At the end of 2002, when United Air Lines declared bankruptcy, its employees (both active and retired) owned more than half the airline's common stock through an ESOP (employee stock ownership plan). In re UAL Corp., 412 F.3d 775, 777 (7th Cir.2005). ESOPs are subject to ERISA, the federal pension law, 29 U.S.C. §§ 1104(a)(2), 1107(b), (d)(6); Armstrong v. LaSalle Bank National Ass'n, 446 F.3d 728, 730 (7th Cir.2006), and this is a suit under ERISA. The plaintiffs represent a class consisting of United's employees. The principal defendant is State Street Bank & Trust, a cofiduciary of the UAL Corporation ESOP Committee. The Committee—which has six members, all appointed by the unions that represent United's employees—is the only fiduciary named in the plan. 29 U.S.C. § 1102(a)(1). The plan directs the Committee "to establish an investment policy and objective for the Plan, except that it is understood that the Plan is designed to invest exclusively in Company Stock." The Committee accordingly established a policy and goal of owning United stock, and appointed State Street to be the plan's trustee, that is, to manage the ESOP's assets, initially consisting of that stock. § 1103(a). The class charges State Street with imprudent management—specifically with failing to sell United stock as its market price plummeted en route to the airline's bankruptcy— and is appealing from the grant of summary judgment to State Street, which has in turn cross-appealed on an unrelated issue that we discuss at the end of this opinion.

State Street is what is called a "directed" trustee, because the Committee (the fiduciary named in the plan), in accordance with the plan language that we have quoted, directed State Street to invest the ESOP's assets exclusively in stock of United Air Lines. Directed trustees are permitted by ERISA: if an ERISA plan "provides that the trustee or trustees are subject to the direction of a named fiduciary [in this case it is the UAL Corporation ESOP Committee] who is not a trustee,...the trustees shall be subject to proper directions of such fiduciary which are made in accordance with the terms of the plan and which are not contrary to [ERISA]." 29 U.S.C. § 1103(a)(1); see Maniace v. Commerce Bank, N.A., 40 F.3d 264, 267 (8th Cir.1994); May Dept. Stores Co. v. Federal Ins. Co., 305 F.3d 597, 599 (7th Cir.2002); LaLonde v. Textron, Inc., 369 F.3d 1, 5 (1st Cir.2004). Like other ERISA trustees, a directed trustee has a statutory duty of prudence. § 1104(a)(1)(B). But as an ESOP fiduciary, he does not have the further (really, the included) duty to diversify the trust assets, § 1104(a)(1)(C)—we call it "included" in the duty of prudence because diversification is normally an essential element of prudent investing by a fiduciary. Armstrong v. LaSalle Bank National Ass'n, supra, 446 F.3d at 732. A directed trustee appointed under an ERISA plan does not have that duty because the very purpose of an ESOP is to invest in a single stock, that of the employer of the ESOP's participants.

We must first decide whether a directed trustee of an ESOP has any fiduciary duty with respect to the choice of trust assets, specifically any duty ever to replace the employer's stock—the normal holding of an ESOP—with some other security. The Maniace case that we cited, along with Herman v. NationsBank Trust Co., 126 F.3d 1354, 1361-62 (11th Cir.1997), says no, but FirsTier Bank, N.A. v. Zeller, 16 F.3d 907, 911 (8th Cir.1994), says yes, as does In re WorldCom, Inc. ERISA Litigation, 354 F. Supp. 2d 423, 444-45, 449 (S.D.N.Y.2005). The split reflects a confusing statutory picture. One provision of ERISA states that the directed trustee cannot be liable for obeying the directions of the fiduciary named in the plan, § 1105(b)(3)(B), but other provisions impose liability on a fiduciary for breaches of fiduciary duty by a cofiduciary if he knows of the breach and takes no reasonable efforts to prevent it, or if by his own failure to exercise prudence he enables the breach. §§ 1105(a)(2), (3). And recall that the directed trustee "shall be subject to proper directions of [the named] fiduciary which are made in accordance with the terms of the plan and which are not contrary to [ERISA]." § 1103(a)(1) (emphasis added). An imprudent direction cannot be a proper direction since the trustee has an express statutory duty of prudence.

