Elliotts, Inc. (“Taxpayer”), challenges the Tax Court’s determination of deficiencies in its 1975 and 1976 tax returns. It argues that the Tax Court erred in finding that part of the compensation paid Taxpayer’s chief executive and sole shareholder during those years constituted a dividend distribution and thus was not deductible under section 162(a)(1) of the Internal Revenue Code. We reverse.

I

BACKGROUND

Taxpayer is an Idaho corporation that sells equipment manufactured by John Deere Co. and services equipment made by Deere and several other manufacturers. Its principal place of business is Burley, Idaho. During the period relevant here, Taxpayer had business locations in both Burley and Idaho Falls, Idaho. The Idaho Falls location dealt only in industrial equipment; the Burley office sold and serviced a wide range of agricultural and industrial equipment. Out of approximately 168 John Deere agricultural dealers in the zone comprising seven western states, Taxpayer remained, at the time of trial, one of only three dealers handling both agricultural and industrial equipment.

Taxpayer was incorporated in 1952. During its first year, it grossed $500,000 in agricultural equipment sales in the Burley area. It employed about eight people at that time. By 1975, Taxpayer was employing 40 people, selling both agricultural and industrial equipment throughout southeast Idaho, and achieving gross annual sales in excess of $5 million.

Edward G. Elliott has been Taxpayer’s chief executive officer since its incorporation and he has also been its sole shareholder since 1954. He has always had total managerial responsibility for Taxpayer’s business. In addition to being Taxpayer’s ultimate decision and policy maker, he has performed the functions usually delegated to sales and credit managers. It is undisputed that he works about 80 hours each week.

For several years, Taxpayer has paid Elliott a fixed salary of $2000 per month plus a bonus at year’s end. Since Taxpayer’s incorporation, Elliott’s bonus has been fixed at 50% of net profits (before subtraction of taxes and management bonuses).

On its return for the fiscal year ending February 28, 1975, Taxpayer claimed a $181,074 deduction for total compensation paid Elliott. It claimed a similar $191,663 deduction on its return for the fiscal year ending February 28, 1976. The Commissioner of Internal Revenue (“Commissioner”) found these deductions to be in excess of the amounts Taxpayer properly could deduct as reasonable salary under section 162(a)(1). On June 16, 1978, the Commissioner issued Taxpayer a notice of deficiency which limited deductions for Elliott’s salary to $65,000 for each fiscal year.

Taxpayer petitioned the Tax Court for a redetermination of liability. The court, after reviewing the testimony and statistical evidence, concluded that the payments to Elliott, in addition to providing compensation for personal services, were intended in part to distribute profits. Although the Tax Court acknowledged that it could not determine what amounts paid Elliott actually were dividends, it found that the total amounts paid him were in excess of reasonable compensation. It determined that $120,000 was reasonable compensation for the year 1975 and that $125,000 was reasonable for 1976. The deficiencies assessed to Taxpayer by the Commissioner were reduced accordingly. Taxpayer appeals the Tax Court’s determination of reasonable compensation.

II

THE SHAREHOLDER-EMPLOYEE PROBLEM

The issue presented by this case concerns the deductibility by a corporation of payments ostensibly made as compensation for services to an employee who is also a shareholder. If the payments are reasonable compensation for services rendered, the corporation may deduct them. 26 U.S.C. § 162 (a)(1). If, however, they are actually dividends, they are not deductible. Thus, it will normally be in a corporation’s interest to characterize such payments as compensation rather than dividends.

The general problem is that of distinguishing between dividends and compensation for services received by a shareholder-employee of a closely held corporation. What makes this situation troublesome is that the shareholder-employee and the corporation are not dealing with each other at arm’s length. It is likely to be in the interests of both the corporation and the shareholder-employee to characterize any payments to the shareholder-employee as compensation rather than dividends. For this reason, a taxpayer’s characterization of such payments may warrant close scrutiny to ensure that a portion of the purported compensation payments is not a disguised dividend. See Nor-Cal Adjusters v. Commissioner, 503 F.2d 359, 361 (9th Cir.1974).

