We apply to the resolution of the complex matter of this ease relatively simple rules. We undertake a lengthy examination of the history, as well as the content, of the attacked regulation of the Director of the Department of Employment to show that he did not violate the precept that he must not act arbitrarily, capriciously or without evidentiary support. We explain, too, why the director properly applied the regulation to existing plans of disability insurance, *828and in so doing did not effectuate an unlawful retroactivity.
The attacked action of the director comprises two regulations (Cal. Admin. Code, tit. 22, §§ 3254(i)-1, 3254(i)-2) which he adopted pursuant to Unemployment Insurance Code sections 3254, subdivision (i), and 3270 to prevent private insurance companies from substantially selecting risks for unemployment compensation disability insurance that would adversely affect the State Disability Fund. Some of those companies brought an action in the Superior Court of Sacramento County for declaratory relief to determine the validity of the regulations and to enjoin their enforcement. The court issued a preliminary injunction enjoining the director from enforcing the regulations pending a determination of the action.
Thos. L. Pitts, individually and as secretary-treasurer of the California Labor Federation, AFL-CIO, then commenced this proceeding by filing a petition for a writ of mandate with the District Court of Appeal, Third District, seeking to compel the director to enforce his regulations. After that court denied the petition, without opinion, petitioner requested this court for a hearing, alleging that the director has not, and will not, enforce his regulations so long as the superior court’s preliminary injunction remains in effect. Petitioner asserts an interest in the enforcement of the regulations because a substantial number of members affiliated with the labor federation pay contributions to, and receive benefits from, the State Disability Fund. In a letter filed with this court the director stated that he did not oppose the petition for the writ but that the private insurance companies would do so. Through their counsel these companies informed us by letter that, if the case were an appropriate one, they had no objection to this court’s assuming jurisdiction, although they urged that the matter could be more satisfactorily determined by the superior court.
We granted a hearing, issued an alternative writ of mandamus ordering respondent to place the regulations into full force pending the determination of their validity, and stayed the proceedings in the superior court, including the enforcement of the preliminary injunction. We did so because of compelling reasons based on public interest. (See Perry v. Jordan (1949) 34 Cal.2d 87, 90-91 [207 P.2d 47].)
In his letter to this court respondent asserted that if substantial adverse selection by the private insurance companies did not terminate on January 1, 1963, the State Disability *829Fund would suffer additional losses of several million dollars, and, unless the Legislature increased the mandatory statutory contributions required of the workers, the fund would be unable to pay benefits in April 1965. Other reasons persuading us to assume jurisdiction rested upon the Legislature’s direction to the director that he adopt regulations to be effective January 1, 1962 (Stats. 1961, eh. 2154, p. 4458, §24), which date has, of course, passed. We were likewise influenced by the fact that the parties stipulated that no new evidence would be introduced before the superior court.1 Finally, we concluded that petitioner could properly assert his standing to maintain this action since he is a citizen with a substantial interest in the enforcement of the director’s public duty. (Hollman v. Warren (1948) 32 Cal.2d 351, 357 [196 P.2d 562].)
The Legislature in 1946 enacted the Unemployment Compensation Disability, or UCD, Act. The act is now contained in part 2 of the Unemployment Insurance Code and provides benefits for persons who are unable to work because of nonoceupational illness or injury. It thus fills the gap between workmen’s compensation, which provides benefits for disability incurred during employment, and unemployment insurance, which affords benefits for involuntary unemployment suffered by persons who are physically able to work. The act is financed by a 1 per cent contribution of each worker’s wages, presently limited to the first $4,100 paid to him during the calendar year. (Unemp. Ins. Code, §§ 984, 985.) It authorizes coverage under the State Disability Fund or under voluntary plans, which can be self-insured or insured by a qualified insurance company. Only a very small percentage of employees are covered by self-insured plans, but approximately 20 per cent are presently covered by insured voluntary plans.
The act set up a number of requirements for voluntary plans upon which the director’s approval depended. Thus a voluntary plan must provide greater benefits to covered employees than the Disability Fund (§ 3254, subd. (a)); it must be made available to all of the employees of the employer, or to all employees working in a distinct, separate establishment of the employer (§ 3254, subd. (b)); and approval of the plan must “not result in a substantial selection of risks adverse to the Disability Fund” (§3254, subd. (i)).
*830All of these requirements, except that as to substantial selection of adverse risks, have been in continuous operation since the enactment of the law. The adverse selection provision sought to prevent private insurers from insuring only the better risks and relegating the less desirable ones to the Disability Fund. (Report of Social Insurance Section of Interim Committee on Finance and Insurance, Assembly Interim Committee Reports, 1959-1961, Volume 15, No. 24, p. 63.) [Hereinafter referred to as the “Report.”] First by an informal procedure and later by a formal regulation, the director gave the statutory provision specific effect and required voluntary plans to include 20 per cent females in the total number of employees covered. (Report, p. 64.)
