EASTERBROOK, Circuit Judge.

Although the Dow Jones Industrial Average may be the world's most famous stock market index, the Dow Jones Transportation Average is its most venerable, having been established in 1884. The Dow Jones Utilities Average, which dates from 1929, is another well known indicator. An index uses a few stocks to approximate the performance of a market segment. For example, the 20 stocks in the transportation index are designed to track a portfolio of approximately 145 transportation stocks with a capitalization exceeding $200 billion. The 15 stocks in the utilities index stand in for a utilities segment of 145 firms with a capitalization near $300 billion. When Charles Dow designed these indexes, long before instantaneous worldwide networks, a "computer" was a person who calculated tables of artillery trajectories in longhand on foolscap. In that era a reference to a few stocks as an approximation of many was a valuable time-saving device. Today it is easy to follow the average value-weighted price of a whole market, which an electronic computer can produce at the touch of a button. Each investor can specify and follow the portfolio that seems most interesting or important. Still, indexes have retained their fascination with the media and the public, and they have developed a new use — as the base of futures contracts. Our case presents the question whether futures exchanges may trade contracts based on the Dow Jones Utilities Average and the Dow Jones Transportation Average. For many years Dow Jones was unwilling to license its indexes (rather, the trademarks used to denote them) for use in futures contracts. In 1997 it changed its mind and set in train these proceedings.

"A futures contract, roughly speaking, is a fungible promise to buy or sell a particular commodity at a fixed date in the future. Futures contracts are fungible because they have standard terms and each side's obligations are guaranteed by a clearing house. Contracts are entered into without prepayment, although the markets and clearing house will set margin to protect their own interests. Trading occurs in `the contract', not in the commodity." Chicago Mercantile Exchange v. SEC, 883 F.2d 537, 542 (7th Cir. 1989). See also Dunn v. CFTC, 519 U.S. 465 (1997). The classic futures contract involves a commodity such as wheat, but in principle any measure of value can be used. Financial futures usually take the form of a contract that depends on the value of an index at some future date. Thus, for example, the buyer (the "long") of a futures contract based on the Standard Poor's 500 Index future might promise to pay 100 times the value of that index on a defined future date, and the seller (the "short") will receive that price. Either side may close the position by buying or selling an offsetting obligation before the expiration date of the contract.

Financial futures contracts are useful for hedging or portfolio adjustment. They facilitate risk management — that is, assignment of the inevitable risks of markets to those best able to bear them. See generally Merton H. Miller, Merton Miller on Derivatives 79-100 (1997); Robert C. Merton, Applications of Option-Pricing Theory: Twenty-Five Years Later, 88 Am. Econ. Rev. 323 (1998); Myron S. Scholes, Global Financial Markets, Derivative Securities, and Systemic Risks, 12 J. Risk Uncertainty 271 (1996). Someone who owns a mutual fund containing all of the Standard Poor's 500 stocks can cut risk in half by selling a futures contract based on the SP 500 index, or double the market return (and the risk of loss) for the same financial outlay by buying a SP 500 futures contract. A futures contract based on a market segment (such as utilities) also may be used for portfolio adjustment. Suppose the investor wants to hold a diversified portfolio of stocks that does not include utilities. This investor might own a broadly representative mutual fund and then sell a futures contract based on a utilities index. Similarly, a person who wants to obtain the returns (and take the risks) of particular market segments that do not have their own mutual fund — for example, a combination of utilities and transportation stocks, but no industrials — could purchase an appropriate combination of futures contracts. Using these contracts for portfolio adjustment is attractive because the transactions costs of trading futures are much smaller (by an order of magnitude) than the costs of trading the underlying stocks in equivalent volumes. A pension fund that wants to move from stocks to the equivalent of a mixed stock-and-bond portfolio, without incurring the costs of trading the stocks, can do so by selling a futures contract on an index. See Merton Miller on Derivatives at 86-89.

