State Farm Mutual Automobile Insurance v. Quackenbush

Opinion PAKRILLI, J. This appeal arises from administrative proceedings to determine whether the State Farm insurance companies (collectively, State Farm) will be required to roll back their 1989 premium rates under Proposition 103 (Ins. Code, § 1861.01 et seq.). Insurance Commissioner Charles Quackenbush (Commissioner) rejected a proposed decision by an administrative law judge, who concluded State Farm owed no refund to its policyholders. The Commissioner ordered State Farm to roll back its 1989 rates by 3.7644 percent, and to refund over $200 million. State Farm challenged the Commissioner’s ruling by petitioning the San Francisco Superior Court for a writ of administrative mandamus. The court issued an order partly granting and partly denying the petition. The parties agree the effect of the court’s ruling is to preclude any refund from State Farm to its policyholders. Both parties have appealed. However, State Farm concedes we need not consider its cross-appeal if we affirm the trial court’s order. In his appeal, the Commissioner contends the trial court misconstrued Department of Insurance regulations when (1) it ruled State Farm had provided reliable data justifying certain loss adjustment expense reserves for California; and (2) it determined that an item on State Farm’s annual statement was not subject to a regulation governing the treatment of federal income tax data. We conclude the trial court’s order is supported by substantial evidence and fully consistent with the regulatory scheme governing rate rollbacks. Therefore, we affirm the order and do not reach the issues raised by the cross-appeal. *70 Discussion 1. Proposition 103 and the Rate Regulations Proposition 103, enacted November 8, 1988, mandated an immediate 20 percent rollback of insurance rates for one year, and required the Commissioner’s prior approval for rate increases in subsequent years. (Ins. Code, § 1861.01, subds. (a) and (c).) Insurers could raise rates during the rollback year only if they could show a substantial threat of insolvency. (Ins. Code, § 1861.01, subd. (b).) Our Supreme Court ruled that this provision for obtaining relief from the rollback, which precluded adjustments to achieve fair and reasonable rates, was unconstitutional but severable from the rest of the initiative measure. (Calfarm Ins. Co. v. Deukmejian (1989) 48 Cal.3d 805, 819-822 [258 Cal.Rptr. 161, 771 P.2d 1247].) The Calfarm decision allowed insurers to seek the Commissioner’s approval of higher rates for 1989 and to charge those higher rates pending the Commissioner’s determination, which is what State Farm did in this case. If the Commissioner finds that a lesser rate is fair and reasonable, the insurer must refund the excess premiums it has collected, with interest. (Id. at pp. 825-826.) The Commissioner issued regulations to govern both the rollback determination and “prior-approval” ratemaking under Proposition 103. (Cal. Code Regs., tit. 10, subch. 4.8; 1 see 20th Century Ins. Co. v. Garamendi (1994) 8 Cal.4th 216, 249-256 [32 Cal.Rptr.2d 807, 878 P.2d 566].) Article 4 of the regulations provides the ratemaking formula and the factors governing its implementation in prior-approval ratemaking. (§§ 2644.1-2644.23.) Article 5 (§§ 2645.1-2645.9) governs the determination of rates for the rollback period, which extended from November 8, 1988, to November 7, 1989. (§ 2645.1.) Except as otherwise specified in article 5, “rates for the rollback period shall be evaluated in the manner specified in article 4.” (§ 2645.3.) A principal difference between rollback calculations and prior-approval ratemaking is stated in section 2645.2, subdivision (d): “Historical data shall be used in lieu of forecasts for the rollback period because actual historical data exist for many of the figures that are forecast in prior-approval rate-making, making it possible to avoid the errors inherent in forecasting and resulting in greater accuracy.” (See 20th Century Ins. Co. v. Garamendi, supra, 8 Cal.4th at p. 252.) The Commissioner’s position on the most significant legal issues in this case is undermined by the requirement that actual historical data be used in rollback calculations. *71 2. Standard of Review The standard of review is a critical factor in this appeal. Insurance Code section 1861.09, enacted as part of Proposition 103, states: “Judicial review shall be in accordance with [Insurance Code] Section 1858.6.” Insurance Code section 1858.6 provides that in proceedings to review an administrative decision by the Commissioner, the trial court “is authorized and directed to exercise its independent judgment on the evidence and unless the weight of the evidence supports the findings, determination, rule, ruling or order of the commissioner, the same shall be annulled.” The independent judgment standard requires the trial court to accord a strong presumption of correctness to the Commissioner’s findings, and the burden of proof rests on the party challenging those findings, but ultimately the trial court is free to reweigh the evidence and substitute its own findings. (Fukuda v. City of Angels (1999) 20 Cal.4th 805, 816-819 [85 Cal.Rptr.2d 696, 977 P.2d 693].) On appeal, we apply the substantial evidence test to the trial court’s factual findings, but review legal determinations independently. (20th Century Ins. Co. v. Garamendi, supra, 8 Cal.4th 216, 271-272; National Indemnity Co. v. Garamendi (1991) 233 Cal.App.3d 392, 402-403 [284 Cal.Rptr. 278].) The parties dispute the degree of judicial deference owed to the Commissioner’s interpretation of the regulations implementing Proposition 103. The Commissioner contends his interpretations are entitled to great weight. State Farm argues that the Commissioner’s interpretations are merely litigating positions, and as such are entitled to no deference at all. We disagree with State Farm on this point. The Commissioner’s interpretations are to be respected, though they are not binding on us. An administrative agency’s interpretation of its own regulation deserves substantial weight, even if it amounts to a “litigating position.” On the other hand, it is well settled that the interpretation of a regulation, like the interpretation of a statute, is a question of law ultimately decided by the courts. (Culligan Water Conditioning v. State Bd. of Equalization (1976) 17 Cal.3d 86, 93 [130 Cal.Rptr. 321, 550 P.2d 593]; Yamaha Corp. of America v. State Bd. of Equalization (1998) 19 Cal.4th 1, 12-13 [78 Cal.Rptr.2d 1, 960 P.2d 1031].) The level of deference due to an agency’s regulatory interpretation turns on a legally informed, commonsense assessment of its merit in the context presented. (Yamaha Corp. of America v. State Bd. of Equalization, supra, 19 Cal.4th at p. 14.) 