The problem of determining whether compensation payments contain an element of disguised dividend is exacerbated in a case such as this one where the shareholder-employee is the corporation’s sole shareholder. Not only is a sole shareholder likely to have complete control over the corporation’s operations, he will also be the only recipient of its dividends. If a corporation has multiple shareholders, the existence of a plan which compensates shareholder-employees in proportion to their ownership interests may be evidence that compensation payments contain disguised dividends. In the case of a sole shareholder, such evidence is meaningless.

Section 162(a)(1) of the Internal Revenue Code permits a corporation to deduct “a reasonable allowance for salaries or other compensation for personal services actually rendered.” There is a two-prong test for deductibility under section 162(a)(1): (1) the amount of the compensation must be reasonable and (2) the payments must in fact be purely for services. Treas.Reg. § 1.162-7(a) (1960); Nor-Cal, 503 F.2d at 362 .

Proof of the second prong, which requires a “compensatory purpose,” can be difficult to establish because of its subjective nature. See Note, Reasonable Compensation and the Close Corporation: McCandless, the Automatic Dividend Rule, and the Dual Level Test, 26 Stan.L.Rev. 441, 447 (1974) (hereafter “Note, Reasonable Compensation ”). The existence of a compensatory purpose can often be inferred if the amount of the compensation is determined to be reasonable under the first prong. For these reasons, courts generally concentrate on the first prong — whether the amount of the purported compensation is reasonable. See, e.g., Pacific Grains, Inc. v. Commissioner, 399 F.2d 603 (9th Cir.1968); Note, Reasonable Compensation, 26 Stan.L.Rev. at 447; Coggin, The Status of the McCandless Doctrine, 55 Taxes 720 , 720 (Nov. 1977) (hereafter “Coggin”). Courts have generally not delved into whether a compensatory purpose exists under the second prong except in those rare cases where the Commissioner has come forward with evidence that purported compensation payments, although reasonable in amount, were in fact disguised dividends. See, e.g., Klamath Medical Serv. Bureau v. Commissioner, 29 T.C. 339, 348-49 (1957), affd, 261 F.2d 842 (9th Cir.1958), cert. denied, 359 U.S. 966 , 79 S.Ct. 877 , 3 L.Ed.2d 834 (1959). See generally Note, Reasonable Compensation, 26 Stan.L. Rev. at 447 & n. 35. By and large, the inquiry under section 162(a)(1) has turned on whether the amounts of the purported compensation payments were reasonable.

One court has departed from this practice of restricting the inquiry in most cases to the reasonableness of the payments. In Charles MeCandless Tile Serv. v. United States, 422 F.2d 1336 , 191 Ct.Cl. 108 (1970), the Court of Claims held that ostensible compensation payments paid to two shareholder-employees, even though reasonable in amount, “necessarily” contained disguised dividends because the closely held corporation had been profitable and had not paid out any dividends since its formation. Id. at 1339-40. This has become known as the “automatic dividend rule,” and has been subjected to much criticism. E.g., Coggin, 55 Taxes 720 ; Walthall, MeCandless — Im plications for Compensation Planning and Dividend Policy, 6 Cum.L.Rev. 1 (1975) (hereafter “Walthall”); Note, Reasonable Compensation, 26 Stan.L.Rev. 441.

We reject the automatic dividend rule of MeCandless for several reasons. First, there is no statute requiring profitable corporations to pay dividends. Congress has chosen to handle abuses in this area through the accumulated earnings tax. 26 U.S.C. §§ 531-537 ; see Walthall, 6 Cum.L. Rev. at 16-19; Note, Reasonable Compensation, 26 Stan.L.Rev. at 449-50. Beyond the penalties contained in the accumulated earnings tax, Congress has not indicated that it wants the Commissioner or the courts to require the payment of dividends as a matter of federal tax policy. See Casey v. Commissioner, 267 F.2d 26, 30 (2d Cir.1959); Laure v. Commissioner, 70 T.C. 1087, 1098 (1978).

