Opinion
LUCAS, C. J.I. Introduction and Summary
In 1988 the voters enacted Proposition 73, which was designed to reform financing of statewide and local political campaigns. (See Gov. Code, §§ 82041.5, 85100-85400, 89001.)1 Five provisions of that measure are at issue in this litigation: Section 85301, subdivision (a) (hereafter section 85301(a)) limits to $1,000 per fiscal year, contributions by a “person” to a candidate or to committees controlled by the candidate. 2 Section 85302 limits to $2,500 per fiscal year, contributions by a “person” to a “political committee, broad based political committee, or political party.”3 Section 85303, subdivision (a) (hereafter section 85303(a)) limits to $2,500 per fiscal year, contributions by a “political committee” to a “candidate or any committee controlled by that candidate.” Subdivision (b) (hereafter section 85303(b)) of the same section limits to $5,000 per fiscal year, contributions by a “broad based political committee or political party” to a “candidate or any committee controlled by that candidate.”4 Finally, section 85304 bans *614transfer of contributions between a candidate’s own committees and between candidates.5
In 1990, in a suit in which petitioners in the present litigation were allowed to intervene on behalf of the defendant (respondent herein Fair Political Practices Commission), the federal district court held the “fiscal year” measure of sections 85301-85303 unconstitutional, and enjoined enforcement of those sections and section 85304. (Service Employees v. Fair Political Practices (E.D.Cal. 1990) 747 F.Supp. 580 (Service Employees I).) The United States Court of Appeals for the Ninth Circuit affirmed the judgment of the district court (Service Emp. Intern, v. Fair Political Prac. Com’n (9th Cir. 1992) 955 F.2d 1312 (Service Employees II)), and the high court denied certiorari review. (505 U.S. 1230 [120 L.Ed.2d 922, 112 S.Ct. 3056, 3057].)
In this original proceeding (see Cal. Const., art. VI, § 10; Cal. Rules of Court, rule 56(a)) brought by petitioners State Senator Quentin Kopp and Assemblyman Ross Johnson, cosponsors of Proposition 73, we issued an alternative writ of mandate to respondent Fair Political Practices Commission directing it to show cause why a peremptory writ of mandate should not issue ordering respondent to enforce sections 85301-85304. Respondent filed an answer taking a neutral position on the issue. We granted the motions of Common Cause to intervene on behalf of petitioners, and of the California Legislature and four legislators6 to intervene on behalf of respondent. In addition we accepted amicus curiae briefs from other interested entities and legislators.7
The issue is one of state law: assuming enforcement of the challenged sections as enacted would violate the federal Constitution, may, and if so, should, the statutes be judicially reformed in a manner that avoids the fiscal year measure?
*615As explained below, we reject claims by interveners on behalf of respondent that the doctrine of res judicata or considerations of comity bar consideration of the issue raised. Nor does the federal appeals court’s judgment affirming the federal district court’s injunction against enforcement of sections 85301-85304 render those sections incapable of reformation by this court in this litigation.
We also reject the view that a court lacks authority to rewrite a statute in order to preserve its constitutionality or that the separation of powers doctrine, which vests legislative power in the Legislature and judicial power in the courts (Cal. Const., art. IV, § 1; id., art. VI, § 1), invariably precludes such judicial rewriting. Under established decisions of this court and the United States Supreme Court, a reviewing court may, in appropriate circumstances, and consistently with the separation of powers doctrine, reform a statute to conform it to constitutional requirements in lieu of simply declaring it unconstitutional and unenforceable, The guiding principle is consistency with the Legislature’s (or, as here, the electorate’s) intent: a court may reform a statute to satisfy constitutional requirements if it can conclude with confidence that (i) it is possible to reform the statute in a manner that closely effectuates policy judgments clearly articulated by the enacting body, and (ii) the enacting body would have preferred such a reformed version of the statute to invalidation of the statute. We conclude, however, that we should not reform section 85304 and, under this test, we may not reform sections 85301(a), 85302, or 85303(a) and (b).
We will not reform the “inter-candidate” aspect of section 85304’s transfer ban because the federal appeals court found that section unconstitutional on First Amendment and overbreadth grounds unrelated to reformation of the fiscal year measures of sections 85301(a) and 85303(a) and (b), and hence its order in this regard will not be implicated by our judgment herein, whether or not we reform the latter two sections.
Nor will we reform section 85301(a) or section 85303(a) and (b)—each of which regulates contributions to individual candidates—because, as illustrated by the starkly divergent positions of petitioners and intervener on their behalf, on one hand, and the justices joining Justice Baxter’s concurring and dissenting opinion, on the other hand, the statutes cannot be reformed in a fashion that closely effectuates policy judgments clearly articulated by the electorate. Specifically, because the “per election” approach advocated by petitioners and Common Cause would allow candidates less funding than the electorate contemplated, and because the so-called “modified election cycle” reformation advocated by Justice Baxter’s concurring and dissenting opinion would allow candidates more funds than the electorate planned (and would *616remove any regulation of the pace of contributions for nonpartisan offices), neither reformation would closely effectuate policy judgments clearly expressed by the electorate, and hence neither reformation is permissible.
Finally, for related reasons, we will not reform section 85302, which regulates contributions to political committees or parties.
II. Background: Litigation Concerning Proposition 73
In Service Employees I, supra, 747 F.Supp. 580, the federal district court considered challenges to, inter alia, four sections of Proposition 73. First, the plaintiffs challenged the measure’s three “contribution limitations” provisions: Sections 85301(a) and 85303(a) and (b), which, as noted above, regulate the maximum dollar amount of contributions to candidates “in any fiscal year,” and section 85302, which, as noted above, regulates the maximum dollar amount of contributions to political committees or parties “in any fiscal year.”8 In addition, the plaintiffs challenged the measure’s ban on transfer of contributions between a candidate’s own committees (the “intracandidate transfer ban”) and between candidates (the “inter-candidate transfer ban”) (§ 85304).
The federal district court found the fiscal year provisions of the three contribution limitation sections (§§ 85301-85303) unconstitutional because they would have the effect of discriminating in favor of incumbents. Relying on expert testimony presented at trial, the court noted that incumbents typically begin to solicit campaign contributions during each of the years of incumbency, but that challengers generally do not (and, as a practical matter, cannot) do so, because unlike incumbents, they typically do not decide to run for office years in advance of an election. (747 F.Supp. at p. 588.) As a result, the district court found, a “fiscal year” measure for contribution limitations would tend to favor incumbents over challengers,9 and it concluded this disparate treatment violated the First and Fourteenth Amendments to the federal Constitution. (Id. at p. 590.) Having reached this *617conclusion, the court found it unnecessary to address additional challenges to those statutes raised by the plaintiffs. (Ibid.)
After resolving this federal constitutional issue the court turned to an issue of state law, i.e., whether the constitutionally invalid “fiscal year” provision of the contribution limitation sections (§§ 85301-85303) might be severed from those sections. (747 F.Supp. at p. 590.) The court noted that Proposition 73 contained a severance clause, and it cited our decision in Calfarm Ins. Co. v. Deukmejian (1989) 48 Cal.3d 805, 821-822 [258 Cal.Rptr. 161, 771 P.2d 1247] (see 747 F.Supp. at p. 590, fn. 22), in which we expressly recognized that the presence of such a clause “ ‘normally calls for sustaining the valid part of the enactment . . . ” (Calfarm, supra, 48 Cal.3d at p. 821.) The federal district court summarily concluded, however, that the fiscal year provisions could not be severed from the contribution limitations themselves, and it did not consider whether the provisions might, under state law, be judicially reformed to avoid the unconstitutional fiscal year measure. Accordingly, the court found sections 85301 through 85303 violate the federal Constitution. (747 F.Supp. at p. 590.)10
Next, the federal district court addressed 85304’s transfer bans. It found the intra-candidate transfer ban to be an unconstitutional spending limitation, and permanently enjoined its enforcement. (See Service Employees I, supra, 747 F.Supp. 580, 591-594; see generally, Buckley v. Valeo (1976) 424 U.S. 1, 54-59 [46 L.Ed.2d 659, 707-710, 96 S.Ct. 612] [spending, as opposed to contribution limitations, are constitutionally permissible only if the candidate agrees to be bound by them].) This aspect of section 85304 is not at issue in this proceeding. Finally, the court addressed the plaintiffs’ challenge to section 85304’s ban on inter-candidate transfers of contributions, i.e., “transfers of funds between candidates.” The district court held this aspect of section 85304 unconstitutional on the ground that no legitimate governmental interest justified the burden on the right of a candidate to contribute to *618another candidate. The court noted that the defendants “maintain that the transfer ban is simply a device to prevent those who desire to avoid the contribution limits from doing so by the simple expedient of using another candidate as a conduit for the contribution” (747 F.Supp. at p. 593), but reasoned that because the contribution limits were themselves constitutionally invalid, the inter-candidate transfer ban, “to the extent it is premised on the need to prevent subversion of the fiscal year limitations, must also fail.” (Ibid., fn. omitted.) The court thus permanently enjoined enforcement of the contribution limitation provisions (§§85301-85303) and the transfer ban provision (§ 80304). (Service Employees I, supra, 747 F.Supp. at p. 593.)11
A month after the federal district court’s decision in Service Employees I, supra, we held that Proposition 73 prevailed over Proposition 68, an alternative campaign reform measure that had garnered a lesser majority vote at the 1988 Primary Election. (Taxpayers to Limit Campaign Spending v. Fair Pol. Practices Com. (1990) 51 Cal.3d 744 [274 Cal.Rptr. 787, 799 P.2d 1220].) Thereafter, the United States Court of Appeals for the Ninth Circuit affirmed the judgment of the district court in Service Employees I, supra, (Service Employees II, supra, 955 F.2d 1312.)
The federal appeals court agreed that the fiscal year measure employed in sections 85301 through 85303 would have the effect of discriminating in favor of incumbents, and it found a violation of the First Amendment because the state failed to show that “the discrimination itself is necessary to serve a substantial governmental interest.” (Service Employees II, supra, 955 F.2d at p. 1320.) The federal appeals court also concluded that the “fiscal year feature” of sections 85301-85303 rendered them constitutionally infirm and not salvageable by severance or reformation. The court reasoned: “[Plaintiffs] have given us no reason to believe that ‘the legislation [i.e., sections 85301-85303] would have been enacted if it had not included the unconstitutional provision[].’ National Advertising Co. v. Town of Babylon, 900 F.2d 551, 557 (2d Cir. [1990]) (citing United States v. Jackson, 390 U.S. 570, 585 n. 27, 88 S.Ct. 1209, 1218 n. 27, 20 L.Ed.2d 138 (1968)), cert, denied, 498 U.S. 852 111 S.Ct. 146, 112 L.Ed.2d 112 (1990). As the district court pointed out, were we simply to strike the word ‘fiscal’ from the statute, contribution limits would still be measured on an annual basis, raising the *619same problems of discrimination. 747 F.Supp. at 590. Were we to rewrite the statute to limit contributions on an election cycle basis, we would be at a loss to know what the dollar amounts of the limitations should be. In short, to save the statute, we would have to rewrite it substantially, ‘a practice that is decidedly disfavored.’ [900 F.2d at p. 557] (citing Thornburgh v. American College of Obstetricians & Gynecologists, 476 U.S. 747, 764-765, 106 S.Ct. 2169, 2180-81, 90 L.Ed.2d 779 (1986)).” (Service Employees II, supra, 955 F.2d at p. 1321.)12
The federal appeals court then addressed section 85304’s two transfer bans. It affirmed the federal district court’s invalidation of the intra-candidate transfer ban, and, as noted above, that aspect of section 85304 is not implicated by the present litigation. Finally, the federal appeals court affirmed the district court’s permanent injunction against enforcement of section 85304’s inter-candidate transfer ban. It first acknowledged the defendants’ claim that the ban was necessary in order to prevent circumvention of the contribution regulations, but concluded, as had the district court, that the ban “cannot serve this purpose in the absence of valid contribution limits.” (955 F.2d at p. 1322.) It then addressed and rejected the defendants’ alternative justification for the ban, namely that it served “the state’s interest in preventing corruption or the appearance of corruption by ‘political power brokers.’ ” (Id. at p. 1323.) The court held: “Even if we assume this to be an important state interest, the ban is not ‘closely drawn to avoid unnecessary abridgment of associational freedoms.’ Buckley [v. Valeo, supra, 424 U.S. 1,] 25 [46 L.Ed.2d 659, 691]. The potential for corruption stems not from campaign contributions per se but from large campaign contributions. Id. at 28 [46 L.Ed.2d at p. 693], The inter-candidate transfer ban prohibits small contributions from one candidate to another as well as large contributions. We hold, therefore, that the inter-candidate transfer ban is unconstitutional because it fails the ‘rigorous’ test used in Buckley, 424 U.S. at 29 [46 L.Ed.2d at pp. 693-694].” (Service Employees II, supra, 955 F.2d at p. 1323.)
We subsequently held in Gerken v. Fair Political Practices Com. (1993) 6. Cal.4th 707 [25 Cal.Rptr.2d 449, 863 P.2d 694] (Gerken), that Proposition 73 was not “invalidated” by the federal litigation in Service Employees II, supra, and thus declined to revive Proposition 68, which, we explained, remained “inoperative.” (6 Cal.4th at pp. 719, 720.)
Our three opinions in Gerken, supra, 6 Cal.4th 707 (see id. at pp. 712, fn. 6 (lead opn. by Lucas, C. J.); id. at p. 721 (conc. opn. of Baxter, J.); and id. at p. 736, fn. 3 (dis. opn. of Arabian, J.)) observed that the assumption underlying some of the reasoning of the federal appeals court—i.e., that the *620contribution limitation sections could not properly be judicially reformed to conform with constitutional principles—was in fact an untested question of state, and not federal, law. As we recognized long ago, construction of a state statute is purely a matter of state law, and an erroneous construction by a federal court does not preclude a state court from later rejecting the federal court’s conclusion. (See, e.g., City of Oakland v. Buteau (1919) 180 Cal. 83, 89 [179 P. 170]; Bank of Italy etc. Assn. v. Bentley (1933) 217 Cal. 644, 653 [20 P.2d 940].) Nevertheless, interveners for respondent assert we are precluded from even considering whether to reform the statutes at issue in this writ proceeding. Before proceeding to the reformation question, we address these preclusion issues.
III. Whether We May Consider the Issues Raised in This Writ Petition
A. Res judicata!collateral estoppel
“The doctrine of res judicata precludes parties or their privies from relitigating a cause of action that has been finally determined by a court of competent jurisdiction.” (Bernhard v. Bank of America (1942) 19 Cal.2d 807, 810 [122 P.2d 892].) Interveners on behalf of respondent correctly note that the federal district and appellate courts are “courts of competent jurisdiction.” They reason therefrom that the final judgment of the federal appeals court against petitioners, who appeared in the Service Employees litigation as interveners on behalf of the defendant, precludes this court from considering whether the unconstitutional aspects of sections 85301 through 85304 may be reformed under state law. As explained below, we disagree.13
State courts “are the principal expositors of state law.” (Moore v. Sims (1979) 442 U.S. 415, 429 [60 L.Ed.2d 994, 1007, 99 S.Ct. 2371].) Whether a state statute may be reformed or construed in a manner that preserves its constitutionality is a question of state, and not federal, law.14 Indeed, as the high court has often observed, federal courts “lack jurisdiction authoritatively to construe state legislation” (United States v. Thirty-Seven *621Photographs (1971) 402 U.S. 363, 369 [28 L.Ed.2d 822, 830, 91 S.Ct. 1400] (Thirty-Seven Photographs)), and for that reason, federal courts are reluctant to reform or “rewrite” state statutes to preserve constitutionally even when the court would otherwise reform or rewrite the statute if enacted by Congress. (Ibid.; see discussion, post, at pp. 629-631; see also Dombrowski v. Pfister (1965) 380 U.S. 479, 491-492 [14 L.Ed.2d 22, 31-32, 85 S.Ct. 1116] [state court may reform state statute previously invalidated by federal court on federal constitutional grounds].) Nevertheless, as the United States Court of Appeals for the Sixth Circuit recently observed when confronted with a state law reformation question, the role of a federal court is to divine, as best it can, how the state court of last resort would rule on the question of statutory reformation. (Eubanks v. Wilkinson (6th Cir. 1991) 937 F.2d 1118, 1122.)15 (We note that in most jurisdictions, federal courts are considerably assisted in this task by the ability to certify such a question to the state’s highest court. California, however, is one of the few states in the country— and the only one in the Ninth Circuit—that has no procedure for federal courts to certify questions of state law to the state’s highest court.)
Petitioners and intervener on their behalf suggest that because a federal court lacks authority authoritatively to construe or reform a state law (Moore v. Sims, supra, 442 U.S. 415, 428-430 [60 L.Ed.2d 994, 1006-1008]; Thirty-Seven Photographs, supra, 402 U.S. 363, 369 [28 L.Ed.2d 822, 829-830]), principles of res judicata or collateral estoppel can never bar a state court from entertaining the same cause of action that has been resolved by a federal decision which, in turn, rested on the federal court’s determination that it should not construe or reform a state statute to preserve its constitutionality. We decline to so hold. Instead, we conclude that assuming principles of res judicata or collateral estoppel would otherwise apply, we should entertain and resolve the present litigation under the “public interest” exception to those doctrines.
In City of Sacramento v. State of California (1990) 50 Cal.3d 51 [266 Cal.Rptr. 139, 785 P.2d 522], we allowed the state to relitigate the issue of whether extension of the state’s unemployment insurance law to include state and local governments constituted a reimbursable state mandate. We noted that the state was the losing party in the earlier litigation and that it was the only entity legally affected by the earlier judgment. “Thus, strict application of collateral estoppel would foreclose any reexamination of the *622holding of that case. The state would remain bound, and no other person would have occasion to challenge the precedent.” (Id. at p. 64.) We observed, however, that “ ‘when the issue is a question of law rather than of fact, the prior determination is not conclusive either if injustice would result or if the public interest requires that relitigation not be foreclosed. [Citations.]’ ” (Ibid., italics added; see also Greenfield v. Mather (1948) 32 Cal.2d 23, 35 [194 P.2d 1] [recognizing public interest exception to res judicata].)
Applying that rule to the facts before us, we concluded: “Yet the consequences of any error transcend those which would apply to mere private parties. If the result of [the earlier litigation] is wrong but unimpeachable, taxpayers statewide will suffer unjustly the consequences of the state’s continuing obligation to fund [the state mandate]. On the other hand, if the state fails to appropriate the funds to meet this obligation, and [the law extending unemployment insurance requirements to local governments] cannot be enforced [citations], the resulting failure to comply with federal law could cost California employers millions.” (City of Sacramento v. State of California, supra, 50 Cal.3d at pp. 64-65, fn. omitted; accord, Arcadia Unified School Dist. v. State Dept, of Education (1992) 2 Cal.4th 251, 256-259 [5 Cal.Rptr.2d 545, 825 P.2d 438] [public interest exception applied to allow relitigation of whether school districts may charge for school transportation].) By the same reasoning, we conclude this is a matter in which the public interest requires that relitigation not be foreclosed, and hence reject the claim that the doctrines of res judicata or collateral estoppel bar consideration of the state law issue in this litigation.16
*623B. Reformation of “constitutionally invalid” statutory provisions
In a related argument, interveners for respondent assert that when the federal courts declared sections 85301-85304 unconstitutional, those statutes ceased to exist, and hence cannot be judicially reformed because there is nothing left to reform. They cite authority for the proposition that an invalidated statute “is not a law; it confers no rights; it imposes no duties; it affords no protection; it creates no office; it is, in legal contemplation, as inoperative as though it had never been passed.” (Norton v. Shelby County (1886) 118 U.S. 425, 442 [30 L.Ed. 178, 186, 6 S.Ct. 1121] (Norton)-, accord, Reclamation District v. Superior Court (1916) 171 Cal. 672, 676 [154 P. 845] (Reclamation District).) They concede “numerous exceptions to that principle”—including the rule that “the text of an unconstitutional statute can be rendered legally operative by amending it to repair the constitutional defect” (1 Sutherland, Statutory Construction (5th ed. 1994) Limitations on Legislative Power, § 2.07, p. 38, fn. omitted; see County of Los Angeles v. Jones (1936) 6 Cal.2d 695, 708 [59 P.2d 489]), but assert the latter rule is inapplicable here, because petitioners propose repair by judicial reformation, rather than legislative amendment.
