MENARD, INC. and John R. Menard, Jr., Petitioners-Appellants,
v.
COMMISSIONER OF INTERNAL REVENUE, Respondent-Appellee.

*621 Robert E. Dallman (argued), Reinhart, Boerner, Van Deuren, Milwaukee, WI, for Petitioners-Appellants.

Bethany B. Hauser (argued), Civil Div., Immigration Litigation, Gilbert S. Rothenberg, Deputy Assistant Attorney, Department of Justice, Washington, DC, for Respondent-Appellee.

Before EASTERBROOK, Chief Judge, and POSNER and WILLIAMS, Circuit Judges.

POSNER, Circuit Judge.

The Internal Revenue Code allows a business to deduct from its taxable income a "reasonable allowance for salaries or other compensation for personal services actually rendered," 26 U.S.C. § 162(a)(1), or, as Treas. Reg. § 1.162-7(a) adds, for "payments purely for services." Occasionally the Internal Revenue Service challenges the deduction of a corporate salary on the ground that it's really a dividend. A dividend, like salary, is taxable to the recipient, but unlike salary *622 is not deductible from the corporation's taxable income. So by treating a dividend as salary, a corporation can reduce its income tax liability without increasing the income tax of the recipient. At least that was true in 1998, the tax year at issue in this case. As a result of a change in law in 2003, dividends are now taxed at a lower maximum rate than salaries—15 percent, versus 35 percent for salary. 26 U.S.C. § 1(h)(11). This makes the tradeoff more complex; although the corporation avoids tax by treating the dividend as a salary, which is deductible, the employee pays a higher tax.

But depending on its tax bracket, the corporation may still save more in tax than the employee pays, and in that event, if the employee owns stock in the corporation, he may, depending on how much of the stock he owns, prefer dividends to be treated as salary. Menards' tax bracket in 1998 was, its brief tells us without contradiction, 35 percent. Had the new law been in effect then, the corporation, if unable to deduct the $17.5 million bonus, would have paid $6.1 million in additional income tax, while Mr. Menard, had he received the bonus as a dividend and thus paid 15 percent rather than 35 percent of it in tax, would have saved only $3.5 million.

Even before the change in the Internal Revenue Code, treating a dividend as salary was less likely to be attempted in a publicly held corporation, because if the CEO or other officers or employees receive dividends called salary beyond what they are entitled to by virtue of owning stock in the corporation, the other share-holders suffer. But in a closely held corporation, the owners might decide to take their dividends in the form of salary in order to beat the corporate income tax, and there would be no one to complain— except the Internal Revenue Service.

The usual case for forbidding the reclassification (for tax purposes) of dividends as salary is thus that "of a corporation having few shareholders, practically all of whom draw salaries," Treas. Reg. § 1.162-7(b)(1), especially if the corporation does not pay dividends (as such) and some of the shareholders do no work for the corporation but merely cash a "salary" check. A difficult case—which is this case—is thus that of a corporation that pays a high salary to its CEO who works full time but is also the controlling shareholder. The Treasury regulation defines a "reasonable" salary as the amount that "would ordinarily be paid for like services by like enterprises under like circumstances," § 1.162-7(b)(3), but that is not an operational standard. No two enterprises are alike and no two chief executive officers are alike, and anyway the comparison should be between the total compensation package of the CEOs being compared, and that requires consideration of deferred compensation, including severance packages, the amount of risk in the executives' compensation, and perks.

Courts have attempted to operationalize the Treasury's standard by considering multiple factors that relate to optimal compensation. E.g., Haffner's Service Stations, Inc. v. Commissioner, 326 F.3d 1, 3-4 (1st Cir.2003); Eberl's Claim Service, Inc. v. Commissioner, 249 F.3d 994, 999 (10th Cir.2001); LabelGraphics, Inc. v. Commissioner, 221 F.3d 1091, 1095 (9th Cir.2000); Alpha Medical, Inc. v. Commissioner, 172 F.3d 942, 946 (6th Cir.1999); Rutter v. Commissioner, 853 F.2d 1267, 1271 (5th Cir.1988). (Alpha and Rutter each list nine factors.) We reviewed a number of these attempts in Exacto Spring Corp. v. Commissioner, 196 F.3d 833 (7th Cir.1999), and concluded that they were too vague, and too difficult to operationalize, to be of much utility. Multifactor tests with no weight assigned to any *623 factor are bad enough from the standpoint of providing an objective basis for a judicial decision, Farmer v. Haas, 990 F.2d 319, 321 (7th Cir.1993); Prussner v. United States, 896 F.2d 218, 224 (7th Cir.1990) (en banc); Palmer v. Chicago, 806 F.2d 1316, 1318 (7th Cir.1986); United States v. Borer, 412 F.3d 987, 992 (8th Cir.2005); multifactor tests when none of the factors is concrete are worse, and that is the character of most of the multifactor tests of excessive compensation. They include such semantic vapors as "the type and extent of the services rendered," "the scarcity of qualified employees," "the qualifications. . . of the employee," his "contributions to the business venture," and "the peculiar characteristics of the employer's business."

