AMERICAN FINANCIAL GROUP AND CONSOLIDATED SUBSIDIARIES, Plaintiff-Appellee, v. UNITED STATES of America, Defendant-Appellant, 678 F.3d 422

OPINION SUTTON, Circuit Judge. In 1995, an insurance company made three accounting changes to its annuity business. The changes allowed the company to deduct $59 million from its federal taxable income over the next ten years, or so it thought. The Internal Revenue Service rejected the deduction, and the company sued for a refund, which the district court granted. Because the company’s calculations followed the accounting principles in place at the time it issued the annuities in question, we affirm. I.

Unlike the regulation of other financial-services industries, the regulation of insurance traditionally has been handled by the States. See McCarran-Ferguson Act of 19^5, ch. 20, 59 Stat. 33 (codified as amended at 15 U.S.C. §§ 1011-1015). The National Association of Insurance Commissioners helps to coordinate the state-based regulations, creating a series of model statutes and regulations that its members — the top insurance officials from the fifty States — may advance in their home jurisdictions.

The Association also periodically releases “Actuarial Guidelines,” which answer questions about the application of the model laws. See NAIC, Financial Condition Examiners Handbook 9-1, R. 26-16 at 1. In 1995, the Association released one such guideline, Actuarial Guideline 33, describing how insurance companies should handle accounting questions connected to annuities sold after 1980. The new guidance prompted Great American Life Insurance Company to change the way it calculated financial reserves for roughly 200,000 annuity contracts it had issued over the prior fifteen years.

Based on this guidance, Great American increased its reserves by approximately $59 million— about 1.2 percent. The company’s parent (American Financial Group) claimed a deduction for part of that increase on its federal taxes for the following year and sought to do the same for the next nine years after that. See 26 U.S.C. § 807(f) (requiring taxpayers to spread such deductions over ten years). The IRS balked.

It concluded that insurers could not use Guideline 33 in calculating reserves for annuity contracts issued before its effective date because the guidance differed from the accounting method applicable at the time the company issued those annuities. American Financial filed an administrative appeal and, when that failed, paid the disputed taxes under protest. It then filed this lawsuit, seeking to recover $11 million in overpayments and several million more in interest.

The district court concluded that Guideline 33 clarified the pre-1995 requirements rather than changing them, meaning that this method described an approach that had always been available and meaning that American Financial could apply it to existing annuities. The court granted American Financial’s motion for

summary judgment. II. A. When an insurance company issues an annuity contract, it promises to pay the purchaser a stream of revenue tomorrow in exchange for fixed amounts of money today. 28 Bertram Harnett & Irving I. Lesnick, Appleman on Insurance 2d § 173.05[G][1]. To ensure that insurance companies live up to these promises, States require them to hold reserves sufficient to cover their long-term future liabilities. Jeffrey K. Dellinger, The Handbook of Variable Income Annuities 301. In establishing these reserves requirements, the States lean heavily on the National Association of Insurance Commissioners.

In 1976, the National Association adopted a model law for calculating reserves, known as the Standard Valuation Law and eventually enacted by all fifty States. The Association also periodically issues proposed regulations (for States to promulgate) and guidance interpreting the model law. These state-law reserves requirements have federal tax implications. The Internal Revenue Code permits insurance companies to deduct increases in their annuity reserves on their federal income taxes, see 26 U.S.C. § 807(b), and it requires them to treat decreases in reserves as income, id. § 807(a).

In view of the extensive state regulations in this area, it should come as no surprise that the Internal Revenue Code refers to and incorporates the state regulations, sprinkling technical acronyms along the way. The Code requires companies to calculate their reserves using “CARVM,” id. § 807(d)(3)(A)(ii), which stands for “the [State] Commissioners’ Annuities Reserve Valuation Method prescribed by the National Association of Insurance Commissioners which is in effect on the date of the issuance of the contract,” id. § 807(d)(3)(B)(ii).

The point is that, when it comes to the federal-tax consequences of increasing or decreasing their annuity reserves, insurance companies must follow the reserve-valuation method (the CARVM) “prescribed” by the National Association in effect on the date the company issued the annuities. One thing is clear under this language: If the National Association of Insurance Commissioners replaces the existing model for reserves calculations (the Standard Valuation Law) or materially amends it, a company could apply that law for tax purposes only to contracts issued after its effective date.

In this instance, the relevant provision of the Standard Valuation Law did not change after 1976, meaning that it indisputably applies to the annuity contracts at issue in this case, all entered into between 1981 and 1995. Also clear is this: If the National Association materially changes any of the model regulations — relevant to reserves calculations — a company could apply a new regulation for tax purposes only to contracts issued after its effective date. In this instance, the relevant regulations did not materially change between 1981 and 1995.

