Covenant Mutual Insurance v. Young (1986) · DecisionDepot
Covenant Mutual Insurance v. Young
California Court of Appeal Mar 7, 1986 No. B007181Published
Before: Johnson
[320]
Opinion
JOHNSON, J.
Facts and Proceedings Below
The appellant in this case, Covenant Mutual Insurance Company (Covenant), issued a bond guaranteeing a contractor. This contractor was constructing a supermarket on the property of one of the respondents, William, Ruby and Kenneth Young (the Youngs). A dispute arose between the Youngs and the contractor. The Youngs hired as their attorneys the other respondents, McCormick, Barstow, Shepard, Coyle and Wayte (MBSCW) and Oliver Wanger (Wanger).
Covenant’s attorneys attempted to settle the dispute. Eventually they were convinced they had arrived at a binding settlement agreement with Youngs’ attorneys, MBSCW and Wanger. After the Youngs refused to sign the purported agreement, Covenant sued not only the Youngs but their lawyers.
Covenant’s complaint included a cause of action for breach of warranty of authority against the two law firms, MBSCW and Wanger. It alleged MBSCW and Wanger represented they had authority to settle the dispute for the Youngs when in fact they did not. As part of this claim, Covenant asked for attorney fees pursuant to Civil Code section 3318.
On February 16, 1984, in a nonjury trial the court decided against the plaintiff Covenant. It entered judgment for defendants MBSCW and Wanger on March 13, 1984. Meantime the successful defendants filed a cost bill which included attorney fees in the amount of $25,913.75. Covenant moved to strike or tax the attorney’s fees. On June 1, 1984, the court entered a minute order denying Covenant’s motion while reducing MBSCW and Wanger’s attorney fee award to an even $24,000.
Covenant filed a notice of appeal on July 20, 1984.
Discussion
The trial court expressly recognized it was reaching beyond existing precedent when it awarded fees to a defendant under Civil Code section 3318. Indeed during the hearing the judge observed: “The fact that there are no cases one way or the other leaves this court with the responsibility of exercising its responsibility to interpret the law as best it can in a new area .... [T]he only . . . question for the court to determine [is] whether the reciprocity that appears to be the way the legislature is moving should be
applied .... I think it is appropriate. On the other hand, the court recognizes that
a very, very narrow, conservative approach
by an appellate court could go the other way.” (Italics added.)
Fulfilling the trial court’s worst fears, this court now proceeds to adopt “a very, very narrow, conservative approach” to this question. In doing so we do not wish to discourage trial judges from applying their best judgment about the future direction of the law when confronting issues of first impression. Indeed this trial court may prove to have been prescient about what the California Legislature eventually will do in this area. Nonetheless, given the present legislative scheme we feel constrained to limit attorney fee awards to plaintiffs in these breach of warranty of authority actions. Moreover, we believe there are sound policy reasons for continuing to do so.
Section 3318 was first enacted in 1872 and remains unchanged from that date. It reads in full: “Breach of Warranty of Authority. The detriment caused by the breach of a warranty of an agent’s authority, is deemed to be the amount which could have been recovered and collected from his principal if the warranty had been complied with,
and the reasonable expenses of legal proceedings taken, in good faith, to enforce the act of the agent against his
principal. ” (Italics added.)
The italicized clause of this section supplies the authority—and the sole authority, it should be emphasized—for awarding attorney fees in these actions. Otherwise the so-called American rule requires both winners and losers to bear their own legal fees in this as in all litigation.
1
The California version of the “American rule” is found in Code of Civil Procedure section 1021 which reads in pertinent part:
“Except as attorneys’ fees are specifically provided for by statute,
the measure and mode of compensation of attorneys and counselors at law is left to the agreement, express or implied, of the parties . . . .” (Code Civ. Proc., § 1021, italics added.)
2
Neither party contends there was an agreement, express or implied, to the effect the loser would pay the winner’s legal fees. Hence we must look for
specific
statutory authorization for
any
attorney fee award in this litigation.
3
[322]
It already has been held
plaintiffs
in actions for breach of warranty of authority are entitled to attorney fees under Civil Code section 3318. In
Beraksa
v.
Stardust Records, Inc.
