Opinion

HUFFMAN, J.

This appeal of an award of attorney fees and costs to plaintiffs arises out of one of a series of class actions for damages that challenged certain mortgage lenders’ practices of imposing forced order property insurance charges against those of its borrowers who failed to maintain their own insurance on the property securing their loans. In this case, plaintiff borrowers, represented by class representatives Patricia A. Ramos et al. (plaintiffs), reached a stipulated settlement with defendant Countrywide Home Loans, Inc. (Countrywide), providing for the creation of a fund on behalf of plaintiff class members, injunctive relief, and the bringing of an attorney fees application to the trial court. Countrywide now appeals the judgment and postjudgment order of the trial court that granted plaintiffs’ motion for attorney fees and costs in the total amount of $2,171,629.38, plus costs of $95,560.18, pursuant to the attorney fees provision of the settlement agreement.

On appeal, Countrywide generally asserts the trial court abused its discretion in awarding an excessive amount of attorney fees and costs. Specifically, it argues the trial court’s finding of a lodestar amount of fees was *619 presumptively accurate and should not have been exceeded in this factual context, such that any use of a multiplier was per se erroneous. Second, Countrywide argues the multiplier used, 2.5, was excessive. It also contends the trial court reached improper conclusions on fees and costs because of its use of an alternative measure of fees, a percentage of a common fund, to justify the amount of fees and costs awarded pursuant to the fee-shifting settlement agreement.

Although the trial court order on review makes reference to a number of appropriate factors on which the trial court may base such a fee determination, the decision nevertheless does not represent a proper exercise of discretion which would allow us to conduct a meaningful review of the application of all of the competing standards and policies that have been developed in this area. Rather, the order does not articulate how these various factors relate to this case. Accordingly, we deem it necessary to affirm the underlying judgment with the exception that the postjudgment order awarding attorney fees and costs is reversed, with directions to the trial court to make appropriate findings and exercise its discretion anew with respect to setting a proper amount of costs and fees to be awarded.

Factual and Procedural Background

In 1995, this action was filed alleging causes of action that included breach of contract, unfair business practices, negligence, and fraud theories against Countrywide. Similar actions had been filed statewide beginning in 1993, all challenging various lenders’ practices of placing replacement property insurance policies under a forced order program when a borrower’s own property insurance had lapsed. This case, the Ramos case, had a companion federal class action filed in Pennsylvania, entitled Robinson v. Countrywide Credit Industries (E.D.Pa. Oct. 8, 1997, Civ. A. No. 97-2747 HB) 1997 WL 634502. All the plaintiffs were contending that these charges were being used for administrative purposes in a manner that unfairly penalized the 2 percent of homeowners/borrowers who fell into this category.

Extensive discovery, investigation, and motion practice were conducted by all parties in both actions. Two or three months before the scheduled 1997 trial date in Ramos, Countrywide reached a stipulated settlement with plaintiffs in the Ramos and Robinson cases for the creation of a $3.2 million fund to pay class members damages for the amounts overcharged, along with injunctive relief against the continuation of such policies. An expert retained by plaintiffs estimated that the value of this injunctive relief was $2.5 million or more. However, Countrywide disputed that its monetary value *620 could be measured at all accurately. The settlement agreement further provided that Countrywide would pay the settling plaintiffs’ counsel such attorney fees, expenses and costs as awarded by the court, payable from Countrywide funds and not from the established settlement fund. The agreement provided Countrywide would have standing to challenge and oppose any such application for attorney fees.

Plaintiffs then brought their motion for an award of attorney fees pursuant to the settlement agreement, in the base amount of $1,073,882 (plus $116,796.93 costs). (Originally, the notice of settlement to class members had stated an award of up to $5.2 million would be sought.) Both contractual (loan documents and settlement agreement) and statutory (Code Civ. Proc., § 1021.5) claims to attorney fees were asserted. Plaintiffs claimed there had been no significant duplication of efforts among the nine groups of attorneys involved. This action included novel theories (reliance on a lender’s loss payable endorsement) not raised in the two local companion cases or the six or so other cases nationwide that raised similar issues.

In opposition, Countrywide contended that plaintiffs had not shown any reasonable amount of fees, nor any basis to enhance them, nor any benefit from comparing this fees measure to another. It argued that this was not a groundbreaking case or settlement, and although very similar issues had been previously litigated, the main new and different issues here (that the use of a lender’s loss payable endorsement theory did not justify the lender’s policies) did not factor into the settlement.