As petitioners observe, more recent decisions have approached the problem differently from Norton, supra, 118 U.S. 425, and Reclamation District, supra, 171 Cal. 672. “They proceed on the principle that a statute declared unconstitutional is void in the sense that it is inoperative or unenforceable, but not void in the sense that it is repealed or abolished . . . .” (Jawish v. Morlet (App.D.C. 1952) 86 A.2d 96, 97, and cases cited [when decision *624declaring statute unconstitutional is subsequently overruled, statute is restored by overruling decision without necessity of reenactment]; Ballew v. State (1974) 292 Ala. 460 [296 So.2d 206] [construing state statute in manner rendering it enforceable despite federal court’s earlier decision holding same statute unconstitutional, void, and subject to injunction].) In this regard, we find Dombrowski v. Pfister, supra, 380 U.S. 479, persuasive. In that case the high court enjoined enforcement of an unconstitutionally overbroad state statute, but specifically acknowledged the state’s authority thereafter to seek and obtain, in state court, a judicial “narrowing” reformation of the invalidated statute. (Id. at pp. 491 & 492 [14 L.Ed.2d at pp. 31 & 32.)
In any event, we note that neither decision of the federal courts in the Service Employees litigation purported to “invalidate” the statutes at issue here, as interveners on behalf of respondent use that term. The federal district court concluded that “because [sections 85301-85303] are measured by a fiscal year, they violate the Constitution of the United States and are unenforceable.” (Service Employees I, supra, 747 F.Supp. at p. 590, italics added.)17 In turn, the federal appeals court simply agreed with “the district court’s decision that all of Proposition 73’s contribution limits that are measured on a fiscal year basis are constitutionally infirm” (Service Employees II, supra, 955 F.2d at p. 1321), and affirmed the district court’s judgment. (Id. at p. 1323.)18 We therefore reject interveners’ premise; the federal appeals court did not “invalidate” sections 85301-85304; instead, it enjoined enforcement of those sections as written.
To the extent interveners on behalf of respondent suggest a statute that has been labeled “constitutionally invalid” is to be treated “as though it had never been passed,” and hence as not susceptible to judicial reformation, Dombrowski v. Pfister, supra, 380 U.S. 479, and our own cases reject that view. Indeed, the leading authority on the general subject of unconstitutional enactments, cited by both petitioners and interveners for respondent, describes with approval one of our cases (Quong Ham Wah Co. v. Industrial Acc. Com. (1920) 184 Cal. 26 [192 P. 1021, 12 A.L.R. 1190]), in which, it notes, we judicially reformed a workers’ compensation statute by extending benefits to the class expressly excluded by statute (id. at p. 39 et seq.), after *625first finding the statute violated the federal Constitution (184 Cal. at pp. 36-38). (Field, The Effect of an Unconstitutional Statute (1935, reprint ed. 1971), p. 274.) We have recently reaffirmed that same judicial authority to reform constitutionally “invalid” statutes. In Del Monte v. Wilson (1992) 1 Cal.4th 1009 [4 Cal.Rptr.2d 826, 824 P.2d 632], we held two veterans’ benefits statutes ‘‘‘'invalid as violative of the equal protection clause of the Fourteenth Amendment of the United States Constitution.” (Id. at p. 1026, italics added.) Nevertheless, we went on to judicially reform both statutes by extending the statutory benefits to those expressly excluded by the statutes. (Ibid.; see post, at p. 650.)
Numerous other decisions have long recognized the propriety of such judicial action. (See, e.g., In re Edgar M. (1975) 14 Cal.3d 727, 736, 737 [122 Cal.Rptr. 574, 537 P.2d 406] [holding statute as enacted unconstitutional, and then reforming to preserve constitutionality]; see also Davis v. Michigan Dept, of Treasury (1989) 489 U.S. 803, 817-818 [103 L.Ed.2d 891, 906-907, 109 S.Ct. 1500] [holding state benefits statute unconstitutional as enacted, and remanding to state court to elect between reformation and invalidation]; Wengler v. Druggists Mutual Ins. Co. (1980) 446 U.S. 142, 152-153 [64 L.Ed.2d 107, 116-117, 100 S.Ct. 1540] [same]; Orr v. Orr (1979) 440 U.S. 268, 283-284 [59 L.Ed.2d 306, 321-322, 99 S.Ct. 1102] [same]; Stanton v. Stanton (1975) 421 U.S. 7, 17-18 [43 L.Ed.2d 688, 696-697, 95 S.Ct. 1373] [same].) In view of this authority, we reject the position of interveners for respondent, that sections 85301 through 85304 were rendered legally nonexistent (and hence not susceptible to judicial reformation) by the federal appeals court’s judgment affirming the district court’s order enjoining enforcement of those sections.
C. Summary
We summarize our disposition of interveners’ objections to consideration of the state law reformation issue in this writ proceeding as follows: when faced with a question of whether to reform a state statute, the function of a federal court is to divine, to the best of its ability, how the state’s highest court would resolve that state law issue. (Eubanks v. Wilkinson, supra, 937 F.2d 1118, 1122.) As noted above, neither federal court in the Service Employees litigation did so, and instead both relied solely on federal law in concluding the statutes should not be reformed.
Contrary to suggestions of interveners on behalf of respondent, we conclude that a state supreme court is not constrained by principles of res judicata, collateral estoppel, or comity, to keep silent on a state law statutory reformation issue, when the question is presented to it in litigation such as *626this. Nor does the federal appeals court’s judgment affirming the injunction against enforcement of sections 85301 through 85304 render those sections legally nonexistent and hence not susceptible to judicial reformation. Our sovereign duty as a state court of last resort (see Scott v. Bank One Trust Co., N.A., supra, 577 N.E.2d 1077, 1080), consistent with principles of federalism and comity, requires that we not automatically accept the federal court’s ruling on this important state law issue, but consider the reformation question afresh ourselves and reach a different conclusion if state law leads us to that result.
For these reasons we issued an order to show cause in this matter, and now proceed to address the issue presented therein, namely, whether we may, and if so, should reform and order respondent to enforce sections 85301(a), 85302, 85303(a) and (b), and the inter-candidate transfer ban of section 85304.
IV. The Authority of a Court to Reform a Statute to Preserve its Constitutionality
Interveners and amicus curiae on behalf of respondent assert this court lacks authority to reform statutes, and that if we were to claim such authority, we would step out of legitimate judicial bounds and improperly invade the Legislature’s domain. They rely on numerous cases such as Metromedia, Inc. v. City of San Diego (1982) 32 Cal.3d 180, 187 [185 Cal.Rptr. 260, 649 P.2d 902] and Blair v. Pitchess (1971) 5 Cal.3d 258, 282 [96 Cal.Rptr. 42, 486 P.2d 1242, 45 A.L.R.3d 1206]—decisions in which we broadly stated we may not rewrite a statute even to preserve its constitutionality. As we shall explain, those cases, with one antique and unpersuasive exception, are all distinguishable. Moreover, as we shall explain directly below, numerous decisions of the United States Supreme Court and lower federal courts and sister states, and numerous decisions of this court, amply support the propriety of judicial reformation—including “rewriting”—of statutes to preserve constitutionality when (i) doing so closely effectuates policy judgments clearly articulated by the enacting body, and (ii) the enacting body would have preferred such a reformed version of the statute over the invalid and unenforceable statute.
Because much of the jurisprudence of our own cases rests on and flows from decisions of the United States Supreme Court addressing judicial authority to reform statutes to preserve them against constitutional infirmity, we will first survey in some detail decisions of the high court, and to a lesser extent, lower federal and state courts. (Post, pt. IV. A.) Thereafter, we will review California cases on that question (post, pt. IV. B.), and finally, as *627noted, we will consider cases in which we have disclaimed such authority (post, pt. IV. C.).
A. Reformation of enactments by the United States Supreme Court and other courts
1. The jurisprudential!constitutional foundation of a court’s power to reform a statute to preserve its constitutionality: Justice Harlan’s concurring opinion in Welsh v. United States
Modem authority for the judicial power of reformation to preserve constitutionality may be traced to the concurring opinion of Justice John Harlan in Welsh v. United States (1970) 398 U.S. 333, 344-367 [26 L.Ed.2d 308, 321-334, 90 S.Ct. 1792] (Welsh.).19 In Welsh, the high court addressed Congress’s conscientious objector statute, which applied to objectors who, “ ‘by reason of religious training and belief,’ ” were opposed to war. (398 U.S. at p. 336 [26 L.Ed.2d at p. 317].) The statute specifically excluded from its coverage those whose objection to induction arose from “ ‘essentially political, sociological, or philosophical views or merely a personal moral code.’ ” (Ibid.) Despite substantial evidence that petitioner Welsh fell within the excluded class (id. at pp. 341-342 [26 L.Ed.2d at pp. 319-320]), the plurality opinion nevertheless concluded he was covered by the statute (id. at p. 340 [26 L.Ed.2d at pp. 318-319]) and reversed his conviction for refusing induction. (Id. at pp. 343-344 [26 L.Ed.2d at pp. 320-321].)
Justice Harlan concurred separately to disavow the plurality’s reasoning. He asserted the plurality’s judgment effectively eliminated the statute’s exclusion from protection for those with nonreligious beliefs, but he nevertheless concurred in that result, “not as a matter of statutory construction,” but on the ground that under the circumstances it was appropriate for the court to judicially reform the statute. (Welsh, supra, 398 U.S. 333, 345 [26 L.Ed.2d 308, 321-322] (conc. opn. of Harlan, J.).) After explaining that *628Congress intended to exclude from the statute those, like Welsh, whose beliefs were not based on religion (id. at pp. 346-354 [26 L.Ed.2d at pp. 322-327]), Justice Harlan wrote: “If an important congressional policy [i.e., the conscientious objector policy] is to be perpetuated by recasting unconstitutional legislation, as the prevailing opinion has done here, the analytically sound approach is to accept responsibility for this decision. Its justification cannot be by resort to legislative intent, as that term is usually employed, but by a different kind of legislative intent, namely the presumed grant of power to the courts to decide whether it more nearly accords with Congress’ wishes to eliminate the policy altogether or extend it in order to render what Congress plainly did intend, constitutional.” (Welsh, supra, 398 U.S. at pp. 355-356 [26 L.Ed.2d at pp. 327-328] (conc. opn. of Harlan, J.), italics added.)
Justice Harlan determined that, properly construed, the statute ran afoul of the First Amendment’s establishment clause. (Welsh, supra, 398 U.S. 333, 356-361 [26 L.Ed.2d 308, 327-331] (conc. opn. of Harlan, J.).) He then addressed the question of relief (id. at p. 361 et seq. [26 L.Ed.2d at pp. 330-338].), and concluded that petitioner’s conviction for failure to submit to induction in the Armed Forces had to be reversed. He explained: “Where a statute is defective because of underinclusion there exist two remedial alternatives: a court may either declare it a nullity and order that its benefits not extend to the class that the legislature intended to benefit, or it may extend the coverage of the statute to include those who are aggrieved by exclusion.” (398 U.S. at p. 361 [26 L.Ed.2d at p. 331].)
Justice Harlan reasoned that because the statute “created a religious benefit not accorded to petitioner, it is clear to me that his conviction must be reversed under the Establishment Clause of the First Amendment unless Welsh is to go remediless.” (Welsh, supra, 398 U.S. 333, 362 [26 L.Ed.2d 308, 331] (conc. opn. of Harlan, J.).) He maintained that this result, “while tantamount to extending the [conscientious objector] statute, is not only the one mandated by the Constitution in this case but also the approach I would take had this question been presented in an action for a declaratory judgment or ‘an action in equity where the enforcement of a statute awaits the final determination of the court as to validity and scope.’ [Citation.] While the necessary remedial operation, extension, is more analogous to a graft than amputation, I think the boundaries of permissible choice may properly be considered fixed by the legislative pronouncement on severability.” (Id. at pp. 363-364 [26 L.Ed.2d at p. 332].)
Justice Harlan then quoted the statute’s severability clause, and asserted: “In exercising the broad discretion conferred by a severability clause it is, of *629course, necessary to measure the intensity of commitment to the residual policy and consider the degree of potential disruption of the statutory scheme that would occur by extension as opposed to abrogation.” (Welsh, supra, 398 U.S. 333, 365 [26 L.Ed.2d 308, 333] (conc. opn. of Harlan, J.).) He concluded that the policy of exempting conscientious objectors from induction is “one of longstanding tradition in this country” (ibid.), and hence “there is a compelling reason for a court to hazard the necessary statutory repairs if they can be made within the administrative framework of the statute and without impairing other legislative goals, even though they entail, not simply eliminating an offending section, but rather building upon it.” (Id. at p. 366 [26 L.Ed.2d at p. 334], fn. omitted.) Justice Harlan concluded: “Thus I am prepared to accept the prevailing opinion’s conscientious objector test, not as a reflection of congressional statutory intent but as a patchwork of judicial making that cures the defect of underinclusion . . . .” (Id. at pp. 366-367 [26 L.Ed.2d at p. 334].)
As explained below, even before Welsh, supra, and more frequently since that decision, the high court has followed Justice Harlan’s remedial approach in order to preserve the constitutionality of statutes, and in the process it has effectively rewritten various federal and some state statutes. We outline below three general categories of cases in which reformation has occurred: (i) cases concerning procedural safeguards required by the First Amendment and/or principles of procedural due process; (ii) cases concerning classifications underinclusive under the equal protection cause; and (iii) cases concerning otherwise vague or overbroad criminal statutes.
2. Reformation of statutes to avoid First Amendment and procedural due process problems
Thirty-Seven Photographs, supra, 402 U.S. 363, was a proceeding under a federal statute (19 U.S.C. § 1305(a)) providing for the forfeiture of obscene materials imported from a foreign country. The claimant, an importer who intended to publish the photographs in a book, asserted the forfeiture statute was unconstitutional under the First Amendment because it contained no adequate procedural safeguard to ensure prompt judicial review of a customs official’s decision to seize allegedly obscene property and because Congress may not bar importation of obscene material.
The first issue was resolved in an opinion by Justice White, who noted that the court had previously invalidated, as violations of the procedural due process guarantee, three similar state, local, and federal laws that likewise failed to provide for prompt judicial review of “administrative censorship” actions. (Freedman v. Maryland (1965) 380 U.S. 51 [13 L.Ed.2d 649, 85 *630S.Ct. 734]; Teitel Film Corp. v. Cusack (1968) 390 U.S. 139, 141 [19 L.Ed.2d 966, 968-969, 88 S.Ct. 754]; Blount v. Rizzi (1971) 400 U.S. 410 [27 L.Ed.2d 498, 91 S.Ct. 423].) The Thirty-Seven Photographs court acknowledged that it had declined to “rewrite” the statutes involved in those cases (Thirty-Seven Photographs, supra, 402 U.S. at p. 369 [28 L.Ed.2d at pp. 829-830], italics added), but explained that rewriting was impossible because the first two cases involved state enactments, “and we lack jurisdiction authoritatively to construe state legislation.” (Ibid.) The statute in the third case could not be reformed, the court explained, because doing so “would have required its complete rewriting in a manner inconsistent with the expressed intentions of some of its authors.” (Ibid.)
Justice White explained that “[n]o such obstacles confront us in construing § 1305(a). In fact, the reading into the section of the time limits required by Freedman is fully consistent with its legislative purpose.” (Thirty-Seven Photographs, supra, 402 U.S. at p. 370 [28 L.Ed.2d at pp. 830-831].) The court recited legislative history revealing Congress’ intent to require “immediate” review of confiscated material by a prosecuting attorney, and a “prompt” final disposition of the matter by a court. (Id., at pp. 370-371 [28 L.Ed.2d at p. 830].) The court concluded that “Congress’ sole omission was its failure to specify exact time limits within which resort to the courts must be had and judicial proceedings be completed.” (402 U.S. at p. 371 [28 L.Ed.2d at p. 831].)
After reviewing lower court cases involving substantial delays in the commencement and completion of judicial proceedings (Thirty-Seven Photographs, supra, 402 U.S. at pp. 371-372 [28 L.Ed.2d at pp. 381-382]), the court announced that “fidelity to Congress’ purpose dictates that we read explicit time limits into the section. The only alternative would be to hold § 1305(a) unconstitutional in its entirety, but Congress has explicitly directed that the section not be invalidated in its entirety merely because its application to some persons be adjudged unlawful.” (402 U.S. at p. 372 [28 L.Ed.2d at pp. 831-832].) The court noted that reformation of the statute would not “require us to decide issues of policy appropriately left to Congress” because Congress had “already set its course in favor of promptness and we possess as much expertise as Congress in determining the sole remaining question—that of the speed with which prosecutorial and judicial institutions can, as a practical matter, be expected to function in adjudicating § 1305(a) matters. We accordingly see no reason for declining to specify the time limits which must be incorporated into § 1305(a)—a specification that is fully consistent with congressional purpose and that will obviate the constitutional objections raised by claimant. Indeed, we conclude that the legislative history of the section and the policy of giving legislation a saving *631construction in order to avoid . . . constitutional questions require that we undertake this task of statutory construction.” (402 U.S. at pp. 372-373 [28 L.Ed.2d at p. 832].)
The court observed that in many of the lower court cases the “Government in fact instituted forfeiture proceedings within 14 days of the date of seizure of the allegedly obscene goods, [citations]; and judicial proceedings were completed within 60 days of their commencement. [Citations.]” (Thirty-Seven Photographs, supra, 402 U.S. at p. 373 [28 L.Ed.2d at p. 832].) The court took this as evidence that those precise time limits would impose no undue hardship on the government or the lower federal courts, and that a “delay of as much as 74 days” was reasonable for “importers engaged in the lengthy process of bringing goods into this country from abroad.” (Ibid.) The court announced: “Accordingly, we construe § 1305(a) to require intervals of no more than 14 days from seizure of the goods to the institution of judicial proceedings for their forfeiture and no longer than 60 days from the filing of the action to the final decision in the district court.” (402 U.S. at pp. 373-374 [28 L.Ed.2d at p. 832].) The court then applied its reformed “construction” of the statute to the facts of the case, and upheld the seizure. (Id. at pp. 374-375 [28 L.Ed.2d at pp. 832-833].)
Justice Harlan, who less than a year earlier articulated his view of permissible judicial reformation of underinclusive statutes in his concurring opinion in Welsh, supra, 398 U.S. 333, 361 et seq. [26 L.Ed.2d 308, 338], concurred separately. He endorsed Justice White’s analysis, and stated his agreement “that this statutory scheme may and should be construed” as set out in the court’s opinion. (Thirty-Seven Photographs, supra, 402 U.S. 333, 377 [28 L.Ed.2d 822, 834-835] (conc. opn. of Harlan, J.); accord, id. at p. 378 [28 L.Ed.2d at p. 835] (conc. opn. of Stewart, J.).)
Lower federal circuit courts and state courts have followed Thirty-Seven Photographs, supra, 402 U.S. 333, and have extended its reformation approach beyond the context of First Amendment safeguards, to impose time restraints and other conditions mandated by the federal or state Constitutions. (See Lee v. Thornton (2d Cir. 1976) 538 F.2d 27 [reforming federal statute governing forfeiture of vehicles seized as security at the border, and providing precise deadlines omitted by statute]; United States v. Monsanto (2d Cir. 1991) 924 F.2d 1186 [reforming federal forfeiture statute]; Department of Law Enf. v. Real Property (Fla. 1991) 588 So.2d 957 [reading into state forfeiture statute numerous detailed due process requirements needed to render the statute constitutional]; State ex rel. Berger v. McCarthy (1976) 113 Ariz. 161 [548 P.2d 1158] [reforming state forfeiture statute to avoid infirmity by requiring law enforcement officials file notice of intent to *632institute forfeiture proceedings within 20 days after seizing vehicle used to unlawfully transport drugs]; see also Allen v. State, Human Rights Com’n (1984) 174 W.Va. 139 [324 S.E.2d 99] [reading into statute mandatory duty of Human Rights Commission to hold hearing within 180 days from date of filing of complaint]; State v. Book-Cellar, Inc. (1984) 139 Ariz. 525 [679 P.2d 548] [reforming state “red light” nuisance abatement statute by reading into it requirement that court determine within 60 days whether to make permanent a preliminary injunction].)