Second, the automatic dividend rule is based on the faulty premise that shareholders of a profitable corporation will demand dividends. See MeCandless, 422 F.2d at 1339-40. Shareholders are generally concerned with the return on their investment. While some shareholders may prefer to see their return in the form of dividends, others will prefer to have the corporation reinvest its profits so that their return will be in the form of appreciation and the potential of greater future return. See, Note, Reasonable Compensation, 26 Stan.L.Rev. at 450-53. If the shareholders prefer to have their return in the form of appreciation rather than dividends, there is nothing in the law precluding the corporation from reinvesting its profits.

Third, it may well be in the best interests of the corporation to retain and reinvest its earnings. As the Supreme Court has noted, “Directors of a closely held, small corporation must bear in mind the relatively limited access of such an enterprise to capital markets. This may require a more conservative policy with respect to dividends than would be expected of an established corporation with securities listed on national exchanges.” United States v. Byrum, 408 U.S. 125, 140, 92 S.Ct. 2382, 2392 , 33 L.Ed.2d 238 (1972). Not only may retained earnings be the most rational source of financing for a small corporation, “[sjuppliers or outside lenders may insist upon the retention of earnings as a cushion for the credit which they extend or to insure that the firm has sufficient operating funds to continue functioning as a going concern.” Walthall, 6 Cum.L.Rev. at 15.

For these reasons, we will not presume an element of disguised dividend from the bare fact that a profitable corporation does not pay dividends.

In determining the deductibility of compensation payments paid to shareholder-employees, we will continue to concentrate on the reasonableness of those payments. In the rare case where there is evidence that an otherwise reasonable compensation payment contains a disguised dividend, the inquiry may expand into compensatory intent apart from reasonableness. But where, as here, the evidence focuses only on reasonableness and the failure to pay dividends, and there is no other evidence of an intent to hide dividends in compensation payments, our inquiry will be confined to the reasonableness issue. The inquiry into reasonableness is a broad one and will, in effect, subsume the inquiry into compensatory intent in most cases.

In evaluating the reasonableness of compensation paid to a shareholder-employee, particularly a sole shareholder, it is helpful to consider the matter from the perspective of a hypothetical independent investor. A relevant inquiry is whether an inactive, independent investor would be willing to compensate the employee as he was compensated. The nature and quality of the services should be considered, as well as the effect of those services on the return the investor is seeing on his investment. The corporation’s rate of return on equity would be relevant to the independent investor in assessing the reasonableness of compensation in a small corporation where excessive compensation would noticeably decrease the rate of return.

Bearing in mind the preceding discussion,. we now turn to the reasonableness of the compensation paid by Taxpayer to Elliott.

Ill

REASONABLENESS DETERMINATION

Section 162(a)(1) provides that a taxpayer may deduct “all the ordinary and necessary expenses paid or incurred during the taxable year in carrying on any trade or business,” including “a reasonable allowance for salaries or other compensation for personal services actually rendered.” 26 U.S.C. § 162 (a)(1). That Elliott actually rendered services as an employee of Taxpayer is not disputed. At issue is whether the payments made to Elliott are attributable to that employment relationship or to his role as Taxpayer’s sole shareholder. Our inquiry focuses on whether the Tax Court, in finding that a part of the payments made to Elliott could not be attributed to his employment status, correctly defined and applied the factors that determine what is “reasonable compensation” under section 162(a)(1).

Although we accord deference to the Tax Court’s special expertise, definition of the appropriate factors is reviewable by this court as a question of law. See Keller Street Development Co. v. Commissioner, 688 F.2d 675, 678 (9th Cir.1982); Sibla v. Commissioner, 611 F.2d 1260, 1262-63 (9th Cir.1980). The Tax Court’s findings of fact, derived from application of the appropriate factors, must be affirmed unless clearly erroneous. Keller Street, 688 F.2d at 678 ; Estate of Skaggs v. Commissioner, 672 F.2d 756, 757 (9th Cir.1982), cert. denied, - U.S. -, 103 S.Ct. 448 , 74 L.Ed.2d 603 (1982).