For many years the traditional futures markets, such as the Chicago Board of Trade, have been at odds with the traditional stock markets, such as the New York Stock Exchange, about where financial futures would be traded — and whether they would be traded at all. The stock exchanges prefer less competition; but, if competition breaks out, they prefer to trade the instruments themselves. The disagreement has spilled over to the regulatory bodies. The Securities and Exchange Commission, which regulates stock markets, has sided with its clients; the Commodities Futures Trading Commission, which regulates boards of trade, has done the same. In 1982 this court held that institutions within the CFTC's domain are authorized to trade financial futures (including options on these futures), and, because of an exclusivity clause in the Commodity Exchange Act, that the stock markets are not. See 7 U.S.C. § 2; Chicago Board of Trade v. SEC, 677 F.2d 1137 (7th Cir.), vacated as moot, 459 U.S. 1026 (1982). Because the stock exchanges long had traded options, a financial derivative related to futures, a political donnybrook accompanied the regulatory dispute among the markets and agencies. Shortly after our opinion issued, Congress amended the Commodity Exchange Act to reflect a compromise among the CFTC, the SEC, and the exchanges.

Congress allocated securities and options on securities to exchanges regulated by the SEC, futures and options on futures to boards of trade regulated by the CFTC. If an instrument is both a security and a futures contract, then it falls within the CFTC's domain. (This is the basis of Chicago Mercantile Exchange, which held that a novel "index participation" is a futures contract that belongs to boards of trade.) Options on single securities are allowed, but futures contracts on single securities are not. This allocation appears to be a political compromise; no one has suggested an economic rationale for the distinction. Having drawn this line, however, Congress had to make it stick. Futures contracts thus must reflect "all publicly traded equity or debt securities or a substantial segment thereof". 7 U.S.C. § 2a(ii)(III). Finally, both agencies participate in the process of reviewing applications to trade new financial futures contracts. Before a new contract may start trading, both the SEC and the CFTC must certify that it meets the statutory criteria. Regulation of the trading process belongs exclusively to the CFTC.

A year after this statute was enacted, the SEC and the CFTC issued a Joint Policy Statement spelling out the kinds of financial futures that the agencies believed suitable for trading. 49 Fed. Reg. 2884 (Jan. 24, 1984). The Joint Policy Statement is not a regulation and lacks legal force, but for many years the markets observed its limits when proposing new contracts. One element of the Joint Policy Statement is that any index used as the basis of a futures contract contain at least 25 domestic equity issuers. The Dow Jones Transportation Average is based on 20 stocks, the Utilities Average on 15. The second element is that, in a price-weighted index, no single security may have a weight exceeding 10% of the entire index, if its price weighting exceeds its capitalization weighting by a factor of three. In April 1997 Dow Jones replaced one firm in its Utilities Average with Columbia Gas, which accounts for 2.93% of the Utilities Average by capitalization weight, but 12.56% by price weight. The Transportation Average does not contain a stock with a similar disparity.

A brief detour into price-weighting may be helpful. All Dow Jones averages are price weighted, the only practical way to construct an index before electronic calculation. Price weighting means that the prices of all stocks in an index are added together, then divided by a number that the maintainer of the index selects to preserve consistency over time as stocks are split, firms enter or leave the index, and so on. If one stock in a price-weighted index undergoes a rapid rise in price, that stock can come to have an influence disproportionate to its capitalization. (Think of a stock of a small company that doubles in price. This will drive up the value of the index more than a 15% increase in the price of a much larger company, even though the bigger company's increase yields a larger improvement in investors' wealth.) Indexes constructed since the advent of electronic computers are value-weighted (that is, each stock affects the level of the index according to the ratio of its total market capitalization to that of all other securities in the index), and value-weighting in principle leads an index to be a more accurate reflection of the portfolio. See James H. Lorie Mary T. Hamilton, The Stock Market: Theories and Evidence 55-57 (1973). Nonetheless, price-weighted indexes have been highly correlated with value-weighted indexes (and with the market as a whole) through the years. Id. at 60-69. Movements in the Dow Jones Industrial Average, a small (30-stock) price-weighted index, track very closely movements in the Standard Poor's 500, a large value-weighted index. The law of large numbers plays a major role, as does the skill of index maintainers in keeping divisors up to date. Thus both price-weighted and value-weighted indexes continue to be useful to investors.

