Concurring. I concur in the judgment. The City of Los Angeles (the City) requested a jury instruction that virtually guaranteed it would be held liable for the rape by Officer Schroyer. The City should not now be heard to complain that the jury’s verdict was erroneous.
I respectfully disagree, however, with the majority’s reasoning and conclusion on the substantive question of vicarious liability. The majority presents at length its policy views on why governments should be strictly liable for the crimes of their police officers. However, these observations are largely irrelevant. The Legislature has prohibited such liability without fault except where a public employee was acting “within the scope of . . . employment.” (Gov. Code, § 815.2, subd. (a).) The narrow issue in this case is whether an officer who deviates from duty and commits criminal acts entirely unrelated to his law enforcement responsibilities can ever be deemed “in the scope of . . . employment.” For reasons I will explain, the answer to that question is “no.”
*225Invited Error
I. The rule of invited error should bar the City’s attack on the jury’s verdict.
Special instruction No. 3, requested by the City, stated: “An employer is liable for the wrongful acts of a police officer who, as a result of the exercise of his authority, legally causes injury even though the wrongful acts occurred without the employer’s knowledge, were not related to the duties he was hired to perform, were not for the benefit of employee, and were done in violations [sic] of the employer’s rules or grant of authority.” A reasonable jury faced with this instruction would be hard pressed not to find vicarious liability. The components of the instruction bear emphasis. The jury was told the City was vicariously liable for the rape by Officer Schroyer even if:
a. It occurred without the City’s knowledge;
b. It was not related to Officer Schroyer’s duties;
c. It was not for the City’s benefit;
d. It was solely for the personal benefit of Officer Schroyer; and
e. It violated the City’s rules.
Under the City’s jury instruction, almost any imaginable form of police misconduct would support a finding of vicarious liability. If, for example, Officer Schroyer had “exercise[d] ... his authority” by robbing a bank while on duty, his misconduct would equally have met the criteria for vicarious liability set forth in the City’s instruction.
The City acknowledges the well-established rule of invited error. “Under the doctrine of ‘invited error’ a party cannot successfully take advantage of error committed by the court at his request.” (Jentick v. Pacific Gas & Elec. Co. (1941) 18 Cal.2d 117, 121 [114 P.2d 343].) The rule precludes a party from challenging a jury instruction if he proposed it or a similar instruction. (Ibid.; Weirum v. RKO General, Inc. (1975) 15 Cal.3d 40, 50 [123 Cal.Rptr. 468, 539 P.2d 36].) The City does not dispute that special instruction No. 3 was highly unfavorable to the City’s own position at trial.
The City, however, contends it should not be held accountable for special instruction No. 3 because the City was merely trying to “make the best of a bad situation.” According to the City, the trial court had informed counsel for *226both parties that it would instruct the jury on the vicarious liability issue pursuant to the decision in White v. County of Orange (1985) 166 Cal.App.3d 566 [212 Cal.Rptr. 493] (White), in which the Court of Appeal held that a female motorist had stated a valid cause of action for vicarious liability against a county based on an alleged sexual assault by a deputy sheriff. The City asserts it objected on two grounds to an instruction proposed by plaintiff under White: (1) White should not be followed; and (2) the instruction proposed by plaintiff did not accurately reflect the holding in White. The City claims the trial court made clear its intention to give plaintiff’s instruction unless the City submitted its own instruction under White. The City contends it should therefore be excused from having submitted the instruction that virtually assured an unfavorable verdict.
The City’s argument on this point is flawed in two key respects. First, even if we accept as true the City’s recollection of the facts, the City failed to object on the record to any proposed instruction on the vicarious liability issue until after the jury was instructed. A party must not be allowed to submit a crucial jury instruction and then, after the jury has been instructed and retired to deliberate, attempt for the first time on the record to make excuses for its own proposed instruction. Put simply, a party should not be allowed to create an after-the-fact objection under the guise of “clarifying” the record. It is hornbook law that an appellant must affirmatively show error by an adequate record. (9 Witkin, Cal. Procedure (3d ed. 1985) Appeal, § 418, pp. 415-416.)
Faced with a dispute over jury instructions, the City should have provided for a contemporaneous record of the discussions between counsel and the trial court. Alternatively, the City could have moved under rule 36(b) of the California Rules of Court for a settled statement in place of a transcript of the alleged chambers conference regarding jury instructions. (Lipka v. Lipka (1963) 60 Cal.2d 472, 480-481 [35 Cal.Rptr. 71, 386 P.2d 671] [allowing augmentation of record to include unreported chambers conference].) The City did neither. Accordingly, the record does not show that the City made a timely objection to the proposed instruction on vicarious liability.
Second and equally important, the majority fails to grasp the fact that the fatal instruction was drafted by the City. When the City first raised its belated objection to an instruction based on White, supra, 166 Cal.App.3d 566, the trial court correctly observed that, “Now, however, on this instruction that you prepared, and which you felt was the law in accordance to [szc] White, the language was entirely your own and the court had no input nor did plaintiff’s counsel.” Faced with an instruction it did not like, the City had two choices: (1) object to the instruction and stand on its objection without submitting a different instruction; or (2) object to plaintiff’s instruction and *227submit an alternative one that the City believed to be a correct statement of the law. A party may not submit its own instruction and then challenge it as being incorrect. That is the essence of the invited-error rule. (Jentick v. Pacific Gas & Elec. Co., supra, 18 Cal.2d 117, 121.)
The City cites no authority for its novel proposition that the City may challenge its own instruction as an attempt to make the best of a bad situation.” The rule is to the contrary. In Jentick v. Pacific Gas & Elec. Co., supra, 18 Cal.2d 117, the court rejected the vicariously liable defendant’s argument that its request for an erroneous instruction was not willful. “Defendant may not avoid the application of the doctrine by asserting that the error was not deliberately or willfully induced. The good faith of the defendant is immaterial. It is incumbent upon counsel to propose instructions that do not mislead a jury into bringing in an improper verdict. Whether deliberate or not, defendant’s action was responsible for the erroneous instruction and verdict. Defendant must therefore accept them as correct.” (Id., at p. 122.) Either the same result should obtain in this case, or the majority should forthrightly overrule Jentick.
