(dissenting).
Congress could have selected any one of several different tests for deciding when the life insurance portion of a company’s business is sufficient to characterize the enterprise as a “life insurance -company” for tax purposes. It might have used the number of life policies written, the amount of premium income, the face value of its policies, or possibly some combination of different yardsticks.1 Instead, it chose to attach significance to the relative importance of the company’s life insurance reserves.
Perhaps another test would have been preferable, but the reserve-ratio test does have certain advantages. Insurance companies are regulated by state authorities who require them to maintain adequate reserves. There is, therefore, an independent basis for believing that the amount of an insurance reserve is a realistic measure of the insurance risks the company has been paid to assume. For the reserve is not appropriate until the company (a) has assumed the risk, and (b) has been paid for assuming that risk. The amount which it is paid, usually in the form of a premium which has not yet been earned because it applies to future risks, provides the company with the wherewithal to acquire the assets to hold in reserve.
The test supplied by Congress to identify a life insurance company is a rather simple one and, as I understand it, these taxpayers passed it. If over half of an insurance company’s total reserves consist of “life insurance reserves,” 2 it is a *484life insurance company within the meaning of the statute. . In this case about 90% of the taxpayers’ total reserves were life insurance reserves, and therefore they passed the statutory test by a comfortable margin.
The government does not dispute taxpayers’ calculation of the amount of their life insurance reserves. For each taxpayer, however, the government argues “its” life insurance reserves were less than 50% of “its total reserves”3 because the total reserves as reflected on its balance sheet should be increased to include a reserve for unearned premiums on health and accident insurance written by Standard. If I correctly understand the government’s argument, it rests on two incorrect assumptions: (1) an implicit assumption that taxpayers receive unearned premiums on their health and accident business; and (2) an express assumption that taxpayers were at all times “ ‘fully’ on the [health and accident] risk.”
For purposes of anlysis we may treat Standard as the taxpayers’ only customer and note that they sold only two kinds of policies to Standard. We may assume that a prototype of each policy provided insurance protection of $7,000 for a period of 21 months at a premium cost of $210. With respect to the life insurance policy, the full $210 premium was paid by Standard to taxpayer at the beginning of the policy term. At the outset taxpayer was therefore in possession of a $210 asset which had not yet been earned, but which was available to support a reserve liability of no more than $210. Such a reserve could then be amortized as the already-collected premium was earned at the rate of $10 per month.
With respect to the health and accident insurance, however, the full premium was not paid to taxpayer by Standard at the beginning of the term. The premium was paid as it was earned — in our example, at the rate of $10 per month. Since taxpayer did not receive any unearned premium on its health and accident business, it had no occasion to set up any reserve against unearned premiums.
On their life insurance business taxpayers could not terminate their risk by cancellation of a policy before the expiration of its term. They were therefore required by applicable state law to maintain, and did in fact maintain, life insurance reserves against the unearned premiums which they had collected. On the health and accident business, however, they retained an unqualified cancellation right and did not assume any present risk on the unearned or unexpired portions of the policies. They were not required by state law to maintain, and did not maintain, health and accident reserves against risks they did not assume and for which they had collected no premiums.
Unlike the taxpayers, Standard, as the primary insurer, was required to maintain reserves with respect to health and accident policies which it had sold to customers of the finance companies. It was essential for Standard to maintain such reserves because it received the entire premium on such policies when they were issued and, only as those premiums were earned, did Standard use those funds to purchase re-insurance coverage from the taxpayers. Standard was the only insurer on the unearned, future portions of the health and accident risk for which premiums had been paid to Standard but not to taxpayers. Whether we apply the reserve ratio test to reserves which taxpayers actually maintained, to the reserves which they were required by state law and by good ac*485counting practice to maintain, or to risks that they actually assumed at any relevant point in time against which reserves might appropriately have been created, I find no basis for calculating taxpayers’ ratio of life insurance reserves as less than half of their total reserves.