When Dow Jones Co. agreed to license its trademarks for use in financial products based on its market indexes, both stock markets and futures markets sought to trade products based on these indexes. The SEC promptly approved trading (at the stock markets) in options on the Dow Jones Industrial Average, the Dow Jones Transportation Average, and the Dow Jones Utilities Average. But when the futures exchanges sought permission to trade futures contracts based on these indexes, the SEC was less accommodating. It approved a futures contract based on the Dow Jones Industrial Average but blocked trading in the others, which have fewer than 25 stocks in the index. Release No. 34-40216, 1998 SEC Lexis 1454. The SEC recognized that the Joint Policy Statement of 1984 lacks the force of law. Id. at *13. Nonetheless, the SEC concluded, its criteria would be applied as part of the totality of the circumstances that the agency considered. Id. at *42. But of course the circumstances an agency considers must implement the statute. One can't have a totality-of-the-circumstances approach in the abstract. Here is the critical statutory text:

[T]he [CFTC] shall have exclusive jurisdiction with respect to accounts, agreements (including any transaction which is of the character of, or is commonly known to the trade as, an "option", "privilege", "indemnity", "bid", "offer", "put", "call", "advance guaranty", or "decline guaranty") and transactions involving, and may designate a board of trade as a contract market in, contracts of sale (or options on such contracts) for future delivery of a group or index of securities (or any interest therein or based upon the value thereof): Provided, however, That no board of trade shall be designated as a contract market with respect to any such contracts of sale (or options on such contracts) for future delivery unless the board of trade making such application demonstrates and the [CFTC, with the SEC's concurrence,] expressly finds that the specific contract (or option on such contract) with respect to which the application has been made meets the following minimum requirements:

(I) Settlement of or delivery on such contract (or option on such contract) shall be effected in cash or by means other than the transfer or receipt of any security, except an exempted security under section 3 of the Securities Act of 1933 [ 15 U.S.C. § 77c] or section 3(a)(12) of the Securities Exchange Act of 1934 [ 15 U.S.C. § 78c(a) (12)] as in effect on the date of enactment of the Futures Trading Act of 1982 (other than any municipal security, as defined in section 3(a)(29) of the Securities Exchange Act of 1934 on the date of enactment of the Futures Trading Act of 1982);

(II) Trading in such contract (or option on such contract) shall not be readily susceptible to manipulation of the price of such contract (or option on such contract), nor to causing or being used in the manipulation of the price of any underlying security, option on such security or option on a group or index including such securities; and

(III) Such group or index of securities shall be predominately composed of the securities of unaffiliated issuers and shall be a widely published measure of, and shall reflect, the market for all publicly traded equity or debt securities or a substantial segment thereof, or shall be comparable to such measure.

7 U.S.C. § 2a(ii). The futures markets proposed contracts that would be settled in cash, so subsection (I) is satisfied. The SEC did not find that the contracts would be "readily susceptible to manipulation", so subsection (II) also is satisfied. (The SEC did express concerns related to manipulation; we discuss these below; but it did not find that the proposed contracts would be readily susceptible to manipulation.) As for subsection (III), the Dow Jones averages are "predominately composed of the securities of unaffiliated issuers" and are "widely published." They do not measure or reflect the whole market, but transportation and utilities stocks are "substantial segment[s] thereof" — or so at least the SEC assumed. It did not formally define a "substantial segment" of the market, but neither did it deny that utilities and transportation qualify. What the SEC did conclude is that the Dow Jones Utilities and Transportation Averages do not "reflect" the utilities and transportation segments. According to the SEC, the principal shortcoming is that the indexes are not themselves "substantial segments" of the market or, as the SEC summed up, are not "broad-based."

According to the SEC's lawyers, this decision should receive the respect that Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837 (1984), prescribes for an agency's implementation of an ambiguous statute. One potential problem with this approach is that the sense of deference to which the SEC appeals is the one associated with delegation of authority to act, see Atchison, Topeka Santa Fe Ry. v. Pesa, 44 F.3d 437, 445-47 (7th Cir. 1994) (en banc) (concurring opinion), affirmed under the name Brotherhood of Locomotive Engineers v. Atchison, Topeka Santa Fe Ry., 516 U.S. 152 (1996), and Congress has delegated to the CFTC as much as to the SEC the authority to interpret and implement § 2a(ii). Yet the CFTC disagrees with the SEC's view that the index must be "broad-based," and the CFTC contends that its view of the statute's meaning is entitled to respect from the judiciary. At least one court of appeals believes that disagreement between agencies cancels all deference and requires the court to make an independent decision. Rapaport v. OTS, 59 F.3d 212, 216-17 (D.C. Cir. 1995). We are not so sure about this; Chicago Mercantile Exchange, which like today's case grew out of a fundamental disagreement between the SEC and the CFTC, concluded that it is possible to defer simultaneously to two incompatible agency positions. 883 F.3d at 547-48. There's no problem of logical impossibility; the court could accept the position of whichever agency's order is under review. In practice this would mean that the stingier of the two agencies prevails; if either of them wants to block trading in a particular instrument, then the instrument won't trade. Perhaps this is what Congress set out to achieve when requiring the SEC and the CFTC to concur before a new financial futures product could be brought to market. Just as in Chicago Mercantile Exchange, however, it is not necessary to reach a definitive conclusion about how to proceed when these two agencies disagree over an issue under the Commodity Exchange Act. See also Dunn, 519 U.S. at 479 n. 14 (finessing Chevron issues when the CFTC disagreed with the Treasury Department). Section 2a(ii) (III) is not ambiguous — not, at least, when the question is whether an index "reflects" a substantial segment of the market.