II. Reaching the merits of the vicarious liability issue serves little purpose and will create confusion.
Plaintiff has only one interest in this court: obtaining an affirmance of the monetary judgment in her favor. We can, and should, grant her that relief on the basis of the City’s invited error on the jury instruction. Any discussion of whether vicarious liability should arise in future cases serves no purpose for plaintiff. Yet the majority insists on a broad and potentially mischievous holding that local governments may be liable without fault if a police officer commits a crime that is somehow related to the authority wielded by virtue of peace officer status.
One must keep in mind the precise disposition of this case. The verdict against the City was returned pursuant to an instruction that the City was vicariously liable for the rape by Officer Schroyer because it was a “result of the exercise of his authority,” even if it occurred without the City’s knowledge, was not related to Officer Schroyer’s duties, was not for the City’s benefit, was solely for the personal benefit of Officer Schroyer, and violated the City’s rules. There is no claim that the City was negligent in hiring Schroyer or had reason to know that he might take advantage of his position of authority to commit rape. The sole basis on which the City’s liability is predicated is that he acted within the scope of his employment while committing a rape.
Despite the majority’s effort to suggest some limitations on its holding, the practical result is clear: no matter how attenuated the relationship *228between police misconduct and an officer’s employment, if he takes advantage of the authority he acquires as an officer in order to commit the crime, he may be found to be acting within the scope of his employment, and the City will be liable for any damages he causes. This is an unprecedented expansion of liability which is unauthorized by the controlling governmental immunity statutes.
For the foregoing reasons, I would decide this case in plaintiff’s favor solely on the ground of the City’s invited error. Because the majority, however, decides the broader scope of employment issue, I address that too.
Vicarious Liability
I respectfully disagree with the majority’s reasoning and conclusion that the City may be held vicariously liable for the injury caused by Officer Schroyer’s criminal conduct. The majority imposes on the taxpayers of the City the financial responsibility for a rape committed by a police officer for his own gratification. No act can be more unrelated to the duties of a police officer than his rape of a member of the public he is sworn and paid to protect. The majority admits, as does the plaintiff, that the City was blameless. The proposed rule is therefore sweeping. Taxpayers may be strictly liable for almost any abuse of position by a police officer no matter how unrelated it is to his or her proper duties. I share the urge to make plaintiff whole—assuming that money can ever erase her pain. No compassionate person can escape outrage at the harm caused by this errant officer. This court’s proper function, however, is not to search for deep financial pockets regardless of the law or practical consequences.
III. The City’s liability is governed, by statute.
The stated cornerstone of the majority opinion is its view that, “The cost resulting from misuse of that [police] power should be borne by the community . . . .” (Maj. opn., ante, at p. 217.) The question of the City’s liability is not a matter of judicial preference. The Legislature has enacted a comprehensive statutory system regulating the liability of public entities. (Gov. Code, § 810 et seq.) Under this scheme, “[a] public entity is liable for injury proximately caused by [the actionable misconduct] of an employee . . . within the scope of.. . employment . . . .” (Id., § 815.2, subd. (a), italics added.) “Except as otherwise provided by statute[,] . . . [f] [a] public entity is not liable for an injury . . . .” (Id., § 815, subd. (a).) “Governmental] tort liability in California is governed completely by statute.” (Swaner v. City of Santa Monica (1984) 150 Cal.App.3d 789, 797 [198 Cal.Rptr. 208].) “[T]he practical effect of [Government Code section 815] is to eliminate any common law governmental liability for damages arising out *229of torts.” (Sen. legis. committee com., 32 West’s Ann. Gov. Code (1980 ed.) § 815, p. 168 [Deering’s Ann. Gov. Code (1982 ed.) § 815, p. 134], italics added.)
The Legislature’s intent to circumscribe liability is clear. In Muskopf v. Corning Hospital Dist. (1961) 55 Cal.2d 211 [11 Cal.Rptr. 89, 359 P.2d 457], this court attempted to abrogate the entrenched doctrine of governmental tort immunity. The Legislature promptly responded by enacting the Moratorium Act of 1961 (former Civ. Code, § 22.3), which suspended the effect of Muskopf and reinstated the immunity. (Stats. 1961, ch. 1404, pp. 3209-3210.) At the Legislature’s request, the California Law Revision Commission submitted a comprehensive report in 1963, which gave rise to the statutory system that now governs the field of public entity tort liability. (Although not officially titled, the legislation is commonly referred to as the Tort Claims Act.) This history makes clear that the Legislature was unwilling to accept the judicial expansion of tort liability attempted by the Muskopf court.
Professor Arno Van Alstyne was the California Law Revision Commission’s chief consultant and much of his work gave rise to the present statutory system. He has explained that, “These provisions were intended to ensure that applicable immunity provisions of the California Tort Claims Act will generally prevail over its liability provisions.” (Van Alstyne, Cal. Government Tort Liability Practice (Cont.Ed.Bar 1980) § 2.26, p. 67.) We have also noted the restrictive nature of the act. “[T]he intent of the act is not to expand the rights of plaintiffs in suits against governmental entities, but to confine potential governmental liability to rigidly delineated circumstances: immunity is waived only if the various requirements of the act are satisfied.” (Williams v. Horvath (1976) 16 Cal.3d 834, 838 [129 Cal.Rptr. 453, 548 P.2d 1125] (opn. by Mosk, J.).)
