OPINION

WERDEGAR, J.

Sharon J. Hoch (decedent) was killed when she was ejected from her 1987 Ford Bronco II, which was equipped with seat belts made by Allied-Signal, Inc./Bendix Safety Restraints Division (Allied-Signal), during a single-vehicle accident. Her parents, Norma and Carl J. Hoch, sued Ford Motor Company (Ford), Allied-Signal and others for wrongful death, alleging negligence and strict product liability; in addition, Carl J. Hoch, as special administrator of his daughter's estate, sought punitive damages in a survival action.

Plaintiffs went to trial against Allied-Signal after settling with Ford and other defendants. At the close of plaintiffs' case, the trial court granted nonsuit on the issue of punitive damages, finding insufficient evidence Allied-Signal had consciously disregarded decedent's rights. Carl Hoch, as special administrator, seeks review of the nonsuit in his appeal.

The jury, by special verdict, found decedent was wearing her seat belt at the time of the accident, the buckle design was defective, and Allied-Signal was negligent. The jury assigned comparative fault 35 percent to Allied-Signal, 45 percent to Ford, and 20 percent to decedent. The total damages, before subtraction of decedent's comparative fault, were found to be $500,000. The court entered judgment on the special verdict against Allied-Signal for $175,000, from which judgment Allied-Signal cross-appeals. The Product Liability Advisory Counsel, Inc., as amicus curiae, joins Allied-Signal in urging modification of the judgment to account for pretrial settlements of other defendants.

FACTS

Decedent was driving alone on Interstate Highway 280. Her mother saw her get into the car, but did not see or hear whether decedent put on her seat belt. An eyewitness estimated her speed before the accident at about 65 miles per hour. The vehicle began swerving right and left, "teeter-totter[ed]," then rolled over two or three times. Decedent was ejected through the sunroof during the rollover and died from injuries sustained when her head hit the pavement.

Decedent's father and a close friend testified decedent always wore her seat belt. Two expert witnesses called by plaintiffs opined a set of linear markings on the left side of decedent's neck was caused by contact with the chest portion of the seat belt. A defense expert disagreed, stating the marks were caused instead by the edge of the sunroof during a period of partial ejection. Plaintiffs' expert testified to the absence of any discernible marks on the interior of the roof, which he would have expected to be present if decedent had been unrestrained during the first part of the rollover; he also thought it likely decedent would have sustained a neck injury, of which there was no evidence in the autopsy report, had she been unrestrained in the first roll. The defense expert, however, found a bent and scuffed grab handle on the passenger side, indicating to him contact with an unrestrained occupant.

The buckle of the driver's seat belt in decedent's vehicle, referred to at trial as a "type one" buckle, has the push button on the front of the buckle. Plaintiffs' theory of design defect was that such a buckle can unlatch inertially. If the housing of the buckle is rapidly accelerated in the forward direction, the button and attached portions of the latching mechanism may momentarily remain at rest relative to the housing, allowing the buckle to unlatch in the same manner as if the button had been pushed. The same effect could be produced by rapid deceleration of a housing moving backward; the button would tend to continue its backward motion relative to the housing.

Allied-Signal did not and does not dispute inertial unlatching is possible when the buckle is removed from the automobile and manipulated in various ways. The principal disputed technical issue was, rather, whether the necessary acceleration or deceleration could be produced on a buckle in use in an automobile and, in particular, during a rollover accident.

David James Biss, an engineer, testified for plaintiffs as an expert on occupant restraint. Biss stated the acceleration needed to unlatch the buckle inertially was in the range of 100 times the acceleration due to gravity. Because of the light weight of the buckle and the "cabling" effect created when a slack belt is suddenly pulled tight, accelerations in that range could be produced in an automobile accident even though the vehicle itself decelerated at a lesser rate. Because seat belts have some slack in them, in a rollover accident the occupant will tumble against and tug on the belt many times. An acceleration or deceleration to or from a speed of 7 or 8 miles per hour in a distance of .2 to .4 inches would be sufficient to open the buckle and could occur in a rollover accident. Biss opined the potential for inertial unlatching makes the type one buckle defective.

The defense expert, Daniel Davee ("reliability engineer" for Allied-Signal), testified to the results of tests he conducted, during the pendency of this action, to evaluate the possibility of inertial unlatching in an impact between the occupant's hip and the buckle. He determined the stopping time of the buckle hit against his hip was eight to ten milliseconds. The relative velocity needed to unlatch the buckle in a stopping time of 8.4 milliseconds was 19.5 miles per hour. He did not believe it was possible in a rollover accident, where the occupant was wearing the belt with the lap portion snug, to develop relative velocities in this range between the hip and the buckle.