Section 2a(ii)(III) says that the index "shall be a widely published measure of, and shall reflect, the market for all publicly traded equity or debt securities or a substantial segment thereof". The index must "reflect" a substantial segment; § 2a(ii)(III) does not require that the index be a substantial segment. The SEC's rejection of futures contracts based on the Dow Jones Utilities and Transportation Averages because the 15 or 20 stocks in the index are not themselves a substantial segment of the market cannot be reconciled with the language or structure of the statute. It is enough, according to Congress, if the index "reflects" the segment. And on this record it is undisputed that the indexes do reflect the stock-market performance of the industries they are designed to measure. The long-term correlation between the indexes and the larger portfolios of stocks in the industries exceeds 92% for both indexes.

According to the SEC, two principal considerations support a conclusion that the index as well as the segment it represents must be "substantial" or "broad-based." First, this is how it told Congress it would implement the statute; second, this approach is essential to prevent an index from being used as a proxy for the forbidden single-stock futures contract. Neither line of argument is convincing.

The SEC's order contains many statements along the lines of: "We told Congress that, in determining whether a stock index reflected a substantial segment of the market, we would consider `the number of securities comprising the index.'" 1998 SEC Lexis at *34. Footnote 31, attached to this passage, gives as the source of the internal quotation an attachment to a letter that Edward A. Greene, then the SEC's General Counsel, sent to Rep. Timothy Wirth of Colorado on May 12, 1982. All of the SEC's other, similar, statements about what "we told Congress" refer to Greene's letter or something equivalent. Both the CFTC and the Board of Trade contend that the SEC has misunderstood Greene's letter — that what it actually tells Congress is that the SEC will consider the number of securities in the segment, not the number in the index. So far as we are concerned, however, it is irrelevant whose understanding of Greene's letter is superior.

Legislative history is problematic under the best circumstances, and even so reliable a source as the Conference Committee Report may be used only when there is a genuine ambiguity in the statute. Barnhill v. Johnson, 503 U.S. 393, 401 (1992). None of the SEC's lengthy order in this case identifies any portion of the statute that legislative history could help disambiguate. But Greene's letter is not "legislative history" at all; it is an agency's wishful thinking. It does not speak for a single Member of Congress. It may accord with the views of some, but the only way to know is to listen to the Members. Representative Wirth received Greene's letter in silence. If Greene had written to Wirth that the SEC would not approve a futures contract unless it expired in mid-month (to avoid "triple witching hour") and each stock in the index traded for more than $10 (the SEC deprecates penny stocks), it would be neither more nor less authoritative about the meaning of the statute that emerged months later from the legislative give-and-take. (This was not a situation in which Congress enacted, verbatim, a compromise handed to it on a silver platter. Each chamber had its own version; two committees in the House wrote separate reports; and the proviso that has led to this litigation, the SEC's independent approval power, was added in the Conference Committee.)

Our own tour through the committee reports turned up one snippet that might help the SEC — although the SEC does not mention it. One House Report on the statute contains this line: "Briefly, the agreement [between the SEC and the CFTC] stipulates that the SEC will regulate options on securities and on certificates of deposit (and on all groups or indices of securities or certificates of deposit), and the CFTC will regulate futures contracts on exempted securities (other than municipal securities) and on broad-based groups or indices of any securities, as well as options on any such futures contracts." H.R. Rep. 97-565(I), 97th Cong., 2d Sess. 38 (1982). Perhaps the SEC disdained this passage because it describes the SEC-CFTC Accord, rather than the details of the legislation. Perhaps the omission reflects an ambiguity in the passage. Does it mean "(broad-based groups) or indices" or instead "broad-based (groups or indices)"? A passage on the very next page of the report implies the former, for there the Committee, when writing about the statutory text, says that the index "must be a widely published and accurate measure of a broad segment of the corporate or municipal securities markets". This implies that it is the segment, and not the index, that must be "broad," and that the word "reflect" in § 2a(ii)(III) means "accurate." But perhaps in the end all this shows is that even the best legislative history has limited utility. It is enough for us to say that the published reports do more to support the CFTC's position than to assist the SEC.

