We are presented with another version of a not so novel question: Are amounts paid as “insurance premiums” by a parent corporation to its “captive insurance company” 1 subsidiary deductible for purposes of federal income taxation? This latest twist, like the previous case we considered, Carnation Co. v. Commissioner, 640 F.2d 1010 (9th Cir.), cert. denied, 454 U.S. 965 , 102 S.Ct. 506 , 70 L.Ed.2d 381 (1981), aff'g 71 T.C. 400 (1978), involves the purchase by a parent of insurance from an unrelated insurance company and the reinsurance by the unrelated company of the principal portion of that liability with the parent’s captive subsidiary; however, in the case now before us there is no agreement between the parties providing for additional capitalization of the captive by the parent or otherwise ensuring that the captive will be able to perform its obligations under the reinsurance agreement.

Differing fact patterns of the captive insurer issue have appeared before several courts, all of which have held that insurance payments from a parent to its wholly-owned captive are not deductible. 2 Yet the issue seems not to have been addressed with finality, or, perhaps, as Clougherty asserts, we and other courts have not explained our decisions adequately. We affirm the United States Tax Court’s finding of non-deductibility and attempt to provide a more complete explanation of the reasoning underlying our conclusion.

I. Facts

The parties have stipulated fully to the facts. Clougherty Packing Company is a California corporation whose principal business is slaughtering and meat processing. It employs over 1,000 workers, for whom California requires Clougherty either to maintain workers' compensation coverage through an authorized insurer or to self-insure after obtaining consent from the California Director of Industrial Relations. Cal.Lab.Code § 3700 (West 1971). Clougherty faced numerous claims, and from 1971 to 1977, it self-insured a portion of its risk and obtained excess liability coverage for the balance from authorized insurers. As a partial self-insurer, Clougherty was required to deposit securities with the state treasurer as collateral for potential claims; the collateral totaled $857,110 in 1977.

In 1976, an outside consultant recommended that Clougherty form a captive insurance company to insure its workers’ compensation liability. Clougherty incorporated two wholly-owned subsidiaries, Lombardy Insurance Corporation in Colorado and Clougherty Packing Company of Arizona, the latter eventually becoming the sole holder of Lombardy stock; 3 Colorado *1299 permits the incorporation of captive insurers pursuant to the Colorado Captive Insurance Company Act, Colo.Rev.Stat. §§ 10-6-101 to -130 (1973). Clougherty capitalized Lombardy for $1 million, and the Colorado Division of Insurance thereafter issued Lombardy a certificate of authority to conduct business as a captive insurance company in Colorado.

Clougherty then negotiated and reached an agreement for a captive insurance program with Fremont Indemnity Company, an authorized insurance carrier in California unrelated to Clougherty. As of January 1,1978, Clougherty terminated the self-insurance portion of its insurance coverage and purchased all of its workers’ compensation insurance from Fremont. Fremont reinsured with Lombardy the first $100,000 of each claim against Clougherty and ceded to Lombardy 92% of its annual premium. Fremont also charged Clougherty an additional amount equal to five percent of the premium as a fee for providing the captive insurer program. There was no agreement requiring Clougherty to indemnify Fremont, to capitalize Lombardy further, or otherwise to guarantee Lombardy’s obligations. Lombardy engaged in no business other than the reinsuring of Clougherty. The premium charged by Fremont and the reinsurance rate charged by Lombardy were approved by the appropriate state agencies. Fremont remained liable for the payment of insurance claims in the event Lombardy became insolvent or defaulted for any other reason. However, under California law, because Clougherty had purchased an insurance policy from an authorized insurer, Clougherty could not be held liable for covered workers’ compensation claims in the event that Fremont failed to make the required payments. Cal.Lab. Code §§ 3755, 3757 (West 1971).

In calculating its federal income tax liability in its fiscal years ending July 29, 1978, and July 28,1979, Clougherty deducted as necessary business expenses the amounts it paid to Fremont as insurance premiums — $840,000 and $1,457,000 respectively. Of these sums, Lombardy received $772,900 and $1,340,000. In reviewing Clougherty’s tax returns, the Commissioner of Internal Revenue disallowed the portions of premiums received by Lombardy from Fremont and determined income tax deficiencies for Clougherty of $370,944 and $628,202.