The plain language of the act supports a restrictive view. Government Code section 815 provides an immunity except as “provided by statute.” (Ibid., italics added.) The Legislature did not provide for exceptions as provided by “law,” which would have included court decisions. (Gov. Code, § 811.) This limitation also reflects the Legislature’s awareness that questions of public entity liability are policy and fiscal questions better left to the Legislature than to the courts. The Legislature’s intent to restrain judicial expansion of liability is made even clearer by its observation that, “The use of the word ‘tort’ had been avoided, however, to prevent the imposition of liability by the courts by reclassifying the act causing the injury.” (Sen. legis. committee com., 32 West’s Ann. Gov. Code, supra, § 815, p. 168 [Deering’s Ann. Gov. Code, supra, § 815, p. 134], italics added.) The California Law Revision Commission further explained the problem of undue judicial inter*230ference: “Experience in states which have left the limits of liability to be determined by the courts has shown that liability insurance to protect the financial integrity of small public entities is at times prohibitively expensive or impossible to obtain when there is no defined limit to the potential extent of liability.” (Recommendations Relating to Sovereign Immunity, No. 1, Tort Liability of Public Entities and Public Employees (Jan. 1963) 4 Cal. Law Revision Com. Rep. (1963) pp. 808-809 (hereafter Law Revision Commission Recommendations).)
The majority asserts that the statutory phrase “scope of employment” imports “general standards” of respondeat superior law into the Tort Claims Act and has been construed as broadly as the similar test used in private tort litigation. (Maj. opn., ante, at pp. 209-210.) Yet the majority applies to financially pressed local governments a startling and unwarranted expansion of the traditional respondeat superior doctrine.
It is ancient law that “[a] master is not liable for a crime or wilful injury, such as an assault, committed by the servant without his command or encouragement, though it may be in the course of, or in relation to, the service.” (2 Stephen, New Commentaries on the Laws of England (1843) p. 278, italics added.) Moreover, under general tort law, an employee’s injurious conduct arises in the “scope of employment” for purposes of vicarious liability where the conduct was “typical,” “usual,” “broadly incidental,” or “inherent” in the employer’s enterprise, but not where the conduct was so “unusual or startling,” or constituted such a “[substantial deviation] from [the employee’s] duties for personal purposes,” that “ ‘it would seem unfair to include the [resulting] loss . . . among other costs of the employer’s business. . . .’” (Perez v. Van Groningen & Sons, Inc. (1986) 41 Cal.3d 962, 968 [227 Cal.Rptr. 106, 719 P.2d 676]; see Hinman v. Westinghouse Elec. Co. (1970) 2 Cal.3d 956, 960 [88 Cal.Rptr. 188, 471 P.2d 988].)
Aside from his original detention of the intoxicated victim, Officer Schroyer’s criminal attack had no relation whatever to the performance of his law enforcement duties. Rather, he deviated completely from his work assignment, in a manner all must concede was both “startling” and “unusual,” to commit a sexual assault for personal gratification. Absent supportive legislation, such an outrageous sexual attack cannot be deemed an “inherent” or “broadly incidental” risk of law enforcement which the taxpayers should absorb as a cost of government. I cannot square the majority’s radical departure from traditional respondeat superior law with the purposes of the Tort Claims Act.
*231IV. The majority’s extensive reliance on public policy is misplaced and unsupported.
The majority considers at some length various “policy factors.” This reliance on public policy is both unsupported by evidence and legally misplaced. The governing statutory scheme precludes us from imposing vicarious liability on a public employer as a matter of “policy.” Liability may be imposed only for an employee’s actionable misconduct “in the scope of . . . employment.” The only issue presented is when, if ever, a police officer’s intentional criminality can fairly be deemed “in the scope” of the officer’s employment. Intentional criminal conduct entirely beyond the scope of an officer’s law enforcement duties cannot meet that test.
A. Statutory nature of question
As noted above, we are restricted to deciding this case in light of the comprehensive statutory scheme that governs public entity liability. In appropriate cases, consideration of public policy may assist the court in construing a statute. However, because the clear legislative intent was to restrict government’s liability, this court should not impose liability absent a clear indication the Legislature intended such result. The majority does not undertake such an analysis and refers only tangentially to the statutes. Its opinion offers no reasoned basis to conclude that the Legislature intended to bring all criminal misuse of an officer’s status, power, or authority, however flagrantly unrelated to duty, within the “scope of [the officer’s] employment.”
Even assuming we were free to resolve the policy question, I am troubled by the majority’s incomplete discussion of the competing public policies. Whether plaintiff should recover for her injuries is only one side of the equation. The other side is whether the taxpayers of the City should be forced to pay for those injuries. The public fisc is not infinite. To the contrary, in this era of limited public resources, every expenditure for one purpose requires a withdrawal of funds for another purpose. Compensating the plaintiff is a worthy and sympathetic goal. Whether it is more worthy than other public purposes is a question beyond our right or ability to answer. Professor Van Alstyne has testified that, “[T]he costs and the funding problems are one of the most difficult problems in the whole field of tort liability ... in the area of government torts particularly . . . .” (Hearings on Government Liability Before the Joint Com. on Tort Liability (Oct. 31, 1977) p. 33.)
The inescapable truth is that in the modern era, payments from the public purse involve hard choices of priorities. For example, in 1986 the voters *232enacted Civil Code section 1431.1 to restrict liability for noneconomic damages. The voters’ findings and declaration of purpose stated, “Local governments have been forced to curtail some essential police, fire, and other protections because of the soaring costs of lawsuits and insurance premiums.” (Civ. Code, § 1431.1, subd. (c).) The effect of tort judgments on public resources is significant. A court should not ignore fiscal reality when expanding the frontier of tort liability. If a court wishes to sit as a “super-Legislature,” the court should wrestle with the same vexing problems that arise in the legislative arena and should be subject to the same electoral pressures faced by legislators.