3. Judicial reformation of statutes underinclusive under the equal protection clause
Even before Justice Harlan’s concurring opinion in Welsh, supra, 398 U.S. 333, 361 et seq. [26 L.Ed.2d 308, 330-338], the high court tacitly followed a similar remedial approach in a number of cases involving underinclusive or otherwise unconstitutional classifications in which the court rendered judgments that effectively, albeit implicitly, extended benefits statutes to improperly excluded groups.20 After Justice Harlan’s concurring opinion in Welsh, supra, 398 U.S. 333, 361 [26 L.Ed.2d 308, 330-331], and the court’s opinion in Thirty-Seven Photographs, supra, 402 U.S. 333, the high court began a series of equal protection cases in which it tacitly followed Justice Harlan’s remedial approach in order to preserve the constitutionality of various benefits statutes. In the process it effectively, albeit not always candidly, rewrote—by elimination and by addition of words—various federal and some state statutes.
In Graham v. Richardson (1971) 403 U.S. 365 [29 L.Ed.2d 534, 91 S.Ct. 1848], the court affirmed federal district court judgments extending to resident aliens statutory benefits established by state public assistance programs. The effect of the court’s ruling was to rewrite the statutory classification of beneficiaries to include a group originally excluded. In Weber v. Aetna Casualty & Surety Co. (1972) 406 U.S. 164 [31 L.Ed.2d 768, 92 S.Ct. 1400] (Weber), in which a similar constitutional problem was presented in a different procedural posture, the court reversed a state court judgment refusing to extend to dependent children “bom out of wedlock” equal entitlement to benefits from a state workers’ compensation scheme. The effect of the court’s ruling in Weber was to affirm the general propriety of *633judicial extension of statutes, and to force the state court to consider whether to extend the statute to include the excluded class.21
Thereafter, the court continued to affirm judgments extending statutes,22 and to reverse judgments refusing to do so,23 in numerous cases. Subsequently, as the lower courts began to order extension of otherwise underinclusive statutes, the high court increasingly entered judgments affirming such judicial actions. In Weinberger v. Wiesenfeld (1975) 420 U.S. 636 [43 L.Ed.2d 514, 95 S.Ct. 1225], the court affirmed a district court judgment extending to all “parents” a section of the Social Security Act that, as enacted, granted benefits only to “mothers.” In Califano v. Goldfarb (1977) 430 U.S. 199 [51 L.Ed.2d 270, 97 S.Ct. 1021], the court affirmed a district court judgment extending social security benefits to widowers on the same basis as widows. (See generally, Ginsburg, Some Thoughts on Judicial Authority to Repair Unconstitutional Legislation (1979) 23 Clev. St. L.Rev. 301, 310-312 (Ginsburg).)
Then, in Califano v. Westcott (1979) 443 U.S. 76 [61 L.Ed.2d 382, 99 S.Ct. 2655] (Westcott), the high court expressly adopted the approach set out in Justice Harlan’s concurring opinion in Welsh, supra, 398 U.S. 333, 361 et seq. [26 L.Ed.2d 308, 330-338]. (Westcott, supra, 443 U.S. at pp. 89-90 [61 L.Ed.2d at pp. 393-394] (maj. opn.); id. at p. 94 [61 L.Ed.2d at pp. 396-397] (conc. & dis. opn. of Powell, J.).) The court’s opinion in Westcott affirmed a district court judgment extending benefits under a public assistance program to children whose mothers’ unemployment deprived them of parental support. The court cited many of the decisions noted directly above (Westcott, *634supra, 443 U.S. at pp. 89-90 [61 L.Ed.2d at pp. 393-394]), and then stated its explicit approval of what the court had been doing implicitly for the prior decade. It observed that “[t]he District Court ordered extension rather than invalidation by way of remedy here, and equitable considerations surely support its choice.” (Id. at p. 90 [61 L.Ed.2d at p. 394].) The court then unanimously confirmed the authority of federal courts to order extension— i.e., reformation—of statutes otherwise unconstitutional under the equal protection clause. (Id. at p. 91 [61 L.Ed.2d at pp. 394-395].)24
The high court subsequently affirmed and applied its judicial extension doctrine in Wengler v. Druggists Mutual Ins. Co., supra, 446 U.S. 142. There the court found a state benefits statute that established a presumption of dependence for widows but not widowers invalid under the equal protection clause. Addressing the question of remedy—i.e., “extending the presumption of dependence to widowers or . . . eliminating it for widows” (id. at p. 152 [64 L.Ed.2d at p. 116])—the high court deferred to the state court. It reasoned that “[b]ecause state legislation is at issue, and because a remedial outcome consonant with the state legislature’s overall purpose is preferable, we believe that state judges are better positioned to choose an appropriate method of remedying the constitutional violation.” (Id. at pp. 152-153 [64 L.Ed.2d at pp. 116-117]; accord, Orr v. Orr, supra, 440 U.S. 268, 283-284 [59 L.Ed.2d 306, 321-322] [remanding for state court to remedy unconstitutional classification imposing alimony obligations on men only]; Stanton v. Stanton, supra, 421 U.S. 7, 17-18 [433 L.Ed.2d 688, 696-697] [remanding for state court to remedy unconstitutional classification imposing parental support obligation for males up to age 21, and females up to age 18].)25
Thereafter, the high court expressly reaffirmed the propriety of the judicial extension remedy. (Heckler v. Matthews (1984) 465 U.S. 728 [79 *635L.Ed.2d 646, 104 S.Ct. 1387].)26 The court stated in its unanimous opinion: “[W]e have noted that a court sustaining such a claim faces ‘two remedial alternatives: [it] may either declare [the statute] a nullity and order that its benefits not be extended to the class that the legislature intended to benefit, or it may extend the coverage of the statute to include those who are aggrieved by the exclusion.’ Welsh v. United States, [supra,] 398 U.S. 333, 361 [26 L.Ed.2d 308, 330-331] . . . (Harlan, J., concurring in result).” (Heckler v. Matthews, supra, 465 U.S. at p. 738 [79 L.Ed.2d at p. 656].) Immediately thereafter, the court explained: “Although the choice between ‘extension’ and ‘nullification’ is within the ‘constitutional competence of a federal district court,’ Califano v. Westcott, [supra,] 443 U.S., at 91 [61 L.Ed.2d at pp. 394-395], and ordinarily ‘extension, rather than nullification, is the proper course,’ id., at 89 [61 L.Ed.2d at p. 393], the court should not, of course, ‘use its remedial powers to circumvent the intent of the legislature,’ id., at 94 [61 L.Ed.2d at pp. 396-397] (opinion of Powell, J.), and should therefore ‘measure the intensity of commitment to the residual policy and consider the degree of potential disruption of the statutory scheme that would occur by extension as opposed to abrogation.’ Welsh v. United States, 398 U.S. at 365 [26 L.Ed.2d at p. 333] (Harlan, J., concurring in result). See also Califano v. Westcott, supra, [443 U.S.] at 90. . . .” (Heckler v. Matthews, supra, 465 U.S. at p. 739, fn. 5 [79 L.Ed.2d at p. 656].)27
Most recently, in Davis v. Michigan Dept, of Treasury, supra, 489 U.S. 803 (Davis), a tax classification case, the court again endorsed Justice Harlan’s concurring opinion in Welsh, supra, 398 U.S. 333, 361 [26 L.Ed.2d 308, 330], and the cases discussed above. The court explained it was “not in the best position to ascertain the appropriate remedy” (Davis, supra, 489 *636U.S. at p. 817 [103 L.Ed.2d at p. 906]) because doing so raised “a question of state law within the special expertise of the [state] courts. [Citation.] It follows that the [state] courts are in the best position to determine how to comply with the mandate of equal treatment.” (Id., at p. 818 [103 L.Ed.2d at p. 907],)28
In response to the high court’s practice of deferring to state courts to select extension or invalidation of underinclusive state statutes, our sister state courts have, with only one remarkable exception (ante, fn. 25), regularly followed the principles articulated by Justice Harlan in Welsh, supra, 398 U.S. 333 361 et seq. [26 L.Ed.2d 308, 330-338], and have accepted the propriety of judicial extension or “repair” of statutes when, inter alia, doing so is consistent with legislative intent.29 For example, when faced with otherwise unconstitutional alimony statutes, numerous courts have extended to men a statutory right of alimony granted to women only.30 Likewise, many courts faced with otherwise unconstitutional workers’ compensation statutes have extended to widowers the same statutory conclusive presumption of dependency and entitlement to workers’ compensation death benefits granted to widows only.31 In numerous additional cases concerning various *637other statutory schemes courts have extended or otherwise “repaired” statutes in order to avoid invalidity under equal protection principles.32 Courts have similarly extended the reach of underinclusive criminal statutes in order to avoid invalidity under equal protection principles.33 Finally, at least one state court has ventured even further to preserve a statute otherwise invalid under equal protection principles. In Insurance Co. of North America v. Russell (1980) 246 Ga. 269 [271 S.E.2d 178], the Georgia high court determined the legislature would prefer merger and substantial reformation of two constitutionally suspect provisions, rather than outright elimination of one provision and extension of the other.34
*6384. Judicial reformation of otherwise vague or overbroad criminal statutes
Finally, the high court has endorsed the propriety of judicial reformation of statutes in the context of otherwise vague or overbroad criminal statutes— namely, criminal obscenity statutes—and has encouraged state courts to do so as well.
In Miller v. California (1973) 413 U.S. 15 [37 L.Ed.2d 419, 93 S.Ct. 2607] (Miller), the high court confirmed that “obscene” materials are not protected by the First Amendment, but limited the scope of nonprotected materials to those that, inter alia, “portray sexual conduct in a patently offensive way.” (Id. at p. 24 [37 L.Ed.2d at pp. 430-431].) The court required that, in order to be regulated, such conduct be “specifically defined by the applicable state law, as written or authoritatively construed.” (Ibid., fn. omitted.) Thereafter the court gave “a few plain examples of what a state statute could define for regulation” under its standard: “(a) Patently offensive representations or descriptions of ultimate sex acts, normal or perverted, actual or simulated[, and] [1 (b) Patently offensive representation or descriptions of masturbation, excretory functions, and lewd exhibition of the genitals.” (Id. at p. 25 [37 L.Ed.2d at p. 431].)
Justice Brennan, dissenting in a companion case (Paris Adult Theater I v. Slaton (1973) 413 U.S. 49 [37 L.Ed.2d 446, 93 S.Ct. 2628]), asserted the court’s Miller standard would “invalidate virtually every state law relating to the suppression of obscenity” (id. at p. 95, fn. 13 [37 L.Ed.2d at p. 480] (dis. opn. of Brennan, J.).) In response, the Miller court stated: “We do not hold . . . that all States . . . must now enact new obscenity statutes. Other existing state statutes, as construed heretofore or hereafter, may well be *639adequate. See United States v. 12 200-Ft. Reels of Film [(1973) 413 U.S. 123,] at [p.] 130 n. 7 [37 L.Ed.2d 500, 507].” (Miller, supra, 413 U.S. at p. 24, fn. 6 [37 L.Ed.2d at p. 430].)
In the cited footnote 7 of United States v. 12 200-Ft. Reels of Film (1973) 413 U.S. 123, 130 [37 L.Ed.2d 500, 507, 93 S.Ct. 2665] (12 200-Ft. Reels of Film), the high court stated: “[W]hile we must leave to state courts the construction of state legislation, we do have a duty to authoritatively construe federal statutes where ‘ “a serious doubt of constitutionality is raised” ’ and ‘ “a construction of the statute is fairly possible by which the question may be avoided.” ’ United States v. Thirty-Seven Photographs, 402 U.S. 363, 369 (1971) [28 L.Ed.2d 822, 829-830, 91 S.Ct. 1400], (opinion of White, J.) .... If and when such a ‘serious doubt’ is raised as to the vagueness of the words ‘obscene,’ ‘lewd,’ ‘lascivious,’ ‘filthy,’ ‘indecent,’ or ‘immoral’ as used to describe regulated material in 19 U. S. C. § 1305(a) and 18 U. S. C. § 1462, ... we are prepared to construe such terms as limiting regulated material to patently offensive representations or descriptions of that specific ‘hard core’ sexual conduct given as examples in Miller v. California, [supra, 413 U.S.] at [page] 25 [37 L.Ed.2d at page 431]. See United States v. Thirty-Seven Photographs, supra, [402 U.S.] at [pages] 369-374 [28 L.Ed.2d at pages 829-835 (opinion of White, J.). . . .”35
Subsequently, citing footnote 7 of 12 200-Ft. Reels of Film, supra, 413 U.S. 123, 130 [37 L.Ed.2d 500, 507], and Thirty-Seven Photographs, supra, 402 U.S. 363, 369 [28 L.Ed.2d 822, 829-830], the high court imposed its elaborate saving gloss—i.e., the specific “hard core” sexual conduct given as examples in Miller, supra, 413 U.S. at page 25 [37 L.Ed.2d at page 431]—to preserve against a vagueness challenge a federal statute prohibiting mailing of obscene materials. (Hamling v. United States (1974) 418 U.S. 87, 113-116 [41 L.Ed.2d 590, 618-620, 94 S.Ct. 2887].) Thereafter, most state courts (see, e.g., State v. A Motion Picture Entitled “The Bet” (1976) 219 Kan. 64 [547 P.2d 760, 767], and cases cited), including our own (see Bloom v. Municipal Court (1976) 16 Cal.3d 71, 81 [127 Cal.Rptr. 317, 545 P.2d 229]) *640did the same, thereby grafting onto the various states’ statutes the detailed gloss articulated by the high court in Miller. In doing so the state courts determined—most often only by implication—that reformation was an appropriate exercise of judicial power, and that reformation of the statutes, rather than invalidating them, was most consistent with legislative intent.36
More recently, in Brockett v. Spokane Arcades, Inc. (1985) 472 U.S. 491 [86 L.Ed.2d 394, 105 S.Ct. 2794] (Spokane Arcades), the high court approved a similar reformation of a state statute that employed the term “lust” in its definition of regulated obscene matter. The United States Court of Appeals for the Ninth Circuit had invalidated the entire statute on the ground the term was unconstitutionally overbroad because it reached material that merely stimulated “normal sexual responses” in addition to material that was properly subject to regulation—i.e., that evincing a “morbid and shameful interest in sex.” (Id. at p. 495 [86 L.Ed.2d at p. 400], citing J-R Distributors, Inc. v. Eikenberry (9th Cir. 1984) 725 F.2d 482, 490-491.) The high court reversed.
The court first declined to construe the term “lust” as referring only to conduct that could properly be regulated. (Spokane Arcades, supra, All U.S. at pp. 500-501 [86 L.Ed.2d at pp. 403-404].)37 The court then noted that a “ ‘statute may be in part constitutional and in part unconstitutional, and . . . if the parts are wholly independent of each other, that which is constitutional may stand while that which is unconstitutional will be rejected.’ ” (472 U.S. at p. 502 [86 L.Ed.2d at p. 405].) Cautioning that “a federal court should not extend its invalidation of a statute further than necessary to dispose of the case before it,” (ibid.) the court concluded the statute could be saved through a combination of severance and tacit insertion of limiting language. “Unless there are countervailing considerations, the [state] law should have been invalidated only insofar as the word ‘lust’ is to be understood as reaching protected materials.” (Id. at p. 504 [86 L.Ed.2d at p. 406], italics added.) The court observed that state law disfavored invalidating the entire statute and *641favored severance (id. at p. 506 [86 L.Ed.2d at pp. 407-408]), and surmised that the state legislature would prefer a statute that was so severed and limited by a definition of “lust” that excludes material that merely stimulates “normal sexual responses.” Accordingly, it reversed the judgment invalidating the statute. (Id. at p. 507 [86 L.Ed.2d at p. 408].)
Although all decisions in which courts preserve enactments by severance are to some extent examples of judicial reformation, the significance of Spokane Arcades, supra, 472 U.S. 491, lies in the type of severance it employed. The court severed from the statute any meaning of “lust” that would include material that merely stimulates “normal sexual responses.” In other words, the court implicitly introduced into the statute words of limitation—confining the reach of the otherwise overbroad term “lust” to cover only material “whose predominant appeal is to a ‘shameful or morbid interest’ ” in sex (id. at p. 498 [86 L.Ed.2d at p. 402])—in order to uphold the statute’s validity. (See also United States v. Treasury Employees (1995) 513 U.S. _, _ [130 L.Ed.2d 964, 986-988, 115 S.Ct. 1003] (maj. opn. of Stevens, J.); id. at p--[130 L.Ed.2d at pp. 992-994] (conc. & dis. opn. of O’Connor, J.); id. at p. _ [130 L.Ed.2d at pp. 1002-1003] (dis. opn. of Rehnquist, C. J.) [all recognizing authority of court to “rewrite” statute to preserve constitutionality].)
B. Reformation of statutes by California courts
Our own cases reveal that, consistently with Welsh, supra, and its numerous high court predecessors and progeny, it is appropriate in some situations for courts to reform—i.e., “rewrite”—enactments in order to avoid constitutional infirmity, when doing so “is more consistent with legislative intent than the result that would attend outright invalidation.” (Arp v. Workers’ Comp. Appeals Bd. (1977) 19 Cal.3d 395, 407-408 [138 Cal.Rptr. 293, 563 P.2d 849] (Arp).) As explained below, like the high court, we have reformed statutes to preserve their constitutionality in cases concerning classifications otherwise invalid under the equal protection clause, and in cases involving criminal statutes otherwise unconstitutionally vague or overbroad. In addition, our decisions have reformed statutes to confer necessary procedural due process protections, to avoid classifications impermissible under the First Amendment, and to avoid nullification under the judicial powers provision of our own Constitution.
1. Guiding principles: the Arp case
Although Arp, supra, 19 Cal.3d 395, was a case in which we ultimately determined not to reform a constitutionally infirm statute, that decision both *642confirms the propriety of such a judicial role in appropriate cases, and provides guidance on the reformation question posed here.
In Arp, supra, a unanimous opinion by Justice Richardson, we considered a widower’s challenge to a section of the Labor Code governing workers’ compensation benefits. Former Labor Code section 3501, subdivision (a), provided that a widow was conclusively presumed to be totally dependent on her deceased husband, but created no such presumption for a widower. Instead, under the statutory scheme, a widower was forced to establish the fact and extent of dependency on his deceased wife. Following high court decisions cited above (e.g., Frontiero v. Richardson, supra, 411 U.S. 677; Weinberger v. Wiesenfeld, supra, 420 U.S. 636; Califano v. Goldfarb, supra, 430 U.S. 199), we held the statute infirm under the equal protection clauses of the state and federal Constitutions. (Arp, supra, 19 Cal.3d at p. 407.)
We then addressed “the question of remedy. Petitioner urges upon us the course adopted in Goldfarb, Wiesenfeld and Frontiero: extension of statutory benefits to males and females alike, without regard to actual dependency. [Citation.] [f] Although courts do not lack the power to remedy a constitutional defect by literally rewriting statutory language, it is a comparatively drastic alternative, to be invoked sparingly, and only when the result achieved by such a course is more consistent with legislative intent than the result that would attend outright invalidation. [Citations.]” (Arp, supra, 19 Cal.3d at pp. 407-408, italics added.)
We noted that in Weinberger v. Wiesenfeld, supra, 420 U.S. 636, the high court “in effect held that substitution of the word ‘parent’ for the word ‘mother’ was consistent with Congressional intent to subsidize parental care for minor children” (Arp, supra, 19 Cal.3d at p. 408), and that the similar reformation in Califano v. Goldfarb, supra, 430 U.S. 199, was also consistent with Congress’s intent. (19 Cal.3d at p. 408.)
We concluded, however, that “[o]ur own case is somewhat different,” in that there was “clear, if antique” evidence that the Legislature “did not want widowers to receive compensation in excess of their actual, demonstrable financial loss: it repealed the original limited presumption affording surviving husbands total dependency benefits on a showing of only partial dependency. [Citations.]” (19 Cal.3d at pp. 408-409, italics in original.) In addition, we observed that extension in such “benefits” cases posed a special problem that militated in favor of legislative reformation. We observed that “[s]uch action would undoubtedly have some impact on workers’ compensation insurance rates, since the present rate structure presumably has been carefully calculated without reference to the additional risk of maximum *643payout for all cases of female fatalities . . .” (id. at p. 409) and that under those circumstances “legislative preference is uncertain” and hence “judicial caution is appropriate.” (Id. at p. 410.) In closing, we also noted that invalidation rather than extension would not substantially disrupt the overall workers’ compensation scheme, or impose “any unfair hardship to an employee’s survivors” (ibid.), because other valid parts of the scheme remained to protect those interests. (Ibid.)
