually raised any additional capital in the private market to replace the lost regulatory goodwill. Granite’s calculations are based upon purely theoretical models. As set forth in detail in the trial court’s opinion in Home Savings, however, both the acquiring thrift in that case and its parent (who was also a party to that litigation) raised substantial capital following the loss of “regulatory goodwill” under FIRREA. Home Sav. of Am., FSB v. United States, 57 Fed.Cl. 694, 700-02 (2003). Home Savings “obtained new capital” by issuing $950 million in subordinated debt. Its parent company also raised $250 million in subordinated debt and $357.5 million in preferred stock, and “[a]fter each of these capital raisings, [the parent] contributed all the capital raised to Home Savings.” Id. at 702. Subsequently, the parent raised an additional $280.7 million in convertible preferred stock, half of which it “infused into Home Savings.” The following year, the parent also raised an additional $125 million in subordinated debt, $100 million of which “went to Home Savings.” Id.
Based on these facts, Granite’s case is more like Fifth Third Bank than Home Savings. The trial court here correctly rejected Granite’s theories of replacement of capital costs as too speculative.
D. Granite makes two related arguments to avoid the foregoing analysis. First, it invokes the principle that where “a reasonable probability of damage can be clearly established, uncertainty as to the amount will not preclude recovery.” LaSalle, 317 F.3d at 1374 (internal citations and quotations omitted). Second, it contends that even if the court below correctly rejected its theories, “at a minimum, jury verdict damages would be appropriate in this case.” Bluebonnet Sav. Bank v. United States, 339 F.3d 1341, 1346 (Fed.Cir.2003). (Granite also states in its opening brief that the Court of Federal Claims “[held] as a matter of law that FNB suffered no damage whatever from the loss of [1383]*1383over a quarter of a billion dollars in contractually guaranteed regulatory capital” (emphasis in original). The trial court, however, did not hold that the government’s breach of contract caused Granite no damage, but held only the far different point that Granite had failed to prove its damages.)
The essential premise of these two bases for awarding damages is that, if the plaintiff proves that it was damaged, difficulties in calculating the exact amount of such damage will not preclude an award. Here, however, the Court of Federal Claims determined that Granite failed to prove the predicate element that it had suffered any damages, and we uphold those rulings (other than the “lost sale value” damages theory whose rejection we vacate and remand for further proceedings in Part IV, below). Granite, therefore, cannot rely upon these two grounds as a basis for recovering damages.
rv
One of Granite’s experts was Joseph Walker, an investment banker who assisted in the First Nationwide sale and who formerly headed J.P. Morgan’s Mergers and Acquisitions group. Walker testified that he could have structured the sale to preserve the value of the “supervisory goodwill” had it been available, and that the bank could have been sold for more if such goodwill had been included. Walker relied upon another hypothetical preferred stock issuance calculation by Professor James and upon his own “leverage model” and “sensitivity analysis” to show how much more the thrift would have been sold for if it had included “supervisory capital.” Damages Op. at 780. The “value” of the “supervisory capital” under these two analyses was $136.8 million under the James model and $137.1 million under the Walker model, respectively. Damages Op. at 770. The Court of Federal Claims rejected this theory of damages because it was “speculative” and “implausible” that a buyer would have “expended actual cash to acquire the supervisory capital.” Damages Op. at 780 (internal quotation marks omitted).
We conclude that Walker’s evidence sufficed to create an issue of material fact as to whether First Nationwide could have been sold for more if it had included “supervisory goodwill,” and thus precluded summary judgment for the government on that issue. Therefore, the case will be remanded to the Court of Federal Claims for further development of the record on that question. On the remand, Granite should have the opportunity to offer evidence, if it can produce it, on at least the following questions:
1. What is the factual basis for Walker’s conclusion that First Nationwide could have been sold for a higher price if it had included the “supervisory goodwill?” Have there been any sales of thrifts that included such goodwill? If so, did they sell for a higher price than thrifts without such goodwill?
2. Does Walker have any actual factual basis for determining the alleged higher amount for which First Nationwide could have been sold if it had included “supervisory goodwill”? How much additional value could the thrift have brought if the goodwill was included in the deal? What is the basis for that calculation?
The Court of Federal Claims correctly rejected Professor James’s preferred stock theory as too speculative and an unreliable measure of the additional amount the thrift would have commanded. [1384]*1384Because we remand the ease for further proceedings on this issue, we express no opinion on the validity of Walker’s own “leverage model” and “sensitivity analysis,” or the weight, if any, to which it is entitled. That is for the trial court to .consider in the first instance, in light of whatever additional evidence is presented on the remand.
3. The parties disagree whether “supervisory goodwill” may be transferred at all. Walker assumed that “supervisory capital could be sold,” based on advice he had received from counsel. Apparently there is no definitive answer to that question at this time. Uncertainty over the question would have affected the additional amount a purchaser of the thrifts would have paid if such goodwill were included. This factor must be considered in determining whether the thrift could have been sold for a higher amount if it had included “supervisory goodwill,” and, if so, for how much more.
We cannot tell what evidence, if any, may be produced on these factual questions (or on any others the trial court or the parties may raise), or how the Court of Federal Claims would evaluate it and what conclusions it would draw from it. All we can say at this time is that Walker’s evidence raised sufficient factual issues to entitle Granite to go forward with its case on lost sale value damages.
CONCLUSION
The portion of the Court of Federal Claims’ judgment that granted summary judgment for the government on the issue of whether First Nationwide could have been sold for more if it had included “supervisory goodwill” is vacated, and the case is remanded to that court for further proceedings on that issue in accordance with this opinion. In all other respects, the judgment of that court is affirmed.
AFFIRMED IN PART, VACATED IN PART and REMANDED