Of course, the Legislature (or the electorate itself) is best equipped to consider empirical evidence, e.g., the frequency of police rape, and to make the hard choices as to where public money will be spent. (Brown v. Kelly Broadcasting Co. (1989) 48 Cal.3d 711, 740 [257 Cal.Rptr. 708, 771 P.2d 406] [leaving consideration of empirical data to the Legislature]; J. C. Penney Casualty Ins. Co. v. M. K. (1991) 52 Cal.3d 1009, 1028 [278 Cal.Rptr. 64, 804 P.2d 689] [noting importance of Legislature’s consideration of empirical data].) We would do well to pay heed to the observation of an English jurist that public policy “is a very unruly horse, and when once you get astride it you never know where it will carry you.” (Richardson v. Mellish (1824 Bing.) 103 Eng. Rep. 294, 303.) Courts should be extremely reluctant to decide for the public how its money should be spent. (Sands v. Morongo Unified School Dist. (1991) 53 Cal.3d 863, 941 [281 Cal.Rptr. 34, 809 P.2d 809] (dis. opn. of Baxter, J., noting importance of not interfering with community-based decisions).)
The majority’s legislative decision to allocate public funds is especially bothersome in light of the absence of any factual support for many of the majority’s critical assumptions. The majority cites no evidence for its sweeping pronouncements that vicarious employer liability for police sexual misconduct will encourage preventive measures that do not hinder the vital law enforcement function. Indeed, both common sense and prior commentary by this court (see discussion, post) suggest the contrary.
B. Source of the majority’s policy factors
The majority relies almost entirely on policy factors set forth in the lead opinion in John R. v. Oakland Unified School Dist. (1989) 48 Cal.3d 438 [256 Cal.Rptr. 766, 769 P.2d 948] (John R.).) This reliance is flawed, even puzzling, in several respects.
1. The John R. court, supra, 48 Cal.3d 438, declined to impose vicarious liability for a sexual assault by a schoolteacher on a pupil. A *233decision rejecting vicarious liability provides questionable support for an expansion of such liability. Even one of the two dissenters on the liability issue observed that, “[Vicarious liability for sexual assaults should be recognized as the exception, not the rule.” (Id., at p. 465 (conc, and dis. opn. by Kaufman, J.).) Another critic of the John R. decision properly called it “an extraordinarily broad rule” against vicarious liability. (Kimberly M. v. Los Angeles Unified School Dist. (1989) 215 Cal.App.3d 545, 550 [263 Cal.Rptr. 612] (conc. opn. of Johnson, J.).) Moreover, when John R. was decided, we had already granted review in this case. To make clear that we were not prejudging this case, the lead opinion in John R. stressed that we were not deciding whether a prior Court of Appeal decision imposing vicarious liability for a sexual assault by a police officer “. . . was properly decided or whether the job-created authority theory has any validity in evaluating vicarious liability for the torts of police officers.” (48 Cal.3d at p. 452 (lead opn. by Arguelles, J.), italics added.) The lead opinion could not have made more clear that we were not deciding the issue now before us. (Liu v. Republic of China (9th Cir. 1989) 892 F.2d 1419, 1431 [noting that we “specifically declined” in John R., supra, 48 Cal.3d 438, to decide the scope of vicarious liability for police misconduct].)
2. In light of the express disclaimer in John R., supra, 48 Cal.3d 438, 452, that we were not deciding the issue of vicarious liability for police rape, any observations in the lead opinion were the barest of dictum, if even that. We have made clear that, “ ‘[T]he language of an opinion must be construed with reference to the facts presented by the case, and the positive authority of a decision is coextensive only with such facts.’ ” (Brown v. Kelly Broadcasting Co., supra, 48 Cal.3d 711, 734-735, quoting River Farms Co. v. Superior Court (1933) 131 Cal.App. 365, 369 [21 P.2d 643].) This is especially so when a court takes pains to stress the narrowness of its decision. The John R. lead opinion, supra, 48 Cal.3d 438, raised a red flag to warn against subsequent reliance on the decision. The present majority takes that flag and lofts it as a standard for a view the John R. court never adopted.
3. Most important, the policy discussion in John R., supra, 48 Cal.3d 438, was in an opinion signed by only two justices of this court. (There were four separate opinions.) Even the other three justices who agreed there should be no vicarious liability declined to sign the portion of the lead opinion dealing with that issue. Instead, they chose to make clear that they concurred only “in the majority’s holding” of no vicarious liability. (Id., at p. 455 (conc, and dis. opn. of Eagleson, J.), italics added.) Except to its precise holding of no liability, the lead opinion stated a minority view and provides no authority for any proposition in a subsequent case. (County of San Mateo v. Dell J. (1988) 46 Cal.3d 1236, 1241, fn. 5 [252 Cal.Rptr. 478, 762 P.2d 1202]; Farrell v. Board of Trustees (1890) 85 Cal. 408, 416 *234[24 P. 868].) This is hornbook law. “No opinion has any value as a precedent on points as to which there is no agreement of a majority of the court.” (9 Witkin, Cal. Procedure, Appeal, supra, § 808, at p. 788.)
4. Even if we were to look to John R., supra, 48 Cal.3d 438, for the type of general guidance we might seek in a treatise or plurality opinion, it would not support the majority’s expansion of vicarious liability. The policy factors noted in the lead opinion weigh against vicarious liability in this case. The majority, however, applies them erroneously and inconsistently to impose liability.
a. Allocation of the risk
The modern justification for vicarious liability is “ ‘ “a deliberate allocation of a risk.” ’ ” (John R., supra, 48 Cal.3d 438, 450, quoting Hinman v. Westinghouse Elec. Co., supra, 2 Cal.3d 956, 959-960.) The John R. lead opinion acknowledged that society benefits from the authority placed in teachers and noted that, “[I]t can be argued that the consequences of an abuse of that authority should be shared on an equally broad basis.” (48 Cal.3d at p. 452.) The lead opinion, however, concluded this factor weighed against vicarious liability because the connection between the authority conferred on teachers and the abuse of that authority by engaging in sexual misconduct is too attenuated to allocate the risk to the employer. That conclusion equally weighs against liability in this case.