Arp, supra, 19 Cal.3d 395, thus stands for the proposition that courts possess the authority, in appropriate cases, “to remedy a constitutional defect by literally rewriting statutory language” when doing so is “more consistent with legislative intent than the result that would attend outright invalidation,” but that such judicial action is improper when the suggested reformation is inconsistent with the Legislature’s intent, or when that intent cannot be ascertained. As explained below, both before and since Arp, supra, 19 Cal.3d 395, we have reformed statutes to remedy constitutional defects consistently with the principles that governed our resolution of Arp.
2. Reformation of statutes to avoid vagueness, overbreadth, or procedural due process problems
In a substantial number of cases we have imposed saving constructions on otherwise unconstitutionally vague terms, thus preserving statutes while at the same time adding a crucial judicial gloss that, in practical effect, operates as a judicial reformation of the statute. As noted above, one such case was Bloom v. Municipal Court, supra, 16 Cal.3d 71, 81, in which we—like most other state courts—added a substantial textual gloss to our “obscenity” statute (Pen. Code, § 311.2) in order to save it against a claim that it was void for vagueness.' After reviewing the high court decisions described above, including 12 200-Ft. Reels of Film, supra, 413 U.S. 123, and Thirty-Seven Photographs, supra, 402 U.S. 363, we held our Penal Code section “is . . . limited to patently offensive representations or descriptions of the specific ‘hard core’ sexual conduct given as examples in Miller I[, supra, 415 U.S. at page 25 (37 L.Ed.2d at p. 431)], i.e., ‘ultimate sexual acts, normal or perverted, actual or simulated,’ and masturbation, excretory functions, and lewd exhibitions of the genitals.’ ” (Bloom, supra, 16 Cal.3d at p. 81.)
Thereafter, in Pryor v. Municipal Court (1979) 25 Cal.3d 238, 256-257 [158 Cal.Rptr. 330, 599 P.2d 636], we concluded that Penal Code section 647, subdivision (a) (disorderly conduct), was unconstitutionally vague. The statute made criminal one “ ‘[w]ho solicits anyone to engage in or who engages in lewd or dissolute conduct in any public place or in any place *644open to the public or exposed to public view.’ ” (25 Cal.3d at pp. 243-244, italics omitted.) In order to preserve the statute, we revised its scope, “arriv[ing] at the following construction of section 647, subdivision (a): The terms ‘lewd’ and ‘dissolute’ in this section are synonymous, and refer to conduct which involves the touching of the genitals, buttocks or female breast for the purpose of sexual arousal, gratification, annoyance or offense, if the actor knows or should know of the presence of persons who may be offended by his conduct. The statute prohibits such conduct only if it occurs in any public place or in any place open to the public or exposed to public view; it further prohibits the solicitation of such conduct to be performed in any public place or in any place open to the public or exposed to public view . . . .” (25 Cal.3d at pp. 256-257.)
In numerous other cases we have similarly reformed partly overbroad or vague statutes—and in doing so imposed what amounts to a judicial reformation of the statutory terms. (See, e.g., City of Los Angeles v. Belridge Oil Co.(1954) 42 Cal.2d 823, 832-833 [271 P.2d 5];38 In re Kay (1970) 1 Cal.3d 930, 943 [83 Cal.Rptr. 686, 464 P.2d 142];39 In re Bushman (1970) 1 Cal.3d 767, 773 [83 Cal.Rptr. 375, 463 P.2d 727];40 Morrison v. State Board of Education (1969) 1 Cal.3d 214, 225, 232-233 [82 Cal.Rptr. 175, 461 P.2d 375], and cases cited;41 Barrows v. Municipal Court (1970) 1 Cal.3d 821, *645827-828 [83 Cal.Rptr. 819, 464 P.2d 483];42 In re Cox (1970) 3 Cal.3d 205, 223 [90 Cal.Rptr. 24, 474 P.2d 992] ;43 Braxton v. Municipal Court (1973) 10 Cal.3d 138, 151 et seq. [109 Cal.Rptr. 897, 514 P.2d 697];44 Associated Home Builders etc., Inc. v. City of Livermore (1976) 18 Cal.3d 582, 598-599 [135 Cal.Rptr. 41, 557 P.2d 473];45 People v. Freeman (1988) 46 Cal.3d 419, 424 [250 Cal.Rptr. 598, 758 P.2d 1128].46) In each of these cases we reformed the statutes because doing so was more in keeping with the discernible legislative will than striking them down.47
*646Apparently because neither petitioners nor intervener on their behalf discusses these cases, interveners and amicus curiae on behalf of respondent do not discuss them or challenge their applicability to the present litigation. We assume that if pressed, they might attempt to distinguish them on the ground we were not forced in those cases to disregard language and to substitute reformed language; instead, we simply placed a saving “construction” on the statutory language, thereby constricting the reach of the statute. The distinction, in our view, suggests a difference of degree, not kind. In each of the cited vagueness and overbreadth cases we declined to give effect to the “plain words” meaning of the statutes, and instead reformed the statutes to in order to save and make them enforceable.48 In practical effect, in all of these cases, we “rewrote” each statute in order to preserve its constitutionality.
3. Reformation of statutes to avoid violation of state constitutional prohibition
In re Edgar M., supra, 14 Cal.3d 727 (Edgar M.), a unanimous opinion by Chief Justice Wright, illustrates the permissible limits of judicial reformation of a statute in order to conform it to constitutional principles. In Edgar M., we considered a constitutional challenge to Welfare and Institutions Code section 558, which governed a juvenile’s application for rehearing after a referee’s decision declaring the minor a ward of the court and removing him from his home. The statute provided: “ ‘If all of the proceedings before the referee have been taken down by an official reporter, the judge of the juvenile court may, after reading the transcript of such proceedings, grant or deny such application. If proceedings before the referee have not been taken down by an official reporter, such application shall be granted as of right. If an application for rehearing is not granted within 20 days following the date of its receipt, it shall be deemed denied. However, the court, for good cause, may extend such period beyond 20 days, but not in any event beyond 45 days, following the date of receipt of the application, at which time the application for rehearing shall be deemed denied unless it is granted within such period.’ ” (14 Cal.3d at pp. 736-737, italics added.)
*647Because the latter two sentences purported to give binding effect to a juvenile referee’s decision without requiring any action by a trial court, we concluded they violated constitutional restrictions on a referee’s powers. Instead of invalidating the statute, however, we sought a “construction of the [statute] that will eliminate this invalid application and yet preserve the parts and applications of the [statute] which do not violate the constitutional provisions and which the Legislature would have intended to put into effect if it had foreseen the constitutional restriction.” (Edgar M., supra, 14 Cal.3d at p. 736.) We determined that we could best effectuate the Legislature’s intent “by altering the operative effect” of the statute’s language rather than striking the two offending sentences altogether: “To strike the last two sentences from the section would remove the limits on the time during which an application for rehearing could be pending and awaiting action .... The section would then read as it did in . . . 1961 .... However, the Legislature showed its dissatisfaction with this omission by introducing the sentences into the section by amendments in 1963 [citations]. The salutary purpose of the amendments was to prevent indefinite prolongation of uncertainty concerning the status of a referee’s order as the order of a court.” (Edgar M., supra, 14 Cal.3d at p. 737.)
Our opinion explained: “We believe that the legislative intent will be more fully effectuated within the constitutional restraint by altering the operative effect of these sentences rather than striking them altogether. The quoted portion of section 558 provides that if the proceedings before the referee have not been taken down by an official reporter, an application for rehearing must be granted as a matter of right. If the proceedings have been taken down by a reporter but the judge does not... act on the application within the required period, the proceedings should be treated as a practical matter as if they had been unreported. Thus we conclude that we can best harmonize the statutory purpose with the constitutional command by requiring that applications which would be ‘deemed denied’ under the section’s literal wording be instead granted as of right, thereby applying to unacted-upon applications based on reported proceedings the rule which the Legislature has laid down for applications based on unreported proceedings.” (Edgar M., supra, 14 Cal.3d at p. 737.) Clearly, in Edgar M., we substantially reformed —indeed, “rewrote”—two sentences of the statute in order to conform it to constitutional principles, and to effectuate, as closely as possible, the Legislature’s intent, which we gleaned from the statute and its history.49
Amicus curiae on behalf of respondent attempts to dismiss Edgar M. as an aberration—“to say the least, an unusual case . . . [embodying a] ‘forced, *648strained, and unsatisfactory principle of statutory construction.’ ” (Quoting People v. Belton (1979) 23 Cal.3d 516, 531 [153 Cal.Rptr. 195, 591 P.2d 485] (dis. opn. of Jefferson, J.).) Interveners on behalf of respondent, on the other hand, accept the analysis and result in Edgar M., supra, 14 Cal.3d 727, as being consistent with the Legislature’s intent, but attempt to distinguish the present case on the ground that “the electorate’s intent with regard to Proposition 73 is much harder to discern.” We agree with interveners’ implicit concession that Edgar M., supra, cannot be dismissed, as amicus curiae would suggest, as some kind of mutant decision, and that the dispositive inquiry in the present case centers on the electorate’s intent.
4. Reformation of underinclusive enactments to avoid First Amendment problems
In City and County of San Francisco v. Eller Outdoor Advertising (1987) 192 Cal.App.3d 643 [237 Cal.Rptr. 815] (Eller), the Court of Appeal considered the constitutionality of San Francisco’s ordinance regulating both commercial and noncommercial signs. The court found the ordinance constitutional in most part, but held one aspect of the measure, which set out exemptions from regulation, conflicted with the First Amendment.
The suspect categories of exempted signs were contained in four subdivisions of section 603 of the ordinance. Subdivision (c) exempted “ ‘Temporary display posters, without independent structural support, in connection with political campaigns and with civic non-commercial health, safety, and welfare campaigns’ subdivision (d) exempted “ ‘[temporary displays of a patriotic, religious, charitable or civic character’ subdivision (f) exempted “ ‘Commemorative plaques’ and subdivision (h) exempted “ ‘Religious symbols attached to buildings . . . .’” (Eller, supra, 192 Cal.App.3d at pp. 649-650, fn. 3.)
Following decisions of the high court, the Court of Appeal concluded that by these exemptions the city had impermissibly attempted “ ‘in the area of noncommercial speech[,] to evaluate the strength of, or distinguish between, various communicative interests. [Citations.] With respect to noncommercial speech, the City may not choose the appropriate subjects for public discourse. . . .’” (Eller, supra, 192 Cal.App.3d at pp. 661-662, quoting Metro-media, Inc. v. San Diego (1981) 453 U.S. 490, 514-515 [69 L.Ed.2d 800, 819-820, 101 S.Ct. 2882].) The Eller court concluded these exemption provisions were “incompatible with the First Amendment” under Metromedia, supra, 453 U.S. 490, and proceeded to determine “whether the defect we have noted requires invalidation of the entire ordinance or whether the offending provisions can be construed in such a way as to save its constitutionality.” (Eller, supra, 192 Cal.App.3d at pp. 662, 663.)
*649The Eller court noted that in Metromedia, Inc. v. City of San Diego, supra, 32 Cal.3d 180, 187-191, we declined to reform a similar local ordinance principally because the enactors’ intent to impose a comprehensive ban on practically all noncommercial signs was incompatible with reformation of the ordinance to exempt noncommercial signs. (Eller, supra, 192 Cal.App.3d at p. 663.) As the Court of Appeal observed, however, the San Francisco ordinance revealed the opposite intent, i.e., to exempt “the vast majority of noncommercial expression.” (Id. at p. 664.)
The court concluded: “In light of the ordinance’s apparent purpose to allow most forms of noncommercial speech, little violence is done to the legislative purpose to interpret the exceptions created in subdivisions (c) and (d) of section 603 to embrace all categories of noncommercial messages, thereby preserving the ordinance’s neutrality and saving it from the constitutional problem raised by the United States Supreme Court in Metromedia[, Inc. v. San Diego, supra, 453 U.S. 490]. Given the high degree of tolerance for noncommercial communication exhibited by San Francisco’s sign legislation, we believe that, had the [board of supervisors] foreseen the First Amendment difficulties created by section 603, they would have chosen an interpretation which broadened, rather than narrowed permissible areas of ideological expression. [1] Finally, in order to avoid vagueness problems with the term ‘temporary’ [in section 603, subdivision (d)], we construe that phrase in accord with the modifying description set forth in section 603, subdivision (c) itself., i.e., signs ‘without independent structural support.’ ” (Eller, supra, 192 Cal.App.3d at pp. 664-665.)
In other words, the Eller court reformed section 603 of San Francisco’s ordinance by, inter alia, extending the exemption for most noncommercial signs to all noncommercial signs. In practical effect, the court rewrote the section to apply to “All noncommercial signs, including” those listed in the various subdivisions, in order to effectuate the local legislature’s intent.
5. Reformation of underinclusive statutes to avoid equal protection violation
Even before Arp, supra, 19 Cal.3d 395, we recognized a court’s authority to extend underinclusive statutory classifications and thereby reform statutes in order to avoid an equal protection violation. For example, in Hayes v. Superior Court (1971) 6 Cal.3d 216, 224 [98 Cal.Rptr. 449, 490 P.2d 1137], we unanimously reformed a statute that unreasonably discriminated against defendants convicted in California but imprisoned out of state, by extending to those persons the benefits enjoyed by those convicted and imprisoned in California. (Id. at p. 225.) In so reforming the statute, we stressed that doing *650so was most consistent with the apparent legislative intent, and explained: “A statutory classification which arbitrarily excludes some but not all of those similarly situated in relation to the legitimate purposes of the statute does not necessarily invalidate the entire statute. (Skinner v. Oklahoma:[, supra,] 316 U.S. 535, 543 [86 L.Ed. 1655, 1661]; In re King (1970) ... 3 Cal.3d 226, 237 [90 Cal.Rptr. 15, 474 P.2d 983].) In light of the purposes and history of a particular statute or an overall statutory scheme a reviewing court may correct a discriminatory classification by invalidating the invidious exemption and thus extending statutory benefits to those whom the Legislature unconstitutionally excluded.” (Hayes v. Superior Court, supra, 6 Cal.3d at p. 224, italics added.)50
Two years later, in Sykes v. Superior Court (1973) 9 Cal.3d 83 [106 Cal.Rptr. 786, 507 P.2d 90], we quoted and followed Hayes v. Superior Court, supra, 6 Cal.3d 216, and unanimously extended to those who establish a right to retrial by way of a writ the same statutory protections afforded those who establish a right to retrial by other legal processes. (Sykes v. Superior Court, supra, 9 Cal.3d at pp. 92-93.) Then, in In re Kapperman (1974) 11 Cal.3d 542 [114 Cal.Rptr. 97, 522 P.2d 657], we extended a statutory right to “presentence credit” to those imprisoned before the effective date of the legislation, despite the Legislature’s expressed intent to confine the benefit to those imprisoned after that date. In so reforming the statute we again quoted and followed Hayes v. Superior Court, supra, 6 Cal.3d 216, and expressed our conviction that such a “correction” was “consistent with probable legislative intent.” (In re Kapperman, supra, 11 Cal.3d at p. 550.)
*651As noted above, we reaffirmed the authority of a court to “rewrite” a statute in order to avoid invalidity for underinclusion in Arp, supra, 19 Cal.3d 395, 407 et seq. (See ante, pp. 641-643.) Shortly after we decided Arp, supra, the Court of Appeal relied on that decision in Fenske v. Board of Administration (1980) 103 Cal.App.3d 590 [163 Cal.Rptr. 182] to extend a disability retirement benefits statute otherwise unconstitutional under the equal protection clause. The court reasoned that extension rather than invalidation was most consistent with the Legislature’s intent. (Id. at pp. 597-598.)
Most recently, in a unanimous opinion by Justice Mosk, we followed these cases and the high court cases discussed ante, at pages 632-636, and reformed two benefits statutes that we found “invalid” under the equal protection clause. (Del Monte v. Wilson, supra, 1 Cal.4th 1009, 1026.) Military and Veterans Code sections 890 and 980 granted state veterans’ benefits, but limited those benefits to veterans who were natives or residents of California at the time of their entry into active service. In selecting reformation over invalidation, we judicially extended the benefits of the statutes to those who had been excluded by the Legislature, i.e., those who were not natives or residents of California at the time of their entry into active service. (1 Cal.4th at p. 1026.) In so doing we were “guided by the intent of the Legislature to aid veterans.” (Ibid., citing Westcott, supra, 443 U.S. 76, 89 [61 L.Ed.2d 382, 393], and Welsh, supra, 398 U.S. 333, 361 [26 L.Ed.308, 330-331] (conc. opn. of Harlan, J.).)
Amicus curiae for respondent asserts that Del Monte v. Wilson, supra, 1 Cal.4th 1009, like Welsh, supra, 398 U.S. 333, 361 et seq. [26 L.Ed.2d 308, 330-338] (conc. opn. of Harlan, J.), and its predecessors and progeny, is distinguishable from the present case in two respects. First, amicus curiae asserts, in the equal protection cases “the constitutional flaw was easily remedied by inclusion.” (Italics added.) Moreover, amicus curiae claims, those decisions “are representative of a number of cases that have expanded government programs when the only options available are expansion or destruction. All these cases present situations where a large number of people have relied on the existence of a program and would suffer injury if the program were abandoned.”
The first asserted distinction is unpersuasive. Amicus curiae suggests the equal protection/underinclusion cases are distinguishable on the ground that the remedy of extension is different in kind from the reformation under consideration here because extension cases do not involve “rewriting” of statutes. Justice Harlan undermined this argument in his concurring opinion in Welsh, supra, 398 U.S. at page 364 [26 L.Ed.2d at pages 332-333], when *652he acknowledged that “the . . . remedial operation [of] extension, is more analogous to a graft than amputation.” (See also id. at pp. 365, fn. 17 [26 L.Ed.2d at p. 333] [characterizing extension as judicial “amendment by expanding the scope of legislation”], 366 [26 L.Ed.2d at pp. 331-332] [characterizing extension as “building upon” a statute] (conc. opn. of Harlan, J.).) We similarly rejected amicus curiae’s proposed distinction in Arp, supra, 19 Cal.3d at pages 407-408, in which we candidly recognized that extension is synonymous with rewriting of statutes. Indeed, examination of the cases confirms this point and disproves amicus curiae’s theory. In each of the numerous extension cases discussed above, a court rendered a decision that effectively reformed a statute by including within its reach that which the enacting body implicitly—and sometimes expressly—excluded. In many of the decisions, the reformation not only altered operative language of the enactments in question, but also increased, potentially dramatically, the state’s financial burden by expanding the class of persons entitled to receive benefits. (See Ginsburg, supra, 28 Clev. St. L.Rev. at pp. 301, 305 [noting federal government estimated cost of extending social security statute in Weinberger v. Wiesenfeld, supra, 420 U.S. 636, to be $20 million per year].)
Amicus curiae is correct that many of the equal protection/extension cases involve situations in which invalidation would cause injury to “a large number of people.” Indeed, this is a persuasive consideration that often influences a court’s assessment of whether the enacting body would prefer a reformed version of a statute over invalidation of the statute. But, contrary to amicus curiae’s suggestion, there is no reason why the absence of this particular consideration should preclude reformation in appropriate cases outside the equal protection/underinclusion context, so long as a court may discern from other factors that the Legislature (here, the electorate) would have intended reformation over invalidation.
In one respect, the equal protection cases are distinguishable from the present litigation. The distinction, however, does not assist amicus curiae or interveners for respondent. The equal protection cases in which courts have “extended” statutes effectuate broad legislative policy judgments granting benefits to a named class, while declining to respect other policy judgments improperly limiting those benefits to the named class. In other words, these cases reflect a limited judicial encroachment into the Legislature’s policy-making domain, to the extent they elevate one “major” legislative policy judgment (e.g., granting benefits to some veterans) over a related, “minor” policy judgment (e.g., excluding certain veterans from benefits). As amicus curiae for respondent suggests, courts have apparently assumed that this limited encroachment into the legislative policy domain is justified in many equal protection/underinclusion cases on grounds akin to estoppel—i.e., as *653necessary to prevent injury to persons who “relied on the existence of a program and would suffer injury if the program were abandoned.” Assuming such limited judicial policymaking is justified and permissible in the equal protection/underinclusion context, we do not believe such judicial conduct is proper or justified outside that context.
With this caveat, we conclude the equal protection decisions are persuasive examples of judicial “rewriting” and that, in combination with the other categories of cases discussed above, they support the proposition that a court may, in appropriate cases, “rewrite” a statute to preserve its constitutionality.