After taking the matter under submission, the trial court first established the lodestar award of attorney fees in the amount of $868,651.75, which was around $200,000 less than the base amount requested in the motion. The court named seven law firms that would receive specified amounts of fees and costs; eight law firms had been associated at various times with plaintiff’s local lead attorney in this case, Mr. Blumenthal. After using a multiplier in the amount of 2.5, the trial court made a final award of $2,171,629.38 attorney fees to plaintiffs. Costs in the amount of $95,560.18 were also awarded. The order used a cross-check of a comparable method of computing fees, setting forth its reasoning as follows: “The court finds Plaintiffs succeeded in challenging perceived abuses in forced order insurance programs for homeowners, and that the lawsuit sent a message to other lenders in the forced order insurance industry. The Court finds further that an enhancement of the lodestar fee amount is appropriate given the risks undertaken, the complexity of the issues, the skill displayed, the preclusion of other employment and the public interests served. . . . The lodestar fees and costs were calculated as follows [seven individualized awards follow, totaling as *621 above]. [¶] The Court finds the above fee award is fair and reasonable and commensurate with the common fund approach also utilized in cases of this nature. In so ruling, the Court notes that the lodestar necessarily did not include future work which will be required of class counsel to monitor the injunction, administration of settlement, etc.” (Italics added.)

Judgment was prepared reflecting this award. Countrywide appeals the underlying judgment to the extent it incorporates the postjudgment order regarding fees and costs.

Discussion

Standard of Review

In reviewing a challenged award of attorney fees and costs, we presume that the trial court considered all appropriate factors in selecting a multiplier and applying it to the lodestar figure. (Downey Cares v. Downey Community Development Com. (1987) 196 Cal.App.3d 983, 998 [242 Cal.Rptr. 272].) This is in keeping with the overall review standard of abuse of discretion, which is found only where no reasonable basis for the court’s action can be shown. (Citizens Against Rent Control v. City of Berkeley (1986) 181 Cal.App.3d 213, 233 [226 Cal.Rptr. 265].) Such a review standard is adequate to address any argument that an award is excessive on its face. (Ibid.)

However, Countrywide relies on Flannery v. California Highway Patrol (1998) 61 Cal.App.4th 629 [71 Cal.Rptr.2d 632] (Flannery) to argue that although the normal standard of review regarding an attorney fees award is abuse of discretion, “discretion may not be exercised whimsically, and reversal is required where there is no reasonable basis for the ruling or when the trial court has applied the wrong test to determine if the statutory requirements were satisfied.” (Id. at p. 634.) We take this contention to mean that de novo review of this trial court order is warranted to the extent that, in this context, a determination of whether the criteria for an award of attorney fees and costs have been met is a question of law. (See Wilson v. Wal-Mart Stores (1999) 72 Cal.App.4th 382, 389 [85 Cal.Rptr.2d 4], and City of Sacramento v. Drew (1989) 207 Cal.App.3d 1287, 1297 [255 Cal.Rptr. 704].)

Along these same lines, the court in Beasley v. Wells Fargo Bank (1991) 235 Cal.App.3d 1407, 1418 [1 Cal.Rptr.2d 459] cited to Serrano v. Priest *622 (1977) 20 Cal.3d 25 [141 Cal.Rptr. 315, 569 P.2d 1303] (Serrano III) for the applicable standard: In reviewing a challenge to a trial court decision applying a lodestar multiplier to an award of attorney fees, the standard of review is abuse of discretion, since the trial judge was presumably in the best position to determine the value of the services rendered by counsel, “and we may not disturb the judge’s decision on this point unless we are convinced it was clearly wrong.” We agree that more than an ordinary question of abuse of discretion is presented here, and the fees award must be scrutinized for its compliance with all applicable standards. (See also Lealao v. Beneficial California Inc. (2000) 82 Cal.App.4th 19, 25-26 [97 Cal.Rptr.2d 797] (Lealao).) 1

With these approaches in mind, we discuss the arguments made by Countrywide. First, it contends the lodestar amount of fees was presumptively accurate and should not be exceeded in this factual context, such that any use of a multiplier was erroneous. Second, the multiplier used (2.5) is said to be excessive, and in any case, error is claimed in the use of an alternative measure of fees (the common fund approach) as a comparison to justify the amount of fees and costs awarded.