Rather than relying on the result in John R., supra, 48 Cal.3d 438, or even the lead opinion’s application of this factor, the present majority relies heavily on a statement in the lead opinion that the authority of a police officer “plainly surpasses that of a teacher over a student.” (Id., at p. 452.) This statement is unpersuasive: (i) It was a passing observation in dictum, (ii) It was in an opinion of only two justices, (iii) The court was fully aware lilis case was pending when we decided John R. As explained above, the lead opinion expressly stated we were not deciding “whether the job-created authority theory has any validity in evaluating vicarious liability for the torts of police officers.” (Ibid.) In light of these multiple limitations and disclaimers, it would be hard to find a more slender reed on which to conclude that John R. supports vicarious liability in this case.
Moreover, I am not persuaded by the speculation in the John R., supra, 48 Cal.3d 438, lead opinion and the present majority opinion that a police officer’s authority “plainly surpasses that of a teacher over a student.” (Id., at p. 452.) The majority’s discussion, like the lead opinion in John R., fails to provide support for this assertion, and common sense suggests to the contrary. A schoolteacher alone at his home with an impressionable child has as *235much power and opportunity to commit a sexual assault against the child, especially one of tender years, as a police officer has to commit an assault against a citizen. Justice Kaufman pointed out in John R., supra, that the circumstances of the case “virtually guaranteed that the teacher could act with impunity . . . .” (Id., at p. 465 (conc, and dis. opn. of Kaufman, J.), original italics.) A teacher may have even greater apparent “authority” than a police officer. None of the indicia of police power cited by the majority—the uniform, badge, and gun—creates any appearance that the officer has the authority to rape. Plaintiff did not believe Officer Schroyer was authorized to have sexual intercourse with her. To the contrary, she struggled to avoid being raped. A young child, however, may be induced to submit to a teacher’s sexual depravity by being led to believe that the teacher has the authority to commit sex acts.
The allocation of risk, or loss spreading as it is sometimes called, should be reasonable and informed as well as deliberate. The decision whether to impose liability requires a delicate balancing of competing interests, particularly when the defendant at law is a public entity and the defendants in fact are the taxpayers. The determination is best left to the Legislature. Neither of the decisions on which the John R. lead opinion relied for the notion of risk allocation involved governmental entities. (Hinman v. Westinghouse Elec. Co., supra, 2 Cal.3d 956; Perez v. Van Groningen & Sons, Inc., supra, 41 Cal.3d 962.)
In its comprehensive study that gave rise to the Tort Claims Act, the California Law Revision Commission explained, “The problems involved in drawing standards for governmental liability and governmental immunity are of immense difficulty. Government cannot merely be made liable as private persons are, for public entities are fundamentally different from private persons. . . . Private persons do not prosecute and incarcerate violators of the law .... Unlike many private persons, a public entity often cannot reduce its risk of potential liability by refusing to engage in a particular activity, for government must continue to govern and is required to furnish services that cannot be adequately provided by any other agency.” (Law Revision Com. Recommendations, supra, at p. 810.) The California Law Revision Commission and the Legislature required enormous amounts of empirical data and many months of collective consideration to reach difficult decisions. The majority acknowledges no difficulty whatsoever and gives no consideration to the potential effects of imposing strict liability on the City.
The notion of risk allocation merits special mention in another regard. We have long emphasized that one factor to be considered in determining whether to impose a particular type of tort liability is “the availability, cost, *236and prevalence of [liability] insurance for the risk involved.” (Rowland v. Christian (1968) 69 Cal.2d 108, 113 [70 Cal.Rptr. 97, 443 P.2d 561, 32 A.L.R.3d 496].) The lead opinion in John R., supra, 48 Cal.3d 438, reiterated this concern: “The imposition of vicarious liability on school districts for the sexual torts of their employees would tend to make insurance, already a scarce resource, even harder to obtain and could lead to the diversion of needed funds from the classroom to cover claims.” (Id., at p. 451.) The high cost and widespread unavailability of municipal liability insurance have been widely reported and studied. (See, e.g., Hearings on Municipal Liability Insurance (Dec. 1975) Before the Joint Assem. Corns, on Finance, Insurance, and Commerce and Local Government; Hearings on Liability Insurance: Threat to the California Dream (Aug. 1986) Before the Sen. Com. on Insurance, Claims, and Corporations; California Citizens’ Commission on Tort Reform, Staff Background Paper: Government Liability (1977) pp. 24-25.) The unavailability of public liability insurance reached such crisis proportions that in 1986 it became one of the key arguments in favor of Proposition 51, which the voters enacted to restrict the liability of defendants (including public entities) for noneconomic injuries. (See Ballot Pamp., argument in favor of Prop. 51 as presented to the voters, Primary Elec. (June 3, 1986) p. 34.) The majority, however, gives no consideration to this traditionally recognized factor.
Our proper function is not to usurp the Legislature’s budgetary function of allocating risk for public entity torts. Even if the question were ours to answer, we do not have before us sufficient empirical data on which to make the difficult choice between competing fiscal priorities.
b. Imposing liability on the employer to prevent recurrence of the tortious conduct
The John R. lead opinion remarked that vicarious liability can be “a spur toward accident prevention.” (48 Cal.3d at p. 451.) On the other hand, John R. recognized that a public entity must not be presented with such an onerous, impossible, or impractical prevention burden that its proper functions are threatened. (Ibid.) This latter principle was a cornerstone of the Tort Claims Act. The California Law Revision Commission emphasized that, “The basic problem is to determine how far it is desirable to permit the loss distributing function of tort law to apply to public entities without unduly frustrating or interfering with the desirable purposes for which such entities exist." (Law Revision Com. Recommendations, supra, at p. 810, italics added.)
Rape, of course, is no accident. It results from an individual’s conscious decision to commit the outrageous act despite all moral and legal sanctions. *237Hence, it cannot be prevented in the way a city might train its officers in safe driving. Rape is a serious crime punishable by imprisonment (Pen. Code, §§261, 264), and a compensable civil wrong as well. We assume such considerations informed the John R. lead opinion’s observation that prevention and deterrence “[play] little role in the allocation of responsibility for the sexual misconduct of employees generally . . . .” (48 Cal.3d at p. 451.)