The rub is what happens when the National Association offers guidance through an Actuarial Guideline — here Actuarial Guideline 33. Does this kind of guidance amount to a new reserve-valuation method that may apply only to annuities issued after 1995? We think not. In the first place, when the National Association issues a guideline, the guidance does not naturally call to mind a “prescribed” change in the reserve-valuation method.

The Code requires companies to calculate their reserves using “CARVM,” 26 U.S.C. § 807(d)(3)(A)(ii), which means they must follow the method “prescribed by the National Association of Insurance Commissioners which is in effect on the date of the issuance of the contract,” id. § 807(d) (3) (B) (ii). The relevant

definitions of “prescribe” suggest that guidance offered by the National Association does not suffice. “To make an authoritative ruling,” “[t]o lay down rules, laws, etc.; to dictate, direct,” or “to impose authoritatively,” Oxford English Dictionary (3d ed. 2007), all convey something more authoritative than the kind of advice and clarification a guideline gives. These

definitions instead evoke the kinds of hard-and-fast rules established by the National Association’s model laws or regulations. Unlike the guidelines, the model laws and regulations are designed to be enacted by state legislatures or promulgated by administrative agencies, both of which make them “authoritative.” See, e.g., NAIC Model Laws, Regulations and Guidelines 820-1, State Adoption (listing statutory incorporation of the Standard Valuation Law in each State). States do not enact guidelines into law.

They instead appear in an appendix — Appendix C, if you are interested — of the Financial Condition Examiner’s Handbook, a $250 publication used by state examiners when going over an insurance company’s books. See NAIC Store — Financial Regulation Publications, http://www.naic.org/store_ pub_fin_receivership.htm (last visited April 30, 2012). But the reader need not take just a dictionary’s word for it.

The author of the guidelines — the National Association of Insurance Commissioners — stated that they do not have a prescriptive effect, both in an amicus brief filed in this case and in other publications as well.

As far back as 1980, the organization wrote that the guidelines “are not intended to be viewed as statutory revisions but merely as a guide to be used in applying a statute to a specific circumstance.” See 1 Proceedings of the National Association of Insurance Commissioners 515, 789-90 (1981), available on Lexis at 1981-4 NAIC Proc. 515, 789-90. “Such actuarial guidelines are not to be construed as mandatory upon [state insurance] commissioners, but a commissioner may wish to apply them when he or she encounters situations where he or she feels that the underlying statutes are ambiguous or unclear.” 1 Proceedings of the National Association of Insurance Commissioners 515, 556 (1980), available on Lexis at 1980-4 NAIC Proc. 515, 556.

This view suggests a manual of Frequently Asked Questions, not the kind of authoritative directions that “prescribe! ]” intimates. Be that as it may, we need not — and thus do not — hold that guidelines may never “prescribe! ]” a change to the Commissioners’ method. Even if it were possible for a guideline to prescribe a new valuation method in the abstract, Actuarial Guideline 33 does not amount to such a change in the here and now. By its terms, this guideline disclaims any such effect.

In offering this guidance, the National Association explained that the guideline “does not constitute a change of method or basis from any previously used method.” R. 19-4 at 3. At least from the perspective of the standard-setting body, the positions of which Congress codified in the Internal Revenue Code, the possibility that this guideline altered the existing rules submits to one response: asked and answered. The State Commissioners did not then, and do not now, think Guideline 33 displaced any existing requirements.

The government concedes as much, as it does not dispute that the guideline is consistent with CARVM as it existed before the guideline was adopted. With no indication that the guideline permits something CARVM prohibited, the most one can say is that the guideline narrows the range of appropriate practices. But if American Financial’s current approach was acceptable before Guideline 33, the reality that the guideline confirmed its acceptability by saying it was the only correct

interpretation of CARVM is of no moment. We are concerned only with whether CARVM, as it was “in effect on the date of the issuance of the contract,” permitted American Financial’s current approach. 26 U.S.C. § 807(d)(3)(B)(ii). The author of CARVM and Actuarial Guideline 33 says that it did, and the government points to nothing to the contrary.

Section 807(d)(3)(B)(ii) thus permits American Financial to apply the accounting method approved by Guideline 33 to these annuities. A Revenue Ruling issued by the IRS a year before American Financial made these reserve changes confirms the point. See Revenue Ruling 94-74, 1994-2 C.B. 157. There, too, an insurer changed the way it calculated its life insurance reserves. In prior years, the insurer assumed it would receive all of its policy premiums on the first day of a year and pay out all of its benefits on the last day of the year, a valuation approach known infelicitously as a “curtate function.” Id. at 2.