I disagree with the government’s conclusion that taxpayers were fully on the health and accident risk at all times. The government’s conclusion depends on the suggestion that taxpayers’ cancellation rights under the health and accident treaties were “illusory.” I may misunderstand the suggestion, but it seems to rest on two quite different grounds. First, in essence, the government seems to claim that notwithstanding the fact that Standard and taxpayer were independently owned and dealt with one another at arm’s length, the economics of the entire arrangement made it so unlikely that either taxpayer would ever cancel its health and accident re-insurance treaty that its clear right to do so should be equated with a legal obligation not to do so.4 In my opinion, the contractual language should be accepted as controlling.5
Second, the government may read the health and accident treaty as imposing the entire risk on taxpayer in the event that cancellation should occur. It is true that under Article IX the taxpayer accepted some of the risk on claims arising subsequent to cancellation, but the amount of the risk so accepted is limited to 10% of the total health and accident premiums earned by taxpayers during the one-year period prior to cancellation of the treaty. This exposure is, of course, far short of being “fully” on the risk. Indeed, it is quite clear that even if the taxpayers had been required to increase their total reserves in an amount sufficient to reflect 100% of this contingent liability, such an increase would not have been nearly sufficient to make the total reserve twice as large as the life insurance reserve.6
In sum, I am persuaded that the government’s conclusion that life insurance represents less than half of taxpayers’ total insurance business rests on a non-statutory standard. Congress may have acted unwisely in giving preferential tax treatment to life insurance companies, and it may have been unwise to select a reserve-ratio test as the definition of a life insurance company for tax purposes. Nevertheless, we must, of course, apply the test which Congress has specified. Although I do not question the majority’s conviction that it is doing so faithfully, I am afraid, the holding may ac*486tually rest on an amalgam of substitute tests that will be difficult to apply in other cases.7 I therefore respectfully dissent.
. It would surprise me if Congress should select a yardstick defined in terms of the amount of premiums received by a third party, rather than by the taxpayer, without setting forth a further test to indicate when “the arithmetic mean as of the beginning and end of each year” of the third party’s revenues should be treated as though they belonged to the taxpayer.
. The statutory definition also permits “unearned premiums, and unpaid losses (whether or not ascertained), on noncancellable *484life, health, or accident policies” to be treated as life insurance reserves for definitional purposes, but that component of the formula may be disregarded in this case since about 90% of each taxpayer’s total reserves consisted of life insurance reserves.
. The quoted words are from the test prescribed by Congress in § 801(a). See Judge Sprecher’s opinion, text at footnote 13.
. At page 20 of its brief, the government argues that “the cancellation provisions grant taxpayers only an illusory right to avoid the risks.” On the next page of its brief, purportedly stating the same legal conclusion in a different way, the government emphasizes evidence which “negates any reasonable likelihood that taxpayers would ever cancel.”
. Even if we accept the government’s theory that we can safely predict that neither the taxpayers nor Standard will ever cancel the health and accident re-insurance treaties, and therefore taxpayers will continue to assume tiie risks on the health and accident policies as the premiums are earned, I would not ignore the anomaly of creating a hypothetical balance sheet on which taxpayers have a large reserve liability which is not matched by any cash or other asset reflecting the receipt of unearned premiums. Necessarily, the government not only attributes Standard’s reserves to the taxpayers, but also must treat them as though they had received the entire health and accident premiums at the beginning of the policy terms instead of on a monthly basis as the premiums are earned.
. As an illustration, plaintiff’s Exhibit 14 indicates that Standard paid $285,574.23 of health and accident premiums to United in 1961. Under the cancellation provisions of the reinsurance treaty, United therefore had a maximum contingent liability of $28,557.42 on the unearned portion of the health and accident risk. Even if this full amount were added to its reported total reserves of $336,215.63, its life insurance reserves, which amounted to $322,055.00, would be over 88% of the total.
. The principal element in the crucible appears to be an emphasis on the long-term investment income potential associated with “the dual nature of life insurance, serving botli an insurance and a banking function.” I do not understand, however, that term life insurance — -which is the only kind of life insurance involved in this case — serves any such banking function. I am therefore in-dined to believe that any analysis based on an attempt to effectuate the majority’s understanding of the underlying rationale for special tax treatment for life insurance companies — rather than the statutory language selected by Congress — would lead to the disqualification of any company writing nothing but term life insurance.