Clougherty appealed the Commissioner’s disallowance, and a divided panel of the United States Tax Court sitting en banc affirmed. Clougherty Packing Co. v. Commissioner, 84 T.C. 948 (1985). Eight judges subscribed to the plurality opinion, three wrote separate concurrences, and seven joined in dissent. Clougherty’s timely filing brings the matter before this court.

II. Legal Discussion

A. Standard of Review

Our court has jurisdiction to review decisions of the United States Tax Court. 26 U.S.C. § 7482 (a) (1982). We review its decisions on the same basis as decisions following civil bench trials in federal district court. 26 U.S.C. § 7482 (a) (1982); Mayors v. Commissioner, 785 F.2d 757, 759 (9th Cir.1986). The parties here have at all times stipulated to the relevant facts; the only questions are those of law. Our review of the Tax Court’s decision is thus de novo. Vukasovich v. Commissioner, 790 F.2d 1409, 1413 (9th Cir.1986). Although the Tax Court’s judgments in its field of expertise are accorded a presumption that they correctly apply the law, we do not apply a rule of special deference to its decisions. Id.

B. The Definition of Insurance

In calculating taxable income, section 162(a) of the Internal Revenue Code, 26 U.S.C. § 162 (1954), permits the deduction from gross income of all ordinary and necessary expenses incurred in carrying on a *1300 business. Premiums for insurance, including those for workers’ compensation coverage, are deductible business expenses. 26 C.F.R. § 1.162 -l(a) (1986). The insuring taxpayer deducts the amounts paid as premiums but, of course, cannot deduct covered claims because the source of the payments is the insurance carrier. In lieu of purchasing insurance, one may elect to self-insure, paying off claims as they arise or setting aside fixed sums into a reserve account to pay off intermittent losses. While insurance premiums are deductible, amounts placed into self-insurance reserves are not. Steere Tank Lines, Inc. v. United States, 577 F.2d 279, 280 (5th Cir.1978), cert. denied, 440 U.S. 946 , 99 S.Ct. 1424 , 59 L.Ed.2d 634 (1979); Spring Canyon Coal Co. v. Commissioner, 43 F.2d 78, 80 (10th Cir.1930), cert. denied, 284 U.S. 654 , 52 S.Ct. 33 , 76 L.Ed. 555 (1931); Pan-American Hide Co., 1 B.T.A. 1249 (1925). Instead, the self-insuring taxpayer must wait until losses actually occur, at which time the reserve funds actually paid out may be expensed and deducted from gross income. In between the extremes of ordinary insurance and direct self-insurance lies the captive insurer transaction. We must decide whether to treat this transaction as one or the other. If the former, then the amounts paid by Clougherty to its captive Lombardy through Fremont are deductible insurance premiums; if the latter, no deduction is permitted. We hold that the captive insurer arrangement used by Clougherty falls on the self-insurance side of the line.

The appropriate starting point of our analysis is the meaning of “insurance.” Neither the Internal Revenue Code nor tax regulations provide a definition of insurance. The accepted definition for purposes of federal income taxation dates back to Helvering v. Le Gierse, 312 U.S. 531 , 61 S.Ct. 646 , 85 L.Ed. 996 (1941), in which the Supreme Court stated that “[historically and commonly insurance involves risk-shifting and risk-distributing.” Id. at 539 , 61 S.Ct. at 649 ; see Commissioner v. Treganowan, 183 F.2d 288, 291 (2d Cir.), cert. denied, 340 U.S. 853 , 71 S.Ct. 82 , 95 L.Ed. 625 (1950); B. Bittker, 5 Federal Taxation of Income, Estates & Gifts ¶ 127.2 at 127-7 (1984) (“under Le Gierse the shifting and distribution of the risk of death are indispensable elements of life insurance”). 4 Shifting risk entails the transfer of the impact of a potential loss from the insured to the insurer. If the insured has shifted its risk to the insurer, then a loss by or a claim against the insured does not affect it because the loss is offset by the proceeds of an insurance payment. See Beech Aircraft, 797 F.2d at 922; Treganowan, 183 F.2d at 291 ; O’Brien & Tung, Captive OffShore Insurance Corporations, 31 N.Y.U. Inst. 665, 683-84 (1973). Distributing risk allows the insurer to reduce the possibility that a single costly claim will exceed the amount taken in as a premium and set aside for the payment of such a claim. Insuring many independent risks in return for numerous premiums serves to distribute risk. By assuming numerous relatively small, independent risks that occur randomly over time, the insurer smoothes out losses to match more closely its receipt of premiums. See Beech Aircraft, 797 F.2d at 922; Treganowan, 183 F.2d at 291 . Risk distribution incorporates the statistical phenomenon known as the law of large numbers. This law is reflected in the financial world by the diversification of investment portfolios and in the day-to-day world by the adage “Don’t put all your eggs in one basket.” See T. Wonnacott & R. Wonnacott, Introductory Statistics for Business & Economics 31-32, 54-55 (1972). 5