From 1946 through 1956 the Disability Fund balance steadily increased. (Report, p. 74.) During this period, the Legislature suspended the prohibition against substantial adverse selection. (Stats. 1953, ch. 1371, p. 2951, § 11, as amended; Stats. 1955, eh. 957, p. 1855, § 6, as amended; Stats. 1957, ch. 2107, p. 3736, as amended; Stats. 1959, ch. 2155, p. 5209, § 5.) This suspension remained in effect during the calendar years 1954 to 1961 inclusive. Beginning in 1957, however, the Disability Fund balance each year showed a decrease, which resulted, in part, from the increase in statutory benefits, without a corresponding increase in employee contributions, and the suspension of the prohibition against adverse selection. (Report, pp. 73, 85.)
To meet the developing financial difficulties, the Legislature in 1961 transferred $70,000,000 from the Unemployment Trust Fund to the Disability Fund (Unemp. Ins. Code, § 3006), enacted a provision for four progressive annual increases of $500 in the wage base which would be subject to the 1 per cent employee contribution for the years 1962 to 1965 inclusive (Unemp. Ins. Code, § 985), and provided that the adverse selection provision of section 3254, subdivision (i), would again become operative on January 1, 1962 (Unemp. Ins. Code, § 3270). The Legislature gave the following direction, which is the crucial one in the present case: “The Director of Employment, in determining whether the approval of a voluntary plan or group of plans under Section 3254 or 3255 of the Unemployment Insurance Code will result in a substantial selection of risks adverse to the Disability Fund, shall implement such sections by authorized regulations to be effective January 1,1962, prescribing a reasonable test or tests for the measurement of such adverse risks, evaluating and *831taking account of the sex, age, and wage distribution of the employees to be covered by the plan and the employees covered by all other plans insured by the same insurer, and such other factors as may relate to a reasonable test or tests for such purpose, when compared to the incidence or distribution of the same or similar factors with respect to employees in employment under said code.” (Stats. 1961, ch. 2154, p. 4458, § 24.)
Pursuant to this legislation, the director noticed a proposed regulation for a public hearing to be held on October 16, 1961. After the hearing the director modified and adopted the regulation. (Cal. Admin. Code, tit. 22, § 3254(i)-l.) More than 200 appeals were filed with the Unemployment Insurance Appeals Board pursuant to section 309 of the Unemployment Insurance Code. The board held five days of hearing in December 1961 and thereafter on April 6, 1962, by a two-to-one decision, approved the regulation.
Since the original regulation provided that existing plans could be cancelled only at the end of the first calendar quarter of each year, the board’s ruling came too late to permit such cancellations in 1962. Consequently, the director filed and published the emergency regulation (Cal. Admin. Code, tit. 22, §3254(i)-2) which is essentially the same as, but not identical with, the original regulation. Pursuant to Government Code section 11422.1, the director, in order to make the regulation permanent, gave notice that a public hearing on the emergency regulation would be held on May 14, 1962. Before that date, however, the insurance companies filed the aforementioned action for declaratory relief in the superior court and obtained the preliminary injunction prohibiting the enforcement of either regulation. Petitioner, then, as we have stated, instituted this mandamus proceeding in the appellate courts against the director.
The public hearing on the emergency regulation took place on May 14, 15, and 21, 1962. On May 29, the director filed a certificate of compliance with the Secretary of State as required by Government Code section 11422.1. The insurance companies then lodged their appeal with the Unemployment Insurance Appeals Board. After receiving briefs and hearing oral argument, the board, on June 22, dismissed the appeal on the ground that it had no jurisdiction to hear an appeal from an emergency regulation.
The emergency regulation has become permanent in that the director followed the procedures specified in Government *832Code section 11422.1. We need not, therefore, pass upon the contentions of the insurance companies that the director lacked the legal authority to adopt an emergency regulation and that, even assuming his authority, the situation did not warrant that kind of regulation.2 Since, once the director complied with section 11422.1, these issues become moot, our sole question congeals into whether the regulation is valid as a matter of substantive law.3
In analyzing the substantive law we shall separately consider the following subject matters: (1) the scope of review of the action of the director, (2) the application of the regulation to existing plans, (3) the content of the regulation, (4) the insurers’ objections to the standards for approval test, (5) the insurers’ objections to the lack of a weighting formula, (6) the insurers’ objections to the amounts of the tolerances.
(1) The Scope of Review of the Action of the Director.
We confront a situation here which graphically illustrates the wisdom of the general rule that the court should not substitute its judgment for that of an administrative agency which acts in a quasi-legislative capacity. All of the parties to this litigation recognize the intricate and technical nature of the subject matter as well as the expertise and full technical knowledge which its administration requires. It would be presumptuous of a court to claim such skill; it will not, therefore, superimpose its own policy judgment upon the agency in the absence of an arbitrary and capricious decision.
