An Update on the Determination of Fair and Reasonable Unpaid Loss Reserves
On November 21, the U.S. Tax Court issued another decision dealing with the determination of fair and reasonable unpaid loss reserves under section 832(b)(5) of the Internal Revenue Code of 1986, as amended. In Physicians Insurance Co. of Wisconsin v. Commissioner, T.C. Memo 2001-304, 82 T.C.M. (CCH) 918, Judge Michael B. Thornton held that the unpaid loss reserves claimed by the insurance company were not fair and reasonable and, therefore, were not allowable in their entirety under I.R.C. section 832. The case comes on the heels of the recent decisions of Minnesota Lawyers Mutual Insurance Co. v. Commissioner, T.C. Memo 2000-203, 79 T.C.M. (CCH) 2234 (also by Judge Thornton) and Utah Medical Insurance Ass’n v. Commissioner, T.C. Memo 1998-458, 76 T.C.M. (CCH) 1100 (decided by Judge John O. Colvin). With one minor exception, all of the cases resulted in a court determination of the amount of fair and reasonable unpaid loss reserves consistent with contemporaneous actuarial determinations of the taxpayers’ actuaries, despite arguments by the government’s experts that the actuaries in question had not determined the reserves appropriately.1
My article entitled “A Comparison of Utah Medical and Minnesota Lawyers: Fair and Reasonable Unpaid Loss Reserves,” The Insurance Tax Review, October 2000, p. 583, described in detail the facts of both cases and why they illustrate the importance of contemporaneous actuarial determinations of unpaid loss reserves. The article concluded:
Where a property and casualty insurance company’s reserves are actuarially determined, and the company provides persuasive testimony in court by an expert of the reasonableness of the actuarial determination, the company should be successful in sustaining the amount of its unpaid loss reserves. On the other hand, for the Internal Revenue Service to be successful, it either must be able to point to a tangible reason why the reserves are not reasonable or the company must fail to satisfy its burden of proof.2
As described in detail below, in Physicians Insurance, Judge Thornton rejected the taxpayer’s attempt to sustain unpaid loss reserves that “were not actuarial in nature” but instead were based on certain “qualitative concerns.” 82 T.C.M. (CCH) at 927. In this regard, the court noted that, “the taxpayer must be prepared to objectively validate that the methods and assumptions it relied upon to make its estimates are reasonable.” Id. at 925.
In addition, this article describes the rulings in TAM 200115002 (April 13, 2001), which was issued by the Internal Revenue Service after the decisions in Minnesota Lawyers and Utah Medical but before Physicians Insurance. Of course the TAM does not resolve the ultimate factual question — the reasonableness of the taxpayers’ reserves. Nevertheless, in rejecting the arguments put forward by the IRS agents in the case, TAM 200115002 does provide the view of the IRS National Office on the meaning of Treas. Reg. section 1.832-4 and the “fair and reasonable” standard.
Physicians Insurance Co. of Wisconsin v. Comm’r
Overview of the Facts
Physicians Insurance Co. of Wisconsin (Physicians) is a property and casualty insurance company that provides medical malpractice insurance for physicians and hospitals. Physicians issued “claim-made” and to a lesser extent “occurrence-based” medical malpractice insurance policies. The company employed Tillinghast-Towers Perrin (Tillinghast) to provide actuarial services. In 1991, Tillinghast began to use five specific actuarial methods in its annual determination of Physician’s unpaid loss reserves. In 1993 and 1994 (the years at issue), Tillinghast also factored in ultimate loss estimates that it had selected in the preceding year (prior selections). As a result, Tillinghast provided a single point estimate to Physicians which exceeded the estimates arrived at using the five specific actuarial methods. Tillinghast’s 1993 and 1994 estimates of unpaid losses were $74,027,009 and $77,029,796, respectively. For purposes of reporting the unpaid losses on the annual statement, Physicians’ management further increased the estimates to $81,391,000 and $84,559,000, respectively, an increase of almost 10 percent in each year. In its final report for 1993 and 1994, Tillinghast noted the difference between its recommended reserves and the reserves carried on the annual statement, but did not otherwise discuss the discrepancy.