*1301 In Le Gierse , an octogenarian attempted to take advantage of the exemption of insurance proceeds from federal estate taxation by simultaneously purchasing an insurance policy and an annuity contract. The former cost $23,000 and paid $25,000 to her daughter upon Le Gierse’s death, while the latter cost $4,000 and paid Le Gierse approximately $600 annually while she was still living. As a review of these sums demonstrates, the arrangement was merely an attempt to transform what would soon have become a taxable portion of an elderly woman’s estate into non-taxable insurance proceeds. It is also apparent that the two contracts taken together did not shift the risk of loss from the “insured” to the insurance company. The amounts the company received under the two agreements, some $27,000, were almost certain to exceed any payouts on the agreements — some $600 a year while an 80-year old was living and $25,000 upon her death. In the highly unlikely event that Le Gierse defied statistical averages and lived for an extended period, the interest earned by the company on the premiums would more than offset any additional payouts on the annuity. 6 The Court required that the agreements be considered together and found they were not “insurance,” because in combination one agreement neutralized the risk customarily inherent in the other. Le Gierse, 312 U.S. at 540-41 , 61 S.Ct. at 649-50 .

Le Gierse provides not only a definition of insurance — the shifting and distributing of risk — but also a rule for delimiting the transaction to be analysed. Where separate agreements are interdependent, they must be considered together so that their overall economic effect can be assessed. Id. Thus, to answer whether risk has shifted between the parties, we do not review the insurance agreement in isolation; rather, we take into account other agreements and relationships that are interdependent with the insurance agreement. Here the Tax Court found, and we agree, that the insurance agreement between Clougherty and Fremont, and the reinsurance contract between Fremont and Lombardy, were interdependent — both were necessary parts of the captive insurance program Fremont agreed to establish for Clougherty. Thus, we consider the agreements together in addressing whether there is risk shifting.

C. Risk Shifting

The earliest pronouncement on the issue of captive insurers came from the Internal Revenue Service in Revenue Ruling 77-316, 1977-2 Cum.Bul. 53. In it the Service reviewed three hypothetical situations involving a wholly-owned captive insuring only its parent and affiliates — in one case directly and in the others through an unrelated intermediate insurer. Situation 2 of the Ruling is identical in substance to the facts presented here. 7 The Service held in Revenue Ruling 77-316 that an arrangement whereby a wholly-owned subsidiary “insurers” only its parent and related companies, whether directly or through an unrelated third party, is not insurance.

In concluding that there is no economic shifting of risk between parent and captive insurance subsidiary, the Ruling relies on what has come to be known as the “economic family” concept.

Id. at 77:375. According to the Ruling’s description, the economic family theory appears to treat a corporation and its affiliates as a single overall entity and then assess whether the insurance arrangement shifts the risk of loss away from that entity. Under that theory, because an “insured” loss remains within the corporate family there is no risk shifting and no insurance results. The Ruling states that its conclusion recognizes the separate tax status of subsidiary corporations, as required by Moline Properties, Inc. v. Commissioner, 319 U.S. 436 , 63 S.Ct. 1132 , 87 L.Ed. 1499 (1943), “but also examines the economic reality of each situation described.” Rev.Rul. 77-316 at 77:376.