The company decided to adopt a new calculation — the “continuous function” — reflecting a different assumption: that the company would space out premiums and benefit payments evenly. Id. The change from one function to the other led the insurer to increase its reserves. Id.

The IRS permitted the company to deduct the increase, noting that because the National Association of Insurance Commissioners’ “prescribed reserve method ... does not specify the use of either curtate or continuous functions,” the insurer should “apply the same timing assumption ... for tax purposes as it uses in computing the state statutory reserves for the contracts.” Id. at 6. Because the insurer had changed to the continuous function for calculating its state reserves, it could make the same change in calculating its tax reserves.

The parallel to this case is unmistakable: (1) the reserving method prescribed by the Commissioners in effect at the time the contracts were issued permitted either of two approaches; (2) the taxpayer permissibly applied one approach for several years; (3) the taxpayer then changed to the other permissible approach in calculating its state-law reserves, leading to higher reserve figures. As Revenue Ruling 94-74 demonstrates, § 807 permits the taxpayer to make the identical change for federal tax purposes. B. The IRS offers several responses.

It submits that “prescribe” could mean something less definitive than we suggest. One of the ten

definitions offered in Webster’s Third New International Dictionary, it points out, is “to lay down authoritatively as a guide, direction, or rule of action,” Br. at 31 (emphasis added), suggesting that a guideline could contain a new legal prescription. Yes, but even if some guidelines theoretically could make “authoritative!]” changes to these accounting rules, Guideline 33 does not — for the reasons just given.

Trying to leverage the glimmer of ambiguity offered by this one definition, the IRS resorts to legislative history, namely the following paragraph: [I]f the NAIC [National Association of Insurance Commissioners] in 1984 clarifies what ... factors are to be disregarded ... under the CARVM, then this clarification is to be considered in effect on the date of issuance of such contracts.

The conferees recognize that giving retroactive effect to a NAIC recommendation in this instance is an exception to the general rule that reserves must be computed for tax purposes under the method prescribed by the NAIC (or the prevailing state

interpretation thereof) in effect on the date of issuance of the contract. H.R. Conf. Rep. 98-861,1984 U.S.C.C.A.N. 1445, 1740. As the IRS reads this report, it suggests that the committee understood the “general rule” to be that guidelines would apply only to contracts issued after their effective dates. Its thinking proceeds in two steps. One: in the report, the committee acknowledges that the National Association of Insurance Commissioners was considering whether to make a new “recommendation” on whether certain factors should be considered in calculating annuity reserves.

Two: without a special dispensation from the Conference Committee granting an “exception” for Guideline 33, American Financial’s case must fall under the “general rule” that guidelines apply only to contracts issued after their adoption. But it is not that simple. For one thing, the Conference Committee does not say whether the contemplated “recommendation” was a new guideline or a new regulation or a new model law. If the recommendation contemplated a new regulation or model law, it might be true that it could apply only to after-issued contracts.

Yet based on the language of the report, there is no way to know what the word “recommendation” means. There being, alas, no legislative history to this legislative history, we are left with more room for conjecture than reliance on committee reports already produces. Other aspects of this 1984 committee report confirm why it does not apply to this 1995 guideline. According to the report, the committee would grant an exception to the non-retroactivity rule if the Commissioners’ new recommendation came out by the end of 1984 rather than, say, January 1, 1985.

The government offers no explanation for this quirk, but a feature of the tax law unique to 1984 helps to explain the time limit and in turn helps to explain why the “general rule” language in the report would not apply to Guideline 33 in 1995. Congress recognized that the 1984 Deficit Reduction Act would force companies to recalculate their reserves for federal tax purposes, and that the new calculations would generally be lower than the old ones (because CARVM, as required in the 1984 Act, produces lower reserves than the methods previously employed).

Such decreases ordinarily produce taxable income for the insurance companies. 26 U.S.C. § 807(a). Because that “income” was attributable to a change in the accounting method required by the tax code, Congress chose to exempt it from taxation through a “fresh start” provision. Congress instructed companies to calculate their reserves as of the last day of the 1983 tax year using their old reserve method, then to calculate reserves for the first day of the 1984 tax year using the new method. Deficit Reduction Act of 1984, Pub. L. No. 98-369, 98 Stat. 494, § 216(b)(2)(B)(ii).

Any excess of the former over the latter (almost all of which would stem from the change in reserving methods rather than real-world changes in the business) could be disregarded when the company filed its 1984 tax return. Id. § 216(b)(2)(B)®. In saying that new positions taken by the Commissioners could be applied “retroactively],” the committee was explaining that any recommendation issued in 1984 could be taken into consideration to re calculate, for tax purposes, a company’s annuity reserves as of January 1, 1984, and that those re-worked reserves should receive the benefit of the fresh-start provision.