(1963) 215 Cal.App.2d 708 [30 Cal.Rptr. 504], some businessmen purchased a restaurant from a man who occupied the position of president of the corporation which owned the property. However, he had been elected to this position by an illegal board of directors. When the real owners took control of the board they rescinded the sale. The businessmen who had sought to purchase the restaurant sued. One of their causes of action was against the purported corporation president “because of the breach of his warranty that he had authority, as president of Stardust, to sell the Flame Room.” (215 Cal-App-2d at p. 716.)
After trial, the plaintiffs were awarded $1,000 “the amount incurred for attorney fees ‘for the commencement and prosecution’ of this action against Stardust.” (215 Cal.App.2d at p. 716.) The Court of Appeal upheld this part of the judgment in the following words: “This provision of the judgment is warranted by the concluding words of section 3318, Civil Code. In its statement of the measure of damages caused by the breach of an agent’s warranty of authority, the section includes ‘and the reasonable expenses of legal proceedings taken, in good faith, to enforce the act of the agent against his principal.’ Although section 3318 was enacted in 1872, it has been cited only three times, and in none of these cases were attorney fees in issue. We entertain no doubt, however, of the validity of the provision authorizing recovery for them in such a case as this.”
(Beraksa
v.
Stardust Records, Inc., supra,
215 Cal.App.2d 708, 716.)
Had appellant Covenant won in the instant case, it would have been entitled to recover attorney fees from the respondents under section 3318. These fees would have been among the “reasonable expenses of legal proceedings” needed
“to enforce
the act” of the lawyers against their principal, that is, to prove the purported agent had committed a breach of his warranty of authority.
This statute by its terms does not, however, expressly or by implication authorize attorney fees for
defendants
who succeed in showing they did
not
commit a breach of warranty of authority. Section 3318 is a “measure of damages” provision not a general attorney fee statute. Defendants do not receive damages and thus cannot claim an element of a damage award. Moreover, as a matter of linguistic construction it is impossible to place them in the class of litigants entitled to legal fees under this statute. By no stretch of imagination do agents
defending
a breach of warranty of authority action incur legal expenses in order “to
enforce
the act of the agent against his principal.” (Italics added.) If anything they are seeking to
[323]
avoid
enforcement. Thus section 3318 standing alone provides no statutory authority for the award of attorney fees to respondents in this case.
Indeed respondents MBSCW and Wanger do not seriously argue the language of section 3318 by itself supports this award. Instead they urge a broader principle—reciprocity—applies. This argument proceeds as follows. In this category of cases the statute authorizes winning plaintiffs to receive attorney fee awards. Hence equity requires that defendants likewise receive attorney fees when they win. In support of this principle the respondents cite Civil Code section 1717 and the policies it implements.
Section 1717 imposes reciprocity when a
contract
only authorizes attorney fees for one of the two parties to an agreement.
4
If the other party wins a lawsuit, he or she is entitled to an attorney fee award just as if the contract term applied expressly to both. Thus, assume a lease agreement provides the tenant must pay the landlord’s attorney fees if the landlord sues him successfully under the lease. But assume also the lease agreement fails to mention the tenant’s right to recover attorney fees if he wins. The Legislature had intervened to effectively rewrite this lease. No matter what the agreement says section 1717 allows the tenant to recover his attorney fees should he rather than the landlord win.
The actual language of section 1717 offers precious little comfort to respondents. By its terms it confines relief to “actions on contracts” which an action for breach of warranty of authority is not.
(Tri-Delta Engineering, Inc.
v.
Insurance Company of North America
(1978) 80 Cal.App.3d 752 [146 Cal.Rptr. 14];
Beraksa
v.
Stardust Records, Inc., supra,
215 Cal.App.2d 708.) More significantly, section 1717 is limited to agreements where “the contract
specifically
provides that attorney fees . . . shall be awarded ... to one of the parties . . . .” (Italics added.) No one claims a contract existed providing attorney fees for
either
side of this litigation. Nor did Covenant’s claim to legal fees in the event it won rest on any contractual provision. Instead it was based on the express language of Civil Code section 3318.
Respondents ask we follow the trial court and venture beyond the arid language of section 1717 and into the verdant pastures of underlying principle. They urge what’s sauce for the contractual goose should be sauce for
[324]
the statutory gander. Surely if the Legislature wanted to make sure that if one party to a
contract
is entitled to attorney fees the other is too, it—or at least common notions of justice enforced by the courts—must ensure that where a statute allows attorney fees for one side of a lawsuit the other side has the same right, also.