Plaintiffs' negligence claim was based on Allied-Signal's previous failure to test the type one buckle for inertial unlatching. Davee did not know of any previous attempts by anyone at Allied-Signal or its corporate predecessors to evaluate through testing the possibility of inertial unlatching, although he noted that "each time that we run a test, okay, whether it be a sled test or other testing that we may do, it confirms to me that that event didn't occur." Dan M. Davey, director of seat belt sales for Allied-Signal, testified neither Allied-Signal nor Jim Robbins Seat Belt Company (Jim Robbins), which Allied-Signal acquired, had tested seat belts for accidental release.

Davee testified Allied-Signal and Jim Robbins had supplied approximately 130,000,000 seat belts with type one buckles to Ford between 1972 and 1987. Approximately 500,000,000 type one buckles were put into domestic cars from 1972 to 1991. Davee was unaware of any claims of inertial unlatching in a vehicle during the 1972-1987 period. He was also unaware of any lawsuit in which the trier of fact had determined a buckle made by Allied-Signal inertially unlatched in an accident.

According to Davee and Davey, Ford actually designed the type one buckle, sending Allied-Signal or Jim Robbins detailed drawings and requiring the buckle be manufactured to Ford's specifications. In contrast, plaintiffs' witness Thomas Feaheny, former Ford vice-president for car engineering and vehicle research, testified Ford specified only exterior appearance, style and configuration, while the suppliers designed the internal latching mechanism.

The Bronco II seat belt assembly was tested by an independent testing company for compliance with Federal Motor Vehicle Safety Standard 209, which requires in part that seat belts be designed to remain on the occupant's pelvis during an accident. There was no evidence, however, the testing company evaluated or tested for the possibility of inertial unlatching.

Peter Bertelson, former manager of impact dynamics at Ford, supervised occupant-safety crash testing in 1967-1968. His group did about 500 crash tests, usually with 2 restrained dummies in the car. Standard production restraints of the period, with a push button buckle, were used; some of the restraints were made by Jim Robbins. About 15 percent of the dummies were not restrained at the end of the test. Bertelson, however, was unsure whether their belts had released or were not securely coupled at the outset. Dan Davey, who started at Jim Robbins in 1969 as liaison engineer to Ford, never heard data were being lost in crash tests because Jim Robbins's buckles were opening accidentally.

Stephen Syson, an automotive safety engineer, testified that about 1965, Ford, General Motors and the seat belt manufacturers, including Jim Robbins, presented a "joint position paper" to United States Senator Ribicoff "explaining inertial seat belt separation and indicating . . . they would redesign seat belt buckles over the near term so that they would not inertially separate." Syson first testified no such redesign was made, but on cross-examination, he stated Ford attempted to modify the buckles by introducing stiffer springs. Syson did not have the "position paper," and it was not produced by any other witness.

DISCUSSION I. The Estate's Appeal

A. Appealability

(1) Decedent's estate in a survival action sought punitive damages from Allied-Signal in the seventh cause of action of the complaint. At the close of plaintiffs' case, defendant moved for nonsuit as to the punitive damages claim. The court orally granted the motion, but at that time (Oct. 31, 1991) did not dismiss the cause of action. Nor did the court incorporate the nonsuit order in the judgment on special verdict, filed November 13, 1991. On February 4, 1992, Carl Hoch, as special administrator, filed a notice of appeal, purporting to appeal from "the Order dismissing the Seventh Cause of Action of the Complaint." Insofar as the record shows, however, no such order was filed until March 2, 1992, when the court filed a written order, drafted by plaintiffs' attorney, granting the nonsuit and dismissing the seventh cause of action.

Code of Civil Procedure section 581c, subdivision (b) provides that where a nonsuit is granted as to some but not all of the issues in an action, ". . . no final judgment shall be entered prior to the termination of the action, but the final judgment in the action shall, in addition to any matters determined in the trial, award judgment as determined by the motion herein provided for." In the present case, the judgment filed November 13, 1991, was the final judgment; the court should have expressly incorporated in that judgment its earlier ruling granting the nonsuit on punitive damages. Apparently neither the court nor the litigants noticed the omission until plaintiffs sought to appeal from the nonsuit. Whether or not dismissal of the seventh cause of action was the technically proper way to phrase the remedial action, the written order was an appropriate attempt to correct the court's inadvertent omission of its nonsuit ruling from the final judgment. We amend the judgment to award judgment for Allied-Signal on the survival cause of action for punitive damages and construe the notice of appeal as being from the amended judgment. ( Molien v. Kaiser Foundation Hospitals (1980) 27 Cal.3d 916, 920-921 [ 167 Cal.Rptr. 831, 616 P.2d 813, 16 A.L.R.4th 518]; see Cal. Rules of Court, rule 2(c).)

