(dissenting). The crux of the issue raised by this appeal is shown by the testimony of a staff witness for the commission:
[M]y tax rate was developed using the new rates as prescribed by the Tax Reform Act of 1986, namely a 34 percent tax rate effective July 1st, 1987. Therefore, I took the months in the test year after June 30th, 1987, and used a 34 percent tax rate for those months. The months prior to July 1,1987,1 used a 46 percent tax rate. I took an average of those 12 months and come up with a composite rate of 36 percent for federal income taxes.
The tax reform act changed corporate tax rates from 46 percent to 34 percent, effective July 1, 1987. Because the company is a calendar year taxpayer, its 1987 tax rate was neither 46 percent nor 34 percent, but a blended rate. Simplified, that rate for a calendar year taxpayer was the average of one-half year at 46 percent and one-half year at 34 percent. However, income taxes are calculated yearly, not monthly or semiannually. Thus, the Internal Revenue Service required the company to pay income tax on its 1987 net income at a 40 percent rate.
There is no way to avoid this method of calculating federal income taxes. The Internal Revenue Service deems it irrelevant that more or less income might be earned in one month, one quarter, or one six-month period. Therefore, neither the commission nor the majority can dispute that the company paid federal income taxes of 40 percent of its net income earned in 1987. Nor can it be disputed that for the calendar year 1988, the company paid federal income taxes of 34 percent of its net income.
The commission did not concern itself with the tax rate that the company actually would pay on 1987 net income. Instead, it applied tax rates to monthly income, *203a method not possible in the real world. "The ultimate test of a formula is whether it bears a reasonable relationship to reality and conforms to state and federal laws and constitutions." Soo Line R. Co. v. Department of Revenue, 89 Wis. 2d 331, 341, 278 N.W.2d 487, 492 (Ct. App. 1979), aff'd as modified on other grounds, 97 Wis. 2d 56, 292 N.W.2d 869 (1980). The commission's formula assumed that the company would pay federal income taxes at a 34 percent tax rate from July 1, 1987 to December 31, 1987. That assumption bears no relationship to reality. In reality, the company paid 40 percent of its net income during that time as federal income taxes. Merely because the commission and the majority assert that the company paid taxes at a lesser rate does not make it so.
During the company's test year, it paid federal income taxes for eight months of 1987 at a 40 percent rate, and for four months of 1988 at a 34 percent rate. No one explains how this results in a 36 percent composite rate. If the company filed income tax returns every six months, reflecting income and expenses for the preceding six months, the commission's position would make sense. As everyone agrees, however, that method of paying income taxes is fictional.
The majority, faced with the task of sustaining the equivalent of a commission finding that 2+2 = 5, reasons that because the 40 percent rate is a blended rate, use of that rate would overstate the company's tax expense.1 That is no more a reason to sustain the com*204mission than to conclude that its opinion is correct because it was written on Tuesday. It just sounds better. Or, as the Court of Appeals for the District of Columbia Circuit put it, in a case indistinguishable from the one we are deciding:
FERC suggested in its brief, and its counsel asserted repeatedly at argument, that — section 15(a) of the Internal Revenue Code notwithstanding — the tax rate applicable to income earned by CP & L prior to July 1,1987 was 46% and the rate thereafter, 34%. This strikes us as equivalent to attempting to prove that the moon is made of green cheese by asserting it several times in quick succession. CP & L did not— and was plainly not entitled to — file two returns in calendar year 1987, one for the first six months and another for the last six. The petitioner filed one return, and for purposes of that return a single corporate income tax rate of 40% was applicable to income earned before and after the July 1,1987 effective date of the statutory change.
Carolina Power & Light Co. v. F.E.R.C., 860 F.2d 1097, 1100 (D.C. Cir. 1988) (emphasis in original).
Carolina Power is the only appellate decision addressing the issue about which the company and the commission disagree. Because Carolina Power's holding is contrary to the majority's mandate, Carolina Power must be reckoned with. The majority declines to follow Carolina Power for two reasons: (1) the case was unreasoned, because its decision was based on the proposition that the moon is (or is not) made of green cheese; and (2) the Carolina Power case is based on concepts of retroactive ratemaking, an issue not present in this case.
Both reasons the majority uses to distinguish Carolina Power are incorrect.
*205Green Cheese — Unreasoned Decision
The Carolina Power opinion has nothing to do with green cheese. Instead it examines the Federal Energy Regulatory Commission's assertion that Carolina Power & Light paid its 1987 federal income tax at a rate of 46 percent prior to July 1, 1987 and at a rate of 34 percent after that date. Carolina Power, 860 F.2d at 1100. The Carolina Power court observed that the Internal Revenue Service would only permit Carolina Power & Light, a calendar year taxpayer, to pay its 1987 income taxes as sec. 15(a) of the Internal Revenue Code required, that is, at 40 percent of its 1987 income. Id. The Carolina Power court found FERC's assertion, i.e., that Carolina Power & Light paid federal income taxes at a 34 percent rate for the second half of 1987, to be whimsical, and concluded that FERC's order failed for want of an articulated rational basis. Id. at 1104.
I cannot join in the majority's assertion that other than a comment on the moon's composition, "[t]he [District of Columbia Circuit] court offered no other rationale [for its holding]." Majority opinion at 194.1 believe that the court gave a reason for its decision, and I agree with that reason.