2 Here we conclude that most of the Commissioner’s interpretations of the regulations at issue are inconsistent with the terms of the regulations, or with *72 related provisions in the regulatory scheme. Furthermore, in challenging the trial court’s factual findings the Commissioner has not met, or even attempted to meet, the requirements of the substantial evidence standard of review. 3. Loss Adjustment Expenses In his briefs, the Commissioner frames his arguments solely in terms of regulatory interpretation, and requests our independent review. However, in developing his contentions on the reliability and justification of State Farm’s California loss adjustment expense data, the Commissioner resorts to re-arguing the evidence and in doing so fails to set forth all the relevant evidence. In such a case, “we discuss the legal questions involved while presuming that the underlying trial court findings of fact are supported by substantial evidence.” (Trailer Train Co. v. State Bd. of Equalization (1986) 180 Cal.App.3d 565, 588 [225 Cal.Rptr. 717].) “It is neither practical nor appropriate for us to comb the record on [the Commissioner’s] behalf [citations], especially when the record is as extensive as it is in this appeal.” (Ibid.) Nevertheless, we have reviewed the evidence referenced in the parties’ briefs and the trial court’s order. Like the Trailer Train court, we are satisfied that the trial court’s factual findings are supported by substantial evidence, even though issues concerning the sufficiency of the evidence are deemed waived. (Id. at pp. 587-588, and fn. 16.) The most significant legal questions subject to our independent review involve the regulations’ treatment of loss adjustment expenses in calculating an insurer’s 1989 rate rollback. The regulations distinguish between two kinds of loss adjustment expense. Allocated loss adjustment expenses (ALAE) are costs of defense “associated with the adjustment of specific claims.” (§ 2644.8, subd. (a).) They consist mostly of legal fees. ALAE are *73 a factor in the ratemating formula set out in the regulations. (§ 2644.3.) Unallocated loss adjustment expenses (ULAE) are costs of defense that insurers treat as overhead, instead of billing them to specific files. ULAE are an element of “fixed expenses,” another factor in the ratemating formula. (§§ 2644.3, 2644.9, subd. (e).) In determining an insurer’s rollback obligation, the Commissioner uses the insurer’s “incurred” loss adjustment expenses for 1989, which include not only paid expenses but also changes in the reserves set aside by the insurer for anticipated expenses. Incurred loss adjustment expenses are calculated by subtracting the insurer’s loss adjustment expense reserves at the end of 1988 from the reserves at the end of 1989, and adding the expenses actually paid during 1989. The Commissioner rejected State Farm’s 1989 California data for both its ALAE and its ULAE in certain lines of insurance, finding the data were unreliable and unjustified under section 2643.6. This regulation, entitled “Interjurisdictional Allocations,” governs data submitted by insurers with multistate operations. The Commissioner must use “reliable” California-specific data submitted by such insurers. However, if the Commissioner finds the California data unreliable, the insurer’s “multi-state data shall be allocated to California.” (§ 2643.6, subd. (a).) Loss adjustment expenses are allocated “by dollars of incurred losses,” which means they are derived from the percentage of the insurer’s California incurred losses corresponding to the ratio between its countrywide loss adjustment expenses and its countrywide incurred losses. (§ 2643.6, subd. (a)(1).) Section 2643.6, subdivision (b) imposes a justification requirement in addition to the reliability requirement of subdivision (a). When the insurer’s California data vary from the interjurisdictional allocation, “the Commissioner shall require the insurer to demonstrate the accuracy of the data and to justify any deviation from the insurer’s experience in other states.” If the commissioner is not satisfied with the insurer’s explanation, the interjurisdictional allocation is imposed. (§ 2643.6, subd. (b).) ■ The Commissioner determined State Farm’s loss adjustment expenses according to the interjurisdictional allocation prescribed by section 2643.6, subdivision (a)(1). The trial court, however, found that State Farm’s California loss adjustment expense data met the requirements of section 2643.6, and therefore must be used to determine State Farm’s rollback obligation. We address first the Commissioner’s arguments on the ALAE data, which present the most significant regulatory issues and have the greatest impact on the ultimate rollback determination. A. The ALAE Data The Commissioner devotes most of his briefing to an attack on the trial court’s finding that State Farm’s 1989 California ALAE data for its *74 “Fire Lines” (lines of insurance other than automobile insurance) were “reliable” under section 2643.6, subdivision (a). Only the Fire Lines ALAE are at issue on this appeal. The Commissioner does not dispute the reliability of State Farm’s California ALAE actually paid in 1989; he contends only that State Farm’s 1989 ALAE reserves were too high. The administrative law judge (ALJ) determined that approximately $90 million of State Farm’s 1989 California ALAE consisted of expenses actually paid, while approximately $30 million reflected an increase in ALAE reserves. The Department of Insurance (the Department) calculated State Farm’s ALAE at $40 million, using the inter-jurisdictional allocation. The ALJ noted that the $80 million difference between the parties’ ALAE figures was “the most economically significant issue in the case.” The ALJ concluded