Here there is no suggestion that the City negligently failed to screen Officer Schroyer’s background and character, or that it failed to exercise due care in training and supervising him. The majority fails to explain what additional measures the City could or should practically have taken to prevent his intentional sexual misconduct. Nor have we any evidence about the costs or benefits of any such measures. Indeed, as the John R. lead opinion observed, excessive restrictions on contacts between public employees and citizens are likely to undermine the employees’ public function. (48 Cal.3d at p. 451.) Common sense suggests that what was true for education in John R. has equal or greater validity in the context of law enforcement.
The premise that the City should adopt further regulations for police training and conduct also runs afoul of Government Code section 818.2. Section 818.2 provides that “[a] public entity is not liable for an injury caused by adopting or failing to adopt an enactment or by failing to enforce any law.” The term “enactment” includes ordinances and regulations. (Gov. Code, § 810.6.)
The majority’s inability to suggest how vicarious liability might deter sexual misconduct by public employees demonstrates that we are ill equipped to dictate such matters. As the California Law Revision Commission explained, “in many cases decisions made by the legislative and executive branches should not be subject to review in tort suits for damages, for this would take the ultimate decision-making authority from those who are responsible politically for making the decisions.” (Law Revision Com. Recommendations, supra, at p. 810.) “The remedy for officials who make bad law, who do not adequately enforce existing law, or who do not provide the people with services they desire is to replace them with other officials.” (Id., at p. 817.)
Of course, the paradoxical result of the majority’s holding is that no matter what the City does, it may be held responsible for a police officer’s criminal conduct including offenses such as this rape. The City’s police department already has a policy that imposes several reporting requirements on officers who transport members of the opposite sex. (See maj. opn., ante, at p. 218, fn. 10.) The City’s assistant chief of police in charge of personnel and training testified that department policy requires a male officer trans*238porting a female arrestee to record the time and mileage of his police vehicle so that the arrestee’s whereabouts could be monitored and verified. Department policy also prohibited Officer Schroyer from transporting plaintiff to her residence. Obviously, these policies did not prevent the rape in this case.
Under the majority’s reasoning, one purpose of vicarious liability in this case would be to encourage the City to adopt further, undefined measures. By adopting the rules then in effect, the City, however, may have done all that it could reasonably do without imposing an undue burden on the police’s resources and mission—the same concern expressed in John R. Indeed, if the City did not act reasonably, it could have been found negligent. Plaintiff, however, dismissed her cause of action for negligence, thereby indicating that the City had done all it could reasonably be expected to do. Plaintiff fails to propose any regulation that would be effective without being unreasonably restrictive on effective law enforcement.
The majority’s treatment of the regulations adopted by the City is self-contradictory. At one point, the majority approvingly notes a rule adopted by the San Francisco Police Department relating to the transport of females by male officers. The majority asserts this rule “illustrates the type of measure that a law enforcement agency can take to reduce the incidence of sexual assaults by police officers on duty.” (Maj. opn., ante, at p. 215, fn. 7.) Only a few paragraphs later, the majority notes that the City has a similar regulation. Thus, the effect of the majority’s holding is that the City is liable despite its adoption of measures vicarious liability is designed to encourage.1
The proper question is whether vicarious liability would deter future misconduct without undue adverse consequences for the police function. If we impose liability, the City has two choices: (1) It can conclude it has already done all that it can reasonably do and accept the fact that errant officers might on occasion rape citizens, thereby subjecting the City to vicarious liability. If this is the result, vicarious liability has no deterrent effect. (2) Alternatively, the City can take measures beyond those already adopted. It requires little common sense to imagine that such measures might lead to the same result disapproved in John R., supra, 48 Cal.3d 438—undue interference with the City’s ability to perform its mission of providing police protection. In rejecting vicarious liability for a teacher’s sexual molestation of a child, the John R. lead opinion explained that, “Although it is unquestionably important to encourage both the careful selection of these employ*239ees and the close monitoring of their conduct, such concerns are, we think, better addressed by holding school districts to the exercise of due care in such matters and subjecting them to liability only for their own direct negligence in that regard. Applying the doctrine of respondeat superior to impose, in effect, strict liability in this context would be far too likely . . . to induce districts to impose such rigorous controls on activities of this nature that the educational process would be negatively affected.” (Id., at p. 451.) The same reasoning applies with equal force in this case.
Whether vicarious liability will have a deterrent effect without undue impediment to a public function depends on what measures a public entity has already taken, what additional measures it can take, and what the effects of those measures will likely be. The majority’s holding will allow liability in future cases regardless of whether it will help attain the goal of deterrence or whether it will unduly restrict an essential public function. At a minimum, the question whether vicarious liability is appropriate should depend on the particular facts of each case. In John R, even Justice Kaufman, who vigorously dissented in favor of vicarious liability, explained that, “Respondeat superior is a fact-specific determination; a holding adverse to the district would necessarily be limited to the uniquely compelling facts of this case.” (48 Cal.3d at p. 465 (conc, and dis. opn. of Kaufman, J.), italics added.) By contrast, the majority result here is absolute and not tethered to any factual basis.
c. Assurance of compensation to accident victims
The John R lead opinion, supra, 48 Cal.3d 438, concluded that the general goal of compensating accident victims weighed against imposing vicarious liability for a sexual assault. “The [sexual] acts here differ from the normal range of risks for which costs can be spread and insurance sought. [Citation.] The imposition of vicarious liability on school districts for the sexual torts of their employees would tend to make insurance, already a scarce resource, even harder to obtain, and could lead to the diversion of needed funds from the classroom to cover claims.” (Id., at p. 451.) The same reasoning applies equally to the present case. Imposing vicarious liability on cities for employee-committed rapes indisputably will increase the cost of insurance and will also decrease its availability.2
*240Perhaps to avoid this difficulty, the majority makes an elliptical argument as to legislative intent, stating that, by not enacting governmental immunity for violent police misconduct, the Legislature has demonstrated that vicarious liability is an appropriate method for ensuring victim compensation. However, the Legislature’s failure to expressly preclude liability is not a valid indicator that the legislative purpose was to allow such liability.
