This petition for review of a Securities and Exchange Commission order is before us for a second time, on remand from the Supreme Court. The petitioner, New England Electric System (NEES), is a registered holding company controlling fourteen electric utility and eight gas utility companies in the New England states. In August 1957 the Commission instituted proceedings to determine whether NEES was entitled to retain that control under the standards of section 11(b) (1) of the Public Utility Holding Company Act of 1935, 15 U.S.C. § 79k(b) (1). Extensive hearings, resulting in a record of nearly 1,500 pages, culminated in the Commission’s order of March 19, 1964, requiring NEES to divest itself of all its gas companies.

In our earlier opinion, as here, we focused primarily on the Commission’s application to the NEES system of the “substantial economies” test of section 11(b) (1) (A), 15 U.S.C. § 79k (b) (1) (A). Generally, section 11(b) (1) allows a holding company to control only one system of operating utility companies. However, control of an additional integrated public utility system1 is permit*109ted if, in addition to size and location requirements not here in dispute, the Commission finds that the system “cannot be operated as an independent system without the loss of substantial economies which can be secured by the retention of control” by the holding company.

We held that the statutory words, “loss of substantial economies”, called for “a business judgment of what would be a significant loss, not for a finding of total loss of economy or efficiency”.- New England Electric System v. SEC, 1 Cir., 1965, 346 F.2d 399, 406. Since we read the Commission’s opinion as holding incorrectly that lost economies would not be substantial unless divestment would render “economical or efficient operation” impossible, and since we believed that under the appropriate standard the Commission could have found for NEES on the record before it, we ordered the case remanded. The Supreme Court reversed, holding that the Commission’s interpretation of the statute was correct, and remanded the case to us for review of the Commission’s order in light of the proper meaning of the statutory term. Securities and Exchange Commission v. New England Elec. System, 1966, 384 U.S. 176, 86 S.Ct. 1397, 16 L.Ed.2d 456.

In approving the Commission’s interpretation of section 11(b) (1) (A), the Court said that it required a showing that the “additional system cannot be operated under separate ownership without the loss of economies so important as to cause a serious impairment of that system.” 384 U.S. at 179, 86 S.Ct. at 1399. In illustration the Court quoted the language of the House report, that there must be a showing of “a real economic need”; of Senator Wheeler, speaking after the bill was passed, that the exception would apply only to systems “so small that they were incapable of independent operation”; and of the Commission itself, in Philadelphia Co., 1948, 28 S.E.C. 35, 46, that there must be a “situation in which the proven inability of the system to stand by itself would result in substantial hardship to investors and consumers were its relationship with the holding company terminated.” Finally, the Court concluded (1) that the proper test was “much more stringent” than a business judgment of significant loss; (2) that while economy in management was one theme of the act, eliminating restraint of free competition was another; and (3) that offsetting gains to competition, difficult to forecast, were “a matter for Commission expertise on the total competitive situation, not merely on a prediction whether, for example, a gas company in a holding company system may make more for investors than a gas company converted into an independent regime.” 384 U.S. at 185, 86 S.Ct. at 1402.

The only area remaining for definitional dispute is whether the holding company, to justify retention, must prove that severance will result in imminent bankruptcy of the subsidiary, or something less — a condition allowing survival but not on a sound or “healthful continuing” basis, Engineers Pub. Serv. Co. v. SEC, 1943, 78 U.S.App.D.C. 799, 138 F.2d 936, 944. We think the answer lies in the latter definition. This, we think, stems from a careful reading of the Court’s opinion, particularly in the light of the Commission’s opinion which it was interpreting.

We have taken pains to summarize the sequence of quotations cited by the Court, all of which have been relied on by the Commission, because, while they are internally consistent in militating against a test of business judgment of probable significant loss, they suggest that the Commission has not always been clear as to whether it considers “serious impairment” or “inability to survive” the standard. The first two quotations are consistent with the former; the last two, with the latter.