The tension among these provisions is reflected in a pamphlet published by the Labor Department's Employee Benefits Security Administration, which affirms both that the directed trustee has a duty of prudence and that he has no "direct obligation to determine the prudence of a transaction" entrusted by the plan to another fiduciary. "Fiduciary Responsibilities of Directed Trustees" (Field Assistance Bulletin 2004-03, Dec. 17, 2004). "[D]irect" is the critical word, inviting us to resolve the tension by ruling that the trustee can disobey the named fiduciary's directions when it is plain that they are imprudent. (The Labor Department's pamphlet, as we'll see, is actually consistent with this approach.) The trustee physically controls the trust assets; knowingly to invest them imprudently or let them remain invested imprudently is irresponsible behavior for a trustee, whose fundamental duty is to take as much care with the trust assets as he would take with his own property. He is "an agent who is required to treat his principal with utmost loyalty and care—treat him, indeed, as if the principal were himself." Pohl v. National Benefits Consultants, Inc., 956 F.2d 126, 128-29 (7th Cir.1992). And although the creator of an ordinary trust may be able to include a provision in the trust instrument excusing the trustee from complying with the prudent-man rule, John H. Langbein, "The Contractarian Basis of the Law of Trusts," 105 Yale L.J. 625, 659-60 (1995), ERISA as we have seen expressly imposes the duty of prudence on directed trustees and forbids them to comply with directions of the fiduciary named in the plan that are not "proper." That is why Kuper v. Iovenko, 66 F.3d 1447, 1457 (6th Cir.1995), holds that an ERISA plan "may not be interpreted to include a per se prohibition against diversifying an ESOP."

Which brings us to the particulars of this case. The plaintiffs argue that State Street violated its fiduciary duty by failing to sell United stock held by the ESOP until the eve of United's bankruptcy. State Street knew long before then, if the plaintiffs are to be believed, that the UAL Corporation ESOP Committee, the named fiduciary, had unreasonably refused to deviate from the plan and diversify the ESOP's holdings or take other measures to reduce the looming risk to the employee-shareholders. The following chart traces the ups and downs (mostly downs) of United's stock price between January 1, 2001, and December 9, 2002, when United declared bankruptcy:

NOTE: OPINION CONTAINING TABLE OR OTHER DATA THAT IS NOT VIEWABLE

The price of United stock, which on September 10, 2001, was already 25 percent below the price at the beginning of the year, dropped almost 50 percent more in the immediate aftermath of the September 11, 2001, terrorist attacks. The following month, United's CEO sent a gloomy letter to all its employees which said that the company was "in a struggle just to survive" and was not yet in "a financial position that will allow us to continue operating," that it was "hemorrhaging money," and unless the "bleeding [was] stopped...United will perish sometime next year." The price of United stock fell another 20 percent in the two days after the publication of the letter.

The price continued falling, though with occasional upticks, and the financial press began speculating on the possibility of bankruptcy. State Street observed the decline with anxiety, but not until August 15, 2002—by which time the price had fallen to $2.70—did it notify the UAL Corporation ESOP Committee that it might be imprudent for the ESOP to continue holding United stock. In response to this warning the Committee authorized State Street to sell the stock, and it began doing so on September 27. By that time, however, the price had fallen to $2.36. The plaintiffs argue that State Street should have started selling within 30 days after the October 2001 letter of United's then-CEO and that had it done so the ESOP would have avoided $540 million in losses.

The plaintiffs say the letter should have alerted State Street that United was going into the tank. That is wrong. After the market "read" the letter, it valued United stock at $15.05 a share. Had the market thought that United would be bankrupt by the end of 2002, it would not have priced its stock that high in October 2001, implying a market capitalization for the company of more than $800 million. A trustee is not imprudent to assume that a major stock market (United was traded on the New York Stock Exchange) provides the best estimate of the value of the stocks traded on it that is available to him. In re UAL Corp., supra, 412 F.3d at 777-78; see Basic Inc. v. Levinson, 485 U.S. 224, 247, 108 S. Ct. 978, 99 L. Ed. 2d 194 (1988); In re Hovis, 356 F.3d 820, 823-24 (7th Cir.2004); Feder v. Electronic Data Systems Corp., 429 F.3d 125, 138 (5th Cir. 2005); Burton G. Malkiel, "Reflections on the Efficient Market Hypothesis: 30 Years Later," 40 Financial Rev. 1 (2005). At any price at which a sale takes place, the buyer thinks he's getting a bargain, or at least a reasonable deal—otherwise he wouldn't buy. Some investors, it is true, consistently beat the market, but few of them are trustees; it would be hubris for a trust company like State Street to think it could predict United's future more accurately than the market could, and preposterous for a committee of union officials (the named fiduciary) to challenge the market's valuation. Nor was State Street duty-bound to offer the Committee's members a tutorial on the obligations of an ERISA trustee; the Committee had its own lawyers for that.