Nothing comparable is occurring here. American Financial did not re-calculate its prior-year reserves and demand a tax advantage due to a paper-only difference in its business; it prospectively changed its real-world reserves using the

interpretation of CARVM laid out in Guideline 33 and sought the ordinary tax treatment for such increases in reserves for the future. A paragraph of legislative history speaking to the fresh-start provision applicable to 1984 tax returns says little about American Financial’s situation, if indeed anything at all. All of this reminds us of the advice “of the wag who said, when the legislative history is doubtful, go to the statute.” Greenwood v. United States, 350 U.S. 366, 374, 76 S.Ct. 410, 100 L.Ed. 412 (1956).

The reality remains that the statute refers only to “CARVM” changes “prescribed” by the National Association of Insurance Commissioners. 26 U.S.C. § 807(d)(3)(B)(ü). And the reality remains that this guideline does not operate like a prescription and at all events does not alter a prior CARVM practice. Because text and context provide a reasonably clear answer to the question at hand, legislative history susceptible to multiple readings has little work to do.

But even if we adopted the “general rule” that the government extracts from the legislative history, that at most establishes a general approach, not an across-the-board one. Surely it is not the case that every new guideline implicates the anti-retroactivity principle. Were that true, a guideline that did no more than confirm that a particular approach had always been permissible would have the paradoxical effect of making that approach impermissible for all previously issued contracts. Even the government does not contend that Congress created such a rule.

And if some guidelines do not raise antiretroactivity concerns, what is wrong with considering whether a guideline authorized an accounting method that was previously incorrect — whether, that is, the guideline authorized the taxpayer to use a valuation method that was not permissible under the version of CARVM “in effect on the date of the issuance of the contract”? 26 U.S.C. § 807(d)(3)(B)(ii). Under this approach, American Financial surely is entitled to the deduction.

Neither party has demonstrated a prior consensus among the States on the issues Guideline 33 addressed, and as shown the IRS makes no claim that the approach described in the guideline would have been unacceptable before 1995. Even if we accepted the government’s position that companies generally may apply new guidelines only to contracts issued after their adoption, then, that general rule would not apply here.

Yes, the IRS responds, but even if American Financial could have applied the same reserve-valuation method that it has used since 1995 in the years before Guideline 33, “the short answer is that taxpayer did not do so.” Reply Br. at 21. Short? Yes. Sufficient? No.

The government’s sole basis for this argument turns again on legislative history — a sentence from a 1984 House Ways and Means Committee report describing the committee’s intent to “imposte] specific rules for the computation of tax reserves that result in a reserve which approximates the least conservative (smallest) reserve that would be required under the prevailing law of the states.” H.R. Rep. 98-432, 1984 U.S.C.C.A.N. 697, 1042, cited in Reply Br. at 22.

From this sentence in this committee report, the IRS extracts a contingent rule: any time a State accepts a company’s calculation for one year, the company is forbidden to use a different, higher one in future years. Because Ohio accepted American Financial’s smaller reserves in previous years when the company was relying on the old accounting method, the argument goes, American Financial cannot increase the size of its reserves now. The IRS misreads the report.

The committee acknowledged it was embracing “specific rules” — the rules required by the National Association — not a contingent requirement that turned on the smallest reserves the States would accept. The IRS’s own analysis in Revenue Ruling 94-74, as shown above, demonstrates that changes from one acceptable valuation method to another are permissible, even if they result in reserves larger than the ones previously accepted for State purposes. The statute also undercuts the government’s point. Subsection 807(

f) accounts for situations in which companies recalculate their reserves, providing that changes (whether increases or decreases) are to be distributed over a prospective ten-year span, just as American Financial sought to do. The provision thus undermines the government’s argument that, having calculated reserves using one appropriate method in the past, American Financial cannot apply a different appropriate method today. The IRS adds that “deductions are a matter of legislative grace and should therefore be narrowly construed.” Br. at 30 (quoting Nichols v. United States, 260 F.3d 637, 653 (6th Cir.2001)).

It is not clear, as an initial matter, that the canon applies. The relevant language of § 807(

d) does not itself create a deduction. It describes the accounting method companies must use in calculating their reserves, leaving it to other provisions to determine whether the reserves warrant deductions (§ 807(b)), result in income (§ 807(a)) or should be ignored for tax purposes altogether (as with the fresh-start program). For this reason, the meaning of § 807(

d) in the abstract favors neither the taxpayer nor the tax collector. But the more important point, the one essential here, is that the canon comes at the end, not the beginning, of the interpretive process. At this point in the process, we see no additional work for the canon to do, and assuredly not the kind of work that would undo all of the other indicators of statutory meaning apparent here. III. For these reasons, we affirm.