B. Propriety of Nonsuit on Punitive Damages

C. Evidentiary Standard for Grant of Nonsuit on Punitive Damages

(2) 581c Stewart Truck Ins. Exchange17 Cal.App.4th 468 21 Cal.Rptr.2d 338 Barry Raskov232 Cal.App.3d 447 283 Cal.Rptr. 463 Bell Sharp Cabrillo Hospital212 Cal.App.3d 1034 260 Cal.Rptr. 886 (3) Carson Facilities Development Co. 36 Cal.3d 830 206 Cal.Rptr. 136686 P.2d 656Ashcraft King228 Cal.App.3d 604 278 Cal.Rptr. 900 Miller Los Angeles County Flood Control Dist. 8 Cal.3d 689 106 Cal.Rptr. 1505 P.2d 193Carson Facilities Development Co., supra,

(4a) On a claim for punitive damages, however, the plaintiff's burden at trial is to establish oppression, fraud or malice by clear and convincing evidence. (Civ. Code, § 3294, subd. (a).) Allied-Signal contends this more rigorous evidentiary standard also applies on review of a nonsuit granted, i.e., to demonstrate error, the plaintiff-appellant must show the existence of clear and convincing evidence to support a finding of malice, fraud or oppression.

We cannot agree a nonsuit is necessarily proper whenever the court deems the plaintiff's evidence less than clear and convincing. "`The sufficiency of evidence to establish a given fact, where the law requires proof of the fact to be clear and convincing, is primarily a question for the [trier of fact] to determine, and if there is substantial evidence to support its conclusion, the determination is not open to review on appeal.'" ( Crail v. Blakely (1973) 8 Cal.3d 744, 750 [ 106 Cal.Rptr. 187, 505 P.2d 1027], quoting Nat. Auto. Cas. Co. v. Ind. Acc. Com. (1949) 34 Cal.2d 20, 25 [ 206 P.2d 841]; see also Beeler v. American Trust Co. (1944) 24 Cal.2d 1, 7 [ 147 P.2d 583]; Rubin v. Los Angeles Fed. Sav. Loan Assn. (1984) 159 Cal.App.3d 292, 298 [ 205 Cal.Rptr. 455].) Where reasonable minds could differ as to whether the evidence would support punitive damages, the resolution of the conflicting inferences and the weighing of opposing evidence is for the jury; for the court to grant a nonsuit in that circumstance, or the appellate court to affirm a judgment of nonsuit, would be to usurp the jury's function.

This is not to say the higher standard of proof has no relevance to the propriety of granting nonsuit. If the plaintiff is to prevail on a claim for punitive damages, the evidence will have to be clear and convincing to the trier of fact. The claim must be evaluated in that light, and on a motion for nonsuit, both the trial and appellate courts must view the evidence "with that higher burden in mind." ( Stewart v. Truck Ins. Exchange, supra, 17 Cal.App.4th at p. 482, fn. omitted; see also Barry v. Rasko, supra, 232 Cal.App.3d at p. 458 [nonsuit on punitive damages affirmed because "[t]he trial court found, correctly, the evidence [of fraud] was not `clear and convincing' . . ."].)

How the higher standard affects the trial or appellate court's view of the evidence may be understood by analogy to the criminal law's requirement of proof beyond a reasonable doubt. (5) Pursuant to this standard, the evidence is sufficient to support a conviction if ". . . a reasonable trier of fact could have found that the prosecution sustained its burden of proof beyond a reasonable doubt." ( People v. Mincey (1992) 2 Cal.4th 408, 432 [ 6 Cal.Rptr.2d 822, 827 P.2d 388]; see Jackson v. Virginia (1979) 443 U.S. 307, 318-319 [61 L.Ed.2d 560, 573, 99 S.Ct. 2781].) In making this determination, the reviewing court must consider conflicting evidence in a light favorable to the judgment, with the presumption the trier of fact drew all reasonable inferences in support of the verdict. The evidence need not convince the reviewing court beyond a reasonable doubt, but must be substantial enough for a reasonable trier of fact to have convicted on that standard. ( People v. Mincey, supra, at p. 432.) The same standard applies to a motion for acquittal, the criminal law parallel to the motion for nonsuit. ( Id. at p. 432, fn. 2; United States v. Taylor (2d Cir. 1972) 464 F.2d 240, 242-244.) (6) Analogously, a motion for nonsuit in a case requiring clear and convincing evidence should be granted only if the trial court determines no reasonable jury could find the plaintiff has presented clear and convincing evidence on the disputed issue.