Thus, at every point in the long slide of United's stock price, that price was the best estimate available either to State Street or to the Committee of the company's value, In re UAL Corp., supra, 412 F.3d at 777-78, and so neither fiduciary was required to act on the assumption that the market was overvaluing United. See Wright v. Oregon Metallurgical Corp., 360 F.3d 1090, 1099 (9th Cir.2004); Kuper v. Iovenko, supra, 66 F.3d at 1460; In re WorldCom, Inc. ERISA Litigation, supra, 354 F.Supp.2d at 447. What is true, however, though largely (and fatally) ignored by the plaintiffs and mistakenly denied by State Street's lawyers in this court (State Street itself we assume knows better), is that the fall in the market price of United was increasing the risk borne by the owners of the stock, the participants in the ESOP. There is always risk, in the sense of variance of returns, to owning common stock, because the fortunes of a company are uncertain and the stockholders, unlike bondholders and other owners of the company's debt, do not have a fixed entitlement; they are the residual risk bearers. Some companies, however, are riskier than others. Of particular relevance to this case, the higher the ratio of fixed-interest debt to equity is, the riskier is the position of the equity holders (the common stockholders). If a company has no debt, a 5 percent increase in the company's value will increase shareholder equity by 5 percent, and a 5 percent decrease in the company's value will decrease shareholder equity by 5 percent. But if the company has a 1:1 debt-equity ratio, a 5 percent increase in the company's value will increase shareholder equity by 10 percent (.05/.50) and a 5 percent decrease in the company's value will decrease shareholder equity by 10 percent. As the value of United, as reckoned by the stock market, plummeted, the airline's debt-equity ratio soared because its debt wasn't decreasing, and this created an acute threat of bankruptcy.

Such a threat would be of little moment to people who held United stock as part of a diversified portfolio, because the risks of the various components of such a portfolio tend to cancel out; that is the meaning and objective of diversification. The Modern Theory of Corporate Finance 6 (Clifford W. Smith, Jr., ed., 2d ed.1990). Probably, however, shares of United stock were the principal financial assets of most of United's employee-shareholders—at least those who, unlike pilots, have modest salaries—and hence their principal source of retirement income. And probably most of those non-wealthy employee-shareholders were risk averse; that is, they would prefer $10,000 certain to a 1 percent chance of obtaining $1 million, even though these are actuarial equivalents. They would be especially risk averse with regard to provision for their retirement, where they would be more dependent on their financial assets since they would no longer have earned income. Being risk averse, they would have preferred that the bulk of their financial assets not be invested in a single stock that, while worth as much as its market valuation (as far as anyone not possessing inside information could know), was extremely risky. "Because the value of any single stock or bond is tied to the fortunes of one company, holding a single kind of stock or bond is very risky. By contrast, people who hold a diverse portfolio of stocks and bonds face less risk because they have only a small stake in each company." N. Gregory Mankiw, Principles of Economics 546 (1998). That is why ERISA fiduciaries have a duty of diversification as part of their overall duty of prudence—unless as in this case they are directed pursuant to an ESOP to invest everything in stock of the participants' employer.