*833The decisions do not sustain the insurers’ contention that in determining the validity of the regulation this court should exercise its independent judgment and reweigh the proffered evidence. While such a test may apply to the review of the adjudicatory or quasi-judicial rulings of certain agencies (Code Civ. Proe., § 1094.5) it does not pertain to the review of regulations rendered by an agency in its quasi-legislative capacity. As to the quasi-legislative acts of administrative agencies, “judicial review is limited to an examination of the proceedings before the officer to determine whether his action has been arbitrary, capricious, or entirely lacking in evidentiary support, or whether he has failed to follow the procedure and give the notices required by law.” (Brock v. Superior Court (1952) 109 Cal.App.2d 594, 605 [241 P.2d 283]; see also Ray v. Parker (1940) 15 Cal.2d 275, 310-311 [101 P.2d 665]; Vita-Pharmacals, Inc. v. Board of Pharmacy (1952) 110 Cal.App.2d 826, 832-833 [243 P.2d 890].)
Justice Bray in Brock explains the legislative origin of section 1094.5 of the Code of Civil Procedure, which does provide for independent judgment in the review of quasijudieial rulings, stating;“. . . [A] n examination of the legislative history of the section convinces us that it was not intended that the section apply to such acts or any quasi-legislative acts of an administrative body.” (109 Cal.App.2d at p. 598.) The section, as the opinion explains, followed the Tenth Biennial Report of the Judicial Council, and that report shows that its authors did not propose to cover rule-making agencies. It states: “[N] o attempt has been made in this report to include the agencies which are primarily rule-making in nature. . . . The proposed legislation is designed to provide a solution for many of the difficulties and injustices arising in the administrative licensing and disciplining of private citizens.” (Tenth Biennial Report, p. 10.)
The statute and the cases account for the insurers’ contention that, accepting the validity of the distinction between quasi-judicial and quasi-legislative action, the director’s regulation falls within the former classification. They contend that the director’s function here was a narrow one; “the promulgation of the regulation is prescribed by specific legislative direction.” Hence the insurers distinguish it from the broad discretion permitted the administrative officer in denying an application for a permit to conduct a bail bond business in such a case as McDonough v. Goodcell (1939) 13 Cal.2d 741 [91 P.2d 1035,123 A.L.R. 1205]. Here the Legislature *834instructed the director to formulate reasonable tests to prevent substantial adverse selection; the Legislature specifically required a hearing for that purpose. Accordingly, the insurers argue that the director rendered a quasi-judicial determination.
The contention fails because, in the first instance, the distinction between the quasi-legislative and quasi-judicial decision contemplates the function performed rather than the area of performance; the breadth or narrowness of the discretion cannot control. In the second instance, the fact that the validity of the director’s regulation depends upon its reasonableness surely does not in itself make the determination quasi-judicial. All regulations must be “reasonably necessary to effectuate the purpose of the statute” (Gov. Code, § 11374; Duskin v. State Board of Dry Cleaners, ante, pp. 155, 165 [23 Cal.Rptr. 404, 373 P.2d 468]). If the insurance companies’ contention were sustained, it would convert the determinations of all agencies into quasi-judicial ones and effectively destroy the ruling of Brock that Code of Civil Procedure section 1094.5 does not apply to the review of quasi-legislative action.
Because of these basic considerations the cases upon which the insurance companies mainly rely do not sustain their position. Thus Temescal Water Co. v. Department of Public Works (1955) 44 Cal.2d 90 [280 P.2d 1], involved an agency which acted in its adjudicatory capacity; the court there expressly confined the independent judgment rule to such cases. (44 Cal.2d at pp. 101, 105.) In People v. Chevalier (1959) 52 Cal.2d 299 [340 P.2d 598], the court held “that the conclusive effect accorded by the Legislature to the condemning body's findings of necessity” in a matter of eminent domain “cannot be affected by allegations that such findings were made as the result of fraud, bad faith, or abuse of discretion” (p. 307); the ruling does not affect the issue before us. Thomas v. California Emp. Stab. Com. (1952) 39 Cal.2d 501 [247 P.2d 561], passed upon a decision of the commission denying unemployment insurance benefits to striking employees, a determination obviously not quasi-legislative.
We conclude that in determining whether the director has acted arbitrarily or capriciously, this court does not inquire whether, if it had power to draft the regulation, it would have adopted some method or formula other than that promulgated by the director. The court does not substitute its judgment for *835that of the administrative body.4 The rendition of this regulation involved “highly technical matters requiring the assistance of skilled and trained experts and economists and the gathering and study of large amounts of statistical data and information.” (Ray v. Parker, supra, 15 Cal.2d 275, 311.) Under such circumstances, “courts should let administrative boards and officers work out their problems with as little judicial interference as possible.” (Ray v. Parker, supra.) The substitution of the judgment of a court for that of the administrator in quasi-legislative matters would effectuate neither the legislative mandate nor sound social policy.