Respondents point to several cases which have extended the reach of section 1717 and suggest this means the law is evolving toward a general principle of “reciprocity. ” However, at the core of all of these cases was a unilateral
contractual
right to attorney fees not a
statutory
right. What these cases did was broaden the class of litigants who could claim to be parties to the contract and thus entitled to attorney fees when they prevailed
(Reynolds Metals Co.
v.
Alperson
(1979) 25 Cal.3d 124 [158 Cal.Rptr. 1, 599 P.2d 83] [alleged alter egos of corporation entitled to attorney fees even though they did not sign the contract since they would have been liable for attorney fees under contract had they lost];
Lewis
v.
Alpha Beta Co.
(1983) 141 Cal.App.3d 29 [189 Cal.Rptr. 840] [one tenant entitled to attorney fees from other tenant where he successfully sues to enforce restrictive covenant present in both leases and where both leases had attorney fee provision]) and to allow recovery of attorney fees even where the contract was found to be invalid or unenforceable
[Berge
v.
International Harvester Co.
(1983) 142 Cal.App.3d 152 [190 Cal.Rptr. 815];
Manier
v.
Anaheim Business Center Co.
(1984) 161 Cal.App.3d 503 [207 Cal.Rptr. 508];
Reynolds Metals Co.
v.
Alperson, supra, 25
Cal.3d 124). What these cases most definitely did
not
do was sanction attorney fees where no contractual attorney fee provision exists.
Respondents’ entire argument overlooks a salient difference between “one-sided” attorney fee provisions which individuals or institutions insert in private contracts and those which lawmakers enact in public legislation. The former are created by private parties for their own private advantage and almost always favor the party with the greater bargaining power.
5
The latter, on the other hand, are created by legislators as a deliberate stratagem for advancing some public purpose, usually by encouraging more effective enforcement of some important public policy. Through enactment of section 1717 the Legislature has evidenced it regards private imposition of unilateral fee-shifting to be
against
the public interest. At the same time, in
[325]
literally scores of specific areas of the law it has intentionally imposed one-sided fee-shifting as a means of
advancing
the public interest.
6
The fact lawmakers offer a bounty for plaintiffs who sue to
enforce
a right the Legislature has chosen to favor in no sense implies it intends to offer this same bounty to defendants who show they have not violated the right. Indeed the more logical explanation is that the Legislature desires to encourage injured parties to seek redress—and thus simultaneously enforce public policy—in situations where they otherwise would not find it economical to sue.
To illustrate, imagine someone contemplating a lawsuit where the potential recovery is only a few thousand dollars. His own legal fees might eat up a good share of the judgment. So he is unlikely to sue even if success is near certain. Or suppose a litigant’s chances of winning are good but far from certain. If he has to deduct his own legal fees from a less than certain judgment, he may not sue even though his injury was quite severe and his potential gross recovery quite large.
In categories of cases where the Legislature wants to encourage litigation it can intervene to alter the decision-making equation by instituting unilateral fee-shifting. Then the injured person knows he will not have to absorb his own lawyer’s legal fees, at least if he wins. This makes it economical to seek redress not just in the aggravated cases where the potential economic recovery is huge but in modest cases as well. It also means the probability of success does not have to be so high before it makes economic sense to file suit. Thus, as a result of the Legislature’s intervention, more injured parties will be able to file more lawsuits and the public policy behind the substantive statute—whatever it may be—will be enforced more broadly and more effectively. Moreover, as one commentator observes: “Awarding fees as part of a prevailing plaintiff’s relief should also provide
increased and efficient deterrence of wrongful primary conduct
because of the prospect of having to pay the full cost that an injured party incurs in securing compensation for its loss . . . .” (Rowe, Attorney Fee Arrangements and Dispute Resolution, paper prepared for the National Conference on the Lawyer’s Changing Role in Resolving Disputes, Harv. Law Sch., Oct. 14-16, 1982, at p. 41, citing Posner, Economic Analysis of Law (1977) p. 143, italics added.)
Respondents and the trial court ask us to superimpose some judicially manufactured principle of “reciprocity” on the statutes which call for one-
[326]
sided fee shifting.