The employee-shareholders' total wealth included not only their stake in the ESOP but also, and for those not yet retired or approaching retirement this was probably more important, their expected earnings and fringe benefits as employees of United. The financial distress that pushed United's stock price down also jeopardized that expected wealth (insofar as United employees could not expect to land equally good jobs at other companies) and so depressed the employee-shareholders' overall wealth by more than the fall in the price of the stock. Brett McDonnell, "ESOP's Failures: Fiduciary Duties When Managers of Employee-Owned Companies Vote to Entrench Themselves," 2000 Colum. Bus. L.Rev. 199, 207 (2000). State Street's lawyers thus miss the point in arguing that American Airlines was on the verge of bankruptcy too but managed to avoid it and so might United have done so. Someone who owned American Airlines stock, and would have wanted but was unable to diversify, would have been bearing unwanted risk during American's period of peril.

But because an ESOP is at once a permissible form of ERISA trust and nondiversified by definition, the trustee (along with the named fiduciary) is in an awkward position. If he diversifies he violates the plan, but if he doesn't diversify he may be imposing unwanted risk on the employee-shareholders—for it is unrealistic to suppose that the ESOP form was chosen because the employees wanted to bear unnecessary risk. The goal of an ESOP is to give employees not the thrills of gambling but a larger stake in the company's fortunes, see, e.g., United States v. McCord, 33 F.3d 1434, 1440 n. 6 (5th Cir.1994), in hopes of "broadening ownership of capital within a capitalist system" to the end of "encouraging capital formation through an improved form of finance, improved management-labor relations, and increased productivity." McDonnell, supra, at 220-21. Why Congress thought it appropriate to stimulate the adoption of this business form by giving companies tax breaks rather than allowing the free market to determine the merits of the form is a separate question but not one we need answer in order to decide this case, although it is pertinent to note that one of the goals not sought to be achieved by the ESOP form is the funding of pension benefits. Michael A. Conte & Jan Svejnar, "The Performance Effects of Employee Ownership Plans," in Paying for Productivity 143, 143-44 (Alan S. Blinder ed.1990). It is a goal that the form is ill suited to attain because of its underdiversification.

The tension between the goal of protection against risk and the goal of a portfolio dominated by a single stock is not acute if the participants in the ESOP have adequate sources of income or wealth that are not correlated with the risk of that stock, so that the ESOP is not their primary financial asset. But if they don't have substantial other wealth the goals cannot be reconciled, though they can be compromised by requiring fiduciaries to begin diversifying the ESOP's assets at the point at which an increase in the riskiness of the assets, had it been foreseen, would have induced the creators of the ESOP either to have not created it at all or to have required at least partial diversification. (We say begin diversifying because if the stock held by the ESOP is a very large fraction of the outstanding stock of the employer, as in this case, the sell-off would have to be gradual in order to avoid precipitating a sharp drop in price, Louis K. Chan & Josef Lakonishok, "The Behavior of Stock Prices Around Institutional Trades," 50 J. Finance 1147, 1147-48 (1995); the market does not treat the stocks of different companies as being perfectly substitutable. Thus, State Street had a policy of limiting any day's sales of a given stock to 30 percent of the volume of sales that day.) Or, as Kuper v. Iovenko, supra, 66 F.3d at 1459, puts it, "the plaintiff must show that the ERISA fiduciary could not have reasonably believed that the plan's drafters would have intended under the circumstances that he continue to comply with the ESOP's direction that he invest exclusively in employer securities." See also Moench v. Robertson, 62 F.3d 553, 571-72 (3d Cir.1995).

In Steinman v. Hicks, 352 F.3d 1101 (7th Cir.2003), we noted a situation in which the duty of prudence could require diversification of an ESOP's holdings: if the "ESOP was [the employees'] principal retirement asset . . . and was entirely invested in the stock of their employer..., and their employer was bought in a stock-for-stock deal—so that all the assets of the ESOP became stock in the acquirer by a company that had a much higher debt-equity ratio than their (former) employer and as a result its stock price was much more volatile and its bankruptcy risk greater. Then, even if the trustees did not predict the company's `impending collapse,' they might be required in the interest of the participants either to diversify the plan's stockholdings or to exchange the...stock for Treasury bills." Id. at 1106 (emphasis added, citation deleted). The source of the duty to diversify would not be the trustee's disagreement with the market valuation (their failure to predict the company's impending collapse), but the excessive risk imposed on employee-shareholders by the rise in the debt-equity ratio of the employer's stock as a result, in the example given in Steinman, of a merger and in our case of a plummeting stock price. How excessive would depend in the first instance on the amount and character of the employees' other assets, for, as we have already indicated, it is the riskiness of one's portfolio, not of a particular asset in the portfolio, that is important to the risk-averse investor.