Our cases have defined a number of factors that are relevant to this attribution determination, “with no single factor being decisive of the question.” Pacific Grains, 399 F.2d at 606 . For analytical purposes, these factors may be divided into five broad categories.

A. Role in Company

The first category of factors concerns the employee’s role in the taxpaying company. Relevant considerations include the position held by the employee, hours worked, and duties performed, American Foundry v. Commissioner, 536 F.2d 289, 291-92 (9th Cir.1976), as well as the general importance of the employee to the success of the company, American Foundry, 536 F.2d at 291-92 . If the employee has received a large salary increase, comparing past duties and salary with current responsibilities and compensation also may provide significant insights into the reasonableness of the compensation scheme. Pacific Grains, 399 F.2d at 605, 607 .

The Tax Court found that Elliott worked 80 hours per week, performed the functions of general manager, sales manager, and credit manager, and made all policy decisions concerning the parts and service department. Elliotts, Inc. v. Commissioner, 40 T.C.M. (CCH) 802, 804 (1980). These are all appropriate considerations. The Tax Court also considered Elliott’s qualifications and found that although he was a “capable executive” he had no “special expertise.” Id. at 813 . The Tax Court did not seem, however, to consider Elliott’s extreme personal dedication and devotion to his work. See id. at 811 . To the extent that this benefited the corporation, it is surely something for which an independent shareholder would have been willing to compensate Elliott. Because we reverse and remand on other grounds, we point this out so that the Tax Court may consider it on remand.

B. External Comparison

The second set of relevant factors is a comparison of the employee’s salary with those paid by similar companies for similar services. See Hoffman Radio Corp. v. Commissioner, 177 F.2d 264, 266 (9th Cir.1949); E. Wagner & Son v. Commissioner, 93 F.2d 816, 819 (9th Cir.1937).

The Tax Court did compare Elliott’s compensation to that of managers at other John Deere dealers. 40 T.C.M. at 813 . In making these comparisons, it appears that the Tax Court considered that Elliott was performing the functions that two or three people performed at other dealers. This was correct. Such comparisons should be made on the basis of services performed. See E. Wagner, 93 F.2d at 818 ; Treas.Reg. § 1.162-7(b)(3). If Elliott was performing the work of three people, the relevant comparison would be the combined salaries of those three people at another dealer.

C. Character and Condition of Company

The third general category of factors concerns the character and condition of the company. The focus under this category may be on the company’s size as indicated by its sales, net income, or capital value. See E. Wagner, 93 F.2d at 819 ; General Water Heater Corp. v. Commissioner, 42 F.2d 419, 420 (9th Cir.1930). Also relevant are the complexities of the business and general economic conditions. To the extent that they are relevant to this case, the Tax Court did adequately consider these factors.

D. Conflict of Interest

The fourth category focuses on those factors that may indicate a conflict of interest. The primary issue within this category is whether some relationship exists between the taxpaying company and its employee which might permit the company to disguise nondeductible corporate distributions of income as salary expenditures deductible under section 162(a)(1). Such a potentially exploitable relationship may exist where, as in this case, the employee is the taxpaying company’s sole or controlling shareholder, see Nor-Cal, 503 F.2d at 361 ; Pacific Grains, 399 F.2d at 607 ; E. Wagner, 93 F.2d at 817, 819 , or where the existence of a family relationship indicates that the terms of the compensation plan may not have been the result of a free bargain, Harolds Club v. Commissioner, 340 F.2d 861, 865 (9th Cir.1965). Other factors also may point toward an attempt to distribute income through “compensation,” including the existence of a bonus system that distributes all or nearly all of the company’s pre-tax earnings, Nor-Cal, 503 F.2d at 362 ; Klamath Medical, 261 F.2d at 846 ; Sunset Scavenger Co. v. Commissioner, 84 F.2d 453, 455 (9th Cir.1936), that amounts to a disproportionately large percentage of gross income when combined with salary, Pacific Grains, 399 F.2d at 607 ; General Water Heater, 42 F.2d at 420 , or that provides large bonuses to owner-executives, but none to non-owner management, Nor-Cal, 503 F.2d at 361 .