C. Cases broadly asserting courts lack authority to “rewrite” statutes in order to preserve them against invalidity under the Constitution
As noted previously, amicus curiae and interveners on behalf of respondent assert we lack authority to reform a statute in order to make the law consistent with the Constitution. The high court, sister-state, and California cases discussed immediately above (pt. IV. A. & B.) amply refute that assertion. As we explain below, those decisions also serve to distinguish the numerous cases relied on by amicus curiae and interveners on behalf of respondent, in which we (and the high court) have broadly stated that a court lacks authority to “rewrite” a statute even to preserve its constitutionality.
As interveners and amicus curiae on behalf of respondent note, one of our most recent assertions in this regard was in Metromedia, Inc. v. City of San Diego, supra, 32 Cal.3d 180, in which we said: “There are limits, however, to the ability of a court to save a statute through judicial construction. As we explained in Blair v. Pitchess (1971) 5 Cal.3d 258 [96 Cal.Rptr. 42, 486 P.2d 1242, 45 A.L.R.3d 1206], 1 “[t]his court cannot . . . , in the exercise of its power to interpret, rewrite the statute. If this court were to insert in the statute all or any of the . . . qualifying provisions [required to make it constitutional], it would in no sense be interpreting the statute as written, but would be rewriting the statute in accord with the presumed legislative intent. That is a legislative and not a judicial function.” ’ (P. 282, quoting Seaboard Acceptance Corp. v. Shay (1931) 214 Cal. 361, 369 [5 P.2d 882]; see Flood v. Riggs (1978) 80 Cal.App.3d 138, 156-157 [145 Cal.Rptr. 573].” (Metro-media, Inc. v. City of San Diego, supra, 32 Cal.3d at p. 187.) Interveners and amicus curiae on behalf of respondents assert we should be guided by this and similar statements in numerous other cases and conclude we have no authority to reform a statute in order to preserve and conform it to constitutional requirements.
Close examination of the numerous cases cited by interveners and amicus curiae on behalf of respondent reveals, however, that with one antique *654exception, the decisions are all distinguishable from the federal and state cases cited ante, part IV. A. and B. Most involved situations in which reformation was inappropriate under the standard articulated today, because it was not possible to reform the statutes in a manner that closely effectuated policy judgments clearly articulated by the Legislature or enacting body. The remaining cases are distinguishable on a variety of grounds: some involved no constitutional infirmity; others did not involve reformation at all; and some of the decisions and authority affirmatively support the power of courts to reform in order to preserve constitutionality.
Metromedia, Inc. v. City of San Diego, supra, 32 Cal.3d 180, falls into the first category. The City of San Diego enacted an ordinance that banned erection of off-site billboards. After the superior court held the ordinance unconstitutional under the First Amendment, we reversed. (Metromedia, Inc. v. City of San Diego (1980) 26 Cal.3d 848 [164 Cal.Rptr. 510, 610 P.2d 407] (Metromedia I).) The United States Supreme Court in turn reversed our decision, holding that to the extent the ordinance banned noncommercial billboards, it violated the First Amendment. (Metromedia, Inc. v. San Diego, supra, 453 U.S. 490. Significantly, the high court did not declare the ordinance unenforceable, but instead remanded to our court to consider an issue of state law, i.e., whether the ordinance should be “sustain[ed] ... by limiting its reach to commercial speech.” (Id. at p. 521, fn. 26 [69 L.Ed.2d at p. 824].) We construed the high court’s order as requiring us to determine “whether the constitutionality of the ordinance could be saved by a limiting judicial construction of its terms or by severance of unconstitutional provisions from the balance of the enactment” (Metromedia, Inc. v. City of San Diego, supra, 32 Cal.3d at p. 182 (Metromedia III)), and proceeded to address those issues.
We began our discussion with the statement quoted above (and by interveners and amicus curiae on behalf of respondent), to the effect that we may not, in the exercise of our power to interpret or construe a statute consistently with the constitution, “rewrite” it. (Metromedia III, supra, 32 Cal.3d at p. 187.) Thereafter, we stressed that the intent of the ordinance drafters was to ban both commercial and noncommercial off-site billboards. We stated: “It does not appear, however, that the city intended to limit its ban to billboards which carried commercial messages. To the contrary, the city’s concern was not with the message but with the structure.” (Ibid., italics added.) We found it “clear that the San Diego City Council, in enacting the ordinance in question, intended to include noncommercial billboards” (id. at p. 189), but noted that purpose could not be given effect under high court authority. (Ibid.)
*655Any judicial reformation of the San Diego ordinance by limiting its coverage to commercial billboards would have required rewriting in a manner that ignored the city’s principal concern—the physical existence of all off-site billboards—and it would have substituted for that expressed concern an intention patently inconsistent with the rest of the ordinance, i.e., to ban only off-site commercial billboards. As we explained, such a reformed ordinance would not have closely effectuated the legislative purpose of the ordinance, because “the effect of such [a reformed] ordinance would depend on the extent to which persons are willing to purchase billboard space for noncommercial advertising” (Metromedia III, supra, 32 Cal.3d at p. 190), and “it would offer no assurance that a substantial number of billboards, or any particular billboard, would be removed, or that the erection of new billboards would be inhibited.” (Ibid.) We concluded that the city’s “legislative purpose may be better served by an ordinance which bans most off-site billboards than [a judicially reformed] one which draws a distinction based on the content of the billboard’s message.” (Id. at p. 191.)
Metromedia III, supra, was thus a case in which judicial reformation was inappropriate: It was impossible to reform the ordinance in a manner that closely effectuated policy judgments clearly articulated by the local legislature.
All but one of the remaining cases cited by amicus curiae and interveners on behalf of respondent are also distinguishable. In many of the cases, reformation designed to closely effectuate the enacting body’s policy judgments was impossible because the enacting body’s intent was plainly inconsistent with a saving or reformed construction of the enactment.51 In some of the cases, reformation designed to closely effectuate the enacting body’s *656policy judgments was impossible because the enacting body’s intent was unascertainable.52
Other cases cited by interveners and amicus curiae on behalf of respondent illustrate another important limitation on a court’s authority to reform statutes: they concern situations in which crucial policy judgments were not clearly articulated by the enacting body, and in which it was thus impossible to reform unconstitutional enactments because doing so would have required the court to substitute or render policy judgments. In Blair v. Pitchess, supra, 5 Cal.3d 258, for example, the plaintiffs sued to enjoin enforcement of the state’s “claim and delivery” law, a detailed statutory scheme (former Code Civ. Proc., §§ 509-521, enacted 1872) under which the plaintiff in an action to recover possession of property could post a bond, pay a fee, and require a law enforcement officer to seize property from the defendant, all before adjudication of the underlying property claim. We found the scheme so wholly infected with both Fourth Amendment (unreasonable search and seizure) and Fifth Amendment (seizure of property without due process of law) infirmities that, we concluded, the statutes could not be salvaged: “[I]n order to create a constitutional prejudgment replevin remedy, there must be provision for a determination of probable cause by a magistrate and for a hearing prior to any seizure, except in those few instances where important state or creditor interests justify summary process. No such safeguards can by any reasonable construction be found in sections 509 through 521; nor do those sections provide any clue as to which state or creditor interests are sufficiently important to warrant summary procedures. Consequently, we are compelled to invalidate the statute in its entirety and await a legislative redrafting.” (Blair v. Pitchess, supra, 5 Cal.3d at p. 283, italics added.) Thereafter the Legislature overhauled the scheme, making scores of changes in response to our decision. (See Code Civ. Proc., §§ 511.101-514.040.)53
*657Still other cases cited by interveners on behalf of respondent are distinguishable on yet another ground: they involved no constitutional infirmity, and hence did not pose the issue addressed herein, i.e., the conditions under which a court may reform an enactment in order to preserve its constitutionality.54 Indeed, as we implied in People v. One 1940 Ford V-8 Coupe, supra, 36 Cal.2d 471 (One 1940 Ford V-8 Coupe), the prospect that an enactment might be declared unconstitutional and unenforceable is a crucial factor that militates in favor of, and in appropriate situations mandates, judicial reformation of an enactment. In One 1940 Ford V-8 Coupe, supra, we specifically distinguished the simple statutory construction issue before us from the statutory reformation question posed in another case, in which an “exception” was “read into the statute” to avoid constitutional due process problems. (36 Cal.2d at pp. 475-476.) We declined to engage in the same reformation in One 1940 Ford V-8 Coupe, supra, precisely because “no constitutional obstacles” were presented on the facts of that case. (Id. at p. 476.)55
Finally, two cases cited by interveners on behalf of respondent actually hold reformation to be appropriate, and hence do not support their view.56 In the same vein, we question interveners’ reliance on 2A Sutherland, Statutory *658Construction (5th ed. 1992) Intrinsic Aids, section 47.38, page 291. The cited authority strongly supports rather than defeats the propriety of judicial reformation. It asserts that, in appropriate situations, courts may insert or add words “to prevent unconstitutionality” (ibid,, fn. omitted) and that “[a] majority of cases permit the elimination or disregarding of words in a statute” or “the substitution of one word for another if necessary to carry out the legislative intent or express clearly manifested meaning.” (Id., §§ 47.37, p. 283, 47.36, p. 277, fn. omitted.) Indeed, the cited page on which interveners rely contains the following; “Although some courts have been hesitant to supply or insert words, the better practice requires that a court enforce the legislative intent or statutory meaning where it is clearly manifested. The inclusion of words necessary to clear expression of the intent or meaning is in aid of legislative authority. The denial of the power to insert words when the intent or meaning is clear is more of a usurpation of the legislative power because the result can be the destruction of the legislative purpose.” (Id., § 47.38, p. 291, italics added, fn. omitted.)
As we acknowledged above, one antique case cited by interveners and amicus curiae on behalf of respondent supports their view. But, as explained below, it also serves to illustrate why courts have, in the intervening century, cautiously recognized the propriety of judicial reformation in order to preserve the constitutionality of statutes.
People v. Perry (1889) 79 Cal. 105 [21 P. 423] (Perry) was an action by one Davidson, who was appointed in November 1887 by Governor Waterman to be a member of the Board of Health of the City and County of San Francisco. Davidson sued Perry, who had been appointed seven months earlier to the same office by the former Governor, Bartlett. The question posed was whether the appointed office carried a fixed term, or whether the holder served at the pleasure of the Governor. The state Constitution had, since 1863, granted the Legislature authority to set the term, not exceeding four years, for such offices. (Perry, supra, 79 Cal. at pp. 113-114.) Nevertheless, in 1870, the Legislature set the office term at five years. (Id. at pp. 112-113.)
The court, with only four justices participating in the case, held the clause of the statute fixing the term at five years unconstitutional. It rejected Perry’s suggestion that the statute be reformed to set out a four-year term, and concluded that because “no term has been declared by law, . . . [Perry] held *659[his office] at the pleasure of the governor, and [Davidson’s] title is valid.” (Perry, supra, 79 Cal. at p. 114.) In reaching this conclusion, the court explained: “[W]e know of no precedent for holding that a clause of a statute, which as enacted is unconstitutional, may be changed in meaning in order to give it some operation, when admittedly it cannot operate as the legislature intended. [<][] This would, it seems to us, be making a law, and not merely correcting an excess of authority.” (Perry, supra, 79 Cal. at p. 115.)
Although the Perry court’s reluctance to reform the statute in order to preserve its constitutionality was consistent with the jurisprudence of the day (in which, for example, common and nonprejudicial errors in criminal litigation regularly gave rise to reversals of judgments), we conclude Perry’s 19th century view of the permissible judicial role regarding reformation of otherwise unconstitutional statutes is dubious authority today. Indeed, in hindsight, it is apparent that the court’s professed reluctance to invade the legislative domain actually did far more violence to the legislative scheme than would the proposed reformation. (See 2A Sutherland, Statutory Construction, supra, §47.38, p. 291, quoted ante, at p. 658.) The Legislature plainly intended to establish an office of fixed term lasting for as many years as the Constitution allowed. The Perry court’s refusal to reform the statute to provide for a four-year fixed term wholly frustrated the Legislature’s clear and principal intent. Moreover, as explained, ante, at pages 641-653, the result in Perry is inconsistent with the case law of this court over the past two decades. For these reasons, although we agree with interveners that Perry, supra, supports their position, we decline to afford that case any weight.
Although we do not otherwise question the analysis or holdings of the cases relied on by amicus curiae and interveners on behalf of respondent, for the reasons set out above we conclude the various cited statements from Metromedia III, supra, 32 Cal.3d 180, and its predecessors and successors (to the effect that a court may never “rewrite” a statute even to preserve its constitutionality), are distinguishable and overbroad dicta.57 Moreover, as explained above (pt. IV. A. & B.), we—and the high court—have expressly and implicitly contradicted and repudiated those overbroad statements in a *660substantial number of decisions in which we and the high court have reformed—i.e., rewritten—enactments to avoid various types of constitutional infirmities.
D. Summary: the propriety of, and standards for, judicial reformation of statutes
In all of the high court and California cases discussed ante, part IV. A. and B., the underlying principle is the same. Contrary to dictum in cases like Metromedia III, supra, 32 Cal.3d at page 187, and consistent with our statement in Arp, supra, 19 Cal.3d at page 407, a court may reform—i.e., “rewrite”—a statute in order to preserve it against invalidation under the
*661Constitution, when we can say with confidence that (i) it is possible to reform the statute in a manner that closely effectuates policy judgments clearly articulated by the enacting body, and (ii) the enacting body would have preferred the reformed construction to invalidation of the statute. By applying these factors, courts may steer clear of “judicial policymaking” in the guise of statutory reformation, and thereby avoid encroaching on the legislative function in violation of the separation of powers doctrine. (See Cal. Const., art. IV, § 1; id.., art. VI, § l.)58
In articulating a reformation designed to effectuate the Legislature’s or electorate’s intent, courts search for phrasing that would disturb as little as possible the language of the statutes as enacted. At the same time, however, we focus not only on the number of words involved in a proposed reformation, but more important, on the quality of the proposed change. As a general matter, it is impermissible for a court to reform by supplying terms that disserve the Legislature’s or electorate’s policy choices.59 By contrast, when legislative (or the electorate’s) intent regarding policy choice is clear, a revision that effectuates that choice is not impermissible merely because it requires insertion of more words than it removes.
In summary, and to paraphrase Justice Ginsburg, supra, 23 Clev. St. L.Rev. at page 324, we conclude courts may legitimately employ the power to reform in order to effectuate policy judgments clearly articulated by the Legislature or electorate, when invalidating a statute would be far more destructive of the electorate’s will. And, “of course . . . ultimate authority to recast or scrap the law in question remains with the political branches [and, as in this case, the electorate].” (Ibid.)
*662Before deciding the specific reformation question posed in this proceeding, we briefly address generally the reformation of initiative statutes.
V. Reformation of Initiative Statutes
Interveners on behalf of respondent assert courts are precluded from reforming initiative statutes, because such judicial action would operate free from the procedural protections that apply to legislative amendment of initiative statutes. Petitioners and intervener on their behalf, joined by the dissenting justices today, advance the opposite view, asserting a court should be especially willing to reform initiative statutes, especially those that concern legislative elections. We accept neither view.
Interveners on behalf of respondents observe that initiative statutes, “[f]or better or worse ... are meant to be set in stone. The Legislature is prohibited from amending them without going back to the people” unless the initiative provides otherwise. (See Cal. Const., art. II, § 10, subd. (c) [The Legislature “may amend or repeal an initiative statute by another statute that becomes effective only when approved by the electors unless the initiative statute permits amendment or repeal without their approval.”].) Proposition 73 permits legislative amendment “to further its purposes,” but allows such amendment only by a two-thirds vote. (§§ 81012, subd. (a), 85103.) Interveners on behalf of respondent conclude that judicial reformation of the statutes at issue here is tantamount to an amendment of them, and that such action is barred by article II, section 10, subdivision (c) of the state Constitution.
We reject the view that courts are barred from reforming initiative statutes. As described above, we have recognized and applied substantial safeguards to ensure both that any proposed reformation closely effectuates policy judgments clearly articulated by the electorate, and that the electorate would have preferred the reformed version of the statute over the constitutionally invalid and unenforceable version. More important, nothing we might do by way of reformation would impair the Legislature’s authority under article II, section 10, subdivision (c) of the California Constitution, to amend or repeal initiative statutes such as section 85301(a) or section 85303(a) and (b).
Although the concerns raised by interveners on behalf of respondent do not justify a conclusion that courts are precluded from reforming initiative statutes, those same points militate against the opposite view of petitioners and their supporters, i.e., that courts should be especially willing to reform such statutes. Granted, we have said the people’s initiative power is to be *663jealously guarded and liberally construed. (Raven v. Deukmejian (1990) 52 Cal.3d 336, 341 [276 Cal.Rptr. 326, 801 P.2d 1077].) But we perceive no principled basis for a general presumption favoring judicial reformation in such cases. Nor do we find persuasive Justice Baxter’s suggested “special” presumption favoring reformation of initiative statutes that the Legislature may be expected to dislike. (See conc. & dis. opn. of Baxter, J., post, at pp. 686-687.) We conclude that in all cases, reformation should be tested objectively against the standard set out herein.
VI. Reformation of Sections 85301-85304
A. Preliminary observations regarding sections 85302 and 85304
As an initial matter, we narrow substantially the scope of our inquiry. As suggested ante, footnote 10, we view the federal courts’ decisions as reflecting an implied determination that unless the limitations on contributions to candidates set out in sections 85301(a), 85303(a) and (b) are enforceable, section 85302, which regulates contributions to political committees and parties, would itself remain unenforceable. We agree with that determination, and accordingly defer our consideration of whether section 85302 may be reformed.
We further confine the scope of our inquiry by determining, at the threshold, that there is no basis for reforming section 85304, which establishes both intra- and inter-candidate transfer bans. As explained above, the intra-candidate aspect of section 85304, which the courts found to constitute an unconstitutional spending limitation, is not at issue in this proceeding. (See ante, pp. 617 & 619; see generally, Buckley v. Valeo, supra, 424 U.S. 1, 54-59 [46 L.Ed.2d 659, 707-710].) As also explained above, the inter-candidate transfer ban aspect of section 85304, which appears to operate as a contribution limitation, was invalidated by the federal court of appeals on dual grounds, one of which was that the section is overbroad because it prohibits both large and small transferred contributions, and is thus “not ‘closely drawn to avoid unnecessary abridgment of associational freedoms’ ” as required by Buckley v. Valeo, supra, 424 U.S. 1, 25 [46 L.Ed.2d 659, 691]. (Service Employees II, supra, 955 F.2d at p. 1323.) Contrary to the view of petitioners and intervener on their behalf, the federal appeals court’s analysis and conclusion in this latter regard would not be undermined by reformation of the “fiscal year” measure of sections 85301 and 85303, and hence section 85304 will remain as enacted and enjoined.60
We thus focus on whether sections 85301(a) and 85303(a) and (b) (regulating contributions to candidates) may be reformed in a manner that renders them constitutional.
*664B. Whether it is possible to reform sections 85301 and 85303 in a manner that closely effectuates policy judgments clearly articulated by the electorate
The enjoined sections reflect three key policy judgments: First, the sections establish a maximum dollar amount for particular contributions. Second, the sections regulate the pace at which those contributions may be made. Finally, the sections implicitly establish the rights to contribute and to accept a theoretical maximum aggregate amount of funds. A proper reformation must closely effectuate each of these policy judgments.
As noted above, the federal appeals court broached the question whether the sections might be reformed or saved by merely striking the word “fiscal” from the term “fiscal year,” but declined to do so because that would create an annual measure and hence lead to the same discriminatory effect perceived as unconstitutional. We agree.
The federal appeals court also considered striking the term “fiscal year” and replacing it with the term “election cycle,” but declined do so because, it reasoned, “we would be at a loss to know what the dollar amounts of the limitations should be.” (Service Employees II, supra, 955 F.2d at p. 1321.)
Petitioners and intervener on their behalf eschew the “election cycle” approach suggested by the federal appeals court,61 and propose instead that the statutes be reformed by inserting a “per election” measure similar to that employed in an analogous federal statute. (See 2 U.S.C. § 431(1)(A) [“The term ‘election’ means [U . . . a general [or] . . . primary . . . election. . . .”].)