All businesses are different, all CEOs are different, and all compensation packages for CEOs are different. In Exacto, in an effort to bring a modicum of objectivity to the determination of whether a corporate owner/employee's compensation is "reasonable," we created the presumption that "when . . . the investors in his company are obtaining a far higher return than they had any reason to expect, [the owner/employee's] salary is presumptively reasonable." But we added that the presumption could be rebutted by evidence that the company's success was the result of extraneous factors, such as an unexpected discovery of oil under the company's land, or that the company intended to pay the owner/employee a disguised dividend rather than salary. 196 F.3d at 839. The strongest ground for rebuttal, which brings us back to the basic purpose of disallowing "unreasonable" compensation, is that the employee does no work for the corporation; he is merely a shareholder. See id.; cf. General Roofing & Insulation Co. v. Commissioner, T.C. Memo 1981-667, 15-17 (1981). Other types of evidence that might rebut the Exacto presumption include evidence of a conflict of interest, though we'll see that such evidence is not always decisive. Also relevant is the relation between the executive's compensation that is challenged and the compensation of other executives in the company; for useful discussions see Rapco, Inc. v. Commissioner, 85 F.3d 950, 954-55 (2d Cir.1996), and Elliotts, Inc. v. Commissioner, 716 F.2d 1241 (9th Cir. 1983).

Comparison with the compensation of executives of other companies can be helpful if—but it is a big if—the comparison takes into account the details of the compensation package of each of the compared executives, and not just the bottom-line salary. This qualification will turn out to be critical in this case. For the Tax Court acknowledged that the presumption of reasonableness had been established but thought it rebutted by evidence that corporations in the same business as Menards paid their CEOs substantially less than Menards paid its CEO.

Menard, Inc. is a Wisconsin firm that under the name "Menards" sells hardware, building supplies, and related products through retail stores scattered throughout the Midwest. It had 138 stores in 1998 and was the third largest retail home improvement chain in the United States; only Home Depot and Lowe's were larger. It was founded by John Menard in 1962, and, at least through 1998, the tax year at issue in this case, he was the company's chief executive. (All the evidence in the record concerns his activities in that year.) Uncontradicted evidence depicts him as working 12 to 16 hours a day 6 or 7 days a week, taking only 7 days of vacation a year, working even while spending "personal time" with his family, involving himself in every detail of his firm's operations, and fixing everyone's compensation. Under *624 his management Menards' revenues grew from $788 million in 1991 to $3.4 billion in 1998 and the company's taxable income from $59 million to $315 million. The company's rate of return on shareholders' equity that year was, according to the Internal Revenue Service's expert, 18.8 percent—higher than that of either Home Depot or Lowe's.

Menard owns all the voting shares in the company and 56 percent of the nonvoting shares, the rest being owned by members of his family, two of whom have senior positions in the company. Like the other executives of Menards, he is paid a modest base salary but participates along with them in a profit-sharing plan. In 1998, his salary was $157,500 and he received a profit-sharing bonus of $3,017,100, and the Tax Court did not suggest that there was anything amiss with these amounts. But the bulk of his compensation was in the form of a "5% bonus" that yielded him $17,467,800, as a result of which his total compensation for the year exceeded $20 million.

The 5% bonus program (5 percent of the company's net income before income taxes) was adopted in 1973 by the company's board of directors at the suggestion of the company's accounting firm, though there is no indication that the firm suggested a number rather than just advising the board that Mr. Menard should have his own bonus plan because of his commanding role in the management of the company. The board at the time included a shareholder who was not a member of Menard's family and he voted for the plan, which was still in force in 1998 and so far as we know had not been reexamined in the interim. That shareholder departed and the board in 1998 consisted of Menard, a younger brother of his who works for the company, and the company's treasurer.