In this case, where Elliott was the sole shareholder, the sort of relationship existed that warrants scrutiny. The mere existence of such a relationship, however, when coupled with an absence of dividend payments, does not necessarily lead to the conclusion that the amount of compensation is unreasonably high. Further exploration of the situation is necessary.

In such a situation, as discussed earlier, it is appropriate to evaluate the compensation payments from the perspective of a hypothetical independent shareholder. If the bulk of the corporation’s earnings are being paid out in the form of compensation, so that the corporate profits, after payment of the compensation, do not represent a reasonable return on the shareholder’s equity in the corporation, then an independent shareholder would probably not approve of the compensation arrangement. If, however, that is not the case and the company’s earnings on equity remain at a level that would satisfy an independent investor, there is a strong indication that management is providing compensable services and that profits are not being siphoned out of the company disguised as salary.

During the fiscal year ending February 28,1975, Taxpayer reported equity of $415,-133 and net profits (profits less taxes and compensation paid Elliott) of $88,969 — a return of 21%. For fiscal year 1976, Taxpayer reported equity of $513,429 and net profits of $98,297 — a return of 19%. 40 T.C.M. at 804 . Thus, the average rate of return on equity during these years was 20%. The Tax Court failed to consider the significance of this data. It seems clear, however, that this rate of return on equity would satisfy an independent investor and would indicate that Taxpayer and Elliott were not exploiting their relationship.

The Tax Court erred by limiting its analysis in this area to the facts that Elliott was Taxpayer’s sole shareholder and Taxpayer paid no dividends. These are relevant factors, but they cannot be viewed in isolation. Taxpayer’s no-dividend policy does not by itself demonstrate that the relationship between Taxpayer and Elliott was being exploited.

E. Internal Consistency

Finally, evidence of an internal inconsistency in a company’s treatment of payments to employees may indicate that the payments go beyond reasonable compensation. Bonuses that have not been awarded under a structured, formal, consistently applied program generally are suspect, see Nor-Cal, 503 F.2d at 362 , as are bonuses consistently designated in amounts tracking either the percentage of the recipient’s stock holdings, id; General Water Heater, 42 F.2d at 420 , or some type of tax benefit, Pacific Grains, 399 F.2d at 607 (bonus adjusted according to available surtax exemption). Similarly, salaries paid to controlling shareholders are open to question if, when compared to salaries paid non-owner management, they indicate that the level of compensation is a function of ownership, not corporate management responsibility. Sunset Scavenger Co., 84 F.2d at 456 . On the other hand, evidence of a reasonable, longstanding, consistently applied compensation plan is evidence that the compensation paid in the years in question was reasonable.

There was evidence in this case of a longstanding, consistently applied compensation plan. Since Taxpayer’s incorporation, it had paid to Elliott an annual bonus equal to 50% of its net profits. Taxpayer contended before the Tax Court that because the yearly bonuses paid Elliott were derived from a predetermined formula that had been in use for over 20 years, it could be inferred that the bonuses constituted compensation for services rather than a dividend distribution. It noted that under the bonus formula Elliott’s salary in some prior years had been too low to compensate him for the services he had rendered to Taxpayer, so that the higher salaries in the years in issue “resulted in average reasonable compensation during the 10-year period . . . 1968 through 1978.” 40 T.C.M. at 809-10 .

In considering the significance of this longstanding bonus formula, the Tax Court erred in several respects.

The Tax Court failed to consider the reasonableness of the contingent formula itself; it concentrated instead on the amounts paid under the formula in two particular years. Such a formula may overcompensate in good years and undercompensate in bad years. This feature, however, does not necessarily make the formula unreasonable. It is permissible to pay and deduct compensation for services performed in prior years. Lucas v. Ox Fibre Brush Co., 281 U.S. 115, 119 , 50 S.Ct. 273, 274 , 74 L.Ed. 733 (1930).