The doctrine of Moline Properties seems relatively straight-forward: “[S]o long as [its] purpose is the equivalent of business activity ... the corporation remains a separate taxable entity.” 319 U.S. at 439 , 63 S.Ct. at 1134 . While Moline Properties concerned an attempt by the sole shareholder of a corporation to report on his personal return income attributable to the corporation, the rule it enunciates applies as well to a corporation and its subsidiaries. National Carbide Corp. v. Commissioner, 336 U.S. 422 , 69 S.Ct. 726 , 93 L.Ed. 779 (1949). Exceptions exist in areas where Congress has evinced an intent to the contrary or where the corporate form is but a sham. 319 U.S. at 439 , 63 S.Ct. at 1134 . Congress, however, has remained silent with respect to the taxation of captive insurers and the Tax Court found that Clougherty had a valid business purpose for incorporating Lombardy.

Clougherty argues forcefully that the economic family concept espoused in the Ruling violates Moline Properties. Evidently, the Tax Court below sensed a tension between the economic family concept and Moline Properties. It expressed concern that the concept “might foster a theory which would be extended to other areas of the tax law.” Clougherty, 84 T.C. at 959 . The court, however, found it unnecessary to use the term “economic family” in deciding the case because it could be decided “within the parameters of Carnation.” Id. at 956. Given that our holding in Carnation Co. v. Commissioner, 640 F.2d 1010 (9th Cir.), cert. denied, 454 U.S. 965 , 102 S.Ct. 506 , 70 L.Ed.2d 381 (1981), aff'g 71 T.C. 400 (1978), explicitly refers to the Ruling, it seems odd that the Tax Court uses Carnation as a means of avoiding reliance on the Ruling. 8

Carnation was the first judicial review of the captive insurer question and the only review of the question undertaken by this circuit. In that case, Carnation purchased an insurance policy from an unrelated insurance carrier, which then reinsured 90% of its liability with Carnation’s wholly-owned subsidiary. In accepting this arrangement, however, the insurer required Carnation to agree to increase at the insurer’s request the subsidiary’s capitalization, which at the time was substantially less than the annual premium ceded to the subsidiary. We agreed with the Tax Court’s assessment that the agreements between Carnation, its captive and the unrelated carrier were interdependent. Considering the agreements together, we held that, to the extent the unrelated insurer reinsured with Carnation’s subsidiary, the economic *1303 effect was to neutralize any shifting of the risk away from Carnation. 640 F.2d at 1013 . We also stated that the facts in Carnation were “identical” to those described in Situation 2 of Revenue Ruling 77-316. Id.

The operative facts of Carnation are identical to those presented here except for one that Clougherty claims to be of critical significance. Clougherty insists that it was the capitalization agreement that neutralized any risk shifting in Carnation and that the absence of any such agreement requires that we reach an opposite result in this case. For support, Clougherty points to the Tax Court’s opinion in Carnation, and particularly the express statement that the capitalization agreement counteracted the reinsurance agreement. 71 T.C. at 409 . In our opinion affirming the Tax Court we appear to have endorsed that statement. 640 F.2d at 1013 . In our very next paragraph, however, we noted the identity between the facts in Carnation and those in Situation 2 of Revenue Ruling 77-316. Id. at 1013 . Because Situation 2 does not include a capitalization agreement, our statement that the fact situations were identical would seem to suggest strongly that our holding did not hinge on the capitalization agreement.

The dissenters below and several tax commentators argue that either our statement regarding the identity in facts between Carnation and Situation 2 is wrong or else Carnation depends on the validity of the Revenue Ruling, which they claim violates Moline Properties. Clougherty, 84 T.C. at 965-66 ; e.g., Liles, Captive Insurance Companies: Further Confusion over Economic Families?, 26 Tax Mgmt. (BNA) 226, 229 (1985). We addressed the latter point in Carnation, albeit briefly, when we said “[w]e reject Carnation’s contention that [Situation 2 of] this ruling conflicts with recognition of the separate status of corporations.” 640 F.2d at 1013 .

Several courts outside of this circuit have addressed the captive insurance issue, and none has found that a policy provided by a wholly-owned subsidiary that exists solely for the purpose of providing insurance to its parent constitutes insurance, or that such a conclusion violates Moline Properties. The Tenth Circuit has twice addressed this issue. In Beech Aircraft Corp. v. United States, 797 F.2d 920 (10th Cir.1986), aff'g U.S.Tax Cas. (CCH) ¶ 9803, 54 A.F.T.R.2d (P-H) ¶ 84-5392 (D.Kan. 1984), the parent insured directly with its captive subsidiary, while in Stearns-Roger Corp. v. United States, 774 F.2d 414 (10th Cir.1985), aff'g 577 F.Supp. 833 (D.Colo.1984), the parent insured some of its liabilities directly with its captive, and some indirectly using an unrelated third-party insurer which reinsured with the captive. In both cases, the trial court found there was no risk shifting and the Tenth Circuit affirmed. Two recent lower court decisions that reviewed captive insurer arrangements reached the same result. Mobil Oil Corp. v. United States, 8 Ct.Cl. 555 (1985); Humana Inc. & Subsidiaries v. Commissioner, T.C.M. (P-H) ¶ 85,426 (1985). Both cases involved direct two-party insurance agreements between parent and wholly-owned captive. 9