7
To do so obviously would frustrate the legislative intent to allow more injured people to seek redress and to encourage improved enforcement of public policy. Indeed it is entirely possible bilateral fee-shifting would lead to fewer lawsuits and less effective enforcement than is experienced in the absence of any fee-shifting at all. Injured people contemplating a lawsuit would confront the prospect of having to pay the defendant’s legal fees as well as their own in the event they lost. This would make the bet even less appealing where the potential recovery was modest or where the chances of winning were good but uncertain. Some evidence suggests mutual fee-shifting causes individual as opposed to institutional litigants to be more cautious than they should be and more cautious than they would be in the absence of any fee-shifting provision at all.
8
Economic analysis likewise supports the proposition that two-way fee-shifting will cause fewer claims to be filed than either the American rule of no fee-shifting or one way proplaintiff fee-shifting as provided in Civil Code Section 3318.
9
As one leading authority summarizes:
[328]
lawyer. Under the American rule, the prospective plaintiff does not risk payment of any legal fee—even his own lawyer’s—if he loses. But if he loses under two-way fee-shifting he is faced with the risk he will have to pay his opponent’s fees. Thus, claimants—especially “risk aversive” ones— are apt to be scared away from filing even highly promising claims by fears of an economic loss they don’t have to worry about under the American rule or one way fee shifting. (See, e.g., Rowe,
Predicting the Effects of Attorney Fee Shifting, supra,
47 Law & Contemp. Probs. at pp. 153-154.)
[327]
“The approach that should uniformly encourage the pursuit of claims of all sorts in all situations is a one-way pro-prevailing-plaintiff rule. Such a policy permits plaintiffs to expect greater net recoveries, without adding a counterbalancing threat of loss .... For the middle-income litigant, to whom the risk of having to pay costs would be a major deterrent,
the difference between American and one-way rules on the one hand, and the two-way approach with its threat of substantial costs on the other, should be quite
significant. ” (Rowe,
Predicting the Effects of Attorney Fee Shifting, op. cit. supra,
at pp. 147-148.) Elsewhere this same commentator observes: “[W]ith medium-strength claims, risk aversion by plaintiffs of modest means appears highly likely to play a significant role. The probable result is that the threat of substantial loss—despite the chance of larger gain than under the American rule—would deter the middle class from pursuing plausible but less than highly promising claims, ...” (Rowe, Attorney Fee Arrangements And Dispute Resolution,
op. cit. supra,
at p. 26, see also, Comment,
Distribution of Legal Expense Among Litigants
(1940) 49 Yale L.J. 699, 702-703.)
10
This analysis also compares the effects of various fee-shifting arrangements where the claimant is represented by a contingent fee
[328]
We think the Legislature’s—not the court’s—calculation of when and where unilateral fee shifting serves the public interest should control.
11
In any event, respondents have not demonstrated the public interest is ill-served by unilateral fee-shifting in the instant case. They raise the spectre of frivolous lawsuits. But unilateral fee-shifting only encourages the filing of cases where there is a reasonable opportunity of winning. Victory is the only way to have one’s legal fees paid by the other party. So defeat is still expensive. Moreover, the Legislature already provides a way for the victim of a frivolous lawsuit to shift his legal fees to the offending plaintiff. Code of Civil Procedure section 128.5 authorizes trial courts to “order a party, . . . pay any reasonable expenses,
including attorney’s fees,
incurred by another party as a result of bad-faith
actions .
. . that
axe, frivolous
. . . . ” (Code Civ. Proc., § 128.5, subd. (a), italics added.) This should be enough to discourage frivolous filings even where the Legislature has provided for unilateral fee-shifting. We are not about to add the further sanction of mutual fee-shifting which, as we have seen, will discourage the meritorious as well as the frivolous lawsuit and thereby defeat the legislative goal of encouraging the redress of grievances and enhancing the enforcement of public policy.
[329]
Disposition
The award of attorney fees to respondents is reversed. The remainder of the judgment is not challenged and is affirmed. Appellant to be awarded its costs on appeal.
Lillie, P. J., and Thompson, J., concurred.
1
In civil law and other common law countries the losing party has to pay the winning party’s “reasonable legal fees” in all or at least nearly all categories of litigation.
2
The full text of Code of Civil Procedure section 1021 reads: “Except as attorney’s fees are specifically provided for by statute, the measure and mode of compensation of attorneys and counselors at law is left to the agreement, express or implied, of the parties; but parties to actions or proceedings are entitled to costs and disbursements, as hereinafter provided.”