Whether payments under such a formula are reasonable will depend on whether the formula is reasonable. The reasonableness of a longstanding formula should not be determined on the basis of just one or two years. Taxpayer persuasively argues that, accepting the Tax Court’s determination of reasonable compensation for 1974 and 1975, Elliott was severely undercompensated in terms of constant dollars in six of the seven preceding years.

Also relevant to the reasonableness of the formula used in this case is the return on equity an independent investor would have achieved. A formula which would not allow a reasonable return on equity is likely to be unreasonable. In this case, however, as discussed earlier, the return on equity in the years at issue was about 20%.

Over the long run, such a formula should reasonably compensate for the work done, the performance achieved, the responsibility assumed, and the experience and dedication of the employee. At the same time, it should not stand in the way of a satisfactory return on equity.

The Tax Court also discounted the significance of the bonus formula on the ground that “[n]o special incentive is necessary to insure [Elliott’s] best efforts since he would ‘receive the fruits of success through his status as the majority shareholder.’ ” 40 T.C.M. at 811 , quoting Charles Schneider & Co. v. Commissioner, 500 F.2d 148, 153 (8th Cir.1974), cert. denied, 420 U.S. 908 , 95 S.Ct. 826 , 42 L.Ed.2d 837 (1975). To the extent that this implied that incentive payment plans for shareholder-employees are unreasonable, it was error.

Incentive payment plans are designed to encourage and compensate that extra effort and dedication which can be so valuable to a corporation. There is no reason a shareholder-employee should not also be entitled to such compensation if his dedication and efforts are instrumental to the corporation’s success. In this case, there is no doubt that Elliott’s extreme dedication and hard work were valuable to Taxpayer. If an outside investor would approve of such a compensation plan, that plan is probably reasonable. The fact that the recipient is a shareholder-employee does not make the plan unreasonable.

IV

CONCLUSION

We reverse and remand to the Tax Court for reconsideration in light of this opinion. On remand, the Tax Court should begin its analysis by looking at the reasonableness of the compensation payments and should consider the bonus payments in the context of the reasonableness of the formula used. It should not be assumed solely from Elliott's role as sole shareholder and the absence of dividends that the compensation payments necessarily contained dis guised dividends. These are just two of many factors to be considered..

REVERSED and REMANDED.

. Taxpayer sold its Idaho Falls business on March 1, 1975. It has since done business only from its Burley location. . The recipient is taxed at ordinary income rates on both dividends and wages. If the recipient is a significant shareholder, his interest in the corporation may cause him to prefer to characterize such payments as compensation. Moreover, for payments made between 1971 and 1981, as is the case here, a high-income recipient has a strong incentive to characterize such payments as compensation rather than dividends: pre-1982 dividends are taxable at a maximum rate of 70% while the maximum tax rate for wages received between 1971 and 1981 is 50%. 26 U.S.C. § 1 (amended 1983) and § 1348 (repealed 1981). (Since 1982, the maximum tax rate for both wages and dividends has been 50%. 26 U.S.C. § 1 (1982).) . It appears that the Court of Claims subsequently adopted a narrow interpretation of MeCandless. Giles Indust., Inc. v. United States, 496 F.2d 556, 567 , 204 Ct.Cl. 202 (1974). . This may also be probative of a presence or absence of compensatory intent. . It should be noted that there are situations in which the compensation paid to employees is reasonable and yet the corporation may suffer a loss or an inadequate return on equity. . Over the period from 1968 to 1978, during which the bonus formula was in effect, the average rate of return on equity was 15%, based on corporate profit in relation to net book value. There was also evidence, which was not disputed, that the market value of the corporation had increased from the initial shareholder investment of $19,000 to a current market value at the time of trial of $5,000,000. Utilizing these market value figures, Professor Albrecht, an expert witness, testified that this resulted in an annual compounded return of 56%. This is also probative of Elliott’s management contributions to Taxpayer. . This may also be probative of a presence or absence of compensatory intent. . A formula which is reasonable at its inception may later become unreasonable because of changed circumstances. Where a formula has proved reasonable over a long period of time, however, it should normally not be deemed unreasonable solely because it has overcompensated in one or two years.