Lodestar Measurement of Attorney Fees and Costs; Enhancement

As the court in Flannery, supra, 61 Cal.App.4th 629 explained, California law in this area has developed from the leading case of Serrano III, supra, 20 Cal.3d 25. There, the Supreme Court approved the lodestar adjustment method of calculating the amount of a fees award. “That method requires the trial court first to determine a touchstone or lodestar figure based on a careful compilation of the time spent and reasonable hourly compensation for each attorney. That touchstone figure may then be augmented or diminished by taking various relevant factors into account, including (1) the novelty and difficulty of the questions involved and the skill displayed in presenting them; (2) the extent to which the nature of the litigation precluded other employment by the attorneys; and (3) the contingent nature of the fee award, based on the uncertainty of prevailing on the merits and of establishing eligibility for the award. (20 Cal.3d at pp. *623 48-49.)” 2 (Flannery, supra, 61 Cal.App.4th at p. 639, italics added, fn. omitted; see also Press v. Lucky Stores, Inc. (1983) 34 Cal.3d 311, 322-323 [193 Cal.Rptr. 900, 667 P.2d 704].) This approach is meant to increase objectivity and predictability in fee awards. (Serrano III, supra, 20 Cal.3d at p. 48, fn. 23.) Such a multiplier or enhancement may be used where the court finds the lodestar figure does not provide sufficient compensation. (Flannery, supra, 61 Cal.App.4th at p. 646.)

In Flannery, supra, 61 Cal.App.4th 629, the Court of Appeal set out a valuable summary of the history of attorney fee awards made under a fee shifting approach. There, as here, the defendant-appellant was arguing that the trial court should not have used a multiplier to increase the lodestar fee amount, on the basis that in Burlington v. Dague (1992) 505 U.S. 557 [112 S.Ct. 2638, 120 L.Ed.2d 449], the United States Supreme Court had rejected the use of multipliers to enhance fees under federal fee-shifting statutes. That approach assumes the factors used to justify application of the multiplier were already subsumed within the lodestar. (Flannery, supra, 61 Cal.App.4th at p. 645.)

Countrywide makes the same argument here, with the addition of policy arguments that risk enhancement would encourage nonmeritorious claims to be brought, and risk multipliers can be unfair to various individual defendants. They add that state law appears to show a trend toward approving the federal approach. (Flannery, supra, 61 Cal.App.4th at p. 647; Weeks v. Baker & McKenzie (1998) 63 Cal.App.4th 1128, 1174-1175 [74 Cal.Rptr.2d 510].)

We disagree that there is any state appellate court trend away from applying the approach of Serrano III, supra, 20 Cal.3d 25, toward the fee request in this case. Rather, as recognized by the court in Flannery, supra, 61 Cal.App.4th at page 646, until the Supreme Court revisits the standard set out in Serrano III or the Legislature definitively changes the law, the intermediate appellate courts are bound by that approach. In this case, there has been no challenge to the trial court’s decision to reduce the lodestar amount by some $200,000, thus setting the base amount of fees at $868,651.75, and to reduce the costs in a similar fashion. More importantly, it is not erroneous per se to add a multiplier into the calculation, because Serrano III clearly permits the use of such a factor to augment or diminish the touchstone figure by taking the various specified factors (e.g., attorney skill, difficulty of issues, risk, preclusion of other employment) into account. (Serrano III, supra, 20 Cal.3d at pp. 48-49.)

*624 III

Enhancement Issues: Amount of Multiplier and Cross-check Used

The unresolved issues remaining for our consideration are whether the trial court’s selection and application of this 2.5 multiplier was a proper exercise of discretion on this record, or amounted to a duplicative reweighing of similar factors in both the lodestar and enhancement determinations. A related issue is whether the trial court should have refrained from using a comparison measure of fees as a cross-check of the ultimate award.

To resolve these issues, we first reiterate the standards under which we review this order. A commentator explains, “Unlike the substantial evidence rule, which measures the quantum of proof adduced in the proceedings below [citation], the abuse of discretion standard measures whether, given the established evidence, the lower court’s action ‘falls within the permissible range of options set by the legal criteria.’ [Citation.] [¶] The substantial evidence rule deals with evidentiary proof, while the abuse of discretion standard is concerned with legal principles. But both standards of review entail ‘considerable deference to the fact-finding tribunal.’ [Citations.] [¶] . . . On appeals challenging discretionary trial court rulings, it is appellant’s burden to establish an abuse of discretion. [Citations.]” (Eisenberg et al., Cal. Practice Guide: Civil Appeals and Writs (The Rutter Group 1999) ¶¶ 8:88 to 8:89, p. 8-33.)

