Opinion

We examine, in this case, aspects of the functions served by administrative agencies in the granting of zoning variances and of courts in reviewing these proceedings by means of administrative mandamus. We[*510] conclude that variance boards like the ones involved in the present case must render findings to support their ultimate rulings. We also conclude that when called upon to scrutinize a grant of a variance, a reviewing court must determine whether substantial evidence supports the findings of the administrative board and whether the findings support the board’s action.^[1] We determine in the present case that the last of these requisites has not been fulfilled.

The parties in this action dispute the Mure of approximately 28 acres in Topanga Canyon located in the Santa Barbara Mountains region of Los Angeles County. A county ordinance zones the property for light agriculture and single family residences;^[2] it also prescribes a one-acre minimum lot size. Upon recommendation of its zoning board and despite the opposition of appellant-petitioner—an incorporated nonprofit organization composed of taxpayers and owners of real property in the canyon—-the Los Angeles County Regional Planning Commission granted to the Topanga Canyon Investment Company a variance to establish a 93-space mobile home park on this acreage.^[3] Petitioner appealed without success to the county board of supervisors, thereby exhausting its administrative remedies. Petitioner then sought relief by means of administrative mandamus, again unsuccessfully, in Los Angeles County Superior Court and the Court of Appeal for the Second District.

In reviewing the denial of mandamus below, we first consider the proper role of agency and reviewing court with respect to the grant of variances. We then apply the proper standard of review to the facts of the case in order to determine whether we should sustain the action of the Los Angeles County Regional Planning Commission.

[*511] 1. An administrative grant of a variance must be accompanied by administrative findings. A court reviewing that grant must determine whether substantial evidence supports the findings and whether the findings support the conclusion that all applicable legislative requirements for a variance have been satisfied.

A comprehensive zoning plan could affect owners of some parcels unfairly if no means were provided to permit flexibility. Accordingly, in an effort to achieve substantial parity and perhaps also in order to insulate zoning schemes from constitutional attack,* **^[4] our Legislature laid a foundation for the granting of variances. Enacted in 1965, section 65906 of the Government Code establishes criteria for these grants; it provides: “Variances from the terms of the zoning ordinance shall be granted only when, because of special circumstances applicable to the property, including size, shape, topography, location or surroundings, the strict application of the zoning ordinance deprives such property of privileges enjoyed by other property in the vicinity and under identical zoning classification [1J] Any variance granted shall be subject to such conditions as will assure that the adjustment thereby authorized shall not constitute a grant of special privileges inconsistent with the limitations upon other properties in the vicinity and zone in which such property is situated.”^[5] ,,

Applicable to all zoning jurisdictions except chartered cities (Gov. Code, § 65803), section 65906 may be supplemented by harmonious local legislation.^[6] We note that Los Angeles County has enacted an ordinance which,[*512] if harmonious with section 65906, would govern the Topanga Canyon property here under consideration. Los Angeles County’s Zoning Ordinance No. 1494, section 522, provides:* **^[7] “An exception [variance] may . . . be granted where there are- practical difficulties or unnecessary hardships in the way of carrying out the strict letter of the ordinance, and in the granting of such exception the spirit of the ordinance will be observed, public safety secured, and substantial justice done.”

Both state and local laws thus were designed to establish requirements which had to be satisfied before the Topanga Canyon Investment Company should have been granted its variance. Although the cases have held that substantial evidence must support the award of a variance in order to insure that such legislative requirements have been satisfied^[8] (see, e.g., Siller v. Board of Supervisors (1962) 58 Cal.2d 479, 482 [25 Cal.Rptr. 73, 375 P.2d 41]; Bradbeer v. England (1951) 104 Cal.App.2d 704, 707 [232 P.2d 308]), they have failed to clarify whether the administrative agency must always set forth findings and have not illuminated the proper relationship between the evidence, findings, and ultimate agency action.^[9]

One of the first decisions to emphasize the importance of judicial scrutiny of the record in order to determine whether substantial evidence supported administrative findings that the property in question met the legislative variance requirements was that penned by Justice Molinari in[*513] Cow Hollow Improvement Club v. Board of Permit Appeals (1966) 245 Cal.App.2d 160 [53 Cal.Rptr. 610]. Less than one year later, we followed the approach of that case in Broadway, Laguna etc. Assn. v. Board of Permit Appeals (1967) 66 Cal.2d 767 [59 Cal.Rptr. 146, 427 P.2d 810], and ordered that a zoning board’s grant of a variance be set aside because the party seeking the variance had failed to adduce sufficient evidence to support administrative findings that the evidence satisfied the requisites for a variance set forth in the same San Francisco ordinance.