3
“The Legislature has established that in the absence of an express agreement or statute, each party to a lawsuit is responsible for its own attorney’s fees [citations omitted].”
Section 1717 provides in pertinent part: “(a) In any action on a contract, where the contract specifically provides that attorney’s fees and costs, which are incurred to enforce the provisions of that contract, shall be awarded either to one of the parties or to the prevailing party, then the party who is determined to be the prevailing party,
whether he or she is the party specified in the contract or not,
shall be entitled to reasonable attorney’s fees . . . .” (Civ. Code, § 1717, italics added.)
5
As explained by one Court of Appeal, “It is common knowledge that parties with superior bargaining power, especially in ‘adhesion’ type contracts, customarily include attorney fee clauses for their own benefit. This places the other contracting party at a distinct disadvantage. Should he lose in litigation, he must pay legal expenses of both sides, and even if he wins, he must bear his own attorney’s fees. One-sided attorney’s fees clauses can thus be used as instruments of oppression to force settlements of dubious or unmeritorious claims. [Citations omitted.] Section 1717 was obviously designed to remedy this evil.”
Of 210 fee-shifting statutes listed in Pearl, California Attorney’s Fee Award Practice, appendix B, pages 165-195 (Mar. 1985 supp. IV) fully 169 explicitly limit the award to one party, usually the plaintiff or claimant. The rest authorize the award of attorney fees to the “prevailing party,” whichever side wins. (Pearl,
supra,
at pp. 67-71.)
7
We are aware another division of this court recently endorsed this principle in the contex
t
of a different but similar statute (Civ. Code, § 3083) which imposed unilateral fee-shifting. See in advance sheets
Kelemen Construction, Inc.
v.
American City Bank
(Cal.App.).
However, the Supreme Court ordered this opinion not to be published.
Accordingly, we find it unnecessary to question the conclusion or the reasoning of this opinion.
8
This was the conclusion of an empirical study of mutual fee-shifting in England’s “payment into court” system. (Zander,
Costs of Litigation—A Study in the Queen’s Bench Division
(June 25, 1975) Law Socy. Gazette, at p. 680 et seq.)
9
The question of fee-shifting and its impact on litigation decisions has been examined rather extensively in one of the more legitimate applications of economic analysis to law and legal processes. See, e.g., Mause,
Winner Takes All: A Re-Examination of the Indemnity System
(1969) 55 Iowa L.Rev. 26; Posner,
An Economic Approach to Legal Procedure and Judicial Administration
(1973) 2 J. Legal Stud. 399; DeWees et al.,
An Economic Analysis of Cost and Fee Rules for Class Actions
(1981) 10 J. Legal Stud. 155; Rowe,
The Legal Theory of Attorney Fee Shifting: A Critical Overview,
1982 Duke L.J. 651; Shavell,
Suit, Settlement, and Trial: A Theoretical Analysis Under Alternative Methods for The Allocation of Legal Costs
(1982) 11 J. Legal Stud. 55; Rowe,
Predicting the Effects of Attorney Fee Shifting
(1984) 47 Law & Contemp. Probs. 139.
Although the viewpoints and results are by no means unanimous, one can distill some useful lessons from this form of analysis. The impact of various fee arrangements on a given claim or class of claims generally will be affected by at least three factors: (1)
How the competing litigants—or typical categories of competing litigants—feel about the inevitable risks of litigation.
Which ones like to take risks, which are afraid of risk, and which are in the middle neither liking nor fearing risk. Those who fear risk are termed “risk aversive”; those who neither prefer nor fear risk are called “risk neutral.” A “risk preferrer” will reject $2,500 in cash in favor of a 25 percent chance to gain $8,000 even though that chance is only worth $2,000 (25 percent of $8,000). Someone who is “risk neutral” would take the $2,500 in this situation but would be indifferent if offered a 25 percent chance at a $10,000 payoff since it would be worth the same, that is, $2,500 (25 percent of $10,000). But someone who is “risk aversive” clearly would take the $2,500 in cash over the 25 percent chance of gaining $10,000 and if really “risk aversive” might well take the $2,500 in preference to a 25 percent chance at $20,000 even though that chance is worth $5,000.