In addition to the review by Tillinghast, Coopers & Lybrand (Coopers) also performed a contemporaneous review of Physicians’ 1993 and 1994 loss reserves and annual statements as part of its audit of the company. For 1993, the Coopers actuary noted that the unpaid losses on the annual statement were 9.9 percent higher than the Tillinghast point estimate, but concluded that the deviation was “acceptable from an actuarial perspective, indicating that a reserve range of minus 5 percent to plus 10 percent was common for Tillinghast analyses.” 82 T.C.M. (CCH) at 921. Coopers’ nonactuarial auditors then determined that the annual statement reserves exceeded the acceptable range under Coopers’ in-house guidelines. However, they concluded that no adjustment to the financial statements was necessary because, among other things, the impact on retained earnings was not considered overly significant. Similarly, for 1994, Coopers’ actuaries noted the annual statement reserves were 9.8 percent over the Tillinghast estimate. Coopers’ auditors also concluded the “redundancy” was acceptable from an actuarial perspective and that no adjustment was necessary because: (i) the loss reserves were within 10 percent of the Tillinghast point estimate; (ii) Physicians only wrote medical malpractice, which is extremely volatile; and (iii) Physicians was a young company with adequate, but not extremely significant, amounts of historical results to assess the adequacy of loss reserves.
Finally, the Wisconsin Commissioner of Insurance retained the actuarial firm of AMI Risk Consultants, Inc. (AMI) to review the 1993 annual statement unpaid losses. In November 1994, AMI determined that the reserves were reasonable because they fell within a range that AMI determined had a low end of $81,300,000 and a high end of $93,539,000.
In its audit, the IRS attacked both the additional 10 percent add-on determined by management, as well as the Tillinghast point estimate and argued that the proper reserves were $46,508,000 and $45,549,000 for 1993 and 1994, respectively. These figures represent approximately 57 percent and 54 percent of Physicians’ unpaid loss reserves as reported on the annual statement for those years. The IRS agents’ computations employed a computer program known as Exhibitmaker, which was developed by Coopers.
The issue presented to the court was whether Physicians correctly reported its undiscounted unpaid losses for purposes of computing “losses incurred” under I.R.C. section 832(b)(5). The court held that Tillinghast’s point estimates were fair and reasonable estimates of unpaid losses which were allowable under I.R.C. section 832 and, that the 10 percent management add-on was not fair and reasonable and, therefore, was not allowable. The court noted that the Tax Court previously held that when the annual statement methodology is predicated on estimates, those estimates must be the “best possible.” However, the court stated, “It does not mean that there is (or could be, except in hindsight) a single correct estimate.” Id. at 925. Rather, as stated earlier, “the taxpayer must be prepared to objectively validate that the methods and assumptions it relied upon to make its estimate are reasonable.” Id.
The government’s expert witnesses, Frederick Kilbourne and David Otto, concluded that Tillinghast’s point estimates were too high and contended that Tillinghast’s work violated professional actuarial standards then in place. They also argued that management’s 10 percent add-on contradicted Tillinghast’s actuarial work. (Kilbourne was also an expert witness for the government in Utah Medical.)3 Physicians offered expert testimony of Owen Gleeson, Robert Sanders, and James Hurley. (Gleeson and Hurley both testified for the taxpayer in Utah Medical). Gleeson found the Tillinghast point estimates to be reasonable, but did not specifically address management’s 10 percent add-on or prepare independent estimates of unpaid losses. Sanders found Physicians’ unpaid loss estimates and the range around the Tillinghast point estimates of plus or minus 10 percent to be reasonable. He relied on the volatile nature of the medical malpractice industry, and Physicians’ “relative immaturity” and growing evidence of a deteriorating claims environment as support, but Sanders could not refer to an actuarial stan- dard or practice evidencing the 10 percent tolerance. Finally, Hurley responded to three criticisms that the government’s witnesses made against the analysis conducted by AMI, and concluded that even if the criticisms were correct and resulted in adjustments to AMI’s point estimate of the 1993 unpaid losses, the losses would only be reduced to $71,915,000 (still far in excess of the IRS’s determination of fair and reasonable reserves).4
Physicians’ primary argument was that because it reported the same estimates of unpaid losses on both its annual statement and tax return, the estimates should be accorded deference for federal income tax purposes, provided that it used “good faith business judgment” in setting those reserves. This argument was partially based on the changes to the Internal Revenue Code in 1986 and 1990, which Physicians argued “continued, and in some ways strengthened, deference to the Annual Statement.” Id. at 925. Judge Thornton rejected those arguments because, as illustrated by Hanover Ins. Co. v. Commissioner, 69 T.C. 260, 272 (1977) aff’d, 598 F.2d 1211, 1217 (1st Cir. 1979), acceptance of annual statement reserves for tax purposes could result in the acceptance of clearly unfair or unreasonable reserve estimates. The court also disagreed (without even discussing the technical merits of Physicians’ argument in its opinion) that the code requires such deference and found that, even if there is some deference, it does not preclude the IRS from adjusting the annual statement estimates of unpaid losses for purposes of determining taxable income.