We conclude, however, that such a reformation would not closely effectuate policy judgments clearly articulated by the electorate. Granted, it would retain the maximum amounts for particular contributions established *665by the sections, and it would effectuate the pacing requirement, at least in the closing months of a campaign, when contributions are arguably most important. But it would promote those policy judgments at the expense of reducing the theoretical maximum aggregate amount of funds that could be contributed to and accepted by most candidates. Specifically, although the “per election” approach would generally62 maintain the theoretical maximum aggregate amount of funds that could be contributed to and accepted by most candidates for offices with two-year terms (i.e., Assembly races),63 it would reduce by 50 percent (see ante, fn. 62) the theoretical maximum aggregate amount of funds that could be contributed or accepted for all partisan offices with four-year terms (i.e., Senate races and most statewide offices), and it would reduce by 75 percent (see ante, fn. 62) the theoretical maximum aggregate amount of funds that could be contributed or accepted for many if not most nonpartisan offices with four-year terms (i.e., most local government offices).64
Intervener on behalf of petitioners acknowledges these consequences of the proposed “per election” reformation, but insists this approach is nevertheless consistent with the electorate’s intent. First, focusing on section 85301(a), intervener on behalf of petitioners asserts the section was not intended to “ensure that contributors would be allowed to donate up to the $4,000 . . . maximum amounts theoretically allowed for four-year offices under the initiative’s fiscal year limits. Rather, ... the voters were primarily motivated by a desire to restrict the size of particular contributions to no *666more than $1,000 . . . Second, intervener on behalf of petitioners cites the corresponding provisions of Proposition 68—a rival campaign reform measure that received a lesser number of affirmative votes at the same election at which Proposition 73 was enacted, and which, accordingly, remains “inoperative.” (See Taxpayers to Limit Campaign Spending v. Fair Pol. Practices Com., supra, 51 Cal.3d 744, 771; Gerken, supra, 6 Cal.4th 707, 720.) Intervener observes that under the proposed “per election” reformation of sections 85301(a) and 85303(a) and (b), the maximum contribution to any given statewide or legislative candidate would be the same as the amount set out in the corresponding provisions of Proposition 68.65 Intervener on behalf of petitioners concludes from this that the voters intended to enact the reduced theoretical maximum aggregate contribution amounts that would result under a “per election” reformation of sections 85301(a) and 85303(a) and (b).
As interveners and amicus curiae for respondent suggest, and as Justice Baxter’s concurring and dissenting opinion shows, the argument of intervener for petitioners does not withstand scrutiny. First, assuming for purposes of analysis that the electorate’s “primary” motivation was to restrict the size of particular contributions, this affords no ground to presume that the statute was not also purposefully designed to accomplish precisely what it would have done in practice, i.e., afford each individual or institutional contributor the opportunity to give, and each candidate the opportunity to accept, a theoretical maximum aggregate amount of contributions—in other words, up to $4,000, $10,000, and $20,000, for a Senate candidate under sections 85301(a) and 85303(a) and (b), respectively.
Second, Proposition 68 is inoperative because the voters stated a preference for its rival, Proposition 73. (Gerken, supra, 6 Cal.4th 707, 720.) The fact that a majority of the electorate voted for Proposition 68 and the lesser maximum aggregate contribution amounts contained therein cannot, and does not, negate the fact that a greater majority voted for Proposition 73 and its more generous maximum aggregate contribution amounts. Moreover, because the lower aggregate contribution limitations set out in Proposition 68 were part of a package that also would have established spending limitations and partial public matching funds for campaigns (see inoperative §§ 85400-85405, 85500-85506, proposed by Prop. 68 and approved by electors June 7, 1988), the contribution amounts are to that extent deflated, and hence not “transferable” to Proposition 73, which specifically bans public financing of campaigns. (See § 85300.)
*667We conclude a “per election” reformation of sections 85301(a) and 85303(a) and (b) would not closely effectuate the right of contributors to give, and candidates to accept, the theoretical maximum aggregate amount of contributions contemplated by the statutes as enacted. Accordingly, we decline to reform the statutes in the fashion advocated by petitioners and interveners on their behalf.
We also decline to revise the statutes in the manner suggested by Justice Baxter. He and the justices joining his concurring and dissenting opinion would reform the three statutes “to limit campaign contributions and loans to candidates to no more than $1,000 (for § 85301, subd. (a)), $2,500 (for § 85303, subd. (a)) and $5,000 (for § 85303, subd. (b)) multiplied by the number of years of the term of office sought by the candidate, and in partisan races, to further incorporate a ‘per election’ pacing mechanism so that no more than one-half the total allowable amount may be contributed prior to the primary election and no more than one-half of the total allowable amount may be contributed between the primary election and the general election.” (Conc. & dis. opn. of Baxter, J., post, at p. 689.)
Because no party advocated or anticipated a reformation similar to that suggested by Justice Baxter, we sought briefing on the issues raised by such a reformation. Our letter asked, inter alia, whether “(i) it would be necessary or possible to reform the statutes in a manner that closely replicates the total theoretical maximum contribution amounts that would have been allowed under the statutes as enacted, and (ii) would it be necessary or possible for such a judicial reformation to retain a mechanism regulating the pace of contributions, similar to that existing in the statutes as enacted?”
The opponents of reformation—interveners and amicus curiae on behalf of respondent—answered that both objectives are necessary, but neither is possible without this court engaging in extensive rewriting of the statutes or without imposing its own policy judgments in place of those reflected in the statutes as enacted.
Petitioners, although not clearly addressing the questions posed, significantly declined to embrace any suggestion that reformation must allow candidates to receive the theoretical maximum aggregate contributions that would have been allowed under Proposition 73 as enacted.
Intervener on behalf of petitioners argued it would be improper to reform the statutes in a manner that permits candidates to receive the theoretical *668maximum aggregate amounts allowable under the statutes as enacted. It acknowledged that its own “per election” approach would diminish substantially the theoretical maximum, but asserted that problem is more theoretical than real, because most fundraising occurs toward the close of a campaign— i.e., near the primary and general election stages. Moreover, intervener claimed, a reformation that attempts to replicate the total theoretical maximum aggregate contribution amounts that would have been allowed under the statutes as enacted, would require improper “wholesale rewriting” of the statutes, and would be unnecessary in any event because it would conflict with the voters’ primary intent, which was to “restrict the size of particular contributions . . . .”
The supplemental brief of respondent is perhaps most significant. Respondent reaffirmed its neutrality on the ultimate question of whether the statutes should be reformed. At the same time, it strongly suggested that if we do reform, we should do so in the manner suggested by petitioners, and not in a manner that attempts to replicate the theoretical maximum aggregate contributions that would have been allowed under the statutes as written (i.e., Justice Baxter’s approach) because that would “in effect . . . raise the contribution dollar amount above the level which voters were attempting to impose” and allow last-minute lump-sum payments well over the $1,000, $2,500, and $5,000 per fiscal year limitations, thus benefiting candidates who do not face competitive primaries. In this vein, respondent’s supplemental brief repeatedly asserted that petitioners’ proposed reformation, compared with a scheme (such as proposed by Justice Baxter) that attempts to preserve total campaign contributions in the amounts allowed under Proposition 73, ubest adheres to the original statutory scheme and preserves the original intent of the electorate.” (Italics added.) Respondent also repeatedly asserted that petitioners’ approach, compared with a scheme that attempts to preserve total campaign contributions in the amounts allowed under Proposition 73, is “closest to [that] contemplated by Proposition 73.” (Italics added.)66 Finally, respondent asserted that although it would be possible to administer a scheme that permits candidates to receive the theoretical aggregate máximums that they would have been allowed under the statutes as enacted, such a scheme would be undesirable because it would be neither simple nor uniform and it would be difficult to administer because, inter alia, there would be at least nine different contribution formulas for candidates for various offices.
*669In other words, no party or amicus curiae endorses Justice Baxter’s approach.67
We conclude the reasoning underlying the supplemental briefing forecloses the extensive and novel reformation proposed by Justice Baxter. Indeed, that proposed reformation is impermissible because it is the mirror image of the “per election” approach: Although it would closely effectuate the policy of protecting rights to contribute and to accept a theoretical maximum aggregate amount of funds, as explained below, it would advance that policy at the expense of doubling, and in some cases, quadrupling the maximum amount of particular contributions. Moreover, in a significant category of elections, it would impose no pacing regulation, thus allowing lump-sum contributions barred by the statutes as enacted.
Under Justice Baxter’s approach, in the 12 months before a general election, Senate candidates in partisan races would be allowed to receive $4,000 from individuals—$2,000 for the primary, and $2,000 for the general election. By contrast, under the statutes as enacted, during the same period the same candidates would have been limited to $2,000, i.e., $1,000 during the fiscal year in which the primary is held and $1,000 during the fiscal year in which the general election is held. Similarly, under Justice Baxter’s approach, in the 12 months before an election, candidates in nonpartisan local contests for 4-year terms—i.e., candidates in most mayoral, council, and supervisorial races—would be allowed to receive $4,000 from individuals at any time. Again, by contrast, under the statutes as enacted, during the same period the same candidates would have been limited to $1,000 (with the possibility of another $1,000 in the event of a runoff election held in a subsequent fiscal year).
The $1,000, $2,500, and $5,000 caps on contribution amounts set out in sections 85301(a) and 85303(a) and (b), together with the pacing mechanism limiting contributions by fiscal year, were obviously intended to limit the opportunity for lump-sum contributions at any time during a campaign. By doubling and in some cases quadrupling those contribution amounts, and by allowing a substantial category of candidates to solicit and accept those contributions in a lump sum at any time, Justice Baxter’s proposed reformation would substantially disserve the electorate’s policy choices.68
Having concluded that neither proposed reformation is permissible, we need not decide whether the electorate would have preferred either of them *670to invalidation. We observe, however, that an affirmative response is not so self-evident as petitioners and Justice Baxter suggest. As noted above, petitioners’ “per election” reformation would impose regulations stricter than those set out in the statutes as enacted, and we might assume, at least for purposes of analysis, that the electorate would have preferred that reformation over invalidation. But, as also noted above, Justice Baxter’s “modified election cycle” reformation would produce a scheme considerably more lenient than the statutes as enacted. We would be extremely reluctant to presume that the electorate might have preferred such a “half loaf’ compromise over invalidation. Indeed, we find it at least as likely that the electorate would prefer to start anew in order to create a comprehensive and coherent scheme, rather than settle for a makeshift and ill-fitting law that might actually hamper future campaign financing reform by creating an illusion that full and complete reform had been achieved. Accordingly, we conclude that it would be impossible to determine with confidence that the electorate would have preferred the reformation proposed by Justice Baxter over invalidation.69
VII. Conclusion
Contrary to interveners and amicus curiae on behalf of respondent, the separation of powers doctrine guides, but does not invariably preclude, judicial rewriting of statutes to preserve constitutionality. Consistent with the doctrine, a court has authority to reform statutes by rewriting them to preserve constitutionality when it can conclude with confidence that (i) it is possible to reform the statute in a manner that closely effectuates policy judgments clearly articulated by the enacting body, and (ii) the enacting body would have preferred such a reformed version of the statute to invalidation of the statute. Indeed, in an appropriate case it would be our duty to rewrite in order “to prevent unconstitutionality.” (2A Sutherland, Statutory Construction, supra, Intrinsic Aids, § 47.38, p. 291.) As observed above, the “better practice requires that a court enforce the legislative intent or statutory meaning where it is clearly manifested. The inclusion of words necessary to clear expression of the intent or meaning is in aid of legislative authority. The denial of the power to insert words when the intent or meaning is clear is more of a usurpation of the legislative power because the result can be the destruction of the legislative purpose.” (Ibid., italics added; see also id., § 47.37, p. 283; id., § 47.36, p. 277.) And, as also noted above, when legislative (or the electorate’s) intent regarding policy choice is clear, a *671revision that effectuates that choice is not impermissible merely because it requires insertion of more words than it removes.
We also conclude, however, that reformation is inappropriate here, and cannot be accomplished consistently with the limitations placed on courts by the separation of powers doctrine. We decline to reform section 85304 because the federal appeals court’s analysis of, and conclusion regarding, that section would not be undermined by reformation of the “fiscal year” measure of sections 85301(a) and 85303(a) and (b). In turn, we may not reform sections 85301(a) and 85303(a) and (b) (and, accordingly, we may not reform section 85302) because it is impossible to do so in a manner that closely effectuates policy judgments clearly articulated by the electorate. The “per election” approach would allow persons to give, and candidates to accept, less funding than the electorate contemplated. By contrast, Justice Baxter’s novel and unsupported “modified election cycle” approach would allow candidates more funds than the electorate planned, and it would remove any regulation of the pace of contributions for the numerous nonpartisan municipal and county elections throughout the state. Because neither reformation would closely effectuate policy judgments clearly expressed by the electorate, it follows that either would impermissibly intrude on the policy-making functions reserved to the Legislature and the people (Cal. Const., art. IV, § 1), and hence neither reformation is permissible.
Accordingly, we deny the relief requested by petitioners.
Werdegar, J., concurred.
All further statutory references are to this code unless otherwise indicated.
That section reads in full as follows: “(a) No person shall make, and no candidate for elective office, or campaign treasurer, shall solicit or accept any contribution or loan which would cause the total amount contributed or loaned by that person to that candidate, including contributions or loans to all committees controlled by the candidate, to exceed one thousand dollars ($1,000) in any fiscal year. [<J] (b) The provisions of this section shall not apply to a candidate’s contribution of his or her personal funds to his or her own campaign contribution account.”
That section reads in full as follows: “No person shall make and no political committee, broad based political committee, or political party shall solicit or accept, any contribution or loan from a person which would cause the total amount contributed or loaned by that person to the same political committee, broad based political committee, or political party to exceed two thousand five hundred dollars ($2,500) in any fiscal year to make contributions to candidates for elective office.”
Section 85303 reads in full as follows: “(a) No political committee shall make, and no candidate or campaign treasurer shall solicit or accept, any contribution or loan which would cause the total amount contributed or loaned by that committee to that candidate for elective office or any committee controlled by that candidate to exceed two thousand five hundred dollars ($2,500) in any fiscal year. [^Q (b) No broad based political committee or political party shall make and no candidate or campaign treasurer shall solicit or accept, any contribution or loan which would cause the total amount contributed or loaned by that committee or political party to that candidate or any committee controlled by that candidate to exceed five *614thousand dollars ($5,000) in any fiscal year. [QQ (c) Nothing in this Chapter shall limit a person’s ability to provide financial or other support to one or more political committees or broad based political committees provided the support is used for purposes other than making contributions directly to candidates for elective office.”
That section reads in full as follows: “No candidate for elective office or committee controlled by that candidate or candidates for elective office shall transfer any contribution to any other candidate for elective office. Transfers of funds between candidates or their controlled committees are prohibited.”
The legislator interveners are Assemblyman Willie L. Brown, Jr., and Senators Bill Lockyer, Ruben S. Ayala, and Nicholas C. Petris.
The Howard Jarvis Taxpayers Association and 31 members of the California Assembly are amici curiae on behalf of petitioners. The California Democratic Party is amicus curiae on behalf of respondent.
The term “fiscal year” is defined as covering the period between “July 1 through June 30.” (§ 85102, subd. (a).) Because at the time Proposition 73 was enacted all California primary elections were held in June, the fiscal year scheme permitted a contributor who gave a candidate the maximum allowable before the primary election, to make another maximum contribution between the primary election and the general election. (See Service Employees I, supra, 747 F.Supp. at p. 589; see also Elec. Code, § 23 [moving some California primary elections from June to March].)
An example illustrates the federal district court’s point: Section 85301(a) would allow an incumbent to accept up to $1,000 per fiscal year from an individual, thus effectively allowing an incumbent to a four-year term to accept $4,000 from an individual contributor (i.e., $1,000 per fiscal year) during the term of office. A challenger, on the other hand, although theoretically allowed to solicit that same $4,000 (or slightly more—see post, p. 665, fn. 62) during the same period, would in practice be restricted to soliciting substantially less, i.e., *617$1,000 in each fiscal year immediately preceding the election, which would likely restrict the challenger to about $2,000, depending, on, inter alia, when he or she decided to run and began accepting contributions.
As noted above, the federal district court included section 85302 in its injunction, and the United States Court of Appeals for the Ninth Circuit affirmed that judgment. As intervener on behalf of petitioners impliedly concedes, it is reasonable to assume the courts enjoined section 85302 based on their implicit determination that unless the limitations on contributions to candidates set out in sections 85301(a) and 85303(a) and (b) are enforceable, section 85302, which regulates contributions to political committees and parties, would itself be unenforceable. (Alternatively, the courts might have impliedly determined that, in practical operation, fundraising under section 85302’s “fiscal year” regulation would indirectly produce the same “discrimination against challenger candidates” as would sections 85301(a) and 85303(a) and (b), because incumbents, better than challengers, would be positioned to solicit “section 85302 contributions” and thereafter receive all or part of those funds under section 85303(a) or (b).)
The plaintiffs challenged one additional provision of Proposition 73, i.e., the “carryover” restriction, which prohibits expenditure of campaign funds raised prior to January 1989 (§ 85306). The district court, on partial motion for summary judgment, held that provision to be an unconstitutional expenditure limitation, and permanently enjoined its enforcement. (See Service Employees Intern. Union, AFL-CIO v. FPP (E.D.Cal. 1989) 721 F.Supp. 1172.) The federal appellate court affirmed that aspect of the judgment in Service Employees II, supra, 955 F.2d at page 1323. Section 85306 was not subject to our order to show cause, and is not at issue in this proceeding.
Regarding the three cited cases, see post, footnote 57.
Intervener on behalf of petitioners suggests the present litigation poses an issue of collateral estoppel (issue preclusion), not res judicata (claim preclusion). (See Clark v. Lesher (1956) 46 Cal.2d 874, 880 [299 P.2d 865] [describing collateral estoppel aspect of res judicata].) We need not resolve this characterization question because, as explained below, neither doctrine bars the litigation here.
As the Ohio Supreme Court has observed, a state’s sovereignty “is unquestionably implicated when the federal court construes state law” because, if the federal court errs, “it applies law other than Ohio law, in derogation of the state’s right to prescribe a ‘rule of decision.’ ” (Scott v. Bank One Trust Co., N.A. (1991) 62 Ohio St.3d 39 [577 N.E.2d 1077, 1080].) In litigation such as the present, concerning reformation, under state law, of important and prominent state statutes, that sovereignty interest is heightened.
There the court said: “When a federal court declares part of a state statute unconstitutional, it acts under the authority of the Supremacy Clause. In order to decide the remedial issue before us—whether a federal court should supply limiting language to preserve the constitutionality of a state statute, thus making policy choices usually made by the state legislature—we must take into account a number of factors. One of those factors is the policy of the [state] respecting judicial revision of [state] statutes. . . .” (937 F.2d at p. 1122.)
Justice Kennard asserts we are bound by federal, and not state, law when determining whether an exception to the doctrines of res judicata or collateral estoppel applies. (Cone. & dis. opn. of Kennard, J., post, at pp. 682-683.) For this proposition the concurring and dissenting opinion cites two cases (Hagen v. Utah (1994) 510 U.S. 399 [127 L.Ed.2d 252, 263-264, 114 S.Ct. 958, 964] and Martin v. Martin (1970) 2 Cal.3d 752, 761 [87 Cal.Rptr. 526, 470 P.2d 662]) and the Restatement Second of Judgments (1982) section 87, none of which addresses the question posed, i.e., the law to be applied by a state court when considering a claimed exception to res judicata or collateral estoppel following a federal court judgment. At the same time, the concurring and dissenting opinion plainly misreads Dombrowski v. Pfister, supra, 380 U.S. 479, 491-492 [14 L.Ed.2d 22, 31-32], in which, as noted above, the high court explicitly recognized that a party enjoined by a federal court from enforcing an unconstitutionally overbroad state statute would be permitted thereafter to seek and obtain a state court reformation or narrowing of that same statute. Clearly, Dombrowski v. Pfister, supra, conflicts with the concurring and dissenting opinion’s unsupported assertion that we are precluded by federal law from considering the issues raised in this petition.