Understandably, however, the impact of these opinions remained uncertain. The San Francisco ordinance applicable in Cow Hollow and Broadway explicitly required the zoning board to specify its subsidiary findings and ultimate conclusions; this circumstance raised the question whether a court should require findings and examine their sufficiency in a case in which the applicable local legislation did not explicitly command the administrative body to set forth findings. Indeed language in Broadway intimated that such a case was distinguishable. (Broadway, Laguna etc. Assn. v. Board of Permit Appeals, supra, at pp. 772-773. See also Stoddard v. Edelman (1970) 4 Cal.App.3d 544, 549 [84 Cal.Rptr. 443]. Cf. Friends of Mammoth v. Board of Supervisors (1972) 8 Cal.3d 247, 270 [104 Cal.Rptr. 761, 502 P.2d 1049].) Further, neither Cow Hollow nor Broadway confronted Government Code section 65906, since both cases concerned a chartered city.^[10] There thus also remained uncertainty with respect to cases involving zoning jurisdictions other than chartered cities.

Nevertheless, in an opinion subsequent to Broadway; Hamilton v. Board of Supervisors (1969) 269 Cal.App.2d 64 [75 Cal.Rptr. 106], a Court of Appeal set aside the grant of a variance by a planning commission under circumstances different from those in Broadway and Cow Hollow. The zoning jurisdiction involved in that controversy was a county, not a chartered city, and the court’s opinion did not suggest that any applicable ordinance required administrative findings. Deeming Government Code section 65906 “concededly controlling,” (Hamilton v. Board of Supervisors, supra, at p. 67), the court undertook the task of squaring the findings announced by the commission with the commission’s grant of the variance and concluded that the findings were insufficient to sustain the variance.

Consistent with the reasoning underlying these cases, we hold that[*514] regardless of whether the local ordinance commands -that the variance board set forth findings,^[11] that body must render findings sufficient both to enable the parties to determine whether and on what basis they should seek review and, in the event of review, to apprise a reviewing court of the basis for the board’s action. We hold further that a reviewing court, before sustaining the grant of a variance, must scrutinize the record and determine whether substantial evidence supports the administrative agency’s findings and whether these findings support the agency’s decision. In making these determinations, the reviewing court must resolve reasonable doubts in favor of the administrative findings and decision.

Our analysis begins with consideration of Code of Civil Procedure section 1094.5, the state’s administrative mandamus provision which structures the procedure for judicial review of adjudicatory decisions rendered by administrative agencies. Without doubt, this provision applies to the review of variances awarded by bodies such as the Los Angeles County zoning agencies that participated in the present case.^[12] Section 1094.5 clearly contemplates that at minimum, the reviewing court must determine both whether substantial evidence supports the administra[*515]tive agency’s findings and whether the findings support the agency’s decision. Subdivision (b) of section 1094.5 prescribes that when petitioned for a writ of mandamus, a court’s inquiry should extend, among other issues, to whether “there was any prejudicial abuse of discretion.” Subdivision (b) then defines “abuse of discretion” to include instances in which the administrative order or decision “is not supported by the findings, or the findings are not supported by the evidence.” (Italics added.) Subdivision (c) declares that “in all .. . cases’’ (italics added) other than those in which the reviewing court is authorized by law to judge the evidence independently,^[13] “abuse of discretion is established if the court determines that the findings are not supported by substantial evidence in the light of the whole record.” (See Zakessian v. City of Sausalito (1972) 28 Cal.App.3d 794, 798 [105 Cal.Rptr. 105].)

We further conclude that implicit in section 1094.5 is a requirement that the agency which renders the challenged decision must set forth findings to bridge the analytic gap between the raw evidence and ultimate decision or order. If the Legislature had desired otherwise, it could have declared as a possible basis for issuing mandamus the absence of substantial evidence to support the administrative agency’s action. By focusing, instead, upon the relationships between evidence and findings and between findings and ultimate action, the Legislature sought to direct the reviewing court’s attention to the analytic route the administrative agency traveled from evidence to action. In so doing, we believe that the Legislature must have contemplated that the agency would reveal this route. Reference, in section 1094.5, to the reviewing court’s duty to compare the evidence and ultimate decision to “the findings” (italics added) we believe leaves no room for the conclusion that the Legislature would have been content to have a reviewing court speculate as to the administrative agency’s basis for decision.