The United States Supreme Court made the analogy to the criminal law and drew precisely this conclusion as to summary judgment on an issue requiring proof by clear and convincing evidence: "[T]he judge must view the evidence presented through the prism of the substantive evidentiary burden. This conclusion is mandated by the nature of this determination. The question here is whether a jury could reasonably find either that the plaintiff proved his case by the quality and quantity of evidence required by the governing law or that he did not. . . . [¶] . . . [¶] . . . Consequently, where the . . . `clear and convincing' evidence requirement applies, the trial judge's summary judgment inquiry as to whether a genuine issue exists will be whether the evidence presented is such that a jury applying that evidentiary standard could reasonably find for either the plaintiff or the defendant." ( Anderson v. Liberty Lobby, Inc. (1986) 477 U.S. 242, 254-255 [91 L.Ed.2d 202, 215-216, 106 S.Ct. 2505], italics in original; see also Reader's Digest Assn. v. Superior Court (1984) 37 Cal.3d 244, 252 [ 208 Cal.Rptr. 137, 690 P.2d 610] [in defamation case requiring clear and convincing proof of actual malice, summary judgment to be granted unless it appears plaintiff may be able to meet that standard at trial]; Edwards v. Hall (1991) 234 Cal.App.3d 886, 900 [ 285 Cal.Rptr. 810] [same]; Miller v. Nestande (1987) 192 Cal.App.3d 191, 196 [ 237 Cal.Rptr. 359, 62 A.L.R.4th 301] [same]; Looney v. Superior Court (1993) 16 Cal.App.4th 521, 539-540 [ 20 Cal.Rptr.2d 182] [prima facie case for punitive damages in action against health care providers (Code Civ. Proc., § 425.13) must meet clear and convincing standard].)

(4b) Accordingly, we conclude a nonsuit on the issue of punitive damages is proper when no reasonable jury could find the plaintiff's evidence to be clear and convincing proof of malice, fraud or oppression. On review of a nonsuit order, we apply the same standard.

D. Application to This Case

Malice, for purposes of awarding exemplary damages, includes "despicable conduct which is carried on by the defendant with a willful and conscious disregard of the rights or safety of others." (Civ. Code, § 3294, subd. (c)(1).) (7) To establish conscious disregard, the plaintiff must show "that the defendant was aware of the probable dangerous consequences of his conduct, and that he wilfully and deliberately failed to avoid those consequences. [Citations.]" ( Taylor v. Superior Court (1979) 24 Cal.3d 890, 895-896 [ 157 Cal.Rptr. 693, 598 P.2d 854]; Mock v. Michigan Millers Mutual Ins. Co. (1992) 4 Cal.App.4th 306, 329 [ 5 Cal.Rptr.2d 594].)

(8) Plaintiffs point to the following evidence as supporting a finding of conscious disregard: Syson's testimony the type one buckle is susceptible to inertial unlatching, that the Ford Taurus has a different, nonsusceptible, buckle, and recounting the manufacturer's unfulfilled 1965 promise (the "position paper") to modify buckles so they would not unlatch inertially; Bertelson's testimony regarding the finding of unrestrained dummies in Ford's 1967-1968 crash testing; Davee's testimony, when called as an adverse witness in the plaintiffs' case, that he was unaware of any testing, prior to his own, at Allied-Signal or Jim Robbins, designed to evaluate the possibility of inertial unlatching, and that engineers at Allied-Signal thought it highly unlikely a buckle could open inertially in an accident in the environment of a vehicle.

Even taken as a whole and given its full credit, this evidence could not reasonably be considered clear and convincing proof of malice. Neither the "position paper" (which was not linked specifically to the type one buckle), nor the susceptibility of the buckle to inertial unlatching when manipulated in certain ways outside the vehicle environment, nor the unexplained finding of unrestrained dummies, could be thought clearly to show Allied-Signal or Jim Robbins was aware a probable danger existed that type one buckles would unlatch under the conditions of an actual automobile accident. In the absence of such proof, plaintiffs could not show Allied-Signal acted despicably or in conscious disregard of anyone's safety by failing to conduct a program of testing the type one buckle for inertial unlatching or for selling Ford that buckle for the Bronco II. No reasonable jury could consider this clear and convincing proof of malicious conduct. Nonsuit was therefore proper on this issue.