Along these lines, it is both parties’ task on appeal to assist the court in its review of the record, by pointing out those portions of the record that support their respective positions on the trial court’s exercise of discretion in deciding the issue presented. Because that task was performed in the appellate briefs we have received in a somewhat cursory fashion (e.g., by merely referring this court to reams of lodged exhibits), we have had some difficulty in evaluating the discretionary ruling before us. It is not enough for the parties to refer to the abuse of discretion standard and claim it has or has not been met, as was essentially the case at oral argument. Moreover, our task has been complicated by the terse nature of the trial court’s ruling itself, which gives virtually no explanation for the basis of the substantially enhanced award of fees and costs here. Because it merely lists the enhancement factors used, without a more complete explanation of their applicability in this context, the order is subject to question regarding the factual basis of the exercise of discretion made.

With this general background stated, we turn to Countrywide’s specific challenges to the different aspects of this award.

*625 A

Multiplier; Attorney’s Skill/Preclusion of Other Employment

Countrywide’s attorneys first argue the trial court erred or abused its discretion in using a relatively large multiplier, 2.5, on the base amount of fees. Their arguments are reflective of the comments of the court in Flannery, supra, 61 Cal.App.4th 629. There, the court acknowledged that the doctrine developed in the federal courts restricting enhancements of lodestar fees has many things to recommend it: “[I]n Burlington v. Dague [, supra,] 505 U.S. 557, [citation], the majority adopted [the] holding that enhancement of the lodestar for contingency is not permitted under federal fee-shifting statutes. It reasoned that an enhancement for contingency duplicates factors already subsumed in the lodestar, i.e., the higher number of hours expended to overcome the difficulty of establishing the merits or the higher hourly rate of an attorney skilled and experienced enough to do so. It acknowledged that the relative merits of the claim are not reflected in the lodestar, but stated that a contingent risk enhancement would provide the same incentive to bring relatively meritless claims as those with merit, an unlikely objective of any ‘reasonable fees’ provision. It also reasoned that the statutory language limiting fees to prevailing parties should bar recovery for the risk of loss. To award a contingency enhancement under a fee-shifting statute in effect pays for the attorney’s time or anticipated time in other cases where his or her client does not prevail. (Id. at pp. 562-565 [112 S.Ct. at pp. 2641-2643].)” (Flannery, supra, 61 Cal.App.4th at p. 645, fn. omitted.)

Specifically, the court in Flannery, supra, 61 Cal.App.4th 629 opined, “The United States Supreme Court’s criticisms of the contingency risk enhancement have considerable persuasive force. Also compelling are its observations that factors such as the novelty and complexity of the case and the special skill and experience of counsel are necessarily reflected in the lodestar and do not merit an enhancement of that figure.” (Id. at p. 645, italics added.) However, the court concluded California law does not require a finding of abuse of discretion when a trial court includes in a determination of reasonable attorney fees the use of a multiplier, or takes into account the contingent nature of the case (there, in the Fair Employment and Housing Act context). (Id. at p. 646.) Moreover, as stated in Beasley v. Wells Fargo Bank, supra, 235 Cal.App.3d at page 1418, the appellate court will not disturb the judge’s decision on this point unless it is convinced it was clearly wrong. (Ibid.; Serrano III, supra, 20 Cal.3d at p. 49.)

We adhere to the rule stated in Flannery, supra, 61 Cal.App.4th 629. “We readily acknowledge the discretion of the trial judge to determine the value *626 of professional services rendered in his or her court. (Serrano III, supra, 20 Cal.3d at p. 49.) Nevertheless, the exercise of that discretion must be based on a proper utilization of the lodestar adjustment method, both to determine the lodestar figure and to analyze the factors that might justify application of a multiplier. Whether an award is justified and what amount that award should be are two distinct questions, and the factors relating to each must not be intertwined or merged. [Citation.]” (Flannery, supra, 61 Cal.App.4th at p. 647.) For example, when the record expressly demonstrates that the trial court has considered the same factors twice, and has used them not only to calculate a reasonable hourly rate for purposes of awarding the lodestar award amount but also to enhance it, impermissible double counting or a windfall may result. (Ibid., disallowing duplicate consideration of public, benefit, skill/nature of work performed.)