There is no suggestion that any of the shareholders were disappointed that the company obtained a rate of return of "only" 18.8 percent or that the company's success in that year or any year has been due to windfall factors, such as the discovery of oil under the company's head-quarters. But besides thinking Menard's compensation excessive, the Tax Court thought it was intended as a dividend. It thought this because Menard's entitlement to his 5 percent bonus was conditioned on his agreeing to reimburse the corporation should the deduction of the bonus from the corporation's taxable income be disallowed by the Internal Revenue Service or its Wisconsin counterpart and because 5 percent of corporate earnings year in and year out "looked" more like a dividend than like salary.

These are flimsy grounds. Given the fondness of the IRS and the Tax Court for a "totality of the circumstances" approach to determining excessive compensation, it was prudent (and incidentally not in Menard's personal financial interest) for the company to require him to reimburse it should the IRS successfully challenge the deduction. Nor does 5 percent of net corporate income look at all like a dividend. Dividends generally are specified dollar amounts—so many dollars per share— rather than a percentage of earnings. E.g., William L. Megginson, Scott B. Smart & Brian M. Lucey, Introduction to Corporate Finance 436-37 (2008); Harold Bierman, Jr. & Seymour Smidt, Financial Management for Decision Making 489 (2003); Angela Schneeman, The Law of Corporations and Other Business Organizations 320-21 (3d ed.2002); Erich A. Helfert, Financial Analysis Tools and Techniques: A Guide for Managers 122 (2001); Jae K. Shim & Joel G. Siegel, Financial Management 285 (2000). When earnings fall, dividends may be cut, but they are cut from one fixed amount to another rather *625 than made to vary continuously, as a percentage of earnings would do.

Paying a fixed (though occasionally altered) dividend provides the shareholder with a more predictable cash flow than if the dividend varied directly with corporate earnings, which fluctuate from year to year. It thus reduces the risk (variance) associated with ownership of common stock. Moreover, the reason for varying a manager's compensation with the company's profits is to increase his incentive to work intelligently and hard in order to increase those profits, and that reason has no application to a passive owner. Although tying compensation to the market value of the company's stock is criticized by some economists because of the many factors besides managerial effort and ability that influence the price of a publicly traded stock, stock in Menards is not publicly traded; probably it is not traded at all.

The most insignificant factor that the Tax Court thought indicative of a "concealed" dividend was that Menard's 5 percent bonus is paid at the end of each year. Well, it would have to be paid either at the end of the year, when the earnings for the year are known, or in installments throughout the year. Bonuses are usually paid in a lump, once a year, often at Christmas, but that would not be a feasible course for Menards to follow unless it closed for the following week, because its net income for the year wouldn't be determinable until the close of the last day of the year on which the store was open. Bonuses are more likely to be paid in single payments at or near the end of the year than dividends are; dividends usually are paid quarterly. William A. Rini, Fundamentals of the Securities Industry 13 (2003).

The court thought it suspicious that the board of directors that approved the 5 percent bonus in 1998 was controlled by Menard. But it could not be otherwise, since he is the only shareholder who is entitled to vote for members of the board of directors, as he owns all the voting shares in the company. The logic of the Tax Court's position is that a one-man corporation cannot pay its CEO (if he is that one man) any salary! The Tax Court has flirted with that strange logic, as we shall see.

A slightly better candidate for suspicion is that the board of directors had not sought outside advice on what appropriate compensation for Menard would be. But the only point of doing that would have been to provide some window dressing in the event of a challenge by the IRS. Menard doubtless has a strong opinion of what he is worth to his company and would not pay a compensation consultant to disagree.

It might seem that since Menards paid no formal dividend at all, some of Mr. Menard's compensation (and perhaps that of other executives as well) must be a dividend. But that is incorrect, as noted in the Elliotts case that we cited earlier. 716 F.2d at 1244. Many corporations do not pay dividends but instead retain all their earnings in order to have more capital. One reason that publicly held corporations—that is, corporations in which ownership is diffuse, which may enable managers to pursue personal goals at the expense of shareholder welfare—usually do pay dividends is that it helps to discipline management by making it more likely to have to go outside the company for money for new ventures, thus forcing it to convince the capital markets that the ventures are likely to succeed. That reason for paying dividends has no application to a corporation like Menards in which there is an almost complete fusion of management and ownership.