that the paid ALAE data were reliable and justified, but the reserve data were not. However, the ALJ disagreed with the Department’s argument that all of State Farm’s California ALAE had to be rejected because of the flawed reserve data. The ALJ accepted the paid data and rejected the reserve data, treating State Farm as though it had not changed its year-end reserves at all from 1988 to 1989. The Commissioner adopted the ALJ’s findings on the problems with State Farm’s California ALAE reserve data, but rejected her remedy. The Commissioner ruled that State Farm’s ALAE were subject to the interjurisdictional allocation as calculated by the Department. The trial court annulled the Commissioner’s decision, finding “the weight of the evidence establishes that State Farm’s 1989 reserve changes for Fire Lines ALAE were reasonable and reliable.” The court noted: “The Commissioner argues that State Farm’s reserving process was ‘flawed’ because it utilized ‘paid loss development triangles’ which, the Commissioner argues, only showed actual payments in prior years and did not adequately reflect the future impact of certain legal developments, primarily the California Supreme Court’s decision in Moradi-Shalal v. Fireman’s Fund Ins. Companies, 46 Cal.3d 287 [250 Cal.Rptr. 116, 758 P.2d 58] (1988). The evidence, however, does not support this argument . . . .” On appeal, the Commissioner renews his contention that State Farm’s 1989 ALAE reserve data for its California business were unreliable because the reserves were calculated solely by reference to historical payment patterns, without accounting for the prospective impact of Moradi-Shalal v. Fireman’s Fund Ins. Companies (1988) 46 Cal.3d 287 [250 Cal.Rptr. 116, *75 758 P.2d 58] and other legal trends of the late 1980’s that should have led State Farm to anticipate lower ALAE. The Commissioner asserts that in rejecting his reasoning on this point, “the superior court did not merely assign different weight to conflicting evidence. Rather, it misconstrued the legal requirements of the regulation and misunderstood the evidence to reach conclusions for which there is no record support.” Whether State Farm’s data were reliable is a predominantly factual question subject to the trial court’s independent review. (Ins. Code, §§ 1858.6, 1861.09.) As noted above, the Commissioner’s failure to discuss all the relevant evidence precludes him from pressing the claim that the evidence does not support the trial court’s finding. Even if the Commissioner’s briefing properly reviewed the evidence, we would not substitute our finding on the reliability of the data for the trial court’s, which is sufficiently supported by the record evidence cited in the court’s order and in State Farm’s brief. (20th Century Ins. Co. v. Garamendi, supra, 8 Cal.4th at p. 272 [substantial evidence standard governs review of trial court’s factual findings].) Furthermore, insofar as legal questions of regulatory interpretation are involved, our consideration of section 2643.6 in the context of the regulations as a whole leads us to the conclusion that the Commissioner’s view on the impact of Moradi-Shalal and related developments in determining an insurer’s rollback obligation is unsupportable. We must respect the Commissioner’s expertise in the insurance regulation arena; agency expertise is one factor identified by our Supreme Court as a reason for judicial deference. (Yamaha Corp. of America v. State Bd. of Equalization, supra, 19 Cal.4th at p. 12.) However, a second group of factors noted by the Yamaha court—“those suggesting the agency’s interpretation is likely to be correct”—is less helpful to the Commissioner in this case. (Id. at p. 13.) These factors include (1) indications of careful consideration by senior agency officials, particularly in a formal setting involving public notice and comment; (2) evidence that the agency has consistently maintained the interpretation in question, especially if it is long-standing; and (3) indications that the interpretation was contemporaneous with the enactment of the statute (in our case, a regulation). (Ibid.) We do not doubt the Commissioner and his senior staff have carefully considered the application of the regulations to State Farm’s California ALAE data. However, the theory advanced by the Commissioner regarding the defects in those data has not been formally adopted in a process including public participation. The regulations do touch on the effect *76 of developments like Moradi-Shalal, but in a manner inconsistent with the Commissioner’s argument, as discussed below. Nor does the Commissioner’s interpretation appear to have been contemporaneous with the enactment of section 2643.6. Since, as noted by the ALJ, this is the first contested case to test the requirements of section 2643.6, it cannot be said that the Commissioner has consistently maintained his interpretation. 3 Indeed, it is accurately characterized as a “litigating position.” We give the Commissioner’s interpretation substantial weight, but limited deference, in our construction of the regulations. (Culligan Water Conditioning v. State Bd. of Equalization, supra, 17 Cal.3d at p. 93; Yamaha Corp. of America v. State Bd. of Equalization, supra, 19 Cal.4th at p. 14.) The Commissioner’s theory is that State Farm’s 1989 California ALAE reserve data were not “reliable” under section 2643.6, subdivision (a), due to the insurer’s exclusive reliance on paid loss development triangles, a technique for extrapolating from historical data that failed to reflect recent changes in the California legal environment. The Commissioner insists that his determination of reliability extends to the methods used by an insurer to determine the level of its loss adjustment reserves. We cannot agree. Section 2643.6 applies only to multistate insurers. The reliability determination it prescribes is directed to making sure that California-specific data submitted by a multistate insurer are a reliable representation of the insurer’s California business. Section 2643.6, subdivision (a) states: “the determination whether data are ‘reliable’ shall be made on the basis of whether the Commissioner finds, under the circumstances, that the data are authentic and believable. In making that finding, the Commissioner