In the case before us, the court began its ruling by stating, “Plaintiffs succeeded in challenging perceived abuses in forced order insurance programs for homeowners, and ... the lawsuit sent a message to other lenders in the forced order insurance industry.” Arguably, this reasoning relies on the skilled nature of the representation and the public interest benefits conferred in setting the base amount of fees. However, the same factors are apparently used again in the ruling to justify enhancement, which controverts the principle stated in Flannery, supra, 61 Cal.App.4th 629 that double-counting of factors should not be allowed to create a windfall. If the trial court’s lodestar calculation already reflected such factors, its order does not adequately reveal its reasoning for a further enhancement on this account. More detailed findings are essential to permit meaningful review of the award. Similarly, the order refers to the possibility of future work required to administer the settlement, but without giving a basis for estimating the value of such work.

Moreover, a well-established public policy in this area is to increase the predictability and to reduce the randomness of attorney fees awards in fee shifting cases. (See Burlington v. Dague, supra, 505 U.S. at p. 566 [112 S.Ct. at p. 2643].) By Countrywide’s calculations, some of plaintiffs’ counsel would be paid around $800 per hour under this award. A reasonable exercise of discretion will not allow such a figure, except in truly pioneering and high risk cases. We do not understand the trial court’s apparent conclusion that this is such a case, and require more explanation before we may uphold it here as a valid exercise of discretion.

On a related note, preclusion of other employment was referred to in the trial court’s ruling as a basis for the enhancement of the lodestar fee amount that was made. Plaintiff’s lead counsel’s supporting declaration in this *627 respect refers to an unrelated case in which his firm was unable to perform as much billable work as it would have liked, due to its responsibilities in pursuing the case now before us. Since the trial court relied on this factor to enhance the award to justify its choice of the 2.5 multiplier, more explicit findings would be of great assistance to this court in relating that factor to the figure chosen.

Effect of Settlement

It must be recalled that the award was rendered pursuant to a stipulated settlement, reached shortly before trial, providing for the creation of a $3.2 million fund for the benefit of class members, along with injunctive relief against the continuation of the subject policies. It was further agreed that Countrywide would pay the settling plaintiff’s counsel such attorney fees, expenses, and costs as would be determined by the court (out of its own funds and not those of the settlement fund). As noted by the Court of Appeal in Lealao, supra, 82 Cal.App.4th at p. 52, California courts place a high value upon settlement of litigation, and attorney fee policy should not subvert the settlement goal. Here, for example, it might be said that the parties’ willingness to compromise should be a factor in setting a reasonable fee, including the choice of any multiplier used. The main objective should be to avoid any chilling of settlement efforts under similar circumstances.

A closely related factor in this context is the degree of risk undertaken by plaintiffs’ counsel in this type of litigation. Here, it must be taken into consideration that this was one of a number of similar actions that raised similar theories to challenge the same set of lending practices, not all of which would necessarily have prevailed at trial. At times, plaintiffs’ attorneys were in search of class representatives, some of whom had been disqualified, which required associating more counsel who actually had clients.

When the trial court reduced the lodestar amount requested by $200,000 after analyzing the time records, its considerations could properly have included some likely duplication of efforts among the nine firms associated in this particular piece of litigation, many of which were associated in companion cases also. However, the base award was then greatly enhanced, which is a somewhat inconsistent approach. We are unable to reconcile these items from the tersely worded order as it currently exists.

*628 C

Use of Alternative Fee Standard for Comparison Purposes

Countrywide contends the trial court should not have utilized in its ruling an alternative measure of fees, the common fund approach, as a comparison to justify the amount of fees and costs it awarded under the fee shifting approach. To the extent this led to an incorrect ruling, it claims reversible error.

This settlement agreement expressly provided that Countrywide would pay the settling plaintiffs’ counsel such attorney fees, expenses and costs determined by the court, payable from its own funds, not from the established settlement fund. This was clearly a fee-shifting agreement, not an actual or constructive common fund agreement. (See Serrano III, supra, 20 Cal.3d at p. 34 [“common fund” principle: “[Wjhen a number of persons are entitled in common to a specific fund, and an action brought by a plaintiff or plaintiffs for the benefit of all results in the creation or preservation of that fund, such plaintiff or plaintiffs may be awarded attorney’s fees out of the fund.”].) 3 This common fund exception is inapplicable where, as here, plaintiffs’ efforts have not created an identifiable fund of money out of which attorney fees are sought. (Serrano III, at p. 38.)

