LYNCH, J.

This is a reservation and report of three cases filed and consolidated in the Supreme Judicial Court for the county of Suffolk, each challenging the decision of the Commissioner of Insurance fixing automobile insurance rates for the calendar year 1987. One of the cases has since been dismissed upon stipulation of all parties. The second, brought by a State legislator, alleges that the rates as set are excessive. No brief having been filed in that action, it is hereby dismissed. Mass. R.A.P. 19 (c), 365 Mass. 867 (1974). Thus, we are left with only the third case, brought by the Massachusetts Automobile Rating and Accident Prevention Bureau and ten of its member insurance companies. For simplicity's sake, we shall refer to the plaintiffs as either the bureau or the industry.

We summarize the facts. The Commissioner of Insurance (commissioner) may fix and establish annual insurance rates if he finds, upon investigation and hearing, that (1) competition in the industry will not ensure that rates will not be excessive or, (2) competition will be "so conducted as to be destructive of competition or detrimental to the solvency of insurers. . . ." G.L.c. 175E, § 5 (1986 ed.). The rates set by the commissioner are comprised of three basic components: (1) the loss allowance (expenses from payment of claims); (2) the expense allowance (nonclaim expenses); and (3) the underwriting profit allowance (an estimate of profit solely from the business of selling insurance). Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 389 Mass. 824, 826 (1983).

This year's hearing lasted eighty-six days. The parties included the bureau, the State Rating Bureau of the Division of Insurance (SRB), and the Attorney General. The commissioner attributed the extraordinary length of this year's hearing to the gross disparity between the parties' rate recommendations. According to the commissioner, "[n]early 40 points, amounting to $550 million, separate[d] the industry on one side and the Attorney General and the State Rating Bureau on the other." The industry says it asked for a rate increase of 37.2%, but only got an increase of about 9%. It claims that the rates set are inadequate and argues in support of this claim as set out below.

1. Underwriting Experience.

a. Exclusion of evidence. The bureau sought to introduce evidence of past underwriting results to show that the commissioner's current provision for underwriting loss was likely to be inadequate. "Underwriting results" are the measure of the industry's profit or loss solely from underwriting insurance and from no other activity. The bureau's evidence included three documents: (1) A chart comparing the commissioner's target underwriting profit provision with the actual underwriting profit earned by the State's insurance industry for the years 1975-1986, inclusive. (2) A document relied on by the SRB's witness, Paul Chernick. This document was an excerpt from an industry source book identified by the parties as Best's Aggregates and Averages, showing country-wide underwriting results. (3) A chart compiling the most recent country-wide underwriting results, abstracted from Best's Aggregates and Averages.

The bureau claims that experts from both sides and the commissioner all have relied on similar data in the past. Nonetheless, the items above were all excluded from evidence without explanation by the commissioner. On this appeal, the bureau contends that the exclusion of the evidence this year — at least without an explanation — was error requiring remand. We agree.

With respect to administrative agencies' evidentiary rulings, "'unless the admission [or exclusion] of the evidence resulted in a denial . . . of substantial justice,' the appellants have no valid complaint." Sudbury v. Department of Pub. Utils., 351 Mass. 214, 220 (1966), quoting Mayor of Everett v. Superior Court, 324 Mass. 144, 148 (1949). See Framingham v. Department of Pub. Utils., 355 Mass. 138, 144 (1969); A.J. Cella, Administrative Law § 280 (1986), and cases cited. Cf. K.C. Davis, Administrative Law § 16.11 (2d ed. 1980 1982 Supp.). In another context, we have said that an agency "has a wide discretion in ruling on evidence." Sudbury v. Department of Pub. Utils., supra at 219. Cf. Maddocks v. Contributory Retirement Appeal Bd., 369 Mass. 488, 498 (1976), and cases cited. Therefore, we review the commissioner's evidentiary rulings for errors of law resulting in a denial of substantial justice.