(2) The Application of the Begulations to Existing Plans.
We cannot accept the twofold argument of the insurance companies that the statute did not authorize the director to apply the regulations to existing plans. Their first contention that such application would work an unlawful retroactivity does not find support in the cases; their second argument that in any event the Legislature harbored no intent that the regulation affect previously approved plans does not accord with the legislative history or purpose.
No rule of law holds that the Legislature cannot fix the standards for plans which are to operate in the future even though the director approved such plans in the past. The regulation looks solely to the future; it provides that the future operation of the plans comply with the standards. No sanction or penalty attaches to any past act of the companies ; the companies need only discontinue one or more existing noneomplying plans or establish complying ones.
The cases to which the companies refer in order to demonstrate that legislation will not be given retroactive application unless it expressly so provides deal in attempts to foist upon *836past conduct new and onerous legal consequences. Thus in Di Genova v .State Board of Education (1962) 57 Cal.2d 167, 174 [18 Cal.Rptr. 369, 367 P.2d 865] cited by the insurers, we held that the Legislature did not intend that a statute which provided for the suspension of credentials of teachers convicted of sex offenses apply to persons convicted prior to the effective date of the statute. In Interinsurance Exchange of Auto. Club v. Ohio Cas. Ins. Co., ante, p. 142 [23 Cal.Rptr. 592, 373 P.2d 640], likewise cited by the insurers, we stated that a statutory amendment repealing a rule requiring that a mandatory provision be included in all automobile insurance policies could not affect policies in force at the effective date of the amendment and could not thereby remove the mandatory coverage from previously negotiated contracts. In the cited cases of State of California v. Industrial Acc. Com. (1957) 48 Cal.2d 355, 361 [310 P.2d 1], and Aetna Cas. & Surety Co. v. Industrial Acc. Com. (1947) 30 Cal.2d 388, 393 [182 P.2d 159], we held that amendments to the workmen’s compensation provisions of the Labor Code could not be construed to cover previously incurred injuries.
In all of the above cases the attempted statutory interpretation sought to apply the new law of today to the conduct of yesterday. We determined that, in the absence of clear legislative language to the contrary, the legal consequences of a past act or occurrence, i.e., a conviction, a contract provision, or an injury, could not be altered by the subsequent enactment of a statute. In the instant ease, however, the application of the regulation to existing voluntary plans does not impose new legal sanctions upon the past action of the insurance companies • it imposes standards for the future operation of insurance plans.
We turn, then, to the insurance companies’ second argument that even assuming that the Legislature was legally empowered to establish new standards for the future operation of existing plans, it did not intend to do so. The companies argue that section 24 of the act (Stats. 1961, eh. 2154, p. 4458), which reinstated the prohibition against adverse selection, reads in the future tense, operates prospectively, and contains no reference to withdrawal of approval or requalification of current plans, and that, indeed the Legislature deliberately refused to give the section such “retroactive” effect.
The companies emphasize the deletion of an assembly amendment which specifically applied the section to existing plans. With the exception of a few words not here relevant, section *83724 of the bill, originally introduced as Assembly Bill 234 at the 1961 General Session of the Legislature, contained language identical to that of the enacted bill. The Assembly did, however, later amend the bill to add the following sentence: “The test or tests for determining the measurement of risks adverse to the Disability Fund shall also be applicable to all existing voluntary plans in accordance with fair and reasonable terms and conditions. ...” The Senate then deleted this sentence. The bill as thus amended passed both houses; subsequently the Governor signed it. According to the insurance companies, the quoted language indicates that the Legislature considered the possibility of applying the adverse selection provision to existing plans, and by later deleting the provision, decided not to do so.