Our ruling in this regard finds support in persuasive policy considerations. (See generally 2 Davis, Administrative Law Treatise (1958) § 16.05, pp. 444-449; Forkosch, A Treatise on Administrative Law (1956) § 253, pp. 458-464.) According to Professor Kenneth Culp Davis, the requirement that administrative agencies set forth findings to support their adjudicatory decisions stems primarily from judge-made law (see, e.g., Zieky v. Town Plan and Zon. Com’n of Town of Bloomfield (1963) 151 Conn. 265 [196 A.2d 758]; Stoll v. Gulf Oil Corp. (1958) 79 Ohio L.Abs. 145 [155 N.E.2d 83]), and is “remarkably uniform in both federal and state[*516] courts.” As stated by the United States Supreme Court, the “accepted ideal ... is that ‘the orderly functioning of the process of review requires that the grounds upon which the administrative agency acted be clearly disclosed and adequately sustained.’ (S.E.C. v. Chenery Corp. (1943) 318 U.S. 80, 94.)” (2 Davis, supra, § 16.01, pp. 435-436. See also Saginaw Broadcasting Co. v. Federal C. Com’n (1938) 96 F.2d 554, 559 [68 App.D.C. 282].)

Among other functions, a findings requirement serves to conduce the administrative body to draw legally relevant sub-conclusions supportive of its ultimate decision; the intended effect is to facilitate orderly analysis and minimize the likelihood that the agency will randomly leap from evidence to conclusions. (See 2 Cooper, State Administrative Law (1965) pp. 467-468; Feller, Prospectus for the Further Study of Federal Administrative Law (1938) 47 Yale L.J. 647, 666. Cf. Comment, Judicial Control Over Zoning Boards of Appeal: Suggestions for Reform (1965) 12 U.C.L.A. L.Rev. 937, 952.)^[14] In addition, findings enable the reviewing court to trace and examine the agency’s mode of analysis. (See California Motor Transport Co. v. Public Utilities Com. (1963) 59 Cal.2d 270, 274 [28 Cal.Rptr. 868, 379 P.2d 324]; Swars v. Council of City of Vallejo (1949) 33 Cal.2d 867, 871 [206 P.2d 355].)

Absent such roadsigns, a reviewing court would be forced into unguided and resource-consuming explorations; it would have to grope through the record to determine whether some combination, of credible evidentiary items which supported some line of factual and legal conclusions supported the ultimate order or decision of the agency.^[15] Moreover,[*517] properly constituted findings^[16] enable the parties to the agency proceeding to determine whether and on what basis they should seek review. (See In re Sturm (1974) ante, pp. 258, 267 [113 Cal.Rptr. 361, 521 P.2d 97]; Swars v. Council of City of Vallejo, supra, at p. 871.) They also serve a public relations function by helping to persuade the parties that administrative decision-making is careful, reasoned, and equitable.

By setting forth a reasonable requirement for findings and clarifying the standard of judicial review, we believe we promote the achievement of the intended scheme of land use control. Vigorous and meaningful judicial review facilitates, among other factors, the intended division of decision-making labor. Whereas the adoption of zoning regulations is a legislative function (Gov. Code, § 65850), the granting of variances is a quasi-judicial, administrative one. (See Johnston v. Board of Supervisors (1947) 31 Cal.2d 66, 74 [187 P.2d 686]; Kappadahl v. Alcan Pacific Co. (1963) 222 Cal.App.2d 626, 634 [35 Cal.Rptr. 354].) If the judiciary were to review grants of variances superficially, administrative boards could subvert this intended decision-making structure. (See 1 Appendix to Sen. J. (1970 Reg. Sess.) Final Rep. of the Joint Committee on Open Space Land (1970) pp. 102-103.) They could “[amend] ... the zoning code in the guise of a variance” (Cow Hollow Improvement Club v. Board of Permit Appeals, supra, at p. 181), and render meaningless, applicable state and local legislation prescribing variance requirements.