Clougherty contends that the cases involving only direct insurance between parent and captive are distinguishable because in the transaction before us there was an intermediate insurer, Fremont, which assumed the risk of Lombardy’s insolvency. As for Stearns-Roger, Clougherty argues that the district court expressly embraced the “erroneous” economic family theory in its decision and that in any event the case is distinguishable because the parent had a separate agreement to indemnify the subsidiary for its losses.

The only case that has allowed the deduction of premiums paid directly or indirectly to a captive insurer is Crawford Fitting Co. v. United States, 606 F.Supp. 136 (N.D.Ohio 1985). It is not, however, factually *1304 akin to the case here. While the insurer there was a captive insurance company incorporated in Colorado, the insured company was not the actual parent. Instead, the captive was owned by individuals who were also the owners or officers of the insured company, or relatives of those persons. The court in Crawford Fitting found that the arrangement exhibited both risk shifting and risk distributing and that it survived application of the economic family theory because the captive's owners were not within the corporate family. Id. at 145 . While limiting the economic family concept to corporate entities and not including the individual owners or officers of those corporations may seem artificial, the case does not deviate from the consensus of prior cases, all of which find that a wholly-owned captive cannot “insure” its parent. 10

Despite the judicial consensus, Clougherty has as allies in its challenge to the Commissioner’s position a number of tax commentators as well as seven dissenters on the Tax Court. See, e.g., Bradley & Winslow, Self-Insurance Plans and Captive Insurance Companies —A Perspective on Recent Tax Developments, 4 Am.J.Tax Pol’y 217, 250-51 (1985); Liles, supra, 26 Tax Mgmt. (BNA) at 226. In large part, the criticism continues unabated because the courts have failed to articulate to the satisfaction of captive insurer proponents how the conclusion of no risk shifting is attained without, as the dissent below puts it, “crashing head-on into the holding in Moline Properties.” 84 T.C. at 967 . There is some merit to the complaint that judicial discussions have often glossed over the interplay of the risk shifting analysis with the Moline Properties recognition of separate taxable entities. We agree that an explanation of the rationale underlying the conclusion reached by a majority of the Tax Court is warranted. We do not, however, agree that the conclusion violates Mo-line Properties.

To begin with, we seriously doubt that the use of an economic family concept in defining insurance runs afoul of the Supreme Court’s holding in Moline Properties. The Moline Properties rule that a corporation with a valid business purpose “remains a separate taxable entity,” 319 U.S. at 439 , 63 S.Ct. at 1134 , has consistently been invoked where there is a question whether a corporate entity should be afforded separate tax status and the answer determines whether the tax consequences that would ordinarily attach to a corporate entity are to be enjoyed or suffered by the corporation or by its owners. E.g., Zmuda v. Commissioner, 731 F.2d 1417, 1420-21 (9th Cir.1984); Kleinsasser v. United States, 707 F.2d 1024, 1027 (9th Cir.1983); Britt v. United States, 431 F.2d 227, 234-35 (5th Cir.1970); see National Inv. Corp. v. Hoey, 144 F.2d 466, 468 (2d Cir.1944) (L. Hand, J.) (Moline Properties “merely declares that to be a separate jural person for purposes of taxation, a corporation must engage in some industrial, commercial, or other activity besides avoiding taxation.”). The Commissioner does not challenge the propriety of affording Lombardy separate status for tax purposes; whatever tax liabilities are incurred by Lombardy will be assessed to it and not to its parent. Instead, the Commissioner’s position is that the Moline Properties rule may not be lifted out of its usual context — cases involving the question which of two related éntities should receive a tax benefit or assume a tax burden — and mechanically applied as the basis for deciding an entirely different type of question — what constitutes “insurance.” Thus, he argues, Mo-line Properties does not bar the Service from considering related entities as a single economic family when assessing whether an insurance policy shifts the risk of loss *1305 from the policyholder. The Commissioner’s view, if correct, would place the economic family concept (at least if limited to its present uses) beyond the reach of Mo-line Properties. We find considerable merit in the Commissioner’s argument. However, we need not resolve any arguable conflict that may exist between Moline Properties and the economic family concept because it is not necessary to rely on that concept in order to conclude that Clougherty’s captive insurer program is not insurance.