Yet, prior to the adoption of the present statute, the administrative rulings of the appeals board had disclosed to the Legislature that the board regarded the identical language involved here to authorize withdrawal of approval of existing plans which resulted in adverse selection. When the Legislature reinstated the adverse selection provision (Unemp. Ins. Code, § 3254, subd. (i)) it reenacted the same language which was on the books in 1954. As we have previously explained, the director from 1946 to 1951 informally implemented the adverse selection provision by requiring that voluntary plans include at least 20 per cent females in the total number of employees covered. In 1951 the director embodied this standard in a formal regulation. Certain insurance companies challenged this regulation before the Unemployment Insurance Appeals Board. They argued that the predecessor of code section 3254, subdivision (i) contemplated a “prospective” test of adverse selection and that the regulation substituted a “retrospective” test in that it permitted the withdrawal of approval of plans previously approved by the director. The board rejected this contention.5
As we held in Richfield Oil Corp. v. Crawford (1952) 39 Cal.2d 729, 736 [249 P.2d 600], “past administrative action is evidence of the limits of the power to act . . . particularly *838when a statute is reenacted without change in the light of a settled administrative interpretation.” Applying this principle, we must assume that the Legislature, in reinstating section 3254, subdivision (i), knew and acted upon the following facts: that this section contained exactly the same words as it embodied from 1946 to 1954; that the director adopted only one regulation during this period; that this regulation provided for the withdrawal of approval from existing plans that did not maintain the required female content; that’ the validity of the regulation had been sustained by the appeals board. In light of this past administrative history, the Legislature, if it had meant to prohibit the director from withdrawing approval from plans previously approved, would have expressly so stated. In striking out the sentence of section 24 of Assembly Bill 234, which specifically applied the adverse selection test to existing plans, the Legislature did no more than eliminate surplusage. (See Burks v. Poppy Construction Co. (1962) 57 Cal.2d 463, 469 [20 Cal.Rptr. 609, 370 P.2d 313].)
Finally, the language of the Legislature compels no such unworkable result as the companies would read into it. If they are correct, an existing plan would become frozen into an infinite immunity. No new regulation could affect it; it would indefinitely continue to the everlasting disadvantage of the fund. Despite the fact that experience dictated the need for modification, the directors would be foreclosed from effecting changes in existing plans; only future ones could be controlled. The Legislature would then have created the grotesque situation in which some plans would work adverse selection of risks and others would not; indeed, the carrier which had written existing plans would be subjected to contrasting standards for its other plans. We do not attribute to the Legislature a design to impose such administrative chaos and such discriminatory criteria.
(3) The Content of the Regulation.
Before considering the companies’ characterization of the regulation as arbitrary and capricious, we summarize its provisions. The regulation contains two alternative tests for determining whether the insurers have substantially selected risks adverse to the Disability Fund: the standards for approval test and the risk experience rating test. Both tests emanated from the director’s determination that proportionately higher benefits are paid to females than to males, to persons age 50 and over than to younger persons, and to those *839earning less than $3,600 in annual wages than to those paid more than that amount.
The director faced the task of setting up a means of measuring the risks covered by the fund and by the private insurers in order to prevent the insurers from selecting only the best risks. For the basic test, the standards for approval, the director, rather than require some 300,000 employers in the state to submit new data, used the department’s record. Since the employer takes time to send in his standard reports and since the department too, takes time to process them, the selection of this method entailed a time-lag. This time-lag was, however, ameliorated by the constancy of the percentages of women, older workers, and low-wage earners. Seasonal fluctuations, however, could not be ignored; accordingly the director used the average of four quarters. Again, this factor did not impose a variation of importance because of the relative stability of the labor force.
Turning to the standards for approval test, we find that the director fixes for each year “standards for approval” for sex and age, and for each quarter the standard for wage distribution. He publishes the standards as percentage figures.
To determine the standards for sex and age the director takes a one per cent random sample of the employees in covered total employment in California during each of the four calendar quarters of a year, excluding employees who received less than $300 in wages during that year. Employees earning less than $300 are excluded because such employees are not “at risk”; i.e., the state would not be required to pay benefits tó them if they should become disabled. The director then determines for employees employed in each of the four quarters the percentages of females and employees age 50 and older. These percentages in each classification are averaged; the resulting percentages for sex and for age are reduced by specified tolerances (5 per cent for sex and 10 per cent for age).
To illustrate how the standard for approval for sex would be determined, we point to the following hypothetical and simplified facts: assume that in each of the four relevant quarters the percentages of female employees were 20 per cent, 25 per cent, 40 per cent, and 35 per cent, respectively, of the number of covered employees who received annual wages of $300 or more. The average of these percentages is 30 per cent. This percentage is reduced by the specified tolerance of 5 per cent applied to the 30 per cent; the result*840ing percentage is 1% per cent; deducting the 1% per cent from the 30 per cent the standard for approval for sex becomes 28.5 per cent.
The director pursues a similar procedure for wage distribution, except that he updates the percentage every quarter and that here the specified tolerance is 7 per cent. Thus, if we took the third quarter of 1962 as a point of reference, the director would determine the percentages of persons who earned annually over $300 but less than $3,600 in each of the four quarters for 1961. We have noted, supra, that the director must use statistics from an earlier period because of the delay in getting and compiling the information. The director averages the four percentages and then reduces them by the allowed tolerance.