Moreover, courts must meaningfully review grants of variances in order to protect the interests of those who hold rights in property nearby the parcel for which a variance is sought. A zoning scheme, after all, is similar in some respects to a contract; each party foregoes rights to use its land as it wishes in return for the assurance that the use of neighboring property will be similarly restricted, the rationale being that such mutual restriction can enhance total community welfare. (See, e.g., 1 Appendix to Sen. J. (1970 Reg. Sess.) Final Rep. of the Joint Committee on Open Space Land (1970) p. 91; Bowden, Article XXVIII—Opening the Door to Open Space Control (1970) 1 Pacific L.J. 461, 501.) If the interest of[*518] these parties in preventing unjustified variance awards for neighboring land is not sufficiently protected, the consequence will be subversion of the critical reciprocity upon which zoning regulation rests.

Abdication by the judiciary of its responsibility to examine variance board decision-making when called upon to do so could very well lead to such subversion.^[17] Significantly, many zoning boards employ adjudicatory procedures that may be characterized as casual. (See Comment, Judicial Control over Zoning Boards of Appeal: Suggestions for Reform (1965) 12 U.C.L.A. L.Rev. 937, 950. Cf. Bradbeer v. England (1951) 104 Cal.App.2d 704, 710 [232 P.2d 308].) The availability of careful judicial review may help conduce these boards to insure that all parties have an opportunity fully to present their evidence and arguments. Further, although we emphasize that we have no reason to believe that such a circumstance exists in the case at bar, the membership of some zoning boards may be inadequately insulated from the interests whose advocates most frequently seek variances. (See e.g., 1 Appendix to Sen. J. (1970 Reg. Sess.) Final Rep. of the Joint Committee on Open Space Land (1970) p. 100.) Vigorous judicial review thus can serve to mitigate the effects of insufficiently independent decision-making.

2. The planning commission’s summary of “factual data”—its apparent “findings”—does not include facts sufficient to satisfy the variance requirements of Government Code section 65906.

As we have mentioned, at least two sets of legislative criteria appear applicable to the variance awarded: Government Code section 65906 and Los Angeles County Zoning Ordinance No. 1494, section 522. The variance can be sustained only if all applicable legislative requirements have been satisfied. Since we conclude that the requirements of section 65906 have not been met, the question whether the variance conforms with the criteria set forth in Los Angeles County Zoning Ordinance No. 1494, section 522 becomes immaterial.^[18]

[*519] We summarize the principal factual data contained in the Los Angeles County Regional Planning Commission’s report, which data the commission apparently relied on to award the variance.^[19] The acreage upon which the original real party in interest^[20] sought to establish a mobile home park consists of 28 acres; it is a hilly and in places steep parcel of land. At the time the variance was granted, the property contained one single-family residence. Except for a contiguous area immediately to the southeast which included an old and flood-damaged subdivision and a few commercial structures, the surrounding properties were devoted exclusively to scattered single-family residences.

The proposed mobile home park would leave 30 percent of the acreage in its natural state. An additional 25 percent would be landscaped and terraced to blend in with the natural surroundings. Save in places where a wall would be incompatible with the terrain, the plan contemplated enclosure of the park with a wall; it further called for rechanneling a portion of Topanga Canyon Creek and anticipated that the developers would be required to dedicate an 80-foot-wide strip of the property for a proposed realignment of Topanga Creek Boulevard.

[*520] The development apparently would partially satisfy a growing demand for new, low cost housing in the area. Additionally, the project might serve to attract further investment to the region and could provide a much needed fire break. Several data indicate that construction on the property of single-family residences in conformance with the zoning classification would generate significantly smaller profits than would development of the mobile home park. Single-family structures apparently would necessitate costly grading, and the proposed highway realignment would require a fill 78 feet high, thereby rendering the property unattractive for conventional residential development. Moreover, the acreage is said not to be considered attractive to parties interested in single-family residences due, in the words of the report’s summary of the testimony, to “the nature of the inhabitants” in the vicinity and also because of local flood problems.

These data, we conclude, do not constitute a sufficient showing to satisfy the section 65906 variance requirements. That section permits variances “only when, because of special circumstances applicable to the property, . . . the strict application of the zoning ordinance deprives such property of privileges enjoyed by other property in the vicinity and under identical zoning classification.” (Italics added.) This language emphasizes disparities between properties, not treatment of the subject property’s characteristics in the abstract. (See Minney v. City of Azusa (1958) 164 Cal.App.2d 12, 31 [330 P.2d 255]; cf. In re Michener’s Appeal (1955) 382 Pa. 401 [115 A.2d 367, 371]; Beirn v. Morris (1954) 14 N.J. 529 [103 A.2d 361, 364]; Note, Administrative Discretion in Zoning (1969) 82 Harv.L.Rev. 668, 671-672.) It also contemplates that at best, only a small fraction of any one zone can qualify for a variance. (See generally 3 Anderson, American Law of Zoning (1968) § 14.69, pp. 62-65.)