*110That the Court’s use of the “serious impairment” language reflected the fair import of the Commission’s opinion is clear. That opinion contained only one reference — to Senator Wheeler’s language above quoted — consistent with an “inability to survive” test. Its remaining verbalizations were “real economic need” (House Report language above quoted); “ability of the additional system to operate soundly” (Commission’s own formulation); “important economies” (North American Co. v. SBC, 2 Cir., 1943, 133 F.2d 148, 152) related to “the healthful continuing business and service of the freed utility” (Engineers Pub. Ser. Co. v. SEC, supra); and the conclusion of the Commission in its own words that “a registrant seeking to retain an additional system has the burden of showing * * * that such additional system cannot be operated under separate ownership without the loss of economies so important as to cause a serious impairment of that system.” The Court, therefore, advisedly chose this language as “the Commission’s reading of Clause (A)”.

We also observe that if Congress had meant to allow retention only when the alternative was extinction, the test of section 11(b) (1) (A) could have been shortened to require a finding that the subsidiary “cannot be operated as an independent system”. Furthermore, the Court’s emphasis on the likelihood of offsetting gains to competition presupposes the continued operation of the gas companies — we cannot imagine that competition would gain from the elimination of competitors. Finally, it is as impossible to say that “serious impairment” of an additional utilities system means “inability to survive” as to say that “serious injury” means “death.”

We take the trouble to distinguish between these two remaining possible approaches because of hints in the Commission’s supplemental brief, signed by its General Counsel and argued by its Solicitor, that it sees the test in terms of likely death of a system if divestment is ordered. This brief articulated the “Question Involved” as follows:

“[T]he only question remaining for this Court’s consideration is whether the Commission correctly held that petitioners failed to sustain their burden of proof that NEES’s integrated gas utility system was unable to stand by itself so that substantial hardship to that system’s investors and consumers would result from a termination of its relationship with NEES.” (Emphasis added.)

This manner of stating the question ignores the “serious impairment” language to which the Court gave first attention in its opinion, but draws exclusively from the language quoted from Philadelphia Co., supra. The Commission’s' wording of the “Question” also makes it clear that a system’s inability “to stand by itself” is the event bringing about the “substantial hardship” to investors and consumers. This reading of the test is made more clear by the brief’s truncated quotation of Philadelphia Co. in saying that that case places “upon petitioners the burden of proving the ‘inability of the additional system to stand by itself.’ ” The significance of hardship to investors and consumers is ignored. As final corroboration, it is to be noted that this abbreviated quotation from Philadelphia Co. is immediately followed by the statement, “in the light of this test * * * there is no basis for petitioners’ arguments that [the statute means] ‘such loss [as] would impair the gas companies’ ability to compete effectively * * *.’ ”

It seems too probable to us, therefore, that the Commission, at least in its supplemental brief, now views the statute as requiring divestment unless petitioners can prove that they could not operate as an independent system. Such a test would, we think, read out of the statute the phrase “without the loss of substantial economies” and would distort the Court’s imprimatur on the words “serious impairment”.

*111Even without the burden of proving likely demise, petitioner’s burden is, as the Court said, to meet “a much more stringent test” than that of a probable significant loss. But, if the standard to be applied to petitioner is stringent, so is the level of analysis and expertise to be exercised by the Commission. We have, only after a fresh review of all the evidence in the light of this most stringent practical standard, concluded that the Commission’s opinion does not reveal that application of both reason and experience to facts which merits endorsement as the responsible excercise of expertise.

To make intelligible our difficulties it is necessary to review again the basic facts in evidence and the Commission’s express bases for decision. The eight gas companies involved are operated under the general control of the NEES Gas Division, whose officers are responsible directly to the general officers of the holding company and have no duties regarding the electric companies. Seven of the eight operate in areas overlapping the service areas of one or more NEES electric companies and share some facilities, such as office space, meter readers, and accounting personnel and equipment. Both the gas and the electric companies obtain services such as supervision of accounting, purchasing, insurance, legal counsel, labor relations administration, and handling of rate matters from the New England Power Service Corporation, another subsidiary of NEES. In most other respects the gas companies are operated as an independent system, in competition with NEES and other electric companies and with numerous unregulated oil companies.