shall consider, among other things, whether the data were recorded by persons and under circumstances likely to produce accurate data, whether the data were relied upon by the party in its business, and whether the data are consistent with other data.” The Commissioner does not contend State Farm’s ALAE reserve data are not “authentic,” “believable,” and “accurate” reflections of its actual expenses in California. His claim is that State Farm’s actual reserve levels were inflated due to faulty reserving methods. However, the reliability of reserve data in that sense is an issue common to all insurers, including those who operate in California only, and whose data are therefore not subject to *77 review under section 2643.6. The regulations contemplate that the Commissioner will review such common issues under the same standards. “While companies remain free to formulate their rates under any methodology, the Commissioner’s review of those rates must use a single, consistent methodology.” (§ 2643.1) The regulations include uniformly applicable provisions specifically governing the Commissioner’s review of the loss adjustment expense data submitted by all insurers for rollback purposes. Those provisions leave no room for the kind of scrutiny the Commissioner applied to State Farm’s 1989 ALAE reserve data in this case. Section 2645.4 provides the data requirements for losses and loss adjustment expenses during the rollback period. 4 Consistent with the general emphasis on historical data in article 5 of the regulations, section 2645.4 mandates the use of “actual loss and loss adjustment expense data for calendar year 1989.” In supplemental briefing on the meaning of this requirement, the Commissioner correctly argues that “actual data” are not whatever data the insurer chooses to submit in its rate application; rather, they must be an accurate rendition of the insurer’s expenses in 1989. The Commissioner goes too far, however, when he asserts that “actual data” means “the right number,” which allows him to judge whether the insurer used “a correct process” when it determined its 1989 ALAE reserves. By no stretch can “actual” be understood to mean “correctly determined.” The plain *78 meaning of section 2645.4 authorizes the Commissioner to reject an insurer’s ALAE data on the ground that they misrepresent its actual 1989 reserves, but not on the ground that the actual reserves were improperly calculated in the first place. 5 This conclusion is corroborated by other aspects of the regulatory scheme. Section 2645.7 states: “No adjustment shall be made for credibility in rollback calculations.” The regulations provide for a “credibility adjustment” in prior-approval ratemaking when projected figures submitted by an insurer, including projected ALAE, are based on insufficient data. (§ 2644.23, subd. (a).) In such a case, the Commissioner “add[s] to the insurer’s data sufficient additional data, drawn from an approved source of substitute data ... to provide a total sample size sufficient to meet the applicable credibility criterion.” (§ 2644.23, subd. (c).) The Commissioner attempts to explain section 2645.7’s bar against credibility adjustments in rollback calculations by reasoning that “because rollbacks are calculated on the basis of historical, rather than projected, data, sample-size statistical reliability is not an issue.” If that were the case, section 2645.7 would be superfluous. In fact, rollback calculations are partly based on historical data that are the product of projections made by the insurer in 1988 and 1989 to account for its expected ALAE. That is the nature of reserves. 6 If an insurer submits historically accurate ALAE reserve data that were based on a statistically unreliable *79 sample, section 2645.7 clearly prevents the Commissioner from adjusting those data based on a different sample. The rollback regulations do not include a similar provision expressly precluding adjustments for the effect of changes in the legal environment like the Moradi-Shalal decision. However, a fair reading of the regulations as a whole makes it clear that such trends cannot be invoked by the Commissioner to adjust an insurer’s historical ALAE data for rollback purposes. The impact of legal developments is discussed in article 4, in conjunction with prior-approval ratemaking. Section 2644.7 governs “loss trend,” defined as “the process by which forces not reflected in historical loss data are expected to affect losses in the rating period.” Loss trends are applied to projected ALAE as well as to projected losses. (§ 2644.8.) Section 2644.7, subdivision (d) states: “Changes in the law such as that created by Moradi-Shalal v. Fireman’s Fund Ins. Companies (1988) 46 Cal.Sd 287 can be expected to have a substantial effect on payouts, and consequently the Commissioner shall specify, in accordance with section 2646.3, the manner in which loss data predating Moradi-Shalal shall be adjusted in affected lines to take this change in the law into 7 No such adjustment is contemplated for rollback calculations, although the rollback regulations address the problem of inflated reserves. Section 2645.4, subdivision (b) provides: “Actual incurred losses shall be used, subject to validation by the Commissioner to ensure that there has not been inappropriate reserves-strengthening.” Section 2645.4, subdivision (c) permits the Commissioner to adjust an insurer’s 1989 reserves only if they fail to meet one of three numerical tests. All the tests are based solely on historical data and omit any consideration of trends in the legal environment. (See fn. 4, ante.) The Commissioner insists these tests apply only to loss reserves, not to loss adjustment expense reserves. 