In reaching our holding, we do not disturb the separate legal status of the various corporate entities involved, either by treating them as a single unit or otherwise. Rather, we examine the economic consequences of the captive insurance arrangement to the “insured” party to see if that party has, in fact, shifted the risk. In doing so, we look only to the insured’s assets, i.e., those of Clougherty, to determine whether it has divested itself of the adverse economic consequences of a covered workers’ compensation claim. Viewing only Clougherty’s assets and considering only the effect of a claim on those assets, it is clear that the risk of loss has not been shifted from Clougherty.

A workers’ compensation claim against Clougherty of less than $100,000 is covered under the reinsurance agreement between Fremont and Lombardy. Fremont first pays the claim under the primary insurance policy and then seeks reimbursement under its reinsurance agreement with Lombardy. There is no recourse against Clougherty. Accordingly, there is no immediate effect of a claim on Clougherty’s assets. 11 So far, it appears that risk of loss has shifted from Clougherty to Lombardy: A loss that would otherwise be borne by Clougherty appears to leave its assets unaffected and instead reduces Lombardy’s assets. Clougherty takes the analysis to this point but no further.

The effect on Clougherty of a covered loss, however, does not stop here. Clougherty is the sole shareholder of the company that owns 100% of the Lombardy stock. As a result of reimbursing Fremont for the loss, Lombardy’s income and net worth fall dollar for dollar by the amount of the loss. Because Lombardy does no business other than insuring Clougherty, the value of Lombardy stock declines by exactly the same amount. The decline affects the owners of Lombardy stock irrespective of who those shareholders are and what corporate relationship they have with Lombardy. As indirect owner of every share of Lombardy stock, Clougherty suffers a loss in the value of one of its assets equal to the full amount of the claim paid by Lombardy. Under the captive insurer program, every such claim reduces the value of Clougherty’s assets by exactly the same amount that it would if Clougherty self-insured in the ordinary sense.

It is true that Lombardy bears a risk of loss from a claim against Clougherty because Lombardy suffers a loss when a claim is paid. But it is also true that Clougherty has not divested itself of its risk of loss from such a claim even though it is paid by Lombardy — and it is the effect on Clougherty’s assets that is determinative here. Le Gierse requires that an insurance agreement negate any effect of a covered loss on the insured party’s assets. We thus look at that party’s assets and not at those of its captive subsidiary or some aggregation of the two. Because a covered claim still affects Clougherty’s assets, its captive insurance arrangement does not succeed in shifting its risk of loss. Therefore, under Le Gierse the arrangement is not insurance.

The risk shifting requirement of Le Gierse accurately describes the essence of *1306 insurance. Insurance involves transferring from the insured to the insurer the consequences of a possible future event. The likelihood that a loss will occur is of uncertain but predictable magnitude; the size of the loss is similarly uncertain but predictable. A party deciding whether to insure against the occurrence of a particular event estimates the potential cost to it if the event occurs, discounts that amount by its assessment of the likelihood of occurrence, and then factors in its risk aversion, i.e., its preference for certainty over uncertainty. If the party buys insurance, it has decided to pay in advance a fixed sum — the premium — in order to avoid the possibility of suffering an uncertain but probably larger loss. Accordingly, a true insurance agreement must remove the risk of loss from the insured party. “It must rid the insured of any economic stake in whether the loss occurs or not.” Bradley & Winslow, supra, 4 Am.J.Tax Pol’y at 250 (quoting O’Brien & Tung, supra, 31 N.Y.U.Inst. at 683-84). Unquestionably, despite its tripartite insurance arrangement, Clougherty still has an economic stake in whether the loss occurs, a stake reflected in its own assets. Clougherty simply cannot deny the fact that when a claim against it is paid (in an amount of $100,000 or less) its net worth is reduced by the amount paid. Under our analysis, it is of no consequence that Lombardy’s net worth is also reduced by that amount.