The Legislature has provided that vicarious liability may be imposed only for a public employee’s actionable misconduct “in the scope ... of employment.” Where, as here, the employee’s intentional criminal conduct was a spontaneous personal deviation from duty and bore no relationship to his work performance, the Legislature’s standard for vicarious liability has not been met.
I am not persuaded that ensuring compensation for victims is a dispositive concern in any event. It is a truism to state that ensuring compensation weighs in favor of vicarious liability. The deeper the defendant’s pocket, the easier the plaintiff is compensated. If ensuring compensation were the only goal, vicarious liability should apply against all employers in all cases. However, as the result in John R., supra, 48 Cal.3d 438, demonstrates, the sympathetic desire to compensate the injured is not a sufficient basis on which to impose vicarious liability.
Our decisions in other areas reinforce this principle. For example, prescription drugs occasionally have grievous, even fatal, side effects upon innocent victims. We recently held, however, that a manufacturer of a defectively designed drug cannot be held strictly liable. (Brown v. Superior Court (1988) 44 Cal.3d 1049, 1061 [245 Cal.Rptr. 412, 751 P.2d 470].) Writing for a unanimous court in Brown, Justice Mosk explained that despite occasional “unfortunate consequences” to sympathetic victims, the public interest in the development and availability of prescription drugs weighed against liability without fault. (Id., at pp. 1061-1065.) Similarly, in Belair v. Riverside County Flood Control Dist. (1988) 47 Cal.3d 550, 564-565 [253 Cal.Rptr. 693, 764 P.2d 1070], we held that strict liability was not appropriate in an inverse condemnation action for property damaged by public flood control projects. We found, in effect, that the desire to compensate individual injuries was outweighed by important public need for such projects.
The public has equally compelling interests in adequate law enforcement and preservation of public funds. A ruling that the public must bear the cost *241of all police misconduct merely because the public benefits from law enforcement is inconsistent with the spirit of Brown and Belair.
V. The majority opinion will have adverse practical effects.
The theoretical and practical ramifications of the majority’s holding are sweeping. At a minimum, the majority opinion will permit imposition of vicarious liability whenever an on-duty police officer commits rape or some other sexual assault against a citizen the officer has detained by invoking his official authority. (Maj. opn., ante, at p. 221.) It is difficult to see how a jury could find, consistent with the majority opinion, that a uniformed officer who detains and sexually assaults a motorist was not acting in the course and scope of his employment.
The majority purports to limit its holding to cases in which an officer “exercises” or “misuses” his authority. Indeed, the opinion stresses that vicarious liability is appropriate here because Office Schroyer committed his criminal act while “on duty” and “in uniform.” But the majority’s underlying logic extends far beyond these limited circumstances. Once the majority’s fundamental premise is accepted, its efforts to limit its ruling are largely illusory.
The majority’s conclusions rest on the principle that a police officer’s special power and authority allow him to impose his will on citizens. But this power and authority are limited neither to uniformed officers, nor to on-duty hours. Officers have law enforcement responsibility even when off duty, and their jurisdiction in certain situations is statewide. (See, e.g., Pen. Code, § 830.1, subd. (a)(3).) Moreover, an officer’s special power to intimidate, if any, does not depend on whether he is actually on duty or in full uniform. If the officer acts in uniform, or displays his badge, or brandishes a regulation firearm, or even mentions his or her status, the officer implicitly uses state-conferred power and ability to subjugate the victim. Under the majority’s reasoning, a jury would be hard pressed to find that misconduct committed under such circumstances was outside “the scope of . . . employment.” (See, e.g., Silver, Police Civil Liability (1991) § 6.07, p. 6-12 [“[T]he issue of ‘off’ vs. ‘on’ duty is usually not critical where, for instance, the officer identifies himself or uses a weapon.”].)
Rather than consider or even acknowledge this consequence of its holding, the majority contends that cases in which courts have refused to impose vicarious liability for sexual misconduct are “distinguishable” because the officers were off duty. However, examination of these decisions discloses no such dispositive distinction. In Gambling v. Cornish (N.D.Ill. 1977) 426 F.Supp. 1153, the court did not decide whether the raping officers were on or *242off duty. Rather, the court stated that, even assuming they were on duty, there was no vicarious liability. (Id., at p. 1155.) Similarly, in Bates v. Doria (1986) 150 Ill.App.3d 1025 [502 N.E.2d 454], the court rejected liability, not because the officer was off duty, but because the rape was outrageous and therefore beyond the scope of his employment.
The logical consequence of the majority’s holding is demonstrated by one of the out-of-state cases on which it relies. In Applewhite v. City of Baton Rouge (La.Ct.App. 1979) 380 So.2d 119, the court upheld vicarious liability for a rape committed by an on-duty police officer. The significance for our case is that, in support of its conclusion, the court relied on prior Louisiana decisions that imposed vicarious liability for torts committed by off-duty officers. (Id., at pp. 121-122, citing Cheatham v. Lee (La.Ct.App. 1973) 277 So.2d 513; Borque v. Lohr (La.Ct.App. 1971) 248 So.2d 901.) The majority’s reliance on Applewhite, supra, is curious because the case refutes the majority’s attempted distinction between on-duty and off-duty misconduct.
The majority’s logic will also extend beyond police officers. Part 2, title 2, chapter 4.5 of the Penal Code grants peace officer status to a wide variety of law enforcement officers. They include sheriffs, marshals, constables, and inspectors for district attorneys. Under appropriate circumstances, peace officer status is conferred on dental examiners, voluntary fire wardens, horse racing board investigators, and many other persons. (Pen. Code, § 830.3.) Like police, some of these officers are authorized to carry firearms. (Ibid) If one of these types of officers uses his weapon or asserts his authority in order to facilitate a rape, it is difficult to see how vicarious liability could be denied under the majority opinion.