To support retention, NEES offered the report of a study by a professional utility engineering and consulting firm, Ebasco Services, Inc. (Ebasco), projecting the results of severance on the gas companies, in terms of increased cost of operation. The report presented a detailed description of the organization and operation of the gas companies as of 1958-1959 and as projected after severance, with an estimate of the actual and pro forma costs assigned to each phase of the operations. It concluded that severance would result in recurring additional costs of $1,495,000 per year, on the assumption that the gas subsidiaries would be operated as eight wholly independent companies. A supplemental report reduced the figure to $1,098,600, on the assumption adopted by the Commission that the companies would be operated as a single integrated system.2 The increase represented primarily the increased cost of the personnel required to duplicate the essential functions now being performed by the service company, and the loss of economies of scale now obtained by joint operation with the electric companies. The latter category included the increase in the cost of customer accounting, the loss of quantity discounts in purchasing, and the substantially increased cost of obtaining insurance against disaster losses.

The Commission rejected the Ebasco report’s conclusion, on the grounds (a) that the supplemental report and the witnesses did not satisfactorily explain why the cost of customer billing could not be reduced by using a centralized system for all eight companies, rather than individual billing by each company; (b) that the report projected greater increases in the per-customer costs of customer accounting to some gas companies than to the NEES electric companies sharing the same facilities, without satisfactory explanation; (c) that, since the increased cost attributed to customer accounting ($472,100) constituted a large part of the total increase claimed, NEES’s failure to justify the whole of that item “substantially impaired” the credibility of the entire estimate; and (d) that the estimated additional salaries and/or positions for the top executive *112staff of the gas system were also suspect.

Nothing in the Court’s opinion causes us to change our views on this issue as expressed in 346 F.2d at 407-409.3 We do not necessarily criticize the Commission for its skepticism in the specifics; but we do think that even taken together these items constitute at most a basis for reducing the estimated figure by some amount, not for concluding that no increased costs have been proved. In fact, the record here demonstrates conclusively that some increased costs are inevitable — the only doubt possible concerns the amount. Furthermore, we think that if the Commission is to use its expertise to assess the impact of the costs in “the total competitive situation”, it must attempt to determine some acceptable figure as a predicate for its assessment. Since, however, the Commission here proceeded on the alternative ground that the full amount of the Ebasco loss estimate ($1,098,600) was proven, we pass further discussion of the amount of losses and turn to their impact.

On this issue, NEES presented evidence that, because of their position at the end of the pipelines, the high cost of manufacturing gas locally, and the absence of storage facilities, the gas companies óf New England pay the highest price in the country for the gas they sell, and consequently enjoy little or no price advantage over oil in the important house heating market. Electricity is competitive, apparently, only in a quality market, but has shown better progress than expected. The growth of gas sales in New England, while showing a high percentage increase from a very low base, has lagged substantially behind the rest of the country in saturation of the market. Evidence was offered that, although the NEES gas companies have shown a higher rate of growth, of revenues, of . increase in saturation, of earnings, and of return on sales expenses, and lower operating expenses than other gas companies in Massachusetts, they have also shown a lower sales revenue per customer. NEES officers and representatives of the Massachusetts Department of Public Utilities testified that the NEES companies operate in areas of slow population growth, as compared to other companies in the state.