The plaintiffs never sought to explore these issues. So even if the methods of litigation could feasibly determine the point at which the ESOP trustee should sell in order to protect the employee-shareholders against excessive risk, the plaintiffs have made no effort to establish that point. They think that what State Street did wrong was to fail to second-guess the market; in fact what State Street may well have done wrong was to delay selling its United stock until too late to spare the employee-shareholders from bearing inordinate risk. Of course, if State Street had sold earlier and the stock had then bounced back, as American Airlines' stock did, State Street might well have been sued by the same plaintiffs, though the analysis presented in this opinion would have bailed it out.

There has thus been a failure of proof. But the plaintiffs can take some solace from the fact that determining the "right" point, or even range of "right" points, for an ESOP fiduciary to break the plan and start diversifying may be beyond the practical capacity of the courts to determine. The Department of Labor pamphlet that we cited earlier states that a directed trustee may have a duty to sell "where there are clear and compelling public indicators, as evidenced by an 8-K filing with the Securities and Exchange Commission (SEC), a bankruptcy filing or similar public indicator, that call into serious question a company's viability as a going concern." U.S. Dept. of Labor, supra, at 5-6. That is not an administrable standard; note the hedge in "may" and the fact that selling when bankruptcy is declared will almost certainly be too late.

The time may have come to rethink the concept of an ESOP, a seemingly inefficient method of wealth accumulation by employees because of the underdiversification to which it conduces (though remember that what is important is the diversification of the employee's entire asset portfolio, including his earning capacity, rather than whether an individual asset is diversified). The tax advantages of the form do not represent a social benefit, but merely a shift of tax burdens to other taxpayers. Nor are we aware of an argument for subsidizing the ESOP form, as the tax law does, rather than letting the market decide whether it has economic advantages over alternative forms of business structure. As for the notion that having a stake in one's employer will induce one to be more productive, the evidence for such an effect—see "Motivating Employees with Stock and Involvement," NBER Website, Apr. 25, 2006, http://www.nber.org /digest/may04/ w10177.html; Joseph Blasi, Michael Conte & Douglas Kruse, "Employee Stock Ownership and Corporate Performance Among Public Companies," 50 Indus. & Lab. Rel. Rev. 60 (1996)—is weak and makes no theoretical sense. An employee has no incentive to work harder just because he owns stock in his employer, since his efforts, unless he is a senior executive, are unlikely to move the price of the stock. Nor is employee stock ownership likely to forge sentimental ties between employees and employers that might cause the former to work harder, although it may alleviate union pressures for wages or benefits that would jeopardize solvency.

A second ground of appeal argued by the plaintiffs is that the district judge should not have prevented their expert witness, Lucian Morrison, from testifying about State Street's prudence or lack thereof in failing to sell United stock sooner. The judge thought Morrison unqualified because he lacks either a degree in economics or experience with bankruptcy. That was error. Morrison is a highly experienced trust officer and as such qualified to testify about State Street's management of the assets of the United ESOP. "Rule 702 [of the Federal Rules of Evidence] specifically contemplates the admission of testimony by experts whose knowledge is based on experience." Walker v. Soo Line R.R., 208 F.3d 581, 591 (7th Cir.2000); see also Smith v. Ford Motor Co., 215 F.3d 713, 718 (7th Cir.2000); Hangarter v. Provident Life & Accident Ins. Co., 373 F.3d 998, 1015-16 (9th Cir.2004). But the error is harmless. Morrison was planning to testify that State Street should have seen the handwriting on the wall back in October or, at the latest, November of 2001. That is an echo of the argument by the plaintiffs' lawyers that State Street should have outsmarted the market and is not a correct interpretation of the duty of prudent management of trust funds, whatever a trust officer might think.