The majority opinion is also unlimited in terms of the types of misconduct that will give rise to liability. Rape, robbery, and murder serve no public or police function. Yet the majority’s holding seems to permit imposition of vicarious liability for all these crimes if the perpetrator made any use of official trappings or weapons or if the victim had knowledge of the attacker’s connection to law enforcement and submitted accordingly. The implications of that conclusion are daunting.
In sum, the principles espoused by the majority have the potential to convert blameless public agencies into liability insurers for much, if not all, of the intentional misconduct committed by peace officers in their employ. Unlike commercial insurers, the innocent agencies can neither define the limits of their coverage nor collect premiums to finance it. Moreover, as we have seen, they may be both legally and practically barred from transferring their exposure to a commercial insurer. The majority fails to persuade me that law or public policy warrants such a result.
*243VI. Prior court decisions weigh against vicarious liability.
As the court of last resort on this question of state statutory construction, we are not bound by the decisions of our sister states’ courts. They do, however, provide guidance in determining whether our decision will be consistent with mainstream thinking on this issue. (Delaney v. Superior Court (1990) 50 Cal.3d 785, 799, fn. 9 [268 Cal.Rptr. 753, 789 P.2d 934].) Those decisions weigh against imposing liability. (The thrust of the majority on this point is defensive, i.e, trying to distinguish cases that decline liability.)
In Bates v. United States (8th Cir. 1983) 701 F.2d 737, the court affirmed a summary judgment in favor of the federal government in an action arising from rapes and murders committed by a military policeman. The majority attempts to downplay the decision’s significance by emphasizing that the federal court was applying state law. The attempted distinction is misplaced. The majority does not dispute that the decision was a correct application of Missouri law. The decision is therefore entitled to the same consideration that the majority gives to the Louisiana and Wisconsin state court decisions on which the majority relies.
Bates v. United States, supra, 701 F.2d 737, does not stand alone. In City of Green Cove Springs v. Donaldson (5th Cir. 1965) 348 F.2d 197, the court held as a matter of law that an on-duty police officer was not within the scope of his employment when he arrested and raped a female motorist. (Id., at p. 202.) In Gambling v. Cornish, supra, 426 F.Supp. 1153, the court held that two policemen who raped and committed other sexual acts on a citizen were not within the scope of their employment, regardless of whether they were on or off duty. (Id., at p. 1155.) The court explained that, “[W]hile the doctrine of respondeat superior should be broadly applied when a police officer is involved, the line must be drawn somewhere.” (Ibid.) In Bates v. Doria, supra, 502 N.E.2d 454, the court affirmed a summary judgment in favor of a county in an action arising from a rape committed by an off-duty deputy sheriff. The majority attempts to distinguish the case on the ground that the officer was off duty. This fact played no part in the court’s decision. The rationale of the decision was that the rape was outrageous and was committed solely for the officer’s benefit. (Id., at p. 457.) Those facts would remain true regardless of whether the officer was on duty or off duty.
The two cases on which the majority relies are not persuasive. In Desotelle v. Continental Cos. Co. (1986) 136 Wis.2d 13 [400 N.W.2d 524], the court affirmed a judgment of no vicarious liability on the ground that the jury’s verdict was supported by the evidence. This fact-specific holding provides *244little guidance one way or the other for our case. That the facts of the case supported the jury’s verdict does not mean the facts in our case support this jury’s verdict. In Applewhite v. City of Baton Rouge, supra, 380 So.2d 119, the court affirmed a judgment against a city for rapes committed by two on-duty officers. Although the result supports the majority’s conclusion, the Louisiana Court of Appeal’s discussion of the issue was brief—less than a page—and did not consider the significant issues raised by imposing liability. More important, neither case (Desotelle, supra, 400 N.W.2d 524; Applewhite, supra, 380 So.2d 119) was decided under a comprehensive statutory scheme like California’s that governs public entity liability.
VII. Conclusion
I concur in the judgment reversing the Court of Appeal decision. I do so on the narrow basis of the City’s invited error as to the jury instruction on the vicarious liability issue. I respectfully decline, however, to join the majority’s unnecessary holding that a police officer may act “in the scope of . . . employment,” thus exposing his blameless public employer to strict tort liability, when the officer rapes a citizen. In the absence of contrary legislation, I conclude, an officer may never be deemed within the “scope of employment” when he or she deviates from work duties to commit a crime unrelated to the performance of law enforcement responsibilities. Whether the taxpayers must absorb the cost of such individual misconduct is a subject within the exclusive purview of the Legislature, which cannot have anticipated this result.
Lucas, C. J., concurred.
The majority actually penalizes the City for adopting its regulations by observing that their adoption shows sexual misconduct by officers is not unexpected. (Maj. opn., ante, at p. 218, fn. 10.)
The concern over lack of insurance may apply even more strongly in this case. After John R, supra, 48 Cal.3d 438, we held that Insurance Code section 533 excludes coverage for sexual molestation of a child as a matter of law. (J. C. Penney Casualty Ins. Co. v. M. K., supra, 52 Cal.3d 1009.) The logical corollary is that coverage for rape is also excluded. Like child molestation, rape “is always intentional, it is always wrongful, and it is always harmful.” (Id., at p. 1025, original italics.) In John R, supra, 48 Cal.3d 438, the lead opinion was *240concerned with scarce insurance. In this case, we might be faced with a legally mandated unavailability of insurance because Insurance Code section 533 arguably precludes coverage for a defendant held vicariously liable, despite contrary dictum in an old opinion (Arenson v. Nat. Automobile & Cas. Ins. Co. (1955) 45 Cal.2d 81, 84 [286 P.2d 816]).