In order to determine whether a particular insurer meets the standards for approval it submits a list of persons proposed to be insured who are employed during a given payroll period. The director then computes from the list the percentages of female employees and employees age 50 and over. He also determines from department records how many employees on that list earned less than $3,600 during the specified “four calendar quarters of computation” period. The director incurs another time-lag here; the mid point of the four calendar quarters used for computation of the insurer’s census falls 135 days after the mid point of the four quarter period used to establish the wage standard for approval. In order to be approved, the percentage of females, employees age 50 and over, and employees who earned less than $3,600 in the proposed plan or group of plans, alone or together with previously approved plans, must be no less than the three standard percentages.
The director has devised the alternative risk experience rating test for insurers who do not recanvass their plans or who have one or more plans not included in this tabulation. Since a plan which did not initially reflect adverse selection might, due to changes in plant personnel, do so later, the standards for approval test calls for periodic checking as well as a complete recanvassing after a plan had run for more than two years. To avoid such costs the alternative plan sets up a risk experience rating test. This criterion rests upon a comparison of the benefits paid by an insurer with those paid by the Disability Fund and all voluntary plans. The test allows the same tolerance to insurers as the standards for approval test.
*841If an insurer has not paid as much proportionately in each of the three categories as the “standard risk,” less the specified tolerances, its plans which continue in effect after July 1 and which are not included in the tabulation are subject to withdrawal of approval as of the following January 1. The insurer, however, may canvass those of its plans which are not included in the tabulation and submit a list of persons recently employed by the employer. The director will then retest the plans against the standards for approval. Thus, an insurer who meets the risk experience rating test will be able to retain all plans and not be required to recanvass and be tested against the standards of approval. If it does not meet the risk experience test it must meet the new standards for approval.
(4) The Insurers’ Objections to the Standards for Approval Test.
We turn to the contention of the companies that the standards for approval test is arbitrary and capricious. The companies state that the statistics used to establish the standards for approval cannot properly be compared with the census information which the insurer supplies the director to determine if the plan has qualified for age, sex and wage distribution.
The insurance companies contend that the quarterly figures used to establish the standards produce considerably different percentages in the three categories than the insurer’s census data, which includes only those persons employed during the relevant single payroll period. They further contend that this attempted comparison of noncomp arable figures produces two unsavory results. First, it creates a number of inaccuracies, some of which operate favorably and others which operate unfavorably to the private insurers. Second, the test operates prejudicially to the insurers because it fails to give them credit for female employees and for persons earning less than $3,600 who are included in the quarterly computations used to establish the standards of approval.6
The insurers primarily rely upon mathematical models and statistical data. We shall for present purposes consider the figures contained in the model which the insurers claim most graphically proves their case.
The companies first point out that for every 38 people *842employed during any payroll period, there are 43 persons who are employed at some time during the quarter. In establishing the standards for approval the director takes a representative sample of the 43 people to determine the percentages of the three relevant categories employed. But when the voluntary plans are canvassed, the employees counted will be only 38 who are actually employed during the payroll period. The insurers concede that the two sets of computations are not necessarily incomparable, but contend that they produce a distortion if the five people counted in establishing the standards represent a different composition with regard to age, sex, and wage distribution from the other 38. The insurers reason that these five people must be intermittent employees and maintain that common experience indicates the intermittent group contains higher percentages of women and of employees earning less than $3,600 than the percentages found among regularly employed people.
To the extent of the claimed deviation, the comparison of the percentages based on the two methods of computation would result in an erroneous indication that the voluntary plans covered less than their share of females and the low wage group. To demonstrate the extent of the discrepancy the insurance companies have constructed their model on the assumption that the five employees included in establishing the standards, but not counted in the insurer’s census, include three females and two males, all earning less than $3,600. On the basis of this assumption, the model shows that the discrepancies result in the hypothetical insurer not meeting the standards for approval for sex or wage distribution. The companies conclude that “This survey in miniature suggests that if a single insurance carrier could somehow achieve a monopoly of all UCD business in the State, it is possible that it would not meet the standards for approval, although it would insure during the course of a calendar quarter every worker in the State earning wages in the calendar quarter.”
The director attacks the accuracy of the analysis predicated upon the above statistical model upon the ground that it ignores factors which operate favorably to the insurers and exaggerates the unfavorable effect of the other factors.
First: The discrepancy between the number of persons included in the director’s calculations and those employed during the payroll period is less than the insurers claim. Although the insurers correctly maintain that 43 persons are employed during every quarter for every 38 employed during *843a payroll period, the director in determining the standards for approval does not count all 43. He excludes persons earning less than $300 per year. After such exclusions, the director’s exhibits show that approximately 41, and not 43, persons remain. Thus, in the insurers’ model the discrepancy between those counted under the two computations becomes three, not five.