The goal in setting rates is to reproduce the effects of competitive markets and the rates as ultimately set must leave the industry with at least the opportunity to achieve the average returns earned in competitive markets. See Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 381 Mass. 592, 605 (1980); Fitchburg Gas Elec. Light Co. v. Department of Pub. Utils., 371 Mass. 881, 884 (1977); Attorney Gen. v. Commissioner of Ins., 370 Mass. 791, 813 (1976); Aetna Casualty Sur. Co. v. Commissioner of Ins., 358 Mass. 272, 281 (1970). The commissioner contends here that exclusion of the evidence was nonetheless proper because past underwriting results are irrelevant to a determination whether the present rates are reasonable. Further, the commissioner argues that even if the bureau's evidence did have some probative value, the evidence was properly excluded because any probative value was outweighed by concerns of undue delay and expense.

There is no question that, taken by themselves, underwriting results do not reflect over-all industry profitability. As we have recognized on prior occasions, in the insurance business "money is made on investments, not on underwriting." Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 389 Mass. at 829, quoting Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 381 Mass. at 604-605. But this does not necessarily mean that underwriting results are irrelevant to profitability or to the reasonableness of the rates.

Evidence is relevant if it has any tendency to make the existence of a material fact more probable or less probable than it would be without the evidence. McCormick, Evidence § 185 (3d ed. 1984). Under this definition, the bureau's evidence is relevant. Dr. William Fairley, recognized as an expert in the field by this court, Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 389 Mass. at 831, has written that a direct way to test the reasonableness of "target" underwriting provisions is to compare them with long-term historical data on underwriting profit margins in competitive markets nationwide. W. Fairley, Investment Income and Profit Margins in Property-Liability Insurance, 10 Bell J. of Econ. 192, 205-206 (1979). The SRB's witness, Paul Chernick, testified on cross-examination that, in order to determine whether this year's proposed profit provisions comported with results in competitive markets, "the best way of doing it would be to rerun a model under the conditions that prevailed in each of the years previously through the underwriting cycle and look at the average provided by the model and the average produced by non or less regulated markets and determine how well those fit." Finally, in an offer of proof, the bureau pointed out that, in prior years, the commissioner had tested the reasonableness of the profit provisions against evidence of past underwriting results. See Decision on 1985 Rates at 108; Decision on 1984 Rates at 93.

Once the bureau had shown that this type of evidence had been admitted — and even relied on — in the past, it was up to the commissioner to articulate some reason for ignoring the evidence on the present occasion. A party is entitled to "reasoned consistency" in agency decisionmaking. Boston Gas Co. v. Department of Pub. Utils., 367 Mass. 92, 104 (1975). Furthermore, the commissioner is required to make findings that indicate the over-all basis for his decision. Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 389 Mass. at 828. Without adequate findings, this court is unable to test the decision for reasoned consistency; we are precluded both from determining if the exclusion of the evidence deprives the plaintiffs of substantial justice and from evaluating the industry's argument that the commissioner improperly excluded the evidence as a way of "transmuting a preconception into judgment." Medical Malpractice Joint Underwriting Ass'n v. Commissioner of Ins., 395 Mass. 43, 57 (1985), quoting New Boston Garden Corp. v. Assessors of Boston, 383 Mass. 456, 474 n. 11 (1981).

The commissioner argues that he is not required to make findings on excluded evidence, so long as his decision rests on independent, adequately explained grounds, citing Fitchburg Gas Elec. Light Co. v. Department of Pub. Utils., 395 Mass. 836, 848 (1985). In Fitchburg, the Department of Public Utilities said it would deny the utilities' financing applications until the utilities provided "adequate enforceable assurances" that they would meet certain conditions not here relevant. The utilities declined to give the assurances and appealed from the effective denial of their applications. On appeal, the utilities claimed that the department should have considered evidence from several contemporaneous financing cases related to the main action. The department did not consider this evidence, nor did it make any findings relative to it. We found no error because we saw no need for the department to have reached the issues presented by the excluded evidence. We did "not have the dilemma of not knowing whether the department disbelieved evidence . . . or believed it but considered it not determinative." Id. at 848.