To this the Commission responded (a) that as to the geographical disadvantages of operating in New England, the NEES gas companies were in no worse position than the non-affiliated Massachusetts companies which have “apparently” been able to earn a fair return; (b) that the NEES companies, being the second largest gas utility operation in Massachusetts, should be able to provide itself with management comparable to the others’; (c) that the NEES gas companies’ operating ratio 4 of 76.41 per cent would compare favorably to others in the area; (d) that there was no “specific demonstration” of a causal relation between population growth and gas sales; (e) that the projected losses, even if assumed to be proved, were not substantial within the meaning of section 11(b) (1) (A), because the ratios of the losses to revenues (4.83 per cent), to operating expenses (6.03 per cent), to gross income (23.28 per cent), and to net income (29.94 per cent) — “the *113test of substantiality of the estimated loss” — were lower or not substantially higher than similar ratios in prior cases where divestment had been ordered; and (f) that in any event, despite the relatively independent operation of the gas subsidiaries, the operation did not secure the competitive advantages obtainable from single-minded independent management.

We have taken the trouble of including all of the Commission’s expressed bases for decision because of our conviction that they do not communicate to us the application of agency expertise in this critical area. For it is here that the delicate judgments — informed by experience, sharpened by close analysis, and ripened into the soundest possible intuition about what is likely to happen— that wTe call expertise, come into play. Cost accountants, efficiency experts, and engineers can arrive at figures for the likely increases of costs. But a wider-ranging inquiry must be made to arrive at an educated guess whether the increased burden, falling on a company in the economic and competitive situation of the gas companies involved here, will amount to substantial lost economies— whether its effect will be undue harm to the companies’ investors and customers. The importance of care and discrimination in the conduct of this inquiry is underscored by the Supreme Court’s clarification of the statutory standard. That the line of “serious impairment” is drawn so close to the point of probable business failure makes the holding company’s burden of proof more difficult to carry; but it also makes the risk more substantial that a divestment order based on erroneous findings will result in economic disaster.

With this in mind, we must conclude that the Commission’s analysis in the present case does not adequately address the considerations relevant to the ultimate issue it is required to resolve. To be sure, the evidence summarized above is far from conclusive; and it cannot be denied that there are imponderable factors. At the time of the hearing, the gas companies in New England were considering several ways of reducing their basic costs, such as development of local storage areas, construction of a plant to manufacture liquefied natural gas, and importation of gas by ocean tanker. At the same time they feared that existing oil import restrictions might be loosened. It may be that in the intervening years some of these possibilities have ripened to change the basic competitive circumstances. Nevertheless, we can only conclude on the present record that all gas in New England operates on, as one witness testified, a small cushion. The significance of this is not negated by observing that non-NEES companies in Massachusetts seem to be surviving, for the focus must be on the specific characteristics of the NEES companies, the only ones affected by the Commission’s order. Similarly, that the NEES gas companies together would be the second largest system in Massachusetts is a descriptive, not an analytical proposition. It offers little help in assessing ability to survive. Indeed, the record here is silent on the economic health of the largest system.

A more precise basis for the Commission’s decision lies in its use of ratios of the estimated losses to revenues, expenses, and income, in comparison with those adduced from previous divestment cases. The use of these ratios as a test is posited on the undeniable observation that, “in the context of clause (A), the ‘substantiality’ of asserted losses of economies cannot be tested in absolute terms but rather must be evaluated in relation to total revenues, expenses and income.” Philadelphia Co., 1948, 28 S.E. C. 35, 48-49, aff’d, 1949, 85 U.S.App. D. C. 327, 177 F.2d 720. But we observe further that, where the statutory test of substantiality is understood in terms of potential harm to the investor and consumer, even these ratios are significant only as they affect the investment structure of the companies in the particular case, and different companies may be compared only on the assumption that both operate at the same level.

*114For example, suppose two utility companies, one earning a rate of 6 per cent return on its investment and the other a rate of 5 per cent,5 both suffer losses amounting to 15 per cent of net income. The result would be to reduce the rates of return to 5.1 per cent and 4.25 per cent respectively. It is entirely possible that the former company would be able to maintain its investment position and operate without a rate increase, while the latter would be unable to produce what would be considered a fair return. In fact, in the present case the projected loss to the NEES gas companies amounted to 30 per cent of net income. At the hearing the Commission staff adduced evidence showing that this loss would decrease the companies’ rate of return from 6.4 per cent in 1959 to 4.1 per cent on the projected basis. This may be compared with an average rate of 5.9 per cent for the non-affiliated Massachusetts gas companies, and is below the minimum of 4.5 per cent. Yet, despite the staff’s initiative in securing this obviously important data, the Commission failed even to discuss it, except to observe that “apparently” the non-affiliated companies were earning a fair return. We cannot assume, without reason, that a projected rate of return for the NEES companies which is 30 per cent below the average for their counterparts is irrelevant to the issue of “serious economic impairment”.