Clougherty argues that even if insuring directly with a captive insurer does not shift the parent’s risk of loss, insuring through an unrelated intermediate insurer does. According to this argument, Clougherty has shifted its risk because Fremont remains liable in the event Lombardy defaults or becomes insolvent. It is true that in the event of Lombardy’s insolvency, all of Clougherty’s subsequent workers’ compensation claims would be paid by Fremont and would have no direct or indirect effect upon Clougherty’s assets. However, the risk of Lombardy’s insolvency does not affect the risk-shifting inquiry for several reasons.

First, the chance that Lombardy will default is extremely small. In view of the capitalization requirements imposed by the State of Colorado, insolvency could ordinarily occur only after Lombardy had paid far more than the number of claims that could reasonably be anticipated. Second, in reviewing the captive insurer program, we should use the same business assumption that the parties used in undertaking it, namely, that Lombardy will pay all of the claims it agreed to pay. The arrangements for payment of claims by Fremont in the event of insolvency constitute only a secondary aspect of the underlying transaction. Third, it is the payments made to Lombardy that are at issue here. Those payments were made as compensation for Lombardy’s assumption of the risk of Clougherty’s workers’ compensation claims, not for its assumption of the risk of its own insolvency. Fremont is the one that assumes that risk and is paid separately for so doing. The risk that Lombardy will become insolvent is irrelevant to the question whether the amounts paid to Lombardy are insurance premiums. Fourth, Clougherty’s argument would lead to a perverse and absurd result. Every insured party faces the theoretical risk that its insurance carrier will become insolvent and that its losses will not be covered (except where state law provides for outside compensation). That risk cannot be eliminated. If the existence of insolvency risk determined whether insurance existed, then no one who purchased an insurance policy would be able to deduct the premiums he paid: The purchaser would always retain the risk that his insurance company would become insolvent. (The result would be the same whether the policy were purchased directly or through an intermediate insurer, since in the latter case both insurers could theoretically become insolvent.) In short, Clougherty’s argument makes no practical sense. 12

*1307 While we have said earlier that we need not consider whether the economic family concept conflicts with the doctrine of Mo-line Properties, we have no hesitation in concluding that neither the rationale we employ nor the result we reach is inconsistent with that doctrine. Moline Properties requires that related corporate entities be afforded separate tax status and treatment. It does not require that the Commissioner, in determining whether a corporation has shifted its risk of loss, ignore the effect of a loss upon one of the corporation’s assets merely because that asset happens to be stock in a subsidiary. Because we only consider the effect of a covered claim on Clougherty’s assets, our analysis in no way contravenes Moline Properties.

III. Conclusion

The parent of a captive insurer retains an economic stake in whether a covered loss occurs. Accordingly, an insurance agreement between parent and captive does not shift the parent’s risk of loss and is not an agreement for “insurance.” Premiums paid by the parent to the captive, whether directly or through an unrelated insurer, may not be deducted by the parent as insurance premiums. Revenue Ruling 77-316, and our opinion in Carnation, both of which concern captives insuring their parents, reach the correct result — a result that does not conflict with Moline Properties. Because Clougherty is the parent of its captive insurer Lombardy, the amounts paid by Clougherty to Fremont and then to Lombardy are not insurance premiums. As such, they may not be deducted as necessary business expenses under 26 U.S.C. § 162 (a). The decision of the Tax Court is

AFFIRMED.

. A “captive insurance company” is a corporation organized for the purpose of insuring the liabilities of its owner. At one extreme is the case presented here, where the insured is both the sole shareholder and only customer of the captive. There may be other permutations involving less than 100% ownership or more than a single customer, although at some point the term “captive” is no longer appropriate.