The court then focused on the expert testimony and the actuarial basis for the determination of Tillinghast’s point estimates and the 10 percent add-on by management. With respect to the add-on, Physicians contended that its decision to increase Tillinghast’s point estimates were not actuarial in nature but instead were based on certain “qualitative concerns,” particularly “regarding the basic actuarial assumption that past experience will replicate itself in the future.” Id. at 927. Therefore, Physicians argued, the 10 percent add-ons resulted in “an appropriate expression of conservatism based on the implied range around Tillinghast’s unchanged point estimate.” Id.
The court disagreed for several reasons. First, it stated that Physicians had provided no contemporaneous documentary evidence as to how it determined the amount of the add-ons. In that regard, Physicians had represented to Tillinghast that all factors which would materially affect loss reserves had been disclosed to Tillinghast, leaving Physicians with little or no ability to argue that certain factors only known to them had not been taken into account by Tillinghast in determining the reserves. Second, the court refused to give any weight to a list of factors suggested by Physicians as support for its adjustments to Tillinghast’s point estimates, because the company could not establish that the list related to the years at issue.5 Third, the court re- lied on the lack of evidence of any actuarial standard supporting the reasonableness, in all cases, of a range of 10 percent. Although Tillinghast’s actuary (Reichle) had argued that such a range is consistent with the uncertainty inherent in a point estimate, he conceded it was impossible to quantify generally the extent of a range that would be appropriate in every case. The court also found it important that Coopers had never actually concluded that the annual statement reserves were reasonable and that Coopers had required no adjustment to the financial statements largely because it found the difference to be insignificant. Consequently, the court concluded that Physicians had not presented sufficient evidence to support the reasonableness of a 10 percent implied range around the Tillinghast point estimate and that Physicians had failed to establish that its 10 percent add-on was based on reasonable methods or assumptions.
With respect to the remaining reserves, the court found that the Tillinghast point estimates were fair and reasonable. It noted that the government’s expert witnesses’ estimates of unpaid losses differed in amount not only from Tillinghast point estimates but also from the IRS determinations in the statutory notice.6 Distilling the government’s evidence, the court noted that the IRS’s primary criticism of the Tillinghast methodology was that it considered prior selections which resulted in higher unpaid loss estimates. The court did not find Kilbourne and Otto’s criticisms of Tillinghast’s method to be persuasive. Instead, it found the testimonies of Tillinghast and Coopers to show that the use of prior selections was standard practice and was justified in the present circumstances and that the estimates and assumptions were reasonable. In doing so, the court relied on their holding in Utah Medical where the use of an analogous lookback method was found to be proper. The court went further and criticized the government’s witnesses saying that the tone of their reports was more indicative of advocacy than of the detached neutrality demanded of expert witnesses and, consequently, the usefulness and credibility of the government’s experts were given little weight.7 Finally, the court rejected the argument that the reserves were not fair and reasonable because, in hindsight, they turned out to be excessive, again citing the holding in Utah Medical, and that the evidence showed Tillinghast took developing redundancies into account in developing the estimates in ques- tion.
Overview of the Facts
Taxpayers include a holding company and several prop- erty and casualty insurance subsidiaries which have multiple lines of business including auto bodily injury, commercial multiple peril, workers compensation, general liability, and homeowners. Taxpayers’ internal actuaries prepared yearly reports, containing their estimates of subsidiaries’ unpaid losses for annual statement purposes, which were reviewed by several independent actuarial consulting firms. One of the independent actuaries determined a range of reasonable reserve values each year, and the reserve figures which Taxpayers booked for both annual statement and tax purposes were always within that range. The reports of other independent actuaries projected only a point estimate. In some cases Taxpayers’ booked reserve was less than a particular point estimate; in other instances the booked reserve may have exceeded the point estimate of an outside actuarial firm, but (we understand) by a margin of only a few percentage points of the total reserve. IRS agents proposed adjustments to the unpaid loss reserves for Years 2 through 5 based on the analysis of an outside actuary hired by IRS Exam.