Second: A discrepancy which results from the two different methods of computation operates to the benefit of the insurers and is not reflected in their model. Between the four calendar quarters of computation period, which the director uses to determine if the employee content of voluntary plans meets the wage distribution standard for approval, and the time the private insurer submits the list of employees presently employed, a time-lag accrues. A department study indicates that 13.3 per cent to 17.9 per cent of those employed when the census is submitted and included in it did not earn any wages during the four calendar quarters of the computation period, even though at the time of the employer’s census they may have been earning more than $3,600. Since the insurer’s census includes this group as earning less than $3,600, this discrepancy obviously works to the advantage of the insurer.
Third: As we have stated, in computing the standard for approval for wages, the director does not count employees earning less than $300. Tet in compiling the census report, the employer does include such employees. This discrepancy also works to the advantage of the insurer.
Fourth: The model incorrectly assumes that the hypothetical five employees who are allegedly included in determining the standard for approval, but not counted in the insurer’s census, earn less than $3,600. This assertion cannot stand for the following reasons: (1) The claim that these employees are in the class earning’ less than $3,600 is no more than an assumption completely lacking in proof. (2) Between the date of the quarters used for computing the standard and the date of the employer’s payroll a percentage of the five employees will in all probability have retired or changed jobs. (3) The right reposed in the insurer to select, within reasonable bounds, the payroll period used in its census may operate to the advantage of the insurer.
Finally, the director concedes that the model works adversely to the insurer in not reflecting the 135-day time-lag between the mid point of the period used to establish the wage standard for approval and the mid point of the period used *844for measuring the content of the plan. During the 135-day period the general rise in wages may shift employees from the classification of those below $3,600 to those above it. But the shift is insignificant. Those who so rise constitute less than one-half of one percentage point.
In summary, both parties agree that more persons are employed during a quarter than during any given payroll period and that, as a result, the two sets of computations cannot be compared with perfect accuracy. The insurers, in assessing the “distortions” which work for them and those which work against them, say “no one can do much more than hazard a guess as to the extent” which the one group is “offset” by the other. The director, however, has presented evidence tending to show that these discrepancies balance out. That the test fails to reflect an exact measurement of the three risk factors it purportedly delineates does not demonstrate its arbitrary or capricious nature. In view of the showing that these intangible inexactitudes are self-liquidating, they do not prove that the director acted arbitrarily.
The burden rests with the insurers to establish that they suffer prejudice by the test but they have not met this obligation. They have only told us the test may hurt them, that they cannot hazard a guess as to the extent of the errors in the test, and that the test is inexact. In the absence of convincing evidence that the purported tolerances are illusory and that the insurers are thereby prejudiced, the test does not assume the characteristic of caprice or arbitrariness.
(5) Insurers’ Objections to the Lack of a Weighting Formula.
Finally, the companies find arbitrary conduct in the director’s requirement that voluntary plans qualify as to each of the three factors independently under both the standards for approval and risk experience rating tests. This contention may be summarized as follows: In reinstating the adverse selection provision the Legislature intended to prevent private insurers from selecting those risks that would be anticipated to result in substantially lower claims costs than would be expected in the total risk represented by the entire covered labor force. The factors of age, sex, and wage distribution were named by the Legislature as the primary factors affecting those costs. The regulation, however, measures coverage of three specific segments of the labor force, irrespective of risk, thereby treating the three categories, -rather than risk, as the ultimate consideration.
*845To illustrate their argument the insurers point to a hypothetical example of a voluntary plan consisting of all women over the age of 50, but 3 per cent below the wage distribution standard for approval. They argue that although this plan would not constitute a selection of risks adverse to the Disability Fund, substantial or otherwise, the plan would be disapproved under the regulation. This example allegedly demonstrates that the independent application of the three criteria does not serve as a measurement of risks. Since the tolerances are small, the attempt of private insurers to insure more than the minimum in each of the three categories will tend to result in the acquisition of worse than average risks. Since the Legislature did not intend a result so disadvantageous to the companies, the director has exceeded his statutory authority. The insurers argue that it would be a simple matter for the director to adopt a reasonable weighting formula that would allow a surplus coverage in one category to compensate for a deficit coverage in another category.
The director testified that he reluctantly decided not to adopt a weighting formula because it would be difficult to administer. He argues further that if some type of weighting were allowed, substantial adverse selection could not be prevented because of the presence of a number of subgroups within each of the three categories which are worse risks than other subgroups. For example, young women represent much better risks than older women, and employees between 50 and 60 years of age are better risks than those over 60. Risks also vary from one type of industry to another. Thus, if the three risk criteria were weighted, the private insurers would enjoy a greater opportunity to exercise their expertise and to insure the subgroups which have proven to be the best risks.
While the present regulation does not provide for weighting, an alternative approach that did so would necessarily impose upon the insurers more onerous regulations. The statute itself indicates the Legislature’s recognition of the existence of subgroups within the three categories; it authorizes the director to base his regulation on sex, age, wage distribution, “and such other factors as may relate to a reasonable test or tests. ...” (Emphasis added.) This broad authorization would have justified the director’s formulation of an extremely complex regulation designed to prevent private insurers from insuring low risk subgroups. The director’s forebearance from that sweeping and intricate regula*846tion does not bar him from the more simple achievement of the same result by the establishment of only three risk-measuring categories, buttressed by a prohibition of weighting.