Retroactive Ratemaking
The majority asserts that the Carolina Power decision was based on retroactive ratemaking considerations. The short answer to this is that the Carolina Power court rejected this notion:
As petitioner observes, FERC has made no finding that petitioner profited excessively during the first eight months of 1987. Indeed, petitioner suggests that were the Commission to conduct an overall evaluation of CP & L's profitability during early 1987, no *206such finding could be made. This issue, of course, we cannot resolve, for the Commission did not frame its disposition in this case as a refund action, and we are thus not presented with a sufficient record to test the Commission's application of West Texas to CP & L's filing as an instance of retroactive ratemaking.
Carolina Power, 860 F.2d at 1104. (Footnotes omitted.)
A more complete answer is that in the present case, the commission's use of nonexistent tax rates did constitute retroactive ratemaking. Therefore, if the majority is correct that retroactivity considerations are the basis for the Carolina Power decision, this case and Carolina Power are even more alike.
When the company's old service rates were set, its federal income tax expense was estimated using a 46 percent rate. The old rates continued until May 31,1987. Because the company's actual tax rate for the first five months of 1987 was 40 percent, it enjoyed a greater return than predicted, all other income and expense estimates being accurate. This is an unfortunate, but nonetheless a necessary implication of a system that prohibits retroactive ratemaking. However, the theory behind that prohibition is realistic. Unexpected changes in income and expenses will favor a utility on some occasions, and favor customers on other occasions. Over time, inequities will even out, leading to justice for both customers and the utility.
The trial court examined the commission's reasons for using a nonexistent tax rate, and considered whether this procedure implicated retroactive ratemaking:
The PSC briefly argues that "[W]hile it is true that the Commission was concerned about the possibility that overcollection of federal income tax expense that might result from the use of the Company's method, it does not follow that the method *207actually adopted by the Commission involved retroactive ratemaking." (Reply Brief at p. 14) Despite this statement, the clear language of the order is that the commission intended to prevent overearning in calendar year 1987, and that it sought to reach that result by setting rates for the test year that result in a 40% tax recoupment for the calendar year — even though rates already collected during the first part of 1987 were uncontroverably [sic] based on a final order of the PSC that used a 46 percent tax recoupment rate. I am not sure that this action can be better described than by virtue of the language used by the Wisconsin Supreme Court when it reiterated the rein on the PSC's ratemaking power: "[T]he Commission may not. . .make a rate sufficiently low to recapture the excesses." Wisconsin Telephone Co. v. PSC, 232 Wis. 274, 303 (1939).
The commission's decision leaves no question but that the trial court was correct:
The company has requested an effective tax rate of 40% for the period of May 1-December 31,1987. The commission has determined that the use of the company's proposed rate would result in an overcollection from the ratepayers for calendar year 1987. Under current rates, the company will have recovered for income taxes at an effective rate of 46% from January 1 to May 31, 1987 (it is assumed that new rates will go into effect on June 1, 1987). (Emphasis supplied.)
The commission was undoubtedly correct that the company's use of a 40 percent rate for 1987 would result in an overcollection for that year. A previous and now-inaccurate estimate of the company's first-half federal income tax expense caused 1987's overcollection. However, the commission's decision was dated May 28,1987, and became effective on June 1, 1987. Therefore, the *208effects of federal income tax rates in effect before June 1, 1987 could not be used to justify a fictional and lower estimate of the company's second half 1987 federal income taxes. When the commission did so, it set rates retroactively.
If, as the majority asserts, Carolina Power was really a retroactive ratemaking case, then the holding of Carolina Power is that the accounting method used by FERC constituted retroactive ratemaking. In this case, the commission used the identical accounting method used by FERC in the Carolina Power case. The conclusion is inescapable: the commission's treatment of the company's federal income tax expense constituted retroactive ratemaking, an act the commission concedes it may not do.
In reality, the commission was attempting to correct for a flaw inherent in the practice of using a test year to predict income and expenses. Some items of income and expense are necessarily unknowable until they accrue. Thus a utility's income may be altered by unexpected weather patterns, but an estimate of income based on normal weather patterns is acceptable for a test year. This is because no better way of predicting the future can be devised.
Federal income taxes are different. It seems strange to "estimate" taxes for a future test year when, absent action by congress, future rates are a certainty. Yet, if a "test year" is used to predict the future, the "test year" concept requires that actual, not fictional figures be used. Therefore, though the company and the commission both knew that tax rates in the test year would not be the same as in future years, the concept of using a test year as predictive of future years was not abandoned. If the commission wished to change the rules by which the *209game was played, it should not have done so in the middle of the fifth inning.
Because it used the "test year" method of setting rates, but ignored the actual tax rate in effect during the test year, the commission's conclusion is irrational. I agree with the trial court that the commission's reason for its order was in part to recapture what it believed were the company's past excesses. I would therefore affirm the trial court's remand for further commission action.
The majority gives no explanation of why a blended tax rate overstates the company's tax expense. Indeed, the Internal Revenue Service not only accepted, but required that rate. If use of a particular tax rate would overstate an actual expense, one would expect the Internal Revenue Service to prohibit the use of that rate, not require it.