. Id.; Stearns-Roger Corp. v. United States, 577 F.Supp. 833 , (D.CoIo.1984), aff'd, 774 F.2d 414 (10th Cir.1985); Beech Aircraft Corp. v. Commissioner, 84-2 U.S.Tax Cas. (CCH) 9803, 54 A.F. T.R.2d (P-H) ¶ 84-5392 (D.Kan.1984), aff'd, 797 F.2d 920 (10th Cir.1986); Clougherty Packing Co. v. Commissioner, 84 T.C. 948 (1985); Humana, Inc. & Subsidiaries v. Commissioner, T.C.M. (P-H) ¶ 85,426 (1985); Mobil Oil Corp. v. United States, 8 Cl.Ct. 555 (1985). Cf. Crawford Fitting Co. v. United States, 606 F.Supp. 136 (N.D.Ohio 1985) (deductions allowed for amounts paid to captive insurer that was owned by individuals rather than by parent corporation); see text infra at 1303-04.

. That Clougherty was only indirectly the sole shareholder of Lombardy does not affect our analysis of the captive insurer question, nor does either party suggest it should. The analy *1299 sis that underlies the result we reach below, that Clougherty has not insured itself because it retains an economic stake in whether a covered loss occurs, applies regardless of whether Clougherty's ownership of Lombardy is direct or through its Arizona subsidiary.

. Notwithstanding Clougherty’s argument to the contrary, nothing in Consumer Life Insurance Co. v. United States, 430 U.S. 725 , 97 S.Ct. 1440 , 52 L.Ed.2d 4 (1977), requires us to consider a definition of insurance other than the one enunciated by the Court in Le Gierse .

. While the Supreme Court in Le Gierse expressly stated that insurance must exhibit both the shifting and the distributing of risk, the resolution of that case depended only on the absence of risk shifting. Here, too, we need not consider whether Clougherty’s captive insurer arrangement exhibited risk distribution because we conclude that Clougherty did not shift its risk.

. In fact, Le Gierse died less than a month after entering into the insurance and annuity contracts.

. [D]omestic corporation Y and its domestic subsidiaries paid amounts as casualty insurance premiums to M, an unrelated domestic insurance company. This insurance was placed with M under a contractual arrangement that provided that M would immediately transfer 95 percent of the risks under reinsurance agreements to S2, the wholly-owned foreign "insurance” subsidiary of Y. However, the contractual arrangement for reinsurance did not relieve M of its liability as the primary insurer of Y and its domestic subsidiaries; nor was there any collateral agreement between M and Y, or any of Y’s subsidiaries, to reimburse M in the event that S2 could not meet its reinsurance obligations.

Rev.Rul. 77-316 at 77:374. The only differences are the percentage of risk reinsured and the site of incorporation of the captive, which are of no consequence.

. The Tax Court also found that, unlike other captive insurer cases where government experts testified in support of the economic family concept, there was no evidence to support that concept here. 84 T.C. at 956 ; see Stearns-Roger, 577 F.Supp. at 833 ; Beech Aircraft, 84-2 U.S.Tax Cas. (CCH) at ¶ 9803. It is not clear to us why any such evidentiary foundation was considered a prerequisite to addressing the economic family theory.

. One factual difference of note was that the captive in Mobil Oil, while wholly owned, did insure at least some parties other than its parent and affiliates. The claims court, however, found that this difference did not affect the result.

. We distinguish Crawford Fitting on its facts and express no view as to the correctness of its reasoning or result. While the Service argued that the economic family concept applied to the facts in Crawford Fitting, it has indicated its willingness to conclude that insurance exists where a captive insurer has 31 unrelated shareholders, none of whom has a controlling interest, and the captive insures only its 31 shareholders with the limitation that no one policy may exceed five percent of the captive’s total insured risks. Rev.Rul. 78-338, 1978- 2 C.B. 107 .

. The record suggests that payments are made in the manner stated in the text. However, our analysis would be the same whether Clougherty pays claims and then seeks reimbursement from Fremont or Lombardy, Fremont pays them and collects from Lombardy, or Lombardy pays them directly. As we explain in text infra, the critical fact is not which party initially pays claims but rather whether Clougherty avoids the impact of the economic costs. Under any of the methods of paying claims, Clougherty ultimately suffers an adverse economic impact.

. Clougherty also argued below that the risk of loss can shift in noninsurance transactions involving related taxpayers, e.g., sales or loans from a parent company to its subsidiary, and *1307 that by analogy risk of loss must shift in an insurance transaction. On appeal, Clougherty does not pursue this argument. Accordingly, we do not consider it here. We note, however, our agreement with the majority of the Tax Court that cases involving risk of loss in non-insurance transactions between related entities are simply not relevant in determining whether insurance exists.