[327]
If fee-shifting raises the stakes it will tend to discourage those who fear risk more than it does others at least in cases where they perceive any appreciable risk of losing. (2)
How much it will cost to litigate the dispute as a percentage of the amount in dispute.
If the costs are large in relation to the stakes they will be a more important factor in the litigants’ decision making and whether and how these costs are shifted will influence litigation behavior more dramatically than in high stakes-low cost cases. (3)
How the competing litigants view their respective chances of winning—and especially their chances of losing.
If a litigant is certain of victory even two-way fee-shifting will be perceived as operating in only one direction, that is, to encourage the litigant to file suit and pursue his claim to the inevitably successful conclusion and the double reward of a damage recovery and an attorney fee award.
10
The rationale for these conclusions can be found in the three factors identified in footnote 9,
ante.
Each individual dollar is less critical to survival as one gets more of them (law of diminishing marginal value of money). Hence according to this analysis wealthy people tend to be less risk aversive than low and middle income people. Moreover, litigants who engage in a lot of litigation can balance their losses in some cases against their wins in others and thus tend to be “risk neutral.” (Galanter,
Why The “Haves” Come Out Ahead: Speculations on the Limits of Legal Change
(1974) 9 Law & Soc’y Rev. 95, 99-100 and fn. 11, where the author defines the frequent litigator—typically an institution like government or a union or a credit agency—as a “repeat player” and the rest as “one-shotters.”) Thus, this analysis suggests modest income one-shot claimants will be risk aversive. This means they will be less attracted by the uncertain gains and more apprehensive about the potential losses of litigation.
Now assume this risk averter is deciding whether to file a relatively modest claim. Further assume the claimant’s costs of litigating the claim are high and the defendant’s costs about the same. Also assume the claimant’s objective chances of winning are in the 40-80 percent range, not 95-100 percent. Even assume it would still be rational for a claimant to file suit and indeed a “risk neutral” one would. Nonetheless, according to these analysts, our “risk aversive” claimant—who happens to be typical of most individual litigants—would undervalue the sum he might win if successful (the modest damage recovery plus payment of his own high attorney fees) and overvalue the sum he might lose if unsuccessful (both his own attorney fees and his opponent’s). Thus two-way fee-shifting would discourage him from
[328]
pursuing his claim. “The mechanism is simply that, given risk aversion and the diminishing marginal utility of income and wealth, the threat of having to pay the other side’s fee can loom so large in the mind of a person without considerable disposable assets that it deters the pursuit of even a fairly promising and substantial claim or defense.” (Rowe,
Predicting the Effects of Attorney Fee Shifting, supra,
47 Law & Contemp. Probs. 139, 153.)
11
In justifying the essentially one-way nature of fee-shifting in federal civil rights cases, Senator Orrin Hatch made statements which suggest how any legislature might react to well-meaning attempts to tamper with the litigation incentives it deliberately established.
“Some State and local government officials have suggested that a fairer approach, and one which would deter frivilous suits, would be to allow plaintiffs and defendants to recover equally.
“The legislative history of the 1976 Fees Act pointed out clearly, and correctly I think, the need for the dual standard: If the persons seeking to enforce their civil rights were faced with paying their opponents [szc] attorneys’ fees if they simply did not win the case, the Fees Act would create a greater disincentive to bring these civil rights suits than the situation it attempted to remedy.” (128 Cong. Rec. S4878 (daily ed. May 11, 1982).)
AI Brief
AI-generated · verify before citing
Holding. Civil Code section 3318 does not authorize an award of attorney fees to a defendant who successfully defends against a claim for breach of warranty of authority. The statute provides for fee recovery only by a prevailing plaintiff as a measure of damages.
Issues
Whether Civil Code section 3318 authorizes an award of attorney fees to a prevailing defendant in a breach of warranty of authority action.
Whether the principle of reciprocity requires that defendants be awarded attorney fees when a statute provides for such fees to prevailing plaintiffs.
Disposition. reversed
Quotations verified verbatim against the opinion
“This statute by its terms does not, however, expressly or by implication authorize attorney fees for defendants who succeed in showing they did not commit a breach of warranty of authority.”
“Section 3318 is a “measure of damages” provision not a general attorney fee statute.”
“We think the Legislature’s—not the court’s—calculation of when and where unilateral fee shifting serves the public interest should control.”