8 He contends the regulations leave him with wider discretion to review ALAE reserves to determine *80 “whether the claimed quantity was accurate, reliable, and in conformity with the other requirements of the regulations.” According to the Commissioner, allowing the impact of legal trends on loss reserves to escape his review for rollback purposes was the result of a rational policy choice to conduct only a rough numerical review of losses in every case, and an optional, more sophisticated review of other data in selected cases. The regulations, however, make no provision for optional unfettered review of 1989 data in categories other than losses. They require the Commissioner to accept actual historical data. The only data that escape this requirement are those that fail the reserves-strengthening tests of section 2645.4, subdivision (c). While the regulations explicitly recognize that legal trends not reflected in the historical data may have a “substantial effect” on insurers’ expenses (§ 2644.7, subd. (d)), that effect is conspicuously omitted from the provisions governing rollback 9 When an insurer submits historically accurate 1989 loss adjustment expense reserve data, the Commissioner must use them in rollback calculations without regard to whether the insurer adequately accounted for changes in the legal environment in 1989. The Commissioner’s other attack on the trial court’s ALAE determination is based on section 2643.6, subdivision (b), which required State Farm to *81 “justify any deviation” between its California data and its “experience in other states.” Here the Commissioner’s arguments are factual and require no regulatory interpretation. He contends State Farm failed to justify the difference between its ratio of ALAE to incurred losses (the ALAE:IL ratio) in California and its countrywide ALAE:IL ratio. He also points to evidence showing State Farm’s total California loss adjustment expense to incurred loss ratio (including both ALAE and ULAE) was substantially higher than the ratios of insurers doing business only in California. The ALJ found that the two Fire Lines accounting for most of State Farm’s ALAE were Homeowners and Commercial Multi-Peril (CMP). She determined the ALAE:IL ratio for Homeowners was 4.59 times higher in California than in the country as a whole, and the ALAE:IL ratio for CMP was 1.9 times higher. The ALJ noted the difference between State Farm’s total loss adjustment expense ratio and the ratios of California insurers suggested that its 1989 ALAE reserves were unduly inflated. After deleting the California reserve increase from the ALAE calculation, the deviation from State Farm’s nationwide ALAE:IL ratios fell to 3.75 in Homeowners, and to 1.24 in CMP. The ALJ concluded these reduced deviations were justified by the evidence, including the high ratio of lawsuits to incurred losses and the higher prevalence of complex litigation in California. The Commissioner, however, refused to accept State Farm’s paid expenses as the sole component of its ALAE, and deemed the difference in the unadjusted ALAErIL ratios unjustified. The trial court, on the other hand, found the deviation in ALAE:IL ratios was justified by the weight of the evidence. The court noted the higher proportion of lawsuits in California, both absolutely and in relation to incurred losses; the more complex California litigation; the close correlation between State Farm’s ALAE data and the more numerous lawsuits; testimony that State Farm’s ALAE was necessary, prudent, and properly spent; and the fact that the duty to defend is particularly broad in California. Again, we would decline to reweigh the evidence and substitute our judgment for the trial court’s, even if the Commissioner’s briefs fairly summarized all the relevant evidence on this issue. B. The ULAE Data The Commissioner contends the court made legal errors in its findings on State Farm’s ULAE data. The Commissioner insists the ULAE data were not truly California-specific, because they were partly derived by allocating companywide data. He also maintains State Farm failed to justify *82 the fact that its California ULAE were 20 percent higher than it incurred elsewhere. The ALJ considered the evidence supporting the ULAE data at some length, and although she rejected some of State Farm’s arguments she ultimately concluded that the California ULAE data were reliable and justified. As with ALAE, most of the ULAE were actually paid expenses, which amounted to $249,171,004. The reliability and justification of the paid ULAE was not contested. Another $32,633,926 was derived by apportioning expenses from an office that handled both California and Hawaii claims, and by allocating “a relatively small amount of home office expenses to California ULAE.” Regarding the $32,633,926, the ALJ found “[t]he evidence is undisputed that State Farm uses its California ULAE data in the regular course of its business and that its allocation method is reasonable and consistent with the National Association of Insurance Commissioners (‘NAIC’) handbook method.” The third component of ULAE was a $12,947,681 increase in reserves, which was based on State Farm’s allocation of company-wide expenses in proportion to loss reserves. The ALJ noted that this reserve change was substantially lower than in other states, and a third of it was accounted for by the Loma Prieta earthquake and was uncontested. The ALJ reasoned that to disallow the California ULAE data because it was based in part on allocations would be “particularly unfair” because State Farm had allocated its nonpaid ULAE by a lower ratio than would have been applied under section 2643.6, subdivision (a)(1). The ALJ also observed that the Department’s position on this issue was “ironic” because the countrywide ULAE the Department would use to perform the interjurisdictional allocation was itself a product of allocating an undifferentiated loss adjustment expense figure on State Farm’s annual statement to derive numbers for ULAE and ALAE. Regarding justification for the higher ULAE:IL ratio in California than elsewhere, the ALJ determined the auto lines component of the reserves was justified by higher instances of fraud and meritless claims in California, as reflected in a higher rate of claims closed without payment. The ALJ also credited evidence that high ULAE were justified by a high level of customer service. For the non-auto lines, the Department contested only the Homeowners ULAE. The parties agreed at oral argument during the administrative proceeding that this ULAE component would not affect State Farm’s rollback calculation due to the operation of the regulatory efficiency standard for fixed expenses. (§ 2644.12.) In any event, the ALJ found the non-auto *83 lines ULAE data were justified by a higher rate of claims closed without payment. The Commissioner rejected the ALJ’s ULAE findings on the grounds the data were not “recorded California data,” and the reserve changes did not reflect “material factors, especially California-specific developments, not yet reflected in the paid data.” The trial court annulled the Commissioner’s determination and ordered the Commissioner to use State Farm’s California ULAE data to calculate its rollback. The court found the “$12,947,681 in reserve change is modest and is supported by the evidence.” The court found the $32,633,926 “to be a reasonable and reliable statement of expenses actually incurred in California, which was even lower than the allocation using the regulatory method of allocation.” Finally, the court stated that since ULAE are subject to the regulatory efficiency standard, “there is no argument for further disallowance based on grounds of ‘reasonableness.’ ” The Commissioner contends section. 2643.6 simply does not allow the use of multistate data partially allocated to California by an insurer, which are not “fully California-specific.” However, the regulation is not drafted so narrowly. Section 2643.6, subdivision (a) forbids the use of submitted California data “[wjhere data are maintained on a multi-state basis only, or where California data are not reliable.” (Italics added.) Here, only about 16 percent of State Farm’s ULAE were allocated from multistate data, according to the Commissioner’s calculation. Section 2643.6, subdivision (a) does not require that all data must be kept on a California-specific basis to meet the reliability standard. It directs the Commissioner to determine whether data are “authentic and believable,” “recorded by persons and under circumstances likely to produce accurate data,” “relied upon by the party in its business,” and “consistent with other data.” {Ibid.) Section 2643.6, subdivision (b) requires the insurer “to demonstrate the accuracy of the data and to justify any deviation from the insurer’s experience in other states.” None of these regulatory criteria rule out the use of data allocated in part from figures kept on a multistate basis,, so long as the evidence establishes the data are authentic, believable, and accurate. Moreover, to require all ULAE data to be derived solely from California-specific sources would have the practical effect of precluding multistate insurers from using California ULAE data instead of the intequrisdictional allocation. As the Commissioner noted in his 20th Century opening brief, some ULAE, “such as home-office overhead, are inherently national, requiring a choice from among various methods of allocating such expenses to each jurisdiction.” Nevertheless, the Commissioner does not contend that *84 ULAE are categorically excluded from the provisions of section 2643.6 authorizing the use of California data. In the administrative proceedings below, the Department conceded that California ULAE data were not precluded, although the ALJ noted the Department’s attempt to “retreat to a position of neutrality on the issue” in its posthearing brief. (The Commissioner deemed this question moot in his decision, since he found State Farm’s ULAE data unreliable and unjustified.) The Commissioner argues next that the trial court made no finding on justification, and erroneously found that the efficiency standard renders such findings superfluous. According to the Commissioner, an insurer with unusually low ULAE that should translate into lower premiums could inflate its ULAE data without running afoul of the efficiency standard. Regarding the asserted absence of any finding on justification, we note that the trial court declared the ULAE data were both “reliable and justified.” We presume the evidence supported this finding. (Trailer Train Co. v. State Bd. of Equalization, supra, 180 Cal.App.3d at pp. 587-588.) We agree with the Commissioner that the efficiency standard does not necessarily relieve an insurer from the burden of justifying its California ULAE data. However, we are satisfied that the trial court’s finding on justification is supported by substantial evidence, without regard to the court’s reasoning on the effect of the efficiency standard. 4. The Federal Income Tax Factor The “profit factor” figures in both the ratemaking formula and the efficiency standard under the regulations. (§§ 2644.3, 2644.12.) The “federal income tax factor” is an element of the profit factor. (§ 2644.15.) Section 2644.18 defines the federal income tax factor. It provides in relevant part: “(a) ‘Federal income tax factor’ means 1.0 minus the insurer’s effective federal income tax rate reported in the most recent year for which historical data are available, giving full account to all tax credits and offsets used or available to the insurer. Where there has been a change in tax laws between the recorded period and the rating period, the effective tax rate shall be calculated using the historical data and the tax rules for the rating period. “(b) Where the insurer had a net tax credit, or where the insurer had a net tax liability on a net pretax loss, the effective tax rate shall be zero and “(1) if the insurer had a net tax credit, the amount of the credit shall be added, as a positive number, to nationwide projected ancillary income . . . .” (Ibid.) *85 The Commissioner contends the trial court misapplied this regulation in respect to a $40,489,063 reduction in State Farm’s 1989 federal income taxes. The Commissioner characterizes this item as follows: “The $40,489,063 was incurred in 1989 to offset a (positive) tax liability of that amount incurred in the prior year and reported in the annual statement. A subsequent reversal of position by the Internal Revenue Service meant that the $40,489,063 incurred in 1988 would never be paid, so State Farm properly incurred the offset in 1989.” The Commissioner argues that section 2644.18 requires this “offset” to be included in State Farm’s net tax credit and added to its ancillary national income for 1989. The ALJ noted the Department’s tax expert did not address this item, but the head of State Farm’s corporate tax department explained it as follows: “$40.5 million of the reported amount reflects a reconciliation in the 1989 Annual Statement of an error in the previous Statement incorrectly showing that amount