Nor is it sufficient that the loss ratios considered here are “not significantly higher” than the corresponding ratios of companies ordered divested in previous cases. It is obvious from our discussion above that, even if the companies compared were operating at the same level before divestment, the point must come where a small increase in the percentage loss of income crosses the line between a fair return on investment and one so low as to trigger disinvestment or require rate increases. Moreover, where that line is drawn must depend, among other things, on the market position of the company involved. Comparison with figures derived from the situations of other companies in other places at other times cannot be conclusive.6

In short, we think the standard of “substantial economies”, developed by the Commission itself and affirmed by the Supreme Court in this case, requires the Commission — having established its best estimate of the extent of the definable losses likely to result from severance — to address itself to a prediction of the effect of those losses on the economic health of the particular companies involved, in their particular circumstances. Without pretending to define the one true test, we think it clear that relevant factors might include whether the effect of the loss will be (a) to make new equity financing only difficult, or impossible; (b) to reduce the rate of return on investment only slight*115ly, or enough to provoke either disinvestment or a rate increase; (c) to raise the cost of borrowing money a little, or a lot; (d) if rate increases seem probable, to injure deeply, hurt slightly, or affect not at all the companies’ ability to survive. The Commission considered none of these.

We recognize, of course, that the selection of relevant factors is for the Commission’s determination in the first instance, in the exercise of its expertise. But this principle does not isolate the Commission’s conclusions from review to determine whether that expertise has been exercised, and, if exercised, communicated. We do not think that the Commission’s obligation, which is at the root of the respect to which its expertise is entitled, is satisfied by the invocation of largely irrelevant ratios or of other data concerning other companies, at other times, in other areas, facing possibly different conditions. Were such ratios sufficient, the resolution of divestment cases could be as well entrusted to electric desk calculators as to an expert administrative agency.

Finally, the Commission attributed some significance in this case to its doubt that “such a separation of the gas from the electric operations has been effected as to secure the kind of single-minded management for each that would obtain upon actual divestment.” This conclusion was supported by the observation that other Massachusetts companies had been “more effective” in their operations, and by language best considered in the context of its quotation by the Supreme Court:

“In the present case the Commission said on this phase of the controversy:

‘Although the NEES Gas Division handles sales and promotional activities and various other matters for the gas subsidiaries separately from the electric companies, final authority on all important matters rests in the top NEES management. The basic competitive position that exists between gas and electric utility service within the same locality is affected by such vital management decisions as the amount of funds to be raised or allocated to the expansion or promotion of each type of services.’

“Competitive advantages to be gained by a separation are difficult to forecast. The gains to competition might well be in the public interest and might well offset the estimated loss in economies of operation resulting from a separation of the gas properties from the utility system. This is a matter for Commission expertise on the total competitive situation, not merely on a prediction whether, for example, a gas company in a holding company system may make more for investors than a gas company converted into an independent regime.

'3

“The phrase ‘without the loss of substantial economies’ is admittedly not crystal clear. But the Commission’s construction seems to us to be well within the permissible range given to those who are charged with the task of giving an intricate statutory scheme practical sense and application.” 384 U.S. at 184-185, 86 S.Ct. at 1402.