In this case, however, we do have a dilemma. From the record before us, it is impossible for us to determine why the commissioner excluded evidence he formerly considered relevant. The commissioner "should not leave counsel and the courts without the guidance of proper findings . . . to determine from a voluminous record . . . whether his conclusions can be sustained on the evidence." Insurance Rating Bd. v. Commissioner of Ins., 359 Mass. 111, 118 (1971). We will not blindly assume that the commissioner "must have" thought that evidence of a type relied on in the past had for some reason become irrelevant. Adequate explanation from the commissioner will ensure that our review is performed on the basis of what we know rather than what we might opine from a perusal of the record.

This conclusion is in accord with our opinion in Boston Gas Co. v. Department of Pub. Utils., 367 Mass. 92 (1975). In that case, the Department of Public Utilities had permitted Boston Gas to include the undepreciated value of a retired plant in its rate base in three prior rate decisions. But in a subsequent proceeding, the department changed its position and excluded the property from consideration in the company's rate base. The department gave no reasons for its exclusion of the plant except to say that it "questioned" whether the plant should be included. Id. at 100. On appeal, the Attorney General offered several post hoc rationalizations, but could point to no explicit explanation for the department's action. We held that the department could not exclude the retired plant from the rate base in the current year without finding or reporting some facts which would warrant a departure from prior practice. Id. at 104. We said, "A party to a proceeding before a regulatory agency . . . has a right to expect and obtain reasoned consistency in the agency's decisions. This does not mean that every decision of the Department in a particular proceeding becomes irreversible in the manner of judicial decisions constituting res judicata, but neither does it mean that the same issue arising as to the same party is subject to decision according to the whim or caprice of the Department every time it is presented." Id., citing K.C. Davis, Administrative Law §§ 17.01, 17.07, 18.01 and 18.02 (1958 1970 Supp.).

The parallels between Boston Gas and the instant case are obvious. As in Boston Gas, this case presents the problem of an agency ignoring an item that it had deemed relevant in the past. It might be argued that Boston Gas did not involve an evidentiary ruling, so the review there may have been stricter. However, in each case the agency ignored evidence it had relied on in the past. It makes little difference that in Boston Gas, the evidence was admitted and then ignored while in this case the evidence was simply not admitted.

b. Commissioner's imputation of Federal tax liability on instalment income. In computing the underwriting profit allowance, the commissioner utilizes a mathematical model. This year's model was stipulated by the parties to be the same as that used in 1986, but the parties were free to contest factual data entered into the model. That is, the "equation" was set, but some of the "variables" had yet to be quantified. The industry claims that by using the stipulated model and the same data as last year (except for updated interest rates and portfolio composition, as stipulated), it came up with an underwriting profit provision of -4.1%. The SRB recommended an underwriting profit provision of -11.1%. The SRB's recommendation was different because it used updated premium cash flow data which imputed to the industry far more income from finance charges (instalment income) than had previous data.

The SRB's recommendation was also different because the underlying data did not include provision for the industry's Federal income tax liability on the instalment income. As a result, the SRB data overstated cash flow, which led to an increase in estimated investment income, which in turn would support a greater underwriting loss. The commissioner resolved this problem by imputing a 46% tax on the instalment income included in the new data. The industry argues that this resolution was improper because the commissioner made no findings nor was there any evidence to show how the tax was to be imputed. We conclude that there is error requiring remand.

The industry concedes that the tax liability on instalment income should be imputed in some way; evidence indicated that failure to recognize the tax would harm the industry by some forty to fifty million dollars. However, the industry argues that the stipulated profit model made no provision for recognizing the tax and that reversal is required because the commissioner failed to indicate how he had recognized the tax liability within the confines of the model.