II. Allied-Signal's Cross-appeal

A., B.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

C. Denial of Setoff for Pretrial Settlements

Before trial, plaintiffs reached good faith settlements with Ford and other defendants, receiving a total of $382,500. At trial, plaintiffs and Allied-Signal stipulated plaintiffs made no claim of economic loss and any jury award would be deemed purely noneconomic. The trial court gave judgment for $175,000, Allied-Signal's proportionate (35 percent) share of the damages found. Allied-Signal urged the judgment against it be reduced to $17,500, which represented the $500,000 in damages found by the jury, reduced by decedent's 20 percent comparative fault and the $382,500 settlement. That motion was denied.

(9a) Allied-Signal and the amicus curiae contend the reduction was required under Code of Civil Procedure section 877, subdivision (a) (hereafter section 877(a)). Plaintiffs argue the judgment is proper under Civil Code section 1431.2, subdivision (a) (hereafter section 1431.2(a)).

Section 877 provides that a good faith settlement by "one or more of a number of tortfeasors claimed to be liable for the same tort . . . [¶] (a) . . . shall not discharge any other such party from liability unless its terms so provide, but it shall reduce the claims against the others in the amount stipulated by the release . . . or in the amount of the consideration paid for it whichever is the greater." (Italics added.)

Section 1431.2(a) provides: "In any action for personal injury, property damage, or wrongful death, based upon principles of comparative fault, the liability of each defendant for non-economic damages shall be several only and shall not be joint. Each defendant shall be liable only for the amount of non-economic damages allocated to that defendant in direct proportion to that defendant's percentage of fault, and a separate judgment shall be rendered against that defendant for that amount." (Italics added.)

On the surface, the two statutes point to opposite results in this case. While section 877(a) apparently commands the trial court to reduce the claim against Allied-Signal by $382,500, section 1431.2(a) mandates the entry of a judgment for noneconomic damages "in direct proportion" to Allied-Signal's fault, to wit, 35 percent of $500,000, or $175,000. We conclude the trial court correctly calculated the judgment under section 1431.2(a). (Accord, Espinoza v. Machonga (1992) 9 Cal.App.4th 268, 271-277 [ 11 Cal.Rptr.2d 498]; In re Piper Aircraft (N.D.Cal. 1992) 792 F. Supp. 1189, 1191-1193.)

Our conclusion is based, first, on the relevant language of section 877(a), which presupposes the existence of multiple defendants jointly liable for the same damages. The settlement by one or more of several tortfeasors "claimed to be liable for the same tort" reduces "the claims against the others." Under the scheme of purely several liability created by section 1431.2(a), however, ". . . a personal injury plaintiff's valid `claim' against one such tortfeasor for noneconomic damages can never be the liability of `the others.'. . . Thus, there is no longer any such claim `against the others' to `reduce.'" ( Espinoza v. Machonga, supra, 9 Cal.App.4th at pp. 274-275.)

This fact can be seen in practice here. Allied-Signal proposes calculating the judgment by subtracting the $382,500 pretrial settlement from $400,000, the amount of damages not attributable to decedent's own negligence. This $400,000 figure, however, does not in any sense constitute a "claim against the others," i.e., a claim against Allied-Signal; plaintiffs could not, under section 1431.2(a), claim Allied-Signal was liable for any noneconomic damages greater than its own proportionate share. Allied-Signal's maximum liability, under section 1431.2(a), is $175,000, yet even Allied-Signal does not contend the court should have deducted $382,500 from this amount, reducing the judgment to zero.

In support of its claim section 877(a) requires a reduction even though multiple tortfeasors are no longer jointly liable for noneconomic damages, Allied-Signal cites statements in various cases to the effect section 877(a) may apply — that is, the defendants may be "claimed to be liable for the same tort" — even if the defendants are not "joint tortfeasors" ( Barney v. Aetna Casualty Surety Co. (1986) 185 Cal.App.3d 966, 983 [ 230 Cal.Rptr. 215]; Lafayette v. County of Los Angeles (1984) 162 Cal.App.3d 547, 554 [ 208 Cal.Rptr. 668]), so long as their actions have combined to cause "one indivisible injury" ( May v. Miller (1991) 228 Cal.App.3d 404, 409 [ 278 Cal.Rptr. 341]). These cases are inapposite. In Barney, the issue was whether the dismissal pursuant to pretrial settlement of one alleged participant in a civil conspiracy defeated the cause of action against the other. The appellate court held that since civil conspirators were contributory wrongdoers, "jointly and severally liable for the same damage," they were subject to the portion of section 877(a) providing the settlement of one did not discharge the other's liability. (185 Cal.App.3d at pp. 982-983, italics added.) In Lafayette, a setoff for settlement was held proper where the settling defendant (an allegedly malpracticing attorney) was claimed to be liable for the same damages allegedly caused his client by the nonsettling defendant (a county hospital). (162 Cal.App.3d at pp. 555-556.) In May, a setoff was approved where the two defendants (an insurer and insurance broker) were joint tortfeasors and there was no evidence they had caused " separate and distinct damages." (228 Cal.App.3d at pp. 409-410, italics added.) Here, in contrast, pursuant to section 1431.2(a), the noneconomic damages for which the settling and nonsettling defendants could be claimed liable were, by law, separate and distinct, allocated according to the defendants' individual fault.