*626 The main focus of the Tax Court's decision was not on the issue of "concealed dividend" (that is, whether the company was acting in good faith in paying $17.5 million as a bonus rather than as a dividend); it was on whether Menard's compensation exceeded that of comparable CEOs in 1998—that is, whether it was objectively excessive, and hence (in part) functionally if not intentionally a dividend rather than a bonus.

The CEO of Home Depot was paid that year only $2.8 million, though it is a much larger company than Menards; and the CEO of Lowe's, also a larger company, was paid $6.1 million. But salary is just the beginning of a meaningful comparison, because it is only one element of a compensation package. Of particular importance to this case is the amount of risk in the compensation structure. Risk in corporate compensation is significant in two respects. First, most people are risk averse, and the scholarly literature on corporate compensation suggests that risk aversion is actually an obstacle to efficient corporate management because managers tend to be more risk averse than shareholders. Shareholders can diversify the risk of a particular company by owning a diversified portfolio, but a manager tends to have most of his financial, reputational, and "specific human" capital tied up in his job. Robert Yalden et al., Business Organizations: Principles, Policies and Practice 698-99 (2008); Lucian Bebchuk & Jesse Fried, Pay Without Performance: The Unfulfilled Promise of Executive Compensation 19 (2006); Lance A. Berger & Dorothy R. Berger, The Compensation Handbook: A State-of-the-Art Guide to Compensation Strategy and Design 386-87 (4th ed.2000); Frank H. Easterbrook & Daniel R. Fischel, The Economic Structure of Corporate Law 99-100 (1991). (By "specific human capital" economists mean the earnings that a worker obtains by virtue of skills, training, or experience specialized to the specific firm that he is working for.) So the riskier the compensation structure, other things being equal, the higher the executive's salary must be to compensate him for bearing the additional risk.

That is not a critical consideration in this case because, as we said, management and ownership in Menards are not divorced. But a second significance of risk in a compensation structure is fully applicable to this case. A risky compensation structure implies that the executive's salary is likely to vary substantially from year to year—high when the company has a good year, low when it has a bad one. Mr. Menard's average annual income may thus have been considerably less than $20 million—a possibility the Tax Court ignored. Had the corporation lost money in 1998, Menard's total compensation would have been only $157,500—less than the salary of a federal judge—even if the loss had not been his fault. The 5 percent bonus plan was in effect for a quarter of a century before the IRS pounced; was it just waiting for Menard to have such a great year that the IRS would have a great-looking case?

Nor did the Tax Court consider the severance packages, retirement plans, or perks of the CEOs with whom it compared Menard (although it did take account of their stock options), even though such extras can make an enormous difference to an executive's compensation. E.g., Lucian Arye Bebchuk & Jesse M. Fried, "Stealth Compensation via Retirement Benefits," 1 Berkeley Bus. L.J. 291 (2004); Phred Dvorak, "Companies Cut Holes in CEOs' Golden Parachutes—New Disclosure Rules Prompt More Criticism of Guaranteed Payouts," Wall St. J., Sept. 15, 2008, p. B4. Just two years after Menard received his questioned $20 million, Robert Nardelli became *627 CEO of Home Depot. In his slightly more than six years in that post he was paid $124 million in salary, exclusive of stock options; and when fired in 2007 (he was unpopular, and during his tenure the market capitalization of Home Depot increased negligibly—only to jump when his firing was announced), he received a much-criticized severance payment of $210 million (including the value of his stock options). He went on to become the CEO of Chrysler, where he is being paid $1 a year, thought by some observers to be generous. We wonder whether the IRS plans to challenge Menard's compensation for the years 2001 to 2006, using Nardelli's compensation package as a basis for comparison.