The commissioner's ruling on the issue states that "recognition of federal tax liability, which has been traditionally associated with premium income, was contemplated by the terms of the stipulation." According to the bureau, recognition of tax liability on finance charge income involves a complex interaction among various interdependent algebraic components of the profit model. The commissioner does not dispute this contention on appeal.

This court does not require the commissioner to make findings on every controverted issue of law or fact. Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 384 Mass. 333, 337 (1981). However, we do require findings from the commissioner that indicate the over-all basis of his decision and that permit effective appellate review. Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 389 Mass. at 828. Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 384 Mass. at 337. See Medical Malpractice Joint Underwriting Ass'n v. Commissioner of Ins., 395 Mass. 43, 57 (1985); Insurance Rating Bd. v. Commissioner of Ins., 359 Mass. 111, 118 (1971). On the basis of the commissioner's ruling, we have no way of knowing how the tax on instalment income was imputed and, hence, effective review is frustrated.

The commissioner argues that there was no prejudice to the industry because imputation of the tax saved the industry forty to fifty million dollars. But the amount of the savings is only conjecture when the method of computation is not disclosed. At this point, we can only speculate on the effects of the tax because the commissioner has not revealed his methodology.

Furthermore, the commissioner's ruling finds meager, if any, support in the evidence. Experts from both sides testified at the hearing below that a 46% tax should be imputed on finance charge income. However, no expert ever testified as to how the tax should be imputed within the context of all the components of the profit model. The closest any expert came to such testimony was when the SRB's witness, Paul Chernick, testified that the tax on finance charge income was "algebraically equivalent" to the underwriting tax, a tax which apparently is contemplated in the stipulated model. The commissioner may or may not have used this testimony as the basis for imputing the tax. Indeed, that testimony may have been all the evidence necessary to determine how to impute the tax. But without adequate findings, we are unable to determine what the commissioner did. Therefore, the case is remanded for further findings and, if necessary, further hearings, on the issue of how the Federal tax liability was imputed on instalment income.

2. Loss Allowance.

a. Exclusion of evidence. In order to understand the bureau's claims of error with regard to the loss allowance, some background is necessary. In determining the loss allowance the commissioner settles upon a "base year," the most recent year for which reasonably complete claims data are available. The base year for setting 1987 rates was 1985. The commissioner then determines the total dollar amount of claims reported, paid, or incurred by the industry for policies issued during the base year and adds to this amount certain expenses allocable to specific claims in order to come up with "losses." This number is then divided by the total "exposures" — the number of car years for which policies were written — to obtain the "loss pure premium." The loss pure premium is then adjusted in order to determine the loss allowance for the current year. The adjustments may take three forms: (1) An adjustment for lack of complete data for the base year. (2) Adjustments accommodating ways in which the current year is expected to differ from the base year, including an adjustment for differing severity and frequency of claims as between the two years. (3) An adjustment for normative considerations where the commissioner properly determines a need for cost control. See Attorney Gen. v. Commissioner of Ins., 370 Mass. 791, 796-797, 804 (1976).

The bureau claims that the commissioner improperly excluded evidence purportedly showing that loss provisions in the 1986 and 1985 rates were inadequate. The bureau argues that this evidence is relevant to show that the 1987 rates are likely to be inadequate. It also argues that "[r]ates are not adequate, fair and reasonable if a large aggregate loss for the preceding year and a probable greater loss for the year in issue are ignored. . . ." Insurance Rating Bd. v. Commissioner of Ins., 359 Mass. at 115. According to the bureau, this court is precluded from effectively reviewing the adequacy of the rates under the rule above because of the commissioner's erroneous exclusion of evidence.

The bureau does not claim that the commissioner relied on this kind of evidence in the past in order to test the reasonableness of the loss allowance, so the "reasoned consistency" rule of Boston Gas Co. v. Department of Pub. Utils., supra, is not implicated. Therefore, we review only to determine whether the exclusion of the evidence resulted in a denial of substantial justice. Sudbury v. Department of Pub. Utils., supra. Framingham v. Department of Pub. Utils., supra. We discern no error.