Thus, section 877(a) has no clear application here, because there is no amount that represents a common claim for noneconomic damages against the settling and nonsettling defendants. In contrast, section 1431.2(a) speaks directly and unambiguously to the situation here presented. When a jury allocates comparative fault and awards some amount of noneconomic damages, judgment for those damages "shall be rendered" against each defendant in proportion to its fault. The trial court here did no more than follow that clear statutory command.

(10) Section 1431.2, added by initiative (Proposition 51) in 1986, was intended to make the tort system more equitable by partially eliminating the "deep pocket rule" of joint liability, which sometimes required "a tortfeasor who might only be minimally culpable to bear all of the plaintiff's damages." ( Evangelatos v. Superior Court (1988) 44 Cal.3d 1188, 1198 [ 246 Cal.Rptr. 629, 753 P.2d 585].) In the words of the findings and declaration of purpose included in the initiative, it was designed to ensure "defendants in tort actions shall be held financially liable in closer proportion to their degree of fault." (Civ. Code, § 1431.1) (9b) The judgment here, in which Allied-Signal has been held liable in exact proportion to its fault determined by the jury, fully implements the purpose of the initiative.

The trial court's approach is also fairer to plaintiffs, and more conducive to settlement of claims, than that advocated by Allied-Signal and the amicus curiae. Allied-Signal's approach is unfair to plaintiffs because it would require them to bear the risk of divergence between the settlement and the jury's assessment of the settling party's liability, without allowing plaintiffs to reap the potential benefit of such divergence. For the same reason, it would discourage plaintiffs from settling with less than all defendants. (11) The approach we adopt thus accords as well with the goals of section 877(a), to wit, an equitable sharing of costs among the parties at fault and the encouragement of settlements. ( Arbuthnot v. Relocation Realty Service Corp. (1991) 227 Cal.App.3d 682, 687 [ 278 Cal.Rptr. 135] .)

Under a system of joint and several liability, in which nonsettling defendants are liable for all of the damages unsatisfied by pretrial settlement, the nonsettling defendants bear the risk the settlement was "low" compared to the amount the settling defendant would have been liable for according to the jury verdict. The nonsettling defendant may also obtain a benefit if the settlement was "high" and the settling defendant does not obtain equitable partial indemnity. (See Sears, Roebuck Co. v. International Harvester Co. (1978) 82 Cal.App.3d 492, 497 [ 147 Cal.Rptr. 262].) The plaintiff, in the joint and several liability system, can neither lose nor win by divergence between the settlement and the verdict; whether the settlement was "high" or "low," the plaintiff's potential recovery from all solvent defendants is the same — the damages awarded by the jury. This system also provides all parties reasonable incentives to settle. (See American Motorcycle Assn. v. Superior Court (1978) 20 Cal.3d 578, 603-604 [ 146 Cal.Rptr. 182, 578 P.2d 899].)

(9c) Section 1431.2(a), as applied here by the trial court, reallocates the risks and potential benefits of settlement-verdict divergence as to noneconomic damages, but in a manner that remains fundamentally fair. If the settlement was "low," the plaintiff loses; he or she cannot recover the difference in noneconomic damages from the remaining defendants. If the settlement was "high," as here, the plaintiff wins; he or she retains the benefit of the settlement bargain as well as receiving the amounts allocated by the jury to the nonsettling defendants. The nonsettling defendants bear no risk and can reap no benefit from divergence; the settlement does not affect their liability for noneconomic damages.