The SEC's second principal reason is more substantial — at least in theory. Suppose it turned out that a single stock accurately reflected price movements for an entire industry segment. This might be so, for example, if the firm were a holding company or closed-end mutual fund that held shares in every firm of a line of business; or perhaps this could occur just by chance. Suppose, moreover, that this correlation were well known, and that the relation between this stock and the market segment were "widely published." A board of trade then might be tempted to base a futures contract on the price of this stock. Yet such a contract would transgress part of the legislative bargain — the ban on single-firm futures contracts. If the contract must be based on more than one stock, how many more must it have? The SEC did not give a numerical answer (and it recognized that the number 25, from the 1984 Joint Policy Statement, could not be employed as a bright line). Instead the Commission asked whether the numbers in the Dow Jones indexes (15 and 20) are high enough to make surrogate trading difficult. ("Surrogate trading," in the SEC's lexicon, is trading an index future as a surrogate for trading a single stock.) This is a sensible question to ask, but the SEC did not have any evidentiary support for the answer it gave.

Robert J. Mackay, a respected student of futures markets (and director of a center devoted to their analysis), testified in support of the CFTC's position that the Dow Jones Utilities and Transportation Averages are too well diversified to be attractive to investors interested in surrogate trading — even to inside traders, the investors the SEC appears to fear most. (The SEC apparently believes that it is harder to catch investors who use material nonpublic information to take positions in futures markets than those who take positions in equities or options, though why this should be so the SEC does not say.) Mackay also explained that as the number of stocks in an index increases, a trader attempting to use the index as a surrogate for one of its component stocks faces additional risk, caused by the fact that the other stocks in the index might move adversely to the one for which the index is serving as a surrogate. The SEC acknowledged this but asserted in response that this extra risk could be hedged away. 1998 SEC Lexis at *36 n. 34. Let us suppose this is so, though the SEC did not explain how the hedging strategy would work. Mackay's position remains essentially uncontested, and an agency may not disregard evidence in the record by vague references to "our experience in reviewing several early non-diversified (or industry-specific) stock index futures contracts." 1998 SEC Lexis at *37. The "experience" to which the SEC refers comes from 1983 and before, and as the passage says it is the experience in "reviewing . . . contracts" rather than knowledge about how investors use contracts in the market. The SEC does not refer to a single instance in which a single investor used a 15-stock portfolio as a surrogate for an individual stock. Nor does the SEC identify any circumstances under which it would be in an investor's interest to do so.

Columbia Gas accounts for approximately 12% of the Dow Jones Utilities Average. No other single stock accounts for more than 10% of either index. Unless an investor could make a profit by trading a futures contract on the Dow Jones Utilities Average as a surrogate for holding Columbia Gas stock, there is no reason to be concerned about the possibility of surrogate trading in these two indexes. Mackay concluded that surrogate trading would not be profitable, and would incur uncompensated risk. The SEC responded that hedging could reduce the amount of risk incurred, but this does not show (or even assert) that investors would find the strategy profitable. Anyone who wants to receive the risk-return combination offered by Columbia Gas has only to buy Columbia Gas stock. An investor who wants to bear the risk (and obtain the rewards) of this investment without stock ownership can do so by trading in options and forward contracts. Single-stock options trade freely in the stock exchanges, and many options have expiration dates as far into the future as the expiration of futures contracts.

Differences in the margin rules between stock and futures exchanges may affect the profitability of the strategies. Margin in a stock market represents borrowing to pay for the investment; in a futures market, by contrast, there is no present investment, but only a promise to pay according to a future price, so margin serves a completely different purpose: it is a deposit that dealers and clearing corporations require in order to provide some security against default. Virginia G. France, The Regulation of Margin Requirements, in "Mergers and Market Integrity" 15 (Lester G. Telser ed. 1991); Stephen Figlewski, Margins and Market Integrity: Margin Setting for Stock Index Futures and Options, 4 J. Futures Markets 385 (1984). Because the economic function of margin differs across markets, so too does the level required. Futures markets demand less margin because the dealer or clearing corporation can sell the position (or buy an offsetting position) to curtail loss if there is any question about the customer's willingness or ability to perform. Lower margin in the futures markets means that for the same net outlay an investor can take the economic risk of a larger position. That's what makes surrogate trading through futures markets an intriguing possibility to investors. What Mackay concluded is that this benefit is sufficiently small in relation to the risk of price movements introduced by other stocks in the index (or the cost of hedging these risks away) that investors would not find it economically attractive to use any of the Dow Jones indexes as a surrogate for any single stock. Because that conclusion stands unrefuted on this record, the SEC's conclusion that trading in futures contracts on these indexes poses a genuine threat of violating the rule against a futures contract on a single security is not supported by substantial evidence, and must be called arbitrary and capricious.