Insurance rate making is essentially prospective in nature. Massachusetts law no longer makes provision for a "second look" at the rates and does not provide for a rebate to the insurers or to the policyholders if past rate predictions ultimately turn out to be wrong. See St. 1971, c. 977, § 1A; St. 1975, c. 707, § 1A. Many different factors and adjustments go into determining each year's loss allowance. A past year's loss allowance could have been inaccurate because of an adjustment not used this year. Or the 1987 loss allowance might be more accurate than past predictions because it includes data and adjustments not used in the past. Therefore, the relevance of any claim that past rates have turned out to be inadequate is problematic. We have rejected previous attempts by the industry to use past losses as evidence of present inadequacy in the rates. See Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 384 Mass. at 347; Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 381 Mass. at 601. The evidence would have relevance only if it could be shown that the same factors which led to inadequacy in the previous years' predictions were present in the current calculation. The industry did not make such a showing and, hence, the commissioner could properly have excluded the evidence as irrelevant.

Moreover, this case does not fall within the rationale of the rule that "[r]ates are not adequate, fair and reasonable if a large aggregate loss for the preceding year and a probable greater loss for the year in issue are ignored and the preceding year's rates are simply renewed." Insurance Rating Bd. v. Commissioner of Ins., supra at 115. Medical Malpractice Joint Underwriting Ass'n v. Commissioner of Ins., supra at 51-52. In making this argument, the bureau neglects the import of the crucial last phrase from the above rule, that rates are not adequate if "the preceding year's rates are simply renewed" (emphasis added). Id. This phrase indicates that the rationale behind the rule is to prohibit the commissioner from simply renewing a prior year's rates or from arbitrarily determining the rates.

For example, in Medical Malpractice Joint Underwriting Ass'n v. Commissioner of Ins., supra, the commissioner failed to set actuarially sound rates; in fact, he set rates below those indicated by the evidence of both sides. Id. at 51, 54. That was not the case here. In the instant case, the commissioner determined the rates according to a model agreed on by the parties. That model encompassed a year — 1985 — in which the insurance industry suffered some of its worst losses ever. By no means did the model simply renew the prior year's rates; indeed, the commissioner ultimately granted the industry a 9% rate increase. Therefore, there was no error in the commissioner's exclusion of the bureau's evidence of past losses.

b. Commissioner's use of external indicators. The bureau claims that the parties' recommended loss allowances differed by some 320 million dollars. According to the bureau, the disparity resulted because, in formulating recommendations, the bureau relied on "internal" indicators while the SRB relied on "external" indicators. (Internal indicators relate to the actual experience of insurers; external indicators rely on hypothetical models and data from outside the insurance industry.) In coming to his ultimate decision on a loss allowance the commissioner assigned a weighted value of 60% to external indicators and 40% to internal indicators. According to the bureau, the commissioner's reliance on external indicators warrants reversal because the commissioner erroneously invoked "normative ratemaking" as the justification for using the external data. We reject this contention.

It is undisputed that rate making may have both predictive and normative aspects. See Attorney Gen. v. Commissioner of Ins., 370 Mass. 791, 804 (1976). "Normative" rate making simply recognizes the axiom that "regulation should provide cost control as well as cost observation." Id. However, the bureau argues that "there is no sound basis for the Commissioner to engage in normative ratemaking" unless he finds that the internal data reflect excess payments that are within the insurer's control and that the external data "validly" indicate movement in loss costs, citing Medical Malpractice Joint Underwriting Ass'n v. Commissioner of Ins., supra at 53-54. We need not address this argument, however, because we are not persuaded that the commissioner actually engaged in normative rate making; rather, we conclude that his use of external indicators was predicated primarily on his view of the superior predictive accuracy of those indicators.