as incurred, due to the IRS reversing its earlier position on estimated recoveries of salvage and subrogation. Since State Farm filed its 1988 Annual Statement early in the year at a time when the IRS claimed those items should be taxed but did not file its federal income tax return until August or September after the IRS had changed its position, State Farm made the reconciliation on its 1989 Annual Statement.” The ALJ determined “State Farm did not actually enjoy a tax break on this item, which it never actually owed or paid in the first place. To indulge the fiction that it did would have the anomalous effect of penalizing State Farm for the IRS’ change of position.” The Commissioner rejected the ALJ’s resolution of federal tax issues without specifically addressing the correction at issue here. The trial court agreed with the Commissioner’s disposition of the tax issues, except for the correction. The court said “the evidence is uncontradicted that this did not reflect a tax credit of any kind. [Citation.] It does not fall within the term net tax credit. The Commissioner attempts to justify this allocation on grounds of expediency, but that is an insufficient basis where, as here, the evidence is clear that this amount was not in fact a 1989 tax credit or benefit of any kind. [Citation.]” The Commissioner argues the correction “was an undifferentiated part of the $319,755,178 net tax credit incurred by State Farm in 1989, and that amount is the only quantity of relevance under the regulation.” We disagree. Section 2644.18, subdivision (a) addresses situations resulting from a change in federal tax rules, which is essentially what happened here: “Where there has been a change in tax laws between the recorded period and the rating *86 period, the effective tax rate shall be calculated using the historical data and the tax rules for the rating period.” The historical data reflect State Farm’s statement of taxes incurred on its 1988 annual statement, its failure to actually include $40,489,063 of that amount in its 1988 tax filing due to the change of position by the IRS, and its adjustment in the 1989 Annual Statement to reflect these developments. Under the tax rules for 1989, which is the rating period, the $40,489,063 was simply a nontaxable item, not a credit or an offset. It was not a part of State Farm’s effective tax rate because it never appeared in any tax filing and never affected any actual tax credit or liability. . A change in tax rules will usually not affect an insurer’s tax credit or liability as an undifferentiated whole, which is how the Commissioner wishes to view the data used for determining the federal income tax factor. Section 2644.18 contemplates that historical data affected by a rule change will be segregated and evaluated under the rules in effect for the rating period. We conclude the trial court properly applied section 2644.18 in this case. Disposition The trial court’s order is affirmed. State Farm shall recover its costs on appeal. Corrigan, Acting P. J., and Walker, J., concurred. The petition of defendant and appellant for review by the Supreme Court was denied March 22, 2000. Mosk, J., was of the opinion that the petition should be granted. Footnotes [1] Subsequent unspecified section references are to the Department of Insurance regulations found in title 10, subchapter 4.8 of the California Code of Regulations. [2] Our Supreme Court has noted that the Commissioner’s application of the ratemaking regulations in individual cases is a quasi-legislative function. (20th Century Ins. Co. v. Garamendi, supra, 8 Cal.4th at p. 277.) This characterization does not affect the standards of *72 review stated above, however. The 20th Century court made it clear that the Commissioner’s evidentiary findings, even though they are quasi-legislative in character, are subject to the trial court’s independent review, which is in turn subject to substantial evidence review on appeal. (8 Cal.4th at pp. 272, 278 and fn. 12.) The 20th Century court did not consider the standard of review governing the Commissioner’s interpretation of particular regulations, but it is clear that the “qiiasi-legislative” label does not insulate the Commissioner’s legal determinations from the usual level of judicial scrutiny any more than it protects his factual determinations. Even though we are “more likely to defer to an agency’s interpretation of its own regulation than to its interpretation of a statute, since the agency is likely to be intimately familiar with regulations it authored and sensitive to the practical implications of one interpretation over another,” final responsibility for interpreting the law, whether statutory or regulatory, rests with the courts. (Yamaha Corp. of America v. State Bd. of Equalization, supra, 19 Cal.4th at p. 12; Culligan Water Conditioning v. State Bd. of Equalization, supra, 17 Cal.3d at p. 93.) [3] The parties dispute whether the Commissioner took inconsistent positions in his briefing before the Supreme Court in 20th Century Ins. Co. v. Garamendi, supra, 8 Cal.4th 216. We grant the Commissioner’s request for judicial notice of some briefs and other materials submitted in those earlier proceedings, but they have only a limited effect on our decision. The issues presented here are quite narrow in scope and were not explored in any detail in the 20th Century litigation. [4] The relevant provisions of section 2645.4 are as follows: “In lieu of projected losses and loss adjustment expenses, actual loss and loss adjustment expense data for calendar year 1989 shall be used.” “(b) Loss development and trending. Actual incurred losses shall be used, subject to validation by the Commissioner to ensure that there has not been inappropriate reserves-strengthening. Because incurred losses for calendar year 1989 are now known, no loss development is necessary or appropriate and loss and loss adjustment expense trend would have so insubstantial an effect on rates that no adjustment shall be made.” “(c) Reserves-strengthening. The Commissioner will not adjust an insurer’s 1989 reserves when any of the following conditions are found: “(1) Reserves: Premium, One-Year Test. When the year-end 1989 loss reserves to 1989 earned premiums is less than or equal to the year-end 1989 loss