Along the way to its decision, the SEC made a number of other observations that were critical of the Dow Jones indexes, the CFTC, the futures markets, or all three. None of these additional observations bears on the question whether the index "reflects" the substantial segment of the market on which it is based. Having adopted a totality-of-the-circumstances approach, the SEC apparently believed that it need not tie any of the "circumstances" to the statute it was applying. Our view of proper statutory interpretation is otherwise, so we could sweep all of these additional subjects off the table. But for the sake of completeness we discuss each one briefly.

In several places the SEC stressed that the Dow Jones averages are price-weighted, and that price-weighted indexes are inferior to value-weighted indexes in tracking the portfolios they represent. This is true enough as a matter of theory, but unless price-weighting leads to an index that does not "reflect" the segment the index is supposed to measure, it is irrelevant. Evidence received by the SEC shows that in practice the Dow Jones averages closely match their segments, and perhaps do this as closely as a value-weighted index.

To keep a price-weighted index stable in relation to the stocks it represents, its sponsor must change the divisor frequently. Calculating the divisor is an art rather than a science. Moreover, both value-weighted and price-weighted indexes require occasional changes in the component stocks, to reflect mergers and growth or failure in product markets. The SEC observed that after approving a futures contract based on any of the Dow Jones averages, it would be unable to superintend these changes in the index. That's true enough, but what has it to do with § 2a(ii)? It is an argument against any futures contract based on an index (including the futures contract based on the Dow Jones Industrial Average, which the SEC approved), but § 2a(ii) says that such contracts may be traded. If in some future year Dow Jones Co. changes its indexes so that futures contracts based on them no longer satisfy § 2a(ii), then the CFTC bears the responsibility of enforcing the statute. The SEC is not entitled to block trading today on a contract that satisfies § 2a just because the SEC fears that the CFTC will not perform its regulatory tasks tomorrow.

Still other portions of the SEC's opinion reflect disagreements between the SEC and the CFTC about how futures markets should be regulated. For example, the SEC expressed a belief that margin levels in futures markets are too low. Although recognizing that it lacks authority over this subject, it added that the decision to disapprove trading of contracts based on the Dow Jones averages "is based on the totality of the circumstances before us, which includes prospective margin levels." 1998 SEC Lexis at *33 n. 29. What the SEC did not explain is how the "circumstance" of lower margin on futures contracts bears on any of the criteria in § 2a(ii). Forced to guess about this, we conjecture that the SEC thinks that margin has some relation to the possibility of manipulation — but then the SEC did not conclude under § 2a(ii)(II) that any of the proposed contracts would be "readily susceptible to manipulation". We'll come back to manipulation.

Elsewhere in its order, when explaining why it thought that blocking futures contracts based on the Dow Jones Averages is compatible with approval of options trading on these very averages, the SEC expatiated on what it sees as the CFTC's shortcomings. 1998 SEC Lexis at *66-69. Calling the Dow Jones Utilities and Transportation Averages "narrow-based," the SEC stated that it was confident that it could superintend disclosure and maintenance criteria that would allow markets to function smoothly. But because the CFTC handles these matters differently, and the SEC could not require the CFTC or the boards of trade to use the SEC's oversight mechanisms, the SEC decided to block trading altogether. Once again, this approach has nothing to do with § 2a(ii). Differences between the SEC and the CFTC about market oversight and regulation are of long standing. Congress has entrusted the CFTC with the task of regulating futures markets, and the SEC is not entitled to adopt a "my way or the highway" view by using its approval power under § 2a as a lever.