In his decision, the commissioner noted the parties' recommendations as to weighting of external versus internal indicators. He also expressed reservations about the reliability of the internal data presented by the bureau. He determined that the SRB's expert had identified "numerous errors and confusion" in the reporting of the internal data. The commissioner stated that the external data were more recent and "not subject to the phenomena which distort the internal data. . . ." Moreover, the commissioner found that the external indices to which he assigned the 60% weight were the "very ones" that the bureau had "recommended last year." Finally, it may be worth noting that the commissioner had used similar 60%-40% weighted average for the past three years. There were thus sufficient findings and evidence in the record to warrant the commissioner's use of external indicators based on considerations of predictive accuracy alone.

But the bureau contends that, even if the commissioner's use of external factors was not an improper foray into normative rate making, use of external indicators was nonetheless unjustified because the commissioner excluded all of the bureau's evidence relative to the question of predictive accuracy. As noted above, exclusion of evidence generally will not result in reversal unless there has been a denial of substantial justice. Sudbury v. Department of Pub. Utils., supra. Framingham v. Department of Pub. Utils., supra. There was no such error.

The bureau's evidence was offered to show that the commissioner's past rate predictions have been inaccurate and, as a corollary, that this year's prediction is likely to be inaccurate. The rate as ultimately set is the product of many different factors and methodologies. The bureau does not appear to claim that these methodologies were not subject to full examination at the hearing below. Nor does it contend that its proffered evidence had anything to do with methodology. Rather, the evidence was simply aggregate data beginning in 1978 and running through 1986, showing the disparity between the rate set by the commissioner and the actual experience of the industry. Yet there was no showing that the methods used in past years to calculate the rates and the method employed this year were sufficiently comparable that the data from past years would be a meaningful indicator of likely future experience. The bureau is thus apparently arguing that, because the 1985 experience was worse than the 1983 experience, one can expect 1987 results to be worse than 1985. This contention fails when there is no demonstration that the methodology is the same. See Massachusetts Auto. Rating Accident Prevention Bureau v. Commissioner of Ins., 381 Mass. at 601. Therefore, we discern no error in the commissioner's exclusion of the bureau's evidence concerning "predictive accuracy."

The bureau also argues that the exclusion of evidence and other adverse rulings of the commissioner constitute bias. But the mere fact that the commissioner has ruled unfavorably to the industry, even if he did so on every occasion, is not by itself sufficient to show impermissible bias. Proof of bias requires some kind of showing that the hearing was not fair and impartial. See, e.g., Harris v. Board of Registration in Chiropody ( Podiatry), 343 Mass. 536, 540-541 (1962) (personal prejudice or hostility of adjudicator); American Employers' Ins. Co. v. Commissioner of Ins., 298 Mass. 161, 167-168 (1937) (decision based on information known only to agency). Cf. Northampton v. Smith, 11 Met. 390, 394-397 (1846), and cases cited (pecuniary interest of probate judge). See generally A.J. Cella, Administrative Law § § 311-320 (1986). The bureau has not made such a showing.

3. Normative reduction for excess "body shop" payments. The commissioner imposed a normative reduction of 3% on the reported pure premiums for property damage coverages because of the industry's excessive payments for repair and replacement of damaged automobiles and "the failure of the industry to take minimal steps to contain these costs." The commissioner based his decision both on St. 1986, c. 622, a cost containment amendment to G.L.c. 175, § 113B, enacted during the course of the proceedings, and on the "old" G.L.c. 175, § 113B (1984 ed.). The bureau argues that the commissioner unfairly required the bureau to comply with the newly-enacted law. We need not reach that argument, however, because we conclude that there was sufficient basis for the commissioner to invoke normative rate making independent of c. 622.