The data contained in the planning commission’s report focus almost exclusively on the qualities of the property for which the variance was sought. In the absence of comparative information about surrounding properties, these data lack legal significance. Thus knowledge that the property has rugged features tells us nothing about whether the original real party in interest faced difficulties different from those confronted on neighboring land.^[21] Its assurances that it would landscape and terrace parts of the property and leave others in their natural state are all well and good, but they bear not at all on the critical issue whether a variance[*521] was necessary to bring the original real party in interest into substantial parity with other parties holding property interests in the zone. (See Hamilton v. Board of Supervisors, supra, at p. 66.)

The claim that the development would probably serve various community needs may be highly desirable, but it too does not bear on the issue at hand. Likewise, without more, the data suggesting that development of the property in conformance with the general zoning classification could require substantial expenditures are not relevant to the issue whether the variance was properly granted. Even assuming for the sake of argument that if confined to the subject parcel and no more than a few others in the zone, such a burden could support a variance under section 65906, for all we know from the record, conforming development of other property in the area would entail a similar burden. Were that the case, a frontal attack on the present ordinance or a legislative proceeding to determine whether the area should be rezoned might be proper, but a variance would not. (1 Appendix to Sen. J. (1970 Reg. Sess.) Final Rep. of the Joint Committee on Open Space Land (1970) p. 95; Bowden, Article XV111 —Opening the Door to Open Space Control (1970) 1 Pacific L.J. 461, 506.)

Although they dispute that section 65906 requires a showing that the characteristics of the subject property are exceptional, the current real parties in interest would nevertheless have us speculate that the property is unlike neighboring parcels. They point out that the plot has rugged terrain and three stream beds^[22] and that the Topanga Creek Boulevard realignment would bisect the property. Speculation about neighboring land, however, will not support the award of a variance. The party seeking the variance must shoulder the burden of demonstrating before the zoning agency that the subject property satisfies the requirements therefor. (Tustin Heights Association v. Board of Supervisors (1959) 170 Cal.App.2d 619, 627 [339 P.2d 914].) Thus neither an administrative agency nor a reviewing court may assume without evidentiary basis that the character of neighboring property is different from that of the land for which the variance is sought.^[23]

[*522] Moreover, the grant of a variance for nonconforming development of a 28-acre parcel in the instant case is suspect. Although we do not categorically preclude a tract of that size from eligibility for a variance, we note that in the absence of unusual circumstances, so large a parcel may not be sufficiently unrepresentative of the realty in a zone to merit special treatment. By granting variances for tracts of this size, a variance board begins radically to alter the nature of the entire zone. Such change is a proper subject for legislation, not piecemeal administrative adjudication. (See Sinclair Pipe Line Co. v. Village of Richton Park (1960) 19 Ill.2d 370 [167 N.E.2d 406]; Appeal of the Catholic Cemeteries Association (1954) 379 Pa. 516 [109 A.2d 537]; Civil City of Indianapolis v. Ostrom R. & Construction Co. (1931) 95 Ind.App. 376 [176 N.E. 246].) Since there has been no affirmative showing that the subject property differs substantially and in relevant aspects from other parcels in the zone, we conclude that the variance granted amounts to the kind of “special privilege” explicitly prohibited by Government Code section 65906.

We submit, in summary, that this case illumines two important legal principles. First, by requiring that administrative findings must support a variance, we emphasize the need for orderly legal process and the desirability of forcing administrative agencies to express their grounds for decision so that reviewing courts can intelligently examine the validity of administrative action. Second, by abrogating an unsupported exception to a zoning plan, we conduce orderly and planned utilization of the environment.

We reverse the judgment and remand the cause to the superior court with directions to issue a writ of mandamus requiring the Los Angeles Board of Supervisors to vacate its order awarding a variance. We also direct the superior court to grant any further relief that should prove appropriate.

Wright, C. J., McComb, J., Mosk, J., Burke, J., Sullivan, J., and Clark, J., concurred.