Even though the comparison method used in this case was not precisely applicable to these facts, we cannot say the trial court lacked the discretion to perform such a cross-check. In Lealao, supra, 82 Cal.App.4th at p. 45, the Court of Appeal favorably referred to the practice of the federal courts in making fee determinations to cross-check the propriety of a lodestar fee award by using an alternative measure of fees (there, a percentage of the benefit measure, derived from common fund principles).

Here, the settlement fund was created separately from the attorney fee entitlement, so that the fee-shifting nature of this settlement agreement arose quite late in the game in this litigation’s history. For this and many other reasons, there is no reason why a proper exercise of the trial court’s discretion could not include a similar comparison, as long as other applicable rules are applied properly as well.

*629 D

Conclusion

Under all the relevant circumstances, we would not be justified in presuming in this case that the trial court considered only appropriate factors in applying this particular multiplier to the lodestar figure. (Downey Cares v. Downey Community Development Com., supra, 196 Cal.App.3d 983, 998.) Rather, we are authorized to overturn a judge’s decision on this point where we are convinced it was clearly wrong. (Serrano III, supra, 20 Cal.3d 25, 49; Beasley v. Wells Fargo Bank, supra, 235 Cal.App.3d at p. 1418.) When all applicable factors are considered, no reasonable basis for the court’s action is currently evident on this record, but solely with respect to the selection of this relatively large multiplier figure. (Citizens Against Rent Control v. City of Berkeley, supra, 181 Cal.App.3d 213, 233.)

Where discretion has been exercised in a manner that exceeds the applicable legal standards, the proper remedy is to reverse the order and remand the matter to the trial court in order to give it the opportunity to make a ruling that comports with those standards. We adopt this approach here, subject to the following observations we make for the guidance of the trial court.

In general, where the trial court decides to depart from the lodestar attorney fee approach to select and apply a multiplier, it must make appropriate findings on the factors recognized by case law to explain this discretionary determination in such a manner as to make meaningful appellate review possible. Specifically, here, while this case presented risks for plaintiffs’ counsel, it was also resolved without the risks of trial. While it involved some limited novelty of issues, it was also part of a series of related litigation. Further, as we have-explained, the lodestar calculation apparently fully compensated counsel at their respective professional rates for the hours spent in the litigation. Thus, if the trial court on remand wishes to enhance the lodestar amount by a multiplier, it must more precisely articulate why such increment is appropriate. General reference to risk, novelty and skill on this record does not support the relatively large multiplier selected by the trial court. Due to the discretionary nature of this determination, subject to the additional considerations we have identified, we express no opinion as to the ultimate award the court should select on remand. 4

Disposition

The judgment is affirmed with the exception that the order awarding attorney fees and costs is reversed with directions to the trial court to *630 conduct appropriate further proceedings in accordance with the principles expressed in this opinion and to exercise its discretion anew with respect to setting a proper amount of costs and fees to be awarded. Each party shall bear its own costs on appeal.

Work, Acting P. J., and Haller, J., concurred.

A petition for a rehearing was denied August 11, 2000.

The opinion in Lealao, supra, 82 Cal.App.4th 19 was filed July 10, 2000. To the extent it remains controlling authority at the time the trial court in this case receives this matter back upon remand, we commend it to the trial court’s attention for the guidance it may provide stemming from a different set of facts.

Additional factors not pertinent here include considerations of the taxpayers’ interests, and whether charitable or public funding was involved. (Serrano III, supra, 20 Cal.3d at p. 49.)

In a common fund case, the theory is that those who benefit from the creation of the fund must share the wealth with the lawyers whose skills created it. (In re Washington Public Power Supply Sys. Lit (9th Cir. 1994) 19 F.3d 1291, 1300-1301.) Here, in contrast, a separate source of attorney fees was created by the settlement agreement.

Where an attorney fee award has been made at the trial level, the prevailing party may appropriately request fees on appeal. (Serrano v. Unruh (1982) 32 Cal.3d 621, 639, fn. 28 *630 [186 Cal.Rptr. 754, 652 P.2d 985].) No such request has been made in the briefs here and we express no opinion as to any such entitlement.