The Tax Court ruled that any compensation paid Menard in 1998 in excess of $7.1 million was excessive. The $7.1 million figure was arrived at by the following steps: (1) Divide Home Depot's return on investment (16.1 percent) by the compensation of Home Depot's CEO ($2,841,307). (2) Divide Menards' return on investment (18.8 percent) by the result of step (1).(3) Multiply the result of step (2) ($3,317,799) by 2.13, that being the ratio of the compensation of Lowe's' CEO to that of Home Depot's CEO. In words, the court allowed Menard to treat as salary slightly more than twice the salary he supposedly would have had if he had been Home Depot's CEO and if Home Depot had had as high a return on investment as Menards did. The judge's assumption was that rate of return drives CEO compensation except for random factors assumed to have the same effect on Menard's compensation as it did on that of Lowe's' CEO; for Lowe's paid its CEO more than twice as much as Home Depot paid its CEO even though Lowe's was a smaller company and its rate of return was lower.

This was an arbitrary as well as dizzying adjustment. It disregarded differences in the full compensation packages of the three executives being compared, differences in whatever challenges faced the companies in 1998, and differences in the responsibilities and performance of the three CEOs.

We have discussed risk; with regard to responsibilities there is incomplete information about the compensation paid other senior management of Menards besides Mr. Menard himself, and no information about the compensation paid the senior managements of Home Depot or Lowe's other than those companies' CEOs. The relevance of such information is that it might show that Menard was doing work that in other companies is delegated to staff, or conversely that staff was doing all the work and Menard was, in substance though not in form, clipping coupons. The former inference is far more likely, given the undisputed evidence of Menard's workaholic, micromanaging ways and the fact that Menards' board of directors is a tiny dependency of Mr. Menard. He does the work that in publicly held companies like Home Depot or Lowe's is done by boards that have more than two directors besides the CEO. Of course they are larger companies—Home Depot's revenues were seven times as great as Menards' in 1998—so we would expect them to have more staff. But we are given no information on how much more staff they had.

We know that besides Menard himself, Menards—already a $3.4 billion company in 1998—had only three corporate officers. The Tax Court thought it suspicious that they were modestly compensated—their total compensation in 1998 was only $350,000, and the highest-paid employee in the company after Menard himself—the senior merchandise manager—received total compensation of only $468,000. The Tax Court did not consider the possibility, *628 which the evidence supports, that Menard really does do it all himself.

The Tax Court's opinion strangely remarks that because Mr. Menard owns the company he has all the incentive he needs to work hard, without the spur of a salary. In other words, reasonable compensation for Mr. Menard might be zero. How generous of the Tax Court nevertheless to allow Menards to deduct $7.1 million from its 1998 income for salary for Menard!

The Fifth Circuit has commented sensibly on the Tax Court's belief that owners don't need or deserve salaries: "the Tax Court questioned whether an incentive bonus tied to company performance is needed for an employee who is also a shareholder. Apparently, the argument is that such an employee already has sufficient incentive to make the business successful because as a shareholder he will receive the profits of the business anyway. This argument, however, misses the economic realities of the corporate form as taxed under the internal revenue code. For compensation purposes, the shareholder-employee should be treated like all other employees. If an incentive bonus would be appropriate for a nonshareholder-employee, there is no reason why a shareholder-employee should not be allowed to participate in the same manner. In essence, the shareholder-employee is treated as two distinct individuals for tax purposes: an independent investor and an employee." Owensby & Kritikos, Inc. v. Commissioner, 819 F.2d 1315, 1328 (5th Cir.1987); see also Elliotts, Inc. v. Commissioner, supra, 716 F.2d at 1248.

The Tenth Circuit, it is true, remarked in Pepsi-Cola Bottling Co. v. Commissioner, 528 F.2d 176, 182 (10th Cir.1975), that "due to the identity between the predominant shareholder and the employee in our case we cannot accept the applicability of the `incentive compensation' reasoning. Mrs. Joscelyn did not have a lack of such incentive. As owner of 248 of 250 shares she would profit from her hard work even without salary compensation. A bonus contract that might be reasonable if executed with an executive who is not a controlling shareholder may be viewed as unreasonable if made with a controlling shareholder, since incentive to the stockholder to call forth his best effort would not be needed." We do not agree, but we note that the court based its decision on a comparison between Mrs. Joscelyn's compensation and that of executives of other companies, rather than holding that a controlling shareholder may never receive a bonus. That would not make good sense. After all, bonuses do not only, or even primarily, reward motivation; they reward performance.

We conclude that in ruling that Menard's compensation was excessive in 1998, the Tax Court committed clear error, and its decision is therefore

REVERSED.