Justice Kennard’s concurring and dissenting opinion also asserts that “even under state law the ‘public interest’ exception is invoked only in exceptional circumstances when it may be necessary to relieve a public agency of an erroneous earlier determination of the issue." (Post, at p. 684, italics added.) The concurring and dissenting opinion concludes that because three justices of this court find the federal courts’ determinations were not erroneous, the “public interest” exception cannot apply. (Ibid.) But the concurring and dissenting opinion misstates *623the law. Neither Arcadia Unified School Dist. v. State Dept, of Education, supra, 2 Cal.4th 251, 256-259, nor City of Sacramento v. State of California, supra, 50 Cal.3d 51, 64-65, nor any other authority, suggests such a rule. As explained above, the established rule is that the prior legal determination is not conclusive either if injustice would result or if the public interest requires that relitigation not be foreclosed. (City of Sacramento v. State of California, supra, 50 Cal.3d 51, 64.) Although this matter clearly involves a public agency, no authority confines application of the public interest exception to cases concerning public agencies. More important, no authority conditions application of the public interest exception on a finding that the initial determination of an issue was “erroneous,” and Arcadia Unified School Dist. v. State Dept, of Education, supra, 2 Cal.4th 251, which Justice Kennard joined without reservation, is clearly inconsistent with such an approach. In that case we found the public interest exception applied (and only thereafter addressed the merits of the issue) because “[i]f the action were barred from going forward, then the state of the law on a matter of statewide importance would remain permanently unclear and unsettled.” (Id. at p. 259, italics added.) Indeed, the concurring and dissenting opinion proposes an impossibility. It would require a court to resolve the matter on its merits, before deciding whether it should even address the matter on its merits. Tellingly, however, the concurring and dissenting opinion fails to follow its own unsupported and circular position—it rejects application of the public interest exception while simultaneously professing not to address or resolve the merits of the issue presented. (Cone. & dis. opn. of Kennard, J., post, at pp. 678-679, 684.)
The court’s dispositional order recited; “1. All contribution limitations measured on a fiscal year basis violate the provisions of the First Amendment to the Constitution of the United States and are hereby declared unenforceable; accordingly, enforcement of such provisions [is] Permanently Restrained; . . .” (Service Employees I, supra, 747 F.Supp. at p. 593.) The court’s judgment regarding the inter-candidate transfer ban of section 85304 tracked that language and approach. (747 F.Supp. at p. 594.)
As noted above, the federal appeals court found section 85304’s inter-candidate transfer ban “unconstitutional” on dual grounds, one of which was that it was “not ‘closely drawn to avoid unnecessary abridgment of associational freedoms.’ ” (955 F.2d at p. 1323.)
A few prior decisions implicitly recognized that power. (See Levy v. Louisiana (1968) 391 U.S. 68 [20 L.Ed.2d 436, 88 S.Ct. 1509]; Glona v. American Guarantee & Liability Ins. Co. (1968) 391 U.S. 73 [20 L.Ed.2d 441, 88 S.Ct. 1515]; Shapiro v. Thompson (1969) 394 U.S. 618 [22 L.Ed.2d 600, 89 S.Ct. 1322] [all three cases described, post, fn. 20]; Skinner v. Oklahoma (1942) 316 U.S. 535, 543 [86 L.Ed. 1655, 1661, 62 S.Ct. 1110] [described, post, fn. 33]; Nat. Life Ins. Co. v. United States (1928) 277 U.S. 508, 522 [72 L.Ed. 968, 971-972, 48 S.Ct. 591] [extending statutory exemption to excluded class]; Yale & Towne Mfg. Co. v. Travis (S.D.N.Y. 1919) 262 F. 576, 582, affd. 252 U.S. 60, 82 [64 L.Ed. 460, 470, 40 S.Ct. 228] [same]; Burrow v. Kapfhammer (1940) 284 Ky. 753 [145 S.W.2d 1067, 1072] [same]; Quong Ham Wah Co. v. Industrial Acc. Com., supra, 184 Cal. 26, 39-41 [same]; cf. Iowa-Des Moines National Bank v. Bennett (1931) 284 U.S. 239, 247 [76 L.Ed. 265, 272-273, 52 S.Ct. 133] [extending benefit of state error in miscalculation of taxes to entity whose tax was correctly calculated].)
See, e.g., Levy v. Louisiana, supra, 391 U.S. 68 (reversing state court judgment refusing to extend to children bom out of wedlock the right, under state law, to sue for wrongful death of mother); Glona v. American Guarantee & Liability Ins. Co., supra, 391 U.S. 73 (reversing state court judgment refusing to extend to mothers of illegitimate children the right, under state law, to sue for wrongful death of the child); Shapiro v. Thompson, supra, 394 U.S. 618 (affirming district court judgments extending welfare assistance to those who did not satisfy one-year residence requirement); see also cases cited, ante, fn. 19.)
See also Richardson v. Griffin (1972) 409 U.S. 1069 [34 L.Ed.2d 660, 93 S.Ct. 689] (summarily affirming, under Weber, supra, district court judgment extending to dependent children bom out of wedlock equal entitlement to Social Security benefits); Richardson v. Davis (1972) 409 U.S. 1069 [34 L.Ed.2d 659, 93 S.Ct. 678] (same).
U.S. Dept, of Agriculture v. Moreno (1973) 413 U.S. 528 [37 L.Ed.2d 782, 93 S.Ct. 2821] (affirming district court judgment extending full food stamp benefits to households in which members are related).
See Gomez v. Perez (1973) 409 U.S. 535 [35 L.Ed.2d 56, 93 S.Ct. 872] (reversing state court judgment refusing to extend to children of unwed parents a state law right of parental support); New Jersey Welfare Rights Organization v. Cahill (1973) 411 U.S. 619 [36 L.Ed.2d 543, 93 S.Ct. 1700] (reversing district court judgment refusing to extend to families with parents who were not “ceremonially married to each other,” benefits from state program for assistance to families of the working poor); Frontiero v. Richardson (1973) 411 U.S. 677, 691 [36 L.Ed.2d 583, 594-595, 93 S.Ct. 1764] and fn. 25 (reversing district court judgment refusing to extend housing allowance benefits to women military personnel on the same basis that benefits were granted to servicemen, and suggesting extension of statute, rather than invalidation, was proper course); Memorial Hospital v. Maricopa County (1974) 415 U.S. 250 [39 L.Ed.2d 306, 94 S.Ct. 1076] (reversing state court judgment refusing to extend nonemergency medical assistance to those who did not satisfy county’s one-year residence requirement); Jimenez v. Weinberger (1974) 417 U.S. 628 [41 L.Ed.2d 363, 94 S.Ct. 2496] (vacating and remanding district court judgment refusing to extend social security benefits to certain children bom out of wedlock).
The concurring and dissenting justices accepted the general propriety of judicial extension of statutes that would otherwise be invalid under the equal protection clause. (443 U.S. at p. 94 [61 L.Ed.2d at pp. 396-397] (cone, and dis. opn. of Powell, J.).) Their disagreement stemmed from application of the power in the case before the court—they believed extension would “frustrate” and “circumvent” Congress’s intent. (Ibid.)
On remand, the Missouri and Alabama courts extended their statutes. (See Wengler v. Druggists Mut. Ins. Co. (Mo. 1980) 601 S.W.2d 8, 9 (following intervening legislative amendment extending statute); Orr v. Orr (Ala.App. 1979) 374 So.2d 895, 897.) The Utah court originally rebuffed the high court, asserting it (the state court) lacked authority to “legislate a bit on our own” by selecting a sex-neutral age of majority. (Stanton v. Stanton (Utah 1976) 552 P.2d 112, 113.) The high court vacated that judgment and again remanded (Stanton v. Stanton (1977) 429 U.S. 501 [50 L.Ed.2d 723, 97 S.Ct. 717]), but by that time the classification problem had been resolved by the Utah Legislature (id. at p. 501, fn. 2 [50 L.Ed.2d at p. 725]), which amended its statute to provide an age of majority of 18 for both sexes. On second remand, the Utah court concluded that both sexes were to be treated as adults at age 18, but specifically provided that its decision “shall have no retroactive effect.” (Stanton v. Stanton (1977) 564 P.2d 303, 305, italics omitted.)
The court upheld the constitutionality of an amendment to the Social Security law that perpetuated for a five-year grace period the same sex-based classification declared unconstitutional in Califano v. Goldfarb, supra, 430 U.S. 199. Before resolving that constitutional question—and pertinent to our present inquiry—the court first addressed a standing issue, and, in the course of its resolution of that question, it reaffirmed the propriety of judicial extension of statutes, and clarified the circumstances in which courts may do so to remedy a “constitutionally underinclusive scheme.” (465 U.S. at p. 738 [79 L.Ed.2d at p. 656].)
The court plainly implied that, if it were faced with the question, it would decline to order extension in the case before it, not because a court lacks general authority to do so, but because Congress had “clearly expressed its preference for nullification, rather than extension, of the pension offset exception in the event it is found invalid.” (465 U.S. at p. 739, fn. 5 [79 L.Ed.2d at p. 656]; see also id. at pp. 734 742-743 [79 L.Ed.2d at pp. 653, 658-659]; accord, California Federal S. & L. Assn. v. Guerra (1987) 479 U.S. 272, 291-292 & fn. 31 [93 L.Ed.2d 613, 630-631, 107 S.Ct. 683] [reaffirming propriety of extension “in cases where a statute is otherwise invalid,” and reiterating that in such a case “the Court must look to the intent of the state legislature to determine whether to extend benefits or nullify the statute” (italics omitted)].)
On remand the state court extended the statute. (Davis v. Department of Treasury (1989)179 Mich.App. 683 [446 N.W.2d 531, 533] [extending to retired federal employees same favorable tax treatment afforded to retired state and local employees].)
Although some have declined to extend statutes on the ground that on the facts of the case, doing so would frustrate, and not advance, the discernible legislative intent (see, e.g., Kroh v. N.D. Workers Compensation Bureau (N.D. 1988) 425 N.W.2d 899, 900-901; Day v. W.A. Foote Mem. Hospital, Inc. (1982) 412 Mich. 698 [316 N.W.2d 712, 714-717]), even those decisions acknowledge the general propriety of judicial extension to avoid constitutional invalidity when doing so is supported by legislative intent.
A contrary view has occasionally been expressed, but only in dissenting opinions. (See, e.g., Wengler v. Druggists Mut. Ins. Co., supra, 601 S.W.2d 8, 9 (dis. opn. of Donnelly, J.); Hewitt v. Saif (1982) 294 Or. 33 [653 P.2d 970, 984] (cone. & dis. opn. of Peterson, J.) [“There can be no denying that this court is legislating, and I believe that in so doing we violate the separation of powers clause of the Oregon Constitution. . . . [f] The effect of this court’s opinion is to enact a new law. (Fn. omitted.)”]; Padilla v. Industrial Commission (1975) 24 Ariz.App. 42 [535 P.2d 634, 639] (dis. opn. of Wren, J.) [noting court added words to statute and thereby amended it “under the guise of statutory construction”].)
As explained below, California cases as well have acknowledged the propriety of extension to avoid invalidity under equal protection principles. (See post, pp. 641-642, 649-651.)
See, e.g., Orr v. Orr, supra, 374 So.2d 895, 897; Murphey v. Murphey (1982) 103 Idaho 720 [653 P.2d 441, 444-446]; Beal v. Beal (Me. 1978) 388 A.2d 72, 75-76.
See, e.g., Davis v. Aetna Life & Cas. Co. (Tenn. 1980) 603 S.W.2d 718, 721; Portman v. Steveco, Inc. (Ind.Ct.App. 1983) 453 N.E.2d 284, 287-288; Swafford v. Tyson Foods, Inc. (1981) 2 Ark.App. 343 [621 S.W.2d 862, 866]; Passante v. Walden Printing Co. (1976) 53 A.D.2d 8 [385 N.Y.S.2d 178, 181]; Oknefski v. Workmen’s Comp. Appeal Bd. (1981) 63 Pa. Commw. 450 [439 A.2d 846, 849]; Tomarchio v. Township of Greenwich (1977) 75 N.J. 62 [379 A.2d 848, 854-855],
See, e.g., Thomas v. Rutledge (1981) 167 W.Va. 487 [280 S.E.2d 123, 130-131] (extending to “marital quit” claimants the same unemployment benefits afforded other “voluntary quit” claimants); Talley v. Succession of Stuckey (La. 1993) 614 So.2d 55, 59-61 (extending to illegitimate children the benefits of a will revocation statute granted only to legitimate children); Hewitt v. Saif supra, 653 P.2d 970, 980-982 (noting court’s power to “repair” otherwise unconstitutional statutes “in appropriate circumstances” and extending to cohabiting males the same workers’ compensation benefits granted to cohabiting females); Davis v. Department of Treasury, supra, 446 N.W.2d 531, 533 (extending to retired federal employees same favorable tax treatment afforded to retired state and local employees); Padilla v. Industrial Commission, supra, 535 P.2d 634, 636-638 (reforming workers’ compensation statute by extending to child of female wage earner same presumption of dependency afforded to child of male wage earner).
See, e.g., People v. Liberta (1984) 64 N.Y.2d 152 and footnote 15 [485 N.Y.S.2d 207, 218-219, 474 N.E.2d 567], citing additional cases (extending reach of rape law to married men who rape their wives and females who rape males); Pías v. State (Alaska 1979) 598 P.2d 966, 968-969 (extending coverage of prostitution statute to males); State v. Books (Idaho 1975) 225 N.W.2d 322, 325 (extending bribery statute to cover unlawful receipt of gifts to state as well as county employees).
Likewise, the high court earlier approved extension as an alternative to invalidation of a state criminal statute otherwise invalid under the equal protection clause. (Skinner v. Oklahoma, supra, 316 U.S. 535.) Under Oklahoma law, habitual criminals were subject to compulsory sterilization. Embezzlers were specifically exempted from that class, but petty larcenists were included. (Id. at pp. 536-537 [86 L.Ed. at pp. 1657-1658].) The high court found no sufficient state interest in that disparate classification, and found an equal protection violation. (Id. at pp. 541-542 [86 L.Ed. at pp. 1660-1661].) Instead of invalidating the statute, however, the court remanded to state court to allow it to determine whether to invalidate or extend the statute providing for sterilization of the excluded class. (Id. at pp. 542-543 [86 L.Ed. at pp. 1661-1662].) On remand the state court selected invalidation over extension. (Skinner v. State ex rel. Williamson (1945) 195 Okla. 106 [155 P.2d 715, 716].)
Insurance Co. of North America v. Russell, supra, 271 S.E.2d 178, concerned a workers’ compensation statute that established a conclusive presumption of a surviving spouse’s dependency on a deceased spouse, but set out different tests for establishing that presumption for surviving men and women. Subdivision (a) of the statute allowed a surviving wife to establish dependency if she “had not voluntarily deserted or abandoned [him] at the time of the accident.” (Id., at p. 179.) Subdivision (b) of the statute provided that a surviving husband could establish dependency only if he “lived [with his wife] at the time of the accident” and was “then incapable of self-support and actually dependent on her.” (Ibid.) The court acknowledged the disparate classifications violated federal equal protection principles (271 S.E.2d at p. 181), and determined that “the legislative intent can best be accomplished by *638eliminating the conclusive presumption of dependency of widows” (ibid.) rather than extending that presumption to widowers (id. at pp. 181-182). The court noted, however, that unless it made additional adjustments to the statute, widows would be subject to the limited presumption applicable to widowers—a result that, the court believed, the legislature did not intend. (Id. at p. 182.) To resolve that problem, the court decided to scrap the widowers’ test as well, and to merge the two subdivisions into a hybrid provision: “A court has the power to merge into one two provisions invalid under the equal protection clause when the result achieved is more consistent with legislative intent than the result that would attend complete invalidation of one or the other.” (Ibid.) The court then set out explicitly its reformation of the statute: “[U]ntil the General Assembly provides otherwise, as a result of this decision subparagraphs (a) and (b) of [the statute] should be read together, as follows: ‘The following persons shall be conclusively presumed to be the next of kin wholly dependent for support upon the deceased employee: (a, b) A surviving spouse upon a deceased spouse if the survivor was wholly or partially dependent for such support or was in need of such support.’ ” (Ibid., italics added.) Clearly, the Georgia Supreme Court engaged in substantial rewriting of the statute in order to reform and preserve its otherwise constitutionally invalid provisions.
The court has expressly invited reformation by state courts in analogous contexts. See, e.g., Virginia v. American Booksellers Assn. (1988) 484 U.S. 383 [98 L.Ed.2d 782, 108 S.Ct. 636], in which the high court asserted it would “not rewrite a state law to conform it to constitutional requirements” (id. at p. 397 [98 L.Ed.2d at p. 796]), but then certified to the state supreme court the question whether a statute that criminalized commercial display of materials “harmful to children,” was susceptible of a narrowing construction that would render it constitutional. (Id. at pp. 397-398 [98 L.Ed.2d at pp. 796-797].) In response, the state court imposed a saving construction on the statute, holding it barred only “the opportunity [a bookseller] may afford to juveniles to take off the shelves books which they are unable to buy, and to read them in the store.” (Com. v. American Booksellers Assn, Inc. (1988) 236 Va. 168 [372 S.E.2d 618, 624].)
Some decisions addressed this latter question directly. In State v. A Motion Picture Entitled “The Bet," supra, 547 P.2d 760, the Kansas Supreme Court stated: “ ‘We realize that to construe the statute to meet constitutional standards we may be subject to the accusation that we are invading the province of the legislature. However, after considering the manifest intention of the legislature when it passed [the statute], ... we feel fully justified in construing and limiting the present statute to meet constitutional standards. Such was the original intent of the Kansas legislature.” (Id. at p. 767.)
The federal appeals court had declined to so construe the statute. The high court recognized its authority to adopt a construction of the statute different from that of the appeals court, but ultimately “pretermitted” that issue (472 U.S. at p. 501 [86 L.Ed.2d at p. 404]) because it concluded that the federal court of appeals “fell into another error” in failing to save the statute by a form of severance. (Ibid.)
A local ordinance broadly imposed a license tax on all gross receipts. We noted, “[t]o allow a city to levy a license tax based upon gross receipts attributable to selling activities outside the city would be an unreasonable discrimination and a denial of equal protection of the law.” (City of Los Angeles v. Belridge Oil, supra, 42 Cal.2d at p. 832.) In order to preserve the constitutionality of the ordinance, we construed it to provide “that the measure of the tax be limited to those gross receipts attributable to selling activities within the [city’s jurisdiction].” (Id. at p. 833.)
The Penal Code makes it a crime for anyone to “willfully disturb[]” any lawful meeting. (Pen. Code, § 403.) We found the term overbroad because it would criminalize protected speech under the First Amendment, and in order to preserve the statute, we construed it “to require the following showing to establish its transgression: that the defendant substantially impaired the conduct of the meeting by intentionally committing acts in violation of implicit customs or usages or of explicit rules for governance of the meeting, of which he knew, or as a reasonable man should have known.” (In re Kay, supra, 1 Cal.3d 930, 943.)
The Penal Code made it a crime for anyone to “disturb the peace” or “breach the peace.” (Pen. Code, former § 415, repealed by Stats. 1974, ch. 1263, § 1, p. 2742.) We rejected a claim that the terms are overbroad and criminalize protected speech under the First Amendment, by construing the terms “to mean disruption of public order by acts that are themselves violent or tend to incite others to violence.” (In re Bushman, supra, 1 Cal.3d 767, 773.)
The Education Code provided for revocation of a teaching license, on a showing of “any act involving moral turpitude.” (Former Ed. Code, § 13129, subd. (e), repealed by Stats. 1976, ch. 1010, § 1, p. 2384.) We found the term overbroad because it implicated protected privacy interests, and in order to preserve the statute, we construed it to “denote immoral or unprofessional conduct or moral turpitude of the teacher which indicates unfitness to teach.” (Morrison v. State Board of Education, supra, 1 Cal.3d 214, 225.)
The Penal Code makes it a crime for anyone to engage in “lewd or dissolute conduct in any public place open to the public or exposed to public view.” (Pen. Code, § 647, subd. (a).) We noted the statute would, if interpreted according to its literal terms to apply to live theatrical performances, criminalize protected speech under the First Amendment, and pose a “serious equal protection problem.” (Barrows v. Municipal Court, supra, 1 Cal.3d at p. 827.) Accordingly, we construed the statute as inapplicable to such performances. In practical effect, this is analogous to the reformation by severance effectuated by the high court in Spokane Arcades, supra, All U.S. 491.
In order to avoid constitutional vagueness problems, we construed a local trespassing ordinance that proscribed “offensive personal conduct,” to “coincide with those offenses punishable under Penal Code section 646c (obstruction of a street, sidewalk, or other public area).” (In re Cox, supra, 3 Cal.3d at p. 223.)