Allied-Signal's approach, however, places all of the risk on the plaintiffs and gives all the potential benefit to the nonsettling defendants. If the settlement was "low," the plaintiff will recover less than the noneconomic damages awarded by the jury. If the settlement was "high," the nonsettling defendants will reap the benefit, paying less than their fault-share of the noneconomic damages. This would be inequitable and would provide "little incentive for the injured person to settle with one or fewer than all of the tortfeasors." ( Wilson v. Galt (1983) 100 N.M. 227 [ 668 P.2d 1104, 1109]; see also In re Piper Aircraft, supra. 792 F. Supp. at p. 1192 [allowing setoff "would engender situations in which nonsettling tortfeasors would be able to take advantage of the good faith efforts of settling tortfeasors," and hence would "discourage rather than encourage settlement."].)

The amicus curiae and Allied-Signal assert the trial court's approach awards plaintiffs an "impermissible double recovery" or a "windfall," because plaintiffs' total recovery is greater than the amount of recoverable damages awarded by the jury. This is unpersuasive for several reasons.

First, as already discussed, the limitation of total recovery according to the jury verdict makes sense in a joint and several liability situation, where the plaintiff bears no risk the settlement will turn out to be ungenerous in comparison with the jury's assessment of the settling tortfeasor's fault. Where, however, that risk is placed on plaintiffs, so that they are bound by their pretrial estimate of the settling tortfeasor's proportionate liability, equity demands they also be entitled to retain the benefit of their bargain when the settlement is generous.

The fairness of disallowing a setoff in these circumstances has been recognized in cases from other jurisdictions. Holding a nonsettling tortfeasor was required to pay its proportionate share of the damages despite a settlement that "prove[d] to be more generous than the subsequent verdict" (resulting in a total recovery greater than the jury's verdict), the Supreme Court of Pennsylvania, quoting from a Texas case, explained, "`A percent credit necessarily means that settling plaintiffs may recover more than the amount of damages ultimately determined, but they also may recover less. Plaintiffs bear the risk of poor settlements; logic and equity dictate that the benefit of good settlements should also be theirs.'" ( Charles v. Giant Eagle Markets (1987) 513 Pa. 474 [ 522 A.2d 1, 3] (lead opn. of Nix, C.J.), quoting Duncan v. Cessna Aircraft Co. (Tex. 1984) 665 S.W.2d 414, 430; accord, Wilson v. Galt, supra, 668 P.2d at p. 1109.)

Second, the idea of a "double recovery" is inextricably linked to the joint liability of multiple tortfeasors. (12) When multiple defendants are responsible for the same compensatory damages, a setoff is not only mandated under section 877(a), but is required by the fundamental principle that ". . . a plaintiff may not recover in excess of the amount of damages which will fully compensate him for his injury. [Citations.]" ( Jaramillo v. State of California (1978) 81 Cal.App.3d 968, 970 [ 146 Cal.Rptr. 823].) The plaintiff cannot, by proceeding separately against each of several defendants, "`convert a joint into a several [injury], and thereby secure more than one compensation for the same injury.'" ( May v. Miller, supra, 228 Cal.App.3d at p. 410, quoting Butler v. Ashworth (1895) 110 Cal. 614, 618 [43 P. 4].) In a case of joint liability, the damages for which each defendant is responsible to the plaintiff cannot be divided, and a settlement is rationally assumed to be intended to cover the entire damages.

That assumption loses its force under a scheme of several liability allocated by fault. "[T]he one recovery rule was originally adopted because the courts could not conceive of allocating liability. Injuries were considered indivisible; therefore, a settling defendant could only offer to pay for the whole injury, not just his part. . . . [¶] The reasoning behind the one recovery rule no longer applies. . . . Because each defendant's share can now be determined, it logically follows that each may settle just that portion of the plaintiff's suit. The settlement does not affect the amount of harm caused by the remaining defendants and likewise should not affect their liability." ( Duncan v. Cessna Aircraft Co., supra, 665 S.W.2d at p. 431.)

(9d) Although decedent's death is, literally, a single injury, the noneconomic harm caused plaintiffs is not indivisible; rather, by virtue of section 1431.2(a), it must be divided among the multiple tortfeasors according to their shares of fault. As far as the record shows, the settlement agreement did not recite an intent to fully compensate plaintiffs by the settlement; nor did plaintiffs agree any award against the nonsettling defendant, Allied-Signal, would be offset by the settlement amount. (See Wilson v. Galt, supra, 668 P.2d at pp. 1109-1110 [plaintiffs bound by such an agreement even where they would otherwise be entitled to judgment in amount of nonsettling defendant's proportionate liability].) The notes of the settlement conference judge, read into the record by the trial judge, indicate only that the judge regarded the question of a setoff as a "big issue."