reserves to 1988 earned premiums. “(2) Reserves: Premium Four-Year Test. When the 1989 reserves-to-premium ratio is less than or equal to the average of the rfserves-to-premium ratios for 1985 through 1988. “(3) Incurred Versus Paid Test. When the ratio of 1989 incurred losses to 1988 incurred losses was no greater than the ratio of the payments on 1989-incurred losses was to payments on 1988-incurred losses." “(d) If the insurer satisfies none of the three tests in subdivision (c) of this section, unless the Commissioner finds, based on evidence proffered by the insurer and any other party, that some other number better reflects appropriate reserve levels, the insurer’s reserves shall be adjusted ... to the same multiple of earned premium as the 1985 through 1988 reserves had to 1985 through 1988 earned premium. . . .” [5] In a second request for judicial notice, which we grant, the Commissioner submits two decisions involving California-only insurers in which he redistributed data between ALAE and ULAE, in order to accommodate the regulations’ different treatment of the two types of loss adjustment expenses. In one case the Department sought and obtained the adjustment; in the other the insurer successfully requested the adjustment. We note that section 2643.6 would yield different data allocations for a multistate insurer, though the Commissioner describes the reallocation in one case as “analogous” to the inteijurisdictional allocation. Furthermore, the adjustments in those cases did not conflict with the requirement that actual data be used in rollback calculations. A misclassification of loss adjustment expense data provides an inaccurate picture of the insurer’s actual 1989 loss adjustment expenses for purposes of the rollback calculation, and may properly be corrected. But a miscalculation that does not distort the regulatory classifications or the insurer’s actual 1989 expenses is not correctable. We realize that the data available to the Commissioner do not always clearly reflect historical loss adjustment expenses. The Commissioner has broad discretion to determine whether data submitted by an insurer are consistent with its actual 1989 loss adjustment expenses, and to make appropriate changes to improve the historical accuracy of the data. The regulations do not permit him to second-guess 1989 loss adjustment reserving decisions, however. [6] The clause of section 2645.4 requiring the use of actual loss and loss adjustment expense data “[i]n lieu of projected losses and loss adjustment expenses” does not mean that 1989 ALAE data should omit the reserve component. The parties do not dispute that 1989 ALAE are calculated on an incurred basis, which accounts for reserve changes. The reason section 2645.4 calls for a substitution of actual for projected data is that the ratemaking formula is stated in terms of the projected data used in prior-approval ratemaking. (§ 2644.3, subd. (a)(1) *79 and (2).) In rollback calculations, historical data are used in lieu of forecasts, but some of the historical data were themselves derived from projections. [7] The Commissioner informs us that due to limited regulatory resources he has never issued the generic determination required by section 2644.7, subdivision (d). He notes that as a practical matter, the need for a finding on the effect of Moradi-Shalal in prior-approval ratemaking has been diminished by the passage of time and the accumulation of historical data. [8] In the administrative hearing, State Farm contended its ALAE reserves met the reserves-strengthening tests of section 2645.4, subdivision (c), and therefore should be considered reliable and justified under section 2643.6. The ALJ rejected this argument. State Farm renewed the claim in the trial court, but the court did not address it in finding that the evidence established the reliability of the ALAE reserve data. State Farm’s argument on the applicability of the reserves-strengthening tests is an appealing one. The Department’s expert *80 conceded in his testimony that the tests could serve as an appropriate indicator of excessive ALAE reserves. The Commissioner has relied on the provision in section 2645.4, subdivision (b) that “incurred losses shall be used” as support for using incurred ALAE rather than paid ALAE in rollback calculations, even though section 2645.4 never mentions incurred ALAE. Arguably, the same reasoning would support use of the reserves-strengthening tests for ALAE although the regulation specifically mentions only loss reserves. And, as our discussion above makes clear, the reserves-strengthening tests are the only means permitted by the regulations for detecting over-reserving in actual 1989 data. Nevertheless, in supplemental briefing on this issue the Commissioner demonstrates that from their inception the reserves-strengthening tests were not intended to apply to loss adjustment reserves. The Commissioner claims loss data are better suited than ALAE data to review by numerical tests, because unlike ALAE, losses are regularly reported in statutory annual statements, making loss data more reliable. The Commissioner has consistently maintained in this proceeding that the reserves-strengthening tests apply only to loss reserves, and his position finds support in the literal terms of section 2645.4, subdivisions (b) and (c). We defer to the Commissioner’s interpretation on this point. [9] Section 2645.4, subdivision (b) states that “loss adjustment expense trend would have so insubstantial an effect on rates that no adjustment shall be made.” This reference is puzzling in view of the “substantial effect” recognized by section 2644.7, subdivision (d), and the fact that the principal issue in this case concerns the substantial effect on State Farm’s rate of an alleged failure to account for loss adjustment expense trends in 1989. The Commissioner offers the following explanation: The rollback period was November 8, 1988, to November 7, 1989. (§ 2645.1.) However, section 2645.4 requires the use of data “for calendar year 1989.” It might be contended that 1989 calendar year data overstate an insurer’s actual loss adjustment expenses during the rollback period. Section 2645.4, subdivision (b) deems such variations insubstantial and precludes any adjustment for them.