We have quoted at such length because we think it important to understand precisely the meaning of this part of the Court’s opinion. It is possible to read it as saying, in effect, that the Commission’s expertise always entitles it to say that the compensating advantages of increased competitiveness outweigh the loss of economies in a given ease, and that when the Commission has so spoken, its judgment is unreviewable. Were this, however, the Court’s intent, there would have been no occasion for remand for our further deliberation. The order of remand would have directed us to affirm the Commission’s order.

Furthermore, we note that the shared management on which the Commission posits its language quoted above is common to utility holding company systems, not peculiar to NEES. In effect the Commission says that an independent gas company is always likely to *116do better than one affiliated with electric companies. Cf. The North American Co., 1945, 18 S.E.C. 611. That is precisely the sort of empirical judgment that the Commission is best qualified to make in the light of its expertise, acquired from dealing with utility company systems over the decades. And we read the Court’s opinion as saying that the Commission is entitled to take that judgment into account as justifying a stringent test of “substantiality”. But, that done, the general judgment has no independent significance in an individual case. Were this not so, a single statutory standard would be read to set forth two tests: first, is the loss of economies so substantial as to cause serious impairment of the companies’ ability to survive; and, second, even if a loss of such magnitude is probable, is its impact sufficiently offset by efficiencies of independent operation so that it is no longer substantial ? We cannot escape the conclusion that expert judgment on the impact of gains from increased competition must be involved in answering the first question and that a two-step approach is both artificial and unrealistic.

Absent a finding, therefore, that the company to be divested has been suppressed for the benefit of its affiliates, see The North American Co., supra, or that some definable particularized benefit will accrue from divestment, we think the Commission should confine itself to a quantitative analysis of the substantiality of the proved losses. Since we read the Commission’s opinion as improperly attributing independent significance to the generalized competitive advantages here, and since — as we have explained above — we do not read the opinion as expertly evaluating the evidence in the light of the proper meaning of the statutory term, we are compelled to vacate the order of divestment and remand the case for further proceedings and an analysis by the Commission responsive to the difficulties which we have set forth.

APPENDIX

The following table is derived entirely from figures appearing in the Commission’s opinions in five utility divestment cases before 1948.1 ******Its purpose is to indicate the significance of those figures in the Commission’s reasoning in each case, and their reliability as a guide for the decision of later cases. It may be noted, for instance, that in the three cases where the claimed losses constituted relatively large percentages of net income (Virginia Elec. & Pow. Co.; Gulf Gas; St. Louis Co.) the loss estimate was held to be substantially overstated, so that the Commission had no need to consider the “substantiality” of the resulting percentages. Even when the Commission adverted directly to the percentages given it did not attempt a prediction of their economic impact on the companies involved.

We must disapprove entirely of the practice adopted by the Commission after 1948 of comparing the ratios before it with similar ratios derived, not from the opinions in previous cases, but from the papers and records in its files.2 In those circumstances there can be no assurance that the figures relied upon played any part in the decision of the cases.

*117

(concurring).

I concur in the court’s opinion with one possible reservation. The court speaks of “hints in the Commission’s supplemental brief that it sees the test in terms of likely death of a system if divestment is ordered.” The supplemental brief seems to me very explicit; the word “hints” is a judicial euphemism for *118“statements.” If this now stated principle is simply an unwarranted frolic on the part of counsel it is regrettable, but no harm is done. The thought lingers, however, that counsel would not make such statements without reason for so doing, and behind the back of their client. While, like the court, I would not myself read the Commission’s opinion as laying down the rule now asserted in the supplemental brief, if in truth this was the principle that the Commission applied, then its findings are even more vulnerable than we have held them to be.

For example, the matter of the asserted reduction in the new system’s rate of return, discussed in the court’s opinion, which the Commission considered to be of inadequate consequence, might well be insufficient to make bankruptcy imminent, if that is the test the Commission in fact applied. On the other hand,, it could well be of great consequence if the test is simply one of serious impairment in the system’s financial circumstances.

I would hope that on the new review this matter will be cleared up, and if it so happens that the Commission did weigh the evidence in the light of the test it now espouses in its supplemental brief, appropriate reconsideration will be given.