Stock exchanges within the SEC's bailiwick trade options on the New York Stock Exchange Utilities Index. This index includes the gas and electricity businesses, as the Dow Jones Utilities Average does, plus the telecommunications business. Several times the SEC observes that telecommunications firms traditionally have been classed as utilities and implies that the Dow Jones Utilities Average is defective because it omits them. The problem is not that telecommunications firms are missing from the index but included in the segment the index represents; rather, from the SEC's perspective, the problem is that telecommunications stocks are omitted from the segment the index measures, and therefore from the index too. Dow Jones says that this is because telecommunications today is a substantially unregulated market, and the Utilities Average is designed to reflect a portfolio of less-volatile firms that are subject to rate regulation.

While the SEC thinks that the difference between the NYSE and Dow Jones utilities portfolios is bad, an investor may believe that choice is good. The more different proxies for "utilities" are available, the easier it is to hedge or balance a particular portfolio. There's no right definition of the commodity underlying a futures contract. Instead there are many possible definitions. Boards of trade experiment to find good combinations, but most contracts fail to attract enough trading to justify continued listing. See Dennis W. Carlton, Futures Markets: Their Purpose, Their History, Their Growth, Their Successes and Failures, 4 J. Futures Markets 237 (1984). If, as the SEC believes, the NYSE index is superior to the Dow Jones index, then the NYSE index will prevail in competition. Nothing in the SEC's opinion explains why competition between these two indexes must be forbidden, why investors must be denied the opportunity to express a preference. See Daniel R. Fischel, Regulatory Conflict and Entry Regulation of New Futures Contracts, 59 J. Business S85 (1986). If the combination of gas and electricity stocks were too small to be a "substantial segment" of the entire market, then the SEC would be required by § 2a(ii)(III) to block a futures contract based on it. But the Commission did not make this finding, and there can be no serious doubt that the Dow Jones Utilities Average reflects the market segment that Dow Jones has defined — gas, electricity, and other utilities subject to rate regulation.

Actually, the SEC's unfavorable comparison of the Dow Jones Utilities Average to the NYSE utilities index suggests that it does not take seriously its own expressed concern about trading futures contracts as surrogates for individual stocks. The last phrase of § 2a(ii)(III) says that a futures contract may be traded, even if based on an index that does not reflect a substantial segment of the market, if it is "comparable to such measure." After the SEC's staff expressed the view that the utilities segment should include telecommunications, the CFTC and the Chicago Board of Trade replied that, even if the Dow Jones Utilities Average does not directly reflect this larger segment, it is "comparable to" the NYSE index, which does, and thus may be the basis of a futures contract under the comparability clause. The SEC replied that the comparability clause operates only when

the proposed index: (a) "is constructed to mimic" the second index, which itself must be "a widely published measure of the market;" (b) is extremely similar to the second index (e.g., the proposed index has 499 of the 500 stocks in the second index); or (c) is a proportionately reduced or expanded version of the second index, (e.g., an index that is identical in composition to another well-established index but is one-half the value of the well-established index).

1998 SEC Lexis at *54-55. This implies that the Chicago Board of Trade lawfully could offer a futures contract based on the NYSE utilities index, another based on the same index less one stock, and yet a third futures contract based on the same index less a different stock. But then the way to use futures contracts as surrogates for single stocks would be clear. An investor wishing to have the economic equivalent of a single-stock futures contract would go long in the NYSE utilities index future, and short in the "NYSE-utilities-index-comparable" futures contract that was one stock different. Voilé! A single-stock future. Would it not be better, if accomplishing the legislative goal is the end in view, to insist that the "comparable" index be substantially different from its benchmark? That the Dow Jones Utilities Average and the NYSE utilities index differ by an entire industry, telecommunications, means that hedging across the indexes can't be used to approximate a single-firm futures contract. The significant difference between these two measures, which the SEC condemned, actually is a mark in their favor.

Finally, a few words about manipulation. Although the SEC did not find that a futures contract based on either of the Dow Jones averages would be "readily susceptible to manipulation" of the index itself or the market segment it reflects, the SEC's order contains an undercurrent of misgiving. To the extent the SEC disagrees with the legislative judgment — that is, to the extent the SEC believes that trading in a futures contract should be forbidden when manipulation is a possibility, rather than a high probability (which we take the word "readily" to signify) — it should present that view openly to Congress, rather than engage in self-help measures that cannot be reconciled with the statute. What is more, we cannot see either in the record or in the nature of the proposed futures contracts any reason for concern.