Finally, comparison of plaintiffs' total recovery to the jury's award is potentially misleading. "[S]ettlement dollars are not the same as damages. Settlement dollars represent a contractual estimate of the value of the settling tortfeasor's liability and may be more or less than the proportionate share of the plaintiff[']s damages. The settlement includes not only damages, but also the value of avoiding the risk, expense, and adverse public exposure that accompany going to trial. There is no conceptual inconsistency in allowing a plaintiff to recover more from a settlement or partial settlement than he could receive as damages. [Citations.]" ( Duncan v. Cessna Aircraft Co., supra, 665 S.W.2d at pp. 431-432.) That Ford was willing to make a generous settlement in order to avoid a public trial on allegations of defects in its Bronco II does not change either the jury's assessment of total damages or its allocation of fault to Allied-Signal.

The amicus curiae urges us to follow Curtis v. Canyon Highway Dist. No. 4 (1992) 122 Idaho 73 [ 831 P.2d 541]. Although Curtis held a setoff proper under statutes whose wording is in significant respects similar to that of sections 877(a) and 1431.2(a) (831 P.2d at pp. 545-547), we find it unpersuasive. The Curtis court did not discuss any of the equitable or policy considerations that we have considered above. More important, in concluding a statute eliminating joint and several liability in certain classes of cases created only a "cap" on the damages recoverable against an individual defendant, the court failed to give effect to the plain language of the statute, which stated that judgment against each individual defendant "shall be entered" in proportion to comparative fault. ( Id. at p. 547.) Section 1431.2(a), as already discussed, similarly states a judgment for noneconomic damages in proportion to individual fault "shall be rendered," and the trial court correctly followed that command.

D. Allowance of Plaintiffs' Expert Witness Fees

Before trial, on August 8, 1991, plaintiffs offered, pursuant to Code of Civil Procedure section 998 (hereafter section 998), to settle the case against Allied-Signal for $100,000 for each of the three plaintiffs (Carl Hoch, Norma Hoch, and Carl Hoch as special administrator of decedent's estate). The offer was silent as to whether it included a waiver of plaintiffs' costs. The offer was rejected.

(13) The trial court determined plaintiffs Carl and Norma Hoch were entitled to recover their expert witness fees under section 998 because the judgment in their favor, consisting of the verdict of $175,000 plus $30,561.17 in ordinary statutory costs, was greater than the settlement of $200,000 they offered to accept. Allied-Signal contends that ruling was incorrect because the comparison should have been made without addition of costs to the verdict.

Both parties rely on Stallman v. Bell (1991) 235 Cal.App.3d 740 [286 Cal.Rptr. 755], which held the "judgment" to be compared, under section 998, subdivision (d), to the plaintiff's rejected offer should include the costs due the prevailing party under Code of Civil Procedure section 1032, subdivision (b). (235 Cal.App.3d at pp. 747-750.) Because in Stallman the rejected offer included a waiver of costs ( id. at p. 743), Allied-Signal cites it for the proposition costs are to be included in the comparison judgment only where they were waived in the offer. We reject that reading; Stallman, as the trial court here noted, held addition of costs was appropriate despite the offer's cost waiver, not because of it. ( Id. at pp. 748-749.)

Allied-Signal points out that acceptance of a section 998 settlement offer without an express cost waiver does not preclude a plaintiff from seeking statutory costs under Code of Civil Procedure section 1032 ( Rappenecker v. Sea-Land Service, Inc. (1979) 93 Cal.App.3d 256, 261-264 [ 155 Cal.Rptr. 516].) This suggests the presence or absence of a cost waiver should affect the value placed on a settlement proposal for purposes of the comparisons described in section 998, but does not indicate the costs provision affects the value of the judgment actually obtained through trial. As the court said in Stallman v. Bell, supra: "We agree that, under section 998, the costs provision in an offer should be taken into account to determine the amount of the offer for purposes of comparing that amount to the amount of the judgment. It does not follow, however, that the costs provision in the offer should determine what costs are added to the award of damages in order to arrive at the amount of the judgment for purposes of section 998" (235 Cal.App.3d at p. 750.) We conclude the trial court correctly added plaintiffs' ordinary statutory costs to the jury's award in order to determine the amount of the judgment under section 998, subdivision (d).

DISPOSITION

The judgment is modified to award judgment of nonsuit in favor of defendant on the survival cause of action for punitive damages. As amended, the judgment is affirmed. Allied-Signal's application for writ of error coram vobis is denied.

White, P.J., and Merrill, J., concurred.

The petition of defendant and appellant for review by the Supreme Court was denied July 28, 1994. Werdegar, J., did not participate therein.