In order to engage in normative rate making, the commissioner must make findings, supported by substantial evidence, that the payments in question are excessive and that the excess is due to the failure of the insurers to exercise adequate controls. Medical Malpractice Joint Underwriting Ass'n v. Commissioner of Ins., supra at 54. The commissioner's measure of the excess payments must validly indicate an amount which could be eliminated through the exercise of internal controls. Id. The policy behind requiring these findings is to ensure that, in imposing normative considerations on the rates, the commissioner does not "go beyond subjecting a regulated industry to the rigors of simulated competition." Attorney Gen. v. Commissioner of Ins., 370 Mass. at 805. We conclude that the commissioner's 3% normative reduction is supported by the necessary findings and is based on substantial evidence.

The commissioner found that the industry's payments to automobile repair shops (body shops) were excessive because labor rates, labor time, and parts prices at the body shops were all too high. The finding was based in part on a study prepared by SRB witness Robert Moran which was first presented at the hearing on 1986 rates. The study concluded that the third party payment system in Massachusetts for automobile physical damage repair and replacement was inherently anti-competitive. According to the study, the cost of repairs is excessive for three reasons: (1) Labor rates are excessive due to the anti-competitive nature of the system. (2) Labor times are too high because the times are nearly always based on "crash books" — industry guides outlining estimated times to do a number of automobile repair operations — rather than on "flat rate books" — books prepared by automobile manufacturers which list shorter labor times drawn from actual experience. (3) Parts prices are too high because the industry fails to pressure the body shops to pass on to the consumer some of the discount the shops receive on automobile parts.

When the study was first presented, at the hearing on 1986 rates, the commissioner felt he did not have sufficient evidence to quantify a normative reduction in the rates, but he put the parties on notice that he wanted such evidence at the 1987 hearing. The SRB presented an updated version of the 1986 study. It also offered a study of 1,000 repair claims paid by five insurance companies and the testimony of Robert Moran. The industry presented the testimony of two insurance company representatives. The commissioner found that this new evidence "reinforces last year's testimony and conclusions" of excessive payments. There was thus substantial evidence for the commissioner's finding of excess payments.

Substantial evidence likewise supported the commissioner's conclusion that these excess payments were within the control of the insurers. The commissioner found that the companies' cost control efforts in this area were "sadly lacking" and he commented that "[n]o effort has been made to examine body-shop payment practices in a rigorous, business-like fashion with a view towards reducing cost. . . ." The evidence revealed that the insurers simply accede to higher labor rates when no body shop in an area will work for the then prevailing rate. There was also evidence that the insurers do not investigate to determine if a body shop rate increase is justified. The law permits insurers to provide an insured with a list of qualified repair shops at the insured's request, Allstate Ins. Co. v. Auto Damage Appraiser Licensing Bd., 399 Mass. 850 (1987), yet there was no evidence that insurers endeavor to reward shops performing at a lower rate by placing them on such lists. The companies' own witnesses offered no testimony that the insurers pursue a discount on parts prices.

There was testimony from Robert Moran that the industry could use its superior bargaining power to keep costs down. According to Moran, insurers make up over three quarters of the body shops' business. In addition, evidence showed that the body shop industry is relatively unconcentrated, while the insurance industry — where 90% of the market is covered by only twenty-five carriers — is quite concentrated indeed. The above was ample evidence on which the commissioner could conclude that excess body shop payments were within the ability of the insurers to reduce.

Finally, there was substantial evidence for the commissioner to quantify the normative reduction at 3%. The industry argues that there was not sufficient basis for an exact quantification of the excess, so the commissioner should not have made the reduction. Apparently, the industry believes that when an area of excess is difficult to quantify, the presumption should be that the excess will be borne by the policyholders until it can be precisely measured. We cannot agree. This position would reward insurers for neglecting cost control and obfuscating on the issue of measuring excess payments. A 3% reduction is within the range of possible reductions supported by the Moran study and the other evidence adduced at the hearing.

4. Disposition. The case is remanded to the single justice for entry of judgment reversing the decision of the commissioner and directing him to make findings and, if necessary, hold further hearings in order to fix and establish the 1987 automobile insurance rates in accordance with this opinion.

So ordered.