Squeezes, corners, and other forms of manipulation in futures markets for physical commodities depend on hogging the deliverable supply. See Stephen Craig Pirrong, The Economics, Law, and Public Policy of Market Power Manipulation 18-90 (1996). A person who owns a substantial portion of the long interest near the contract's expiration date also obtains control over the supply that the shorts need to meet their obligations. Then the long demands delivery, and the price of the commodity skyrockets. It takes time and money to bring additional supplies to the delivery point, and the long can exploit these costs to force the shorts to pay through the nose. Futures markets deal with the prospect of squeezes and corners by expanding the deliverable supply (allowing, for example, delivery in Kansas City as well as Chicago), by imposing position limits (so that no one long can hold contracts for more than the deliverable supply), and by monitoring positions so that irregularly large positions may be subject to orderly liquidation before the expiration date. But the need for these precautions, like the possibility of manipulation itself, comes from the potential imbalance between the deliverable supply and investors' contract rights near the expiration date. Financial futures contracts, which are settled in cash, have no "deliverable supply"; there can never be a mismatch between demand and supply near the expiration, or at any other time.

Although it is impossible to rule out the use of financial futures contracts to affect the price in the underlying securities, it is hard to see how the effort could be profitable to the would-be manipulator. Financial markets such as all utilities, or all transportation stocks, are so large (many hundreds of billions), liquid, and competitive that their manipulation is almost unimaginable. If someone were determined to drive the price of transportation stocks up or down, buying and selling thousands of futures contracts on the Chicago Board of Trade would be like pouring a Dixie cup of water into Lake Michigan: there would be an effect, but it would not be detectable. One test of the manipulation hypothesis is whether financial futures contracts increase aggregate price volatility, either over the long term or near the expiration date. Economists who have explored this subject agree that on balance stock-index futures contracts have reduced volatility in stock prices, and that the effect on volatility near expiration is trivial. See, e.g., Hendrik Bessembinder Paul J. Seguin, Futures-Trading Activity and Stock Price Volatility, 47 J. Finance 2015 (1992); Richard Roll, Price Volatility, International Market Links and their Implications for Regulatory Policies, 3 J. Fin. Services Res. 211 (1989); George Sofianos, Expirations and Stock Price Volatility, 13 Rev. Futures Markets 39 (1994).

If the design were to manipulate the price of a single stock rather than the index or its market segment, then smaller investments of capital would suffice — but for the reasons covered in discussing surrogate trading, transactions in an index futures contract are lousy ways of approximating transactions in individual stocks. People who want to manipulate the price of Acme Widget Corp. will deal in that stock, or options in it. Data show that options reduce volatility in the security they represent. Aswath Damodaran Marti Subrahmanyan, The Effects of Derivative Securities on the Market for the Underlying Assets in the United States: A Survey, Fin. Markets Institutions Instruments 1 (Dec. 1992). And if one-stock derivatives do not conduce to manipulation, any futures contract where the index is based on two or more stocks must be less readily manipulable. Whether even single-stock options and futures on physical commodities are subject to (reliably profitable) manipulation is an interesting question, see Daniel R. Fischel David J. Ross, Should the Law Prohibit "Manipulation" in Financial Markets?, 105 Harv. L. Rev. 503, 542-52 (1991), but not one we need consider. It is enough to say that manipulation of stock prices through transactions in index futures contracts is hard to do, and a finding that these markets are "readily susceptible to manipulation of the price" would require a great deal more support than the ominous foreboding manifested in the SEC's order — an order that did not begin to discuss the vast empirical literature about the effects of options, futures contracts, and other derivatives on market efficiency. Cf. Bechtel v. FCC, 10 F.3d 875 (D.C. Cir. 1993).

* * * * *

Normally, when a court of appeals concludes that an agency's decision is not adequately supported, it remands so that the agency may enlarge the record or apply correct legal principles to the existing record. South Prairie Construction Co. v. Operating Engineers, 425 U.S. 800 (1976). Section 2a has an unusual proviso that makes the normal approach inappropriate. The SEC has a strict timetable for decision. 2 U.S.C. § 2a(iv)(II). Once the time has elapsed, control passes to the CFTC. This means that if, within the statutory time, the SEC has not given a satisfactory reason for rejecting a proposed futures contract, it does not get a second chance. We have held that the SEC's reasons are not satisfactory. Accordingly, we vacate the SEC's order concluding that the proposed futures contracts do not meet the criteria of § 2a(ii). With the SEC's order vacated, the subject now goes to the CFTC for its decision, just as if the SEC had failed to act within the statutory time.

The order of the Securities and Exchange Commission is vacated.