The Penal Code makes it a crime for anyone who disrupts the orderly operation of a college campus to remain on a campus after being notified that consent to remain has been withdrawn. (Pen. Code, § 626.4.) We found the section overbroad because it would criminalize protected speech under the First Amendment, and in order to preserve the statute, we construed it “to permit exclusion from the campus only of one whose conduct or words are such as to constitute, or incite to, a substantial and material physical disruption incompatible with the peaceful functioning of the academic institution and of those upon its campus.” (Braxton v. Municipal Court, supra, 10 Cal.3d at p. 150.)
We construed a local growth control ordinance “to give specific content to terms that might otherwise be unconstitutionally vague.” (18 Cal.3d at p. 598.) “Following the course suggested by Braxton, [supra, 10 Cal.3d 138],” we reformed the ordinance “to incorporate the standards for determining the overcrowded condition of schools contained in [a resolution adopted by the local school board]. So construed, the ordinance provides a clear and ascertainable educational standard to guide the issuance or denial of a building permit. . . .” (Id. at pp. 598-599.)
The Penal Code makes it a crime for anyone to “procure[] another person for the purpose of prostitution . . . .” (Pen. Code, § 266i.) We noted the statute would, if interpreted to apply to the hiring of actors to perform in a “nonobscene commercial film,” “rather obviously place a substantial burden on the exercise of protected First Amendment rights.” (People v. Freeman, supra, 46 Cal.3d at p. 426.) Accordingly, we construed the statute as inapplicable to such transactions. (Ibid.) Again, in practical effect, this is analogous to the reformation by severance effectuated by the high court in Spokane Arcades, supra, All U.S. 491.
In yet another category of cases we have reformed statutory schemes to make them consistent with due process requirements by reading into enactments the right to a hearing. Most recently, for example, in Traverso v. People ex rel. Dept, of Transportation (1993) 6 Cal.4th 1152 [26 Cal.Rptr.2d 217, 864 P.2d 488], we reaffirmed a long line of cases holding that we will infer the due process right to a hearing—even in the face of statutory silence— when there is some basis on which to conclude that the enacting body would have intended to require a hearing had it foreseen the constitutional necessity of such a hearing. (See id. at pp. 1163-1166 [reading into statute right to hearing before removal of billboard]; People v. Amor *646(1974) 12 Cal.3d 20, 29-30 [114 Cal.Rptr. 765, 523 P.2d 1173] [reading into statute right to hearing, before criminal defendant required to reimburse court costs incurred by appointed counsel]; Carroll v. California Horse Racing Board (1940) 16 Cal.2d 164, 168 [105 P.2d 110] [reading into statute right to hearing before revocation of horse trainer’s license]; Simpson v. City of Los Angeles (1953) 40 Cal.2d 271, 282 [253 P.2d 464] [reading into statute right to hearing before impounded animal may be released to research facilities]; see also Merco Constr. Engineers, Inc. v. Los Angeles Unified Sch. Dist. (1969) 274 Cal.App.2d 154, 168 [79 Cal.Rptr. 23], and cases cited.)
Likewise, in the due process hearing cases (ante, fn. 47), we expanded each statute by reading into it a hearing right necessary under the Constitution, but not set out in the law as enacted.
In Arp, supra, 19 Cal.3d at pages 407-408, we cited Edgar M., supra, 14 Cal.3d 727, as an example of a case in which courts possess the power to “remedy a constitutional defect by literally rewriting statutory language . . . when the result achieved by such a course is more consistent with legislative intent than the result that would attend outright invalidation.”
Thereafter, the Hayes court wrote: “See Baxstrom v. Herold (1966) 383 U.S. 107, 115 [15 L.Ed.2d 620, 625-626, 86 S.Ct. 760] [extending to those civilly committed to mental institution and awaiting expiration of a penal sentence, a statutory right of jury trial granted to civilly committed persons generally]; People v. Smith (1971) ... 5 Cal.3d 313, 319 [96 Cal.Rptr. 13, 486 P.2d 1213] [unanimously extending to wards of the Youth Authority convicted of a criminal offense the same statutory term of commitment imposed on a ward who is committed in juvenile court]; Quong Ham Wah Co. v. Industrial Acc. Com.[, supra, ] 184 Cal. 26, 41, [discussed ante, pp. 624, 625]. . . . ; but see In re Trummer (1964) 60 Cal.2d 658, 664 [36 Cal.Rptr. 281, 388 P.2d 177] [extending statutory right of jury trial on issue of addiction to all persons similarly situated, but asserting court generally lacks authority to so ‘rewrite’ statutes].” (Hayes v. Superior Court, supra, 6 Cal.3d at p. 224, italics added.)
In In re Trummer (1964) 60 Cal.2d 658, 664 [36 Cal.Rptr. 281, 388 P.2d 177], we asserted in dictum that we have neither followed nor cited: “Ordinarily in cases where a substantive right is involved and equal protection has been denied because the Legislature has made an unreasonable classification, the entire statute has been invalidated. [Citation.] In such cases we cannot presume to rewrite the law to make the classification a valid one. . . .” As noted, despite this language, the Trummer court reformed the statute. {Ibid.) As explained elsewhere, the United States Supreme Court, the lower federal courts, the courts of this state, and the courts of our sister states, have all retreated from the view that a court is powerless, in appropriate situations, to “rewrite” statutes otherwise unconstitutionally underinclusive in order to save them.
See Mills v. Superior Court (1986) 42 Cal.3d 951, 957-959 [232 Cal.Rptr. 141, 728 P.2d 211] (term “reasonable efforts” cannot be construed to mean mere “objection” because Legislature demonstrated intent not to equate the two terms); Rockwell v. Superior Court (1976) 18 Cal.3d 420, 445 [134 Cal.Rptr. 650, 556 P.2d 1101] (declining invitation to rewrite death penalty statutes “in manner which we have shown would be contrary to the manifest legislative intent”); Mordecai v. Board of Supervisors (1920) 183 Cal. 434, 443 [192 P. 40] (declining to reform statute in a manner that “would . . . extend the operation of the act to [certain] counties, although the legislature has expressly said they shall not come within its operation”); City of Los Angeles v. Lewis (1917) 175 Cal. Ill, 780-783 [167 P. 390] (declining to reform statute in manner “distinctly the contrary” to scheme that “legislature deliberately and advisedly formulated and passed”); Estate of Mahoney (1901) 133 Cal. 180, 183 [65 P. 389] (declining to reform or sever statute “contrary to the apparent purpose of the legislature”), overruled in part on other grounds by Estate of Johnson (1903) 139 Cal. 532 (73 P. 424)]; McLaughlin v. Superior Court (1983) 140 Cal.App.3d 473, 482-483 [189 Cal.Rptr. 479] (declining to read into statute a “qualifying provision” inconsistent with the Legislature’s apparent intent and constitutional due process principles).
See Lesher Communications, Inc. v. City of Walnut Creek (1990) 52 Cal.3d 531, 541-544 [277 Cal.Rptr. 1, 802 P.2d 317] (no evidence that voters intended local measure, a zoning ordinance, as general plan amendment; hence court could not construe ordinance to amend general plan); Dillon v. Municipal Court (1971) 4 Cal.3d 860, 871 [94 Cal.Rptr. 777, 484 P.2d 945] (declining to reform local ordinance barring parade without permit, and noting city may have intended to allow local authorities to exercise improper discretion in denying permit); People v. Duffy (1947) 79 Cal.App.2d Supp. 875, 888 [179 P.2d 876] (same); Flood v. Riggs (1978) 80 Cal.App.3d 138, 157 [145 Cal.Rptr. 573] (declining proposed reformation, which was based on “sheer conjecture” concerning legislative intent).
A number of the cases cited by interveners and amicus curiae for respondent are akin to Blair v. Pitchess, supra, 5 Cal.3d 258. (See Seaboard Acceptance Corp. v. Shay, supra, 214 Cal. 361, 367-368 [listing five substantial changes required to render statute constitutional]; In re King (1970) 3 Cal.3d 226, 237 [90 Cal.Rptr. 15, 474 P.2d 983] [reformation of statute to preserve constitutionality would require “wholesale rewriting”]; Bayscene Resident Negotiators v. Bayscene Mobilehome Park (1993) 15 Cal.App.4th 119, 134 [18 Cal.Rptr.2d 626] [reformation to preserve constitutionality of local ordinance’s invalid binding arbitration *657provision would require “wholesale rewriting”]; see also Spiritual Psychic Science Church v. City of Azuza (1985) 39 Cal.3d 501, 520 [217 Cal.Rptr. 225, 703 P.2d 1119] [local ban on fortune-telling “invalid in its essence”].) As we explain post, page 661, when legislative intent regarding policy judgments is clear, a revision that effectuates that choice is not impermissible merely because it requires insertion of more words than it removes.
See Solberg v. Superior Court (1977) 19 Cal.3d 182, 198 [137 Cal.Rptr. 460, 561 P.2d 1148]; People v. One 1940 Ford V-8 Coupe (1950) 36 Cal.2d 471, 475-476 [224 P.2d 677]; Anderson v. I.M. Jameson Corp. (1936) 7 Cal.2d 60, 67-68 [59 P.2d 962]; Hopper v. Keys (1907) 152 Cal. 488, 495 [92 P. 1017]; City of Eureka v. Diaz (1891) 89 Cal. 467, 469-470 [26 P. 961]; Carroll v. State Bar (1985) 166 Cal.App.3d 1193, 1202 [213 Cal.Rptr. 305]; People v. Haney (1984) 156 Cal.App.3d 109, 115 [202 Cal.Rptr. 579].
In re Waters of Long Valley Creek Stream System (1979) 25 Cal.3d 339 [158 Cal.Rptr. 350, 599 P.2d 656] is similarly distinguishable. It did not involve reformation, but construction of a statute to “render unnecessary the resolution of a substantial constitutional issue.” In that case we recognized the Legislature intended to grant the state water board broad authority (id. at pp. 348-349), but construed the statutory provisions as precluding the board from extinguishing completely “future riparian rights.” (Id. at pp. 358 & 350, fn. 5.) Similarly, other cited cases did not involve reformation. Mulkey v. Reitman (1966) 64 Cal.2d 529, 543-545 [50 Cal.Rptr. 881, 413 P.2d 825] concerned the inability of the court to mechanically sever unconstitutional aspects of a statewide initiative. In re Blaney (1947) 30 Cal.2d 643, 653-656 [184 P.2d 892] is likewise distinguishable.
As explained ante, pages 648-649, Eller, supra, 192 Cal.App.3d 643, illustrates how the approach set out in Metromedia III, supra, 32 Cal.3d 180, is compatible with the proposition that a reformation is appropriate in a proper case. Similarly, dictum in People v. Skinner (1985) 39 Cal.3d 765, 775 [217 Cal.Rptr. 685, 704 P.2d 752], to the effect that a court will “not undertake to rewrite . . . unambiguous language,” must be read in light to the court’s decision in that case to reform an amendment of Penal Code section 25, subdivision (b), by replacing the word “and” with “or” to correct an apparent drafting oversight in an initiative *658measure and thereby preserve its constitutionality consistently with the electorate’s intent. (See also 39 Cal.3d at pp. 775-776, and cases cited [similar correction of drafting errors].) Indeed, as intervener on behalf of petitioners observes, Skinner, supra, 39 Cal.3d 765, implicitly supports rather than defeats the authority of courts to reform statutes.
There are also numerous cases in which the high court has stated it will not, and may not, reform an enactment even to preserve its constitutionality. For example, an oft quoted passage of United States v. Reese (1876) 92 U.S. (2 Otto) 214 [23 L.Ed. 563] (Reese), states: “The question . . . to be determined, is, whether we can introduce words of limitation into a penal statute so as to make it specific, when, as expressed, it is general only. [1. . . . This would, to some extent, substitute the judicial for the legislative department of the government. . . . [1 To limit this statute in the manner now asked would be to make a new law, not enforce an old one. This is no part of our duty.” (Id. at p. 221 [23 L.Ed. at p. 566]; see also, e.g., Yu Cong Eng v. Trinidad (1926) 271 U.S. 500, 519-520 [70 L.Ed. 1059, 1067-1068, 46 S.Ct. 619]; *660Marchetti v. United States (1968) 390 U.S. 39, 60, fn. 18 [19 L.Ed.2d 889, 904-905, 88 S.Ct. 697].)
These and other cases containing the same or similar statements (e.g., Virginia v. American Booksellers Assn., supra, 484 U.S. 383, 397 [98 L.Ed.2d 782, 796]; Wyoming v. Oklahoma (1992) 502 U.S. 437, 459-460 [117 L.Ed.2d 1, 26, 112 S.Ct. 789]), including those cited by the Ninth Circuit in Service Employees II, supra, 955 F.2d at page 1321 [Thornburgh v. American College of Obstetricians & Gynecologists (1986) 476 U.S. 747, 764 [90 L.Ed.2d 779, 795-796, 106 S.Ct. 2169]; United States v. Jackson (1968) 390 U.S. 570, 585, fn. 27 [20 L.Ed.2d 138, 148-149, 88 S.Ct. 1209]; National Advertising Co. v. Town of Babylon (2d Cir. 1990) 900 F.2d 551, 557], are distinguishable on one or more of the same grounds on which interveners’ state cases are distinguishable, or on the additional and frequently expressed basis that the federal courts are reluctant to (or lack authority to) reform state or local enactments. Reese itself, however, is not so distinguishable.
Reese concerned a criminal prosecution under a federal statute. Two inspectors of a Kentucky municipal election had refused to accept the ballot of “William Gamer, a citizen of the United States of African descent.” (92 U.S. (Otto) at p. 215 [23 L.Ed. at p. 563].) The circuit court upheld the defendants’ demurrers to the indictment, and the high court affirmed, concluding the statute was unconstitutional, as outside Congress’s power to enforce the Fifteenth Amendment by “appropriate legislation.” The court broadly read the statute as an attempt by Congress to go beyond its authority conferred by the Fifteenth Amendment and criminalize conduct unrelated to discrimination against voters on the ground of race.
Contrary to the Reese court’s determination that it lacked authority to “introduce words of limitation into a penal statute so as to make it specific, when, as expressed, it is general only” (92 U.S. at p. 221 [23 L.Ed. at p. 566]), the high court has, in the past two decades, expressly recognized its authority to do precisely that. (See ante, pp. 638-641.) Moreover, contrary to the Reese court’s professed reluctance to invade the province of the legislature, it would appear that the court’s decision to invalidate the voting act, and thereby insulate the defendants from prosecution thereunder, was a far greater intrusion into the legislative domain. (See 2A Sutherland, Statutory Construction, supra, § 47.38, at p. 291, quoted ante, at p. 658.) In conclusion, it seems plain that the relatively minor reformation required to expressly confine the scope of the statute to cover race discrimination only would have been far more respectful of Congress’s intent. (See 92 U.S. at p. 243 [23 L.Ed. at p. 574] (dis. opn. of Hunt, J.) [legislative history disclosed intent to so confine scope of statute].)
For these reasons we conclude that United States v. Reese, supra, 92 U.S. (Otto) 214, 221 [23 L.Ed. 563, 565-566], like our own decision in People v. Perry, supra, 19 Cal. 105, 115, is dubious authority today. The high court has, since 1970, reformed—i.e., rewritten—enactments to preserve their constitutionality in numerous cases, and it has expressly and unanimously affirmed its constitutional authority to do so.
Justice Mosk’s concurring opinion asserts rewriting to preserve constitutionality is barred by the separation of powers doctrine unless the enacting body expressly or by implication “specifically authorize^]” such rewriting. (Cone. opn. of Mosk, J., post, at p. 674.) As explained above, the cases establish a somewhat different standard: Courts have declined to rewrite, even to preserve constitutionality, unless they can conclude with confidence that (i) it is possible to reform the statute in a manner that closely effectuates policy judgments clearly articulated by the enacting body, and (ii) the enacting body would have preferred such a reformed version of the statute to invalidation of the statute.
In view of the prior decisions of this court (cited and described ante, pages 641-653), none of which Justice Mosk claims was incorrectly analyzed or decided, we are inclined to read Justice Mosk’s concurring opinion as simply emphasizing, appropriately in our view, the need for close and careful scrutiny regarding the second part of the traditional test embraced herein, namely, whether the enacting body would have preferred a given reformed construction to invalidation.
As observed ante, pages 652-653, many of the equal protection extension cases effectuate broad legislative policy judgments granting benefits to a named class, while declining to respect other policy judgments improperly limiting those benefits to the named class, and thereby elevate a major legislative policy judgment over a related minor policy judgment. Again, assuming this form of limited judicial policymaking is proper in the equal protection/ underinclusion context, we conclude such judicial conduct is improper outside that context.
Neither petitioners nor intervener on their behalf suggest the inter-candidate transfer ban aspect of section 85304 should be judicially reformed to avoid the overbreadth perceived by *664the federal appeals court. Nor would we grant such a request. As the federal appeals court noted in another context in Service Employees II, supra, 955 F.2d 1312, 1321, “we would be at a loss to know what. . . dollar amounts” should be adopted to distinguish between “small” and “large” contributions. As explained herein, judicial reformation to preserve the constitutionality of a statute is permissible only when, inter alia, a court can conclude with confidence that it is possible to reform the statute in a manner that closely effectuates policy judgments clearly articulated by the enacting body. Neither section 85304 nor the rest of the statutory scheme (nor the ballot materials supporting Proposition 73) suggests the drafters or electorate even considered, much less expressed any intent concerning, the appropriate “large contribution” amount that might be legitimately applied to section 85304’s inter-candidate transfer ban. Accordingly, section 85304 may not be judicially reformed.
As respondent observes, assuming the term “election cycle” were defined to encompass both the primary and general election for an office, use of that definition would frustrate the apparent legislative purpose of allowing contributors an opportunity to contribute before both the primary and the general elections. (See ante, fn. 8.)
The calculations presented herein are premised on the assumption, implicitly embraced by amicus curiae and interveners for respondent, that the statutes as enacted would have allowed two contribution periods for each two-year race, and four contribution periods for each four-year race, etc. Yet, there are at least three, not two, fiscal years prior to the election for the first term of a two-year office, and there are at least five, and not four, “fiscal years” prior to the election for the first term of a four-year office. Although, as a practical matter, it would seem that only those candidates who were not holders of elective office would benefit from this fact, still it must be recognized that the “theoretical maximum aggregate contribution” amounts discussed herein understate slightly the true theoretical maximum aggregate amounts that would have been permitted in some cases under the statutes as enacted.
For example, under section 85301(a) as enacted, a candidate for an Assembly seat could solicit and accept contributions from an individual totaling $2,000 (i.e., $1,000 per “fiscal year”). Likewise, under the proposed “per election” reformation of section 85301(a), a candidate for an Assembly seat could solicit and accept contributions from an individual totaling $2,000 (i.e., $1,000 for primary, and $1,000 for the general election).
Under section 85301(a) as enacted, a candidate for a Senate seat, or a candidate for a local nonpartisan office (e.g., mayor, council, and supervisor races) could solicit and accept contributions from an individual totaling $4,000 (i.e., $1,000 per “fiscal year”). But under the proposed “per election” reformation of section 85301(a), a candidate for a Senate seat could solicit and accept contributions from an individual totaling only $2,000 (i.e., $1,000 for the primary, and $1,000 for the general election). Similarly, under the proposed “per election” reformation of section 85301(a), a candidate for local nonpartisan office could solicit and accept contributions from an individual totaling only $1,000 “per election,” which would effectively reduce contributions to that amount unless a runoff were required.
Under inoperative section 85300 (proposed by Proposition 68 and approved by electors June 7, 1988)—the analog of section 85301 of Proposition 73—contributions by a person to a candidate for a Senate seat would be limited to $1,000 per primary, general, special, or special runoff election.
In view of respondent’s stated position, we question Justice Baxter’s assertion that “respondent has never suggested that [the “modified election cycle”] approach . . . would be . . . improper.” (Conc. & dis. opn. of Baxter, J., at p. 690.)
At oral argument, petitioners, and to a lesser extent intervener on their behalf, Common Cause, agreed that reformation under Justice Baxter’s approach would be “better than no reformation at all.” Neither, however, endorsed Justice Baxter’s “modified election cycle” approach.
For this reason, and contrary to Justice Baxter’s concurring and dissenting opinion, post, at pages 691-692, Buckley v. Valeo, supra, 421 U.S. 1, is distinguishable. In that case, unlike *670this one, judicial reformation properly advanced, and did not frustrate, legislative policy judgments.
Having concluded sections 85301(a) and 85303(a) and (b) may not be reformed, we conclude section 85302 should not be reformed. (See ante, p. 663, and ante, fn. 10.)