Against the backdrop of the legislation and the director’s execution of its policies, the regulation was neither arbitrary nor capricious. We are not called upon to decide whether the provision was perfect; neither do we hold that a better one could not have been fashioned. We do not deal in absolutes or in alternatives; we merely hold the regulation before us does not transgress the pertinent legal limitation.
(6) The Insurers’ Objections to the Amotmts of the Tolerances.
The insurers contend that the Legislature’s proscription of “substantial selection of risks adverse to the Disability Fund” shows, not an absolute prohibition of adverse selection or a complete exclusion of the private carriers from the field, but an intention that some carriers participate. Such intention presupposes the participating carriers operate at a reasonable profit; yet the instant regulations foreclose any profit. Moreover, say the carriers, the Legislature has further exhibited its intent that they not suffer the present unfavorable tolerances in that it inferentially provided for tolerances comparable to those in effect from 1946 to 1953. As we shall explain, we cannot find this manifestation of legislative intent in either of the alleged two respects.
We cannot believe that the Legislature, when reinstating the adverse selection provision, “recognized,” as the insurers state, that “it would have to permit a degree of selective underwriting to provide a reasonable margin of profit, if it continued to provide a dual system.” At that time the Legislature knew that the voluntary plans caused the insurers a $2.8 million deficit in 1960 and that they expected a similar deficit in 1961. Moreover, in 1961 the Legislature enacted Unemployment Insurance Code section 3269, which authorizes the director to assess the voluntary plans for the department’s added administrative expenses attributable to such plans. The Legislature's knowledge of the financial difficulties of the private insurers, coupled with the reinstatement of the adverse selection provision and the enactment of section 3269, hardly comports with the insurers’ argument that the Legislature intended the director to interpret “substantial” in such a manner as to provide to private insurers reasonable profits.
*847The insurers argue that when the Legislature terminated the suspension of the adverse selection requirement in 1961 it directed the director to consider two new risk criteria (age and wage distribution) in addition to sex, but did not intend the director to alter drastically the 33 per cent tolerance which he had allowed prior to 1954. This intention, they argue, becomes clear because the Legislature did not alter the wording of section 3254, subdivision (i). The Legislature thereby acquiesced in the earlier 33 per cent tolerance with regard to sex and, further, implicitly expressed its intent that similar tolerances should be allowed for the two new risk criteria.
Although the 1960 report of the Assembly Interim Committee on Finance and Insurance which prompted the instant amendment specifically pointed out that the “lack of meaningful adverse selection criteria and the suspension of the adverse selection provision . . . have substantially affected the Disability Fund to its detriment” (Report, p. 85), the insurers insist that this language did not contemplate that the earlier tolerances be changed. According to the companies, the Legislature’s reinstatement of adverse selection and its provision for the two additional criteria of age and wage distribution sufficed to meet the criticism of the committee. The insurers say the report indicated no other remedy.
We cannot compress the criticism into so narrow a mould. “The lack of meaningful adverse selection criteria” may well mean not only that additional wage and age tests were needed but that existing sex tests were inadequate. Indeed, the report points out that the director’s adoption of the 20 per cent female content requirement in the 1947-1953 period resulted from an erroneous assumption that the percentage of females in employment in California constituted only 25 per cent of the total labor force. This estimate rested on 1940 census data. The report then explains that the actual female content of the work-force from 1947 to 1953 exceeded 30 per cent. (Report, pp. 64-65.) Implicit in the Committee’s report lies in its rejection of the 33 per cent tolerance and its belief that the liberal allowance resulted from the inaccurate statistics. Obviously the committee intended that more accurate figures would yield a more accurate tolerance; nothing suggests that it certified the preservation of the error. The Legislature’s purposes surely accorded with those of the committee which induced the legislation; neither the *848committee nor the Legislature meant to stay the hand of the director in the manner the insurers suggest.
In summary of the case, we cannot say the director’s method of prohibiting adverse selection by the insurers is arbitrary or capricious. As one of insurers’ counsel stated, “[I]t has certainly been true that the private carriers as they dropped from 52 percent of the total business to 29 percent at the end of 1960 skimmed the cream by dropping their worst risks.” The director adopted regulations reasonably designed to prevent the skimming of the cream. We are neither capable of compounding an alternative nor, if we could, are we called upon to substitute our less skilled judgment for that of the director.
Let the peremptory writ of mandate issue as prayed.
Gibson, C. J., Traynor, J., Schauer, J., McComb, J., Peters, J., and White, J.,* concurred.