State Street has filed a contingent cross-appeal, challenging a settlement between the plaintiffs and its co-fiduciary, the UAL Corporation ESOP Committee. Since we are affirming the dismissal of the plaintiffs' claim against State Street, there is no imperative need to discuss the cross-appeal; it is academic unless this court sitting en banc, or the Supreme Court, reverses our decision in favor of State Street. But the oddity of the situation presented by that appeal calls for some comment. The Committee's members were insured, and the settlement gives the plaintiffs $5.2 million, the limit of the insurance policy after deduction of incurred and expected legal expenses. The settlement provides that if State Street is determined to be liable for some amount of damages, the judge shall determine its fault relative to the Committee's and State Street may then seek contribution from the Committee's members. Suppose that State Street was ordered to pay the plaintiffs $540 million and the judge decided that it and the Committee were equally at fault. Then State Street would be entitled to a contribution of $264.8 million ($540 ÷ 2 = $270-$5.2 = $264.8) from the Committee. But the Committee's members, we are told by their lawyer, have no assets beyond the insurance policy out of which to satisfy a judgment. This is probably an exaggeration, but a slight one. Suppose improbably that the six members have a total of $4.8 million in personal assets that might be used to satisfy a judgment. Then State Street, even though adjudged only 50 percent responsible for the (supposed) wrongful injury to the plaintiffs, would have to pay almost the entire judgment—$530 million ($270 million + $270 million—$ 5.2 million (paid by insurance company)—$4.8 million (paid by the Committee members) = $530 million) out of $540 million.

The judge in approving the settlement entered an order in effect barring State Street from seeking anything from the Committee members beyond what personal assets they may have. The judge gave no reason for this action; if we may judge from the arguments of the parties' lawyers, either he thought this result "fair" or, as argued by the plaintiffs' lawyer, thought it would make no difference because State Street could in no event obtain from the Committee members money they don't have. But if this is what the judge thought, he was wrong. The members' unions have agreed to indemnify them. Therefore if State Street obtained a judgment of $264.8 million (in our example) against the members, the unions would have to pay so much of that judgment as they, not the Committee members, could afford. (So we're surprised that the union has not sought to intervene in the suit.)

The problem for which bar orders or, as they are sometimes called, "settlement bars" are a suggested solution was explained recently in Denney v. Deutsche Bank AG, 443 F.3d 253, 273 (2d Cir.2006): "If a nonsettling defendant against whom a judgment had been entered were allowed to seek payment from a defendant who had settled, then settlement would not bring the latter much peace of mind. He would remain potentially liable to a nonsettling defendant for an amount by which a judgment against a nonsettling defendant exceeded a nonsettling defendant's proportionate fault. This potential liability would surely diminish the incentive to settle." But to the extent that the nonsettling defendant has a valid claim for contribution from the settling defendant, a bar order that extinguished that claim would encourage a race to settle (to the arbitrary disadvantage of the defendants as a group), since each settling defendant's liability would be capped at the amount of his settlement. This court, in a case not cited by the district judge, has declined to adopt the settlement-bar approach on the ground that the hearing necessary to determine the value of the nonsettling defendant's contribution claim would be cumbersome and therefore costly, and "as the costs of settlement rise closer to those of trial, the likelihood of settlement falls—maybe far enough to offset the incentive to settle that a defendant has who knows that settling will enable him to avoid all liability to the other tortfeasors." Donovan v. Robbins, 752 F.2d 1170, 1181 (7th Cir.1985).

The settlement-bar approach assumes, moreover, that there is a right of contribution, and it is unsettled whether ERISA defendants have such a right. Lumpkin v. Envirodyne Industries, Inc., 933 F.2d 449, 464 n. 10 (7th Cir.1991); compare Chemung Canal Trust Co. v. Sovran Bank/Maryland, 939 F.2d 12, 16-17 (2d Cir.1991) (yes, contribution), with Kim v. Fujikawa, 871 F.2d 1427, 1432-33 (9th Cir. 1989) (no). We assumed in Alton Memorial Hospital v. Metropolitan Life Ins. Co., 656 F.2d 245, 250 (7th Cir.1981), that they do, but did not actually discuss the question, which remains an open one in this circuit. This is not the case in which to close it, as it would not affect our result; we mention it merely in the hope of heading off such errors in the future.

The judgment in No. 05-4005 is affirmed, and appeal No. 05-4317 is dismissed.