Taxpayers also were potentially liable for A&E claims attributable to policies written prior to Year 1 under the Comprehensive Environmental Response, Compensation and Liability Act of 1980, Pub. L. No. 96-510, 94 Stat. 2779 (CERCLA). For this liability, Taxpayers estimated loss reserves separately from the other reserves. Taxpayers’ A&E reserves were also reviewed by an outside actuarial firm, which concluded that Taxpayers’ reserves were significantly deficient. A review by the IRS’ actuary determined that the reserves for Years 2 through 4 were reasonable, but that the reserves for Year 5 were overstated and should be reduced.
In TAM 200115002, the IRS National Office addressed several arguments of the IRS agents based on literal readings of select words of Treas. Reg. sections 1.832-4(b) and -4(a)(5) (as it existed prior to 2000).8 In rejecting each of the agents’ arguments, the TAM is consistent with a long history of cases, including the holdings in the three most recent cases discussed in this series of articles. Significantly, the TAM even goes so far as to reject the IRS’s litigating position in Physicians Insurance and Utah Medical (which the court in those cases also rejected) that IRS agents may rely on events subsequent to the tax year in issue to demonstrate that a taxpayer’s reserves were not reasonable. That is, in those cases, the IRS argued that the reserves were not fair and reasonable because they ultimately proved, based on hindsight, to be redundant, but the court found that the reserves were reasonable at the time, despite subsequent events. Likewise, in TAM 200115002, the IRS concludes that the “use of a hindsight analysis is not appropriate,” which may perhaps finally reign in the IRS litigating position.
In TAM 200115002, the IRS first compared the language of Treas. Reg. section 1.832-4(a)(5) (which requires the taxpayer’s estimate of unpaid losses at the close of each year to “represent actual unpaid losses as nearly as it is possible to ascertain them”) to the language in Treas. Reg. section 1.832-4(b) which provides:
Every insurance company to which this section applies must be prepared to establish to the satisfaction of the district director that the part of the deduction for “losses incurred” which represents unpaid losses at the close of the taxable year comprises only actual un- paid losses. . . . These losses must be stated in amounts which, based upon the facts in each case and the company’s experience with similar cases, represents a fair and reasonable estimate of the amount the company will be required to pay. Amounts included in, or added to, the estimates of the unpaid losses which, in the opinion of the district director, are in excess of a fair and reasonable estimate will be disallowed as a deduction. The district director may require any insurance company to submit such detailed information with respect to its actual experience as is deemed necessary to establish the reasonableness of the deduction for “losses incurred.”
While acknowledging that Treas. Reg. section 1.832-4(a)(5) appears to require that reserve estimates be as accurate as possible, the TAM concludes, based on a review of the relevant background behind the two provisions, that both provisions impose the same standard. The IRS also concluded that where a reserve is fair and reasonable in amount, it is assumed that the taxpayer has based the reserve on actual unpaid losses as nearly as it is possible to ascertain them. The TAM interpreted the phrase “must represent actual unpaid losses as nearly as it is possible to ascertain them,” as directing taxpayers to base their estimates of unpaid loss reserves on losses that have actually been incurred. The TAM went on to conclude that the IRS may only adjust a taxpayer’s estimate of unpaid losses if the estimate is not “fair and reasonable.” Therefore, the relevant standard is fair and reasonable.
The TAM also rejected the agents’ contention and concluded that the terms “fair” and “reasonable” are closely related, if not synonymous, and they do not impose two separate requirements on the determination of reserves. Furthermore, the IRS stated that the term “fair” does not mean “more than reasonable” when used in connection with reasonable. The term “will be required to pay” in Treas. Reg. section 1.832-4(b) does not impose a requirement that the taxpayers’ estimates of unpaid losses represent the “most accurate” estimate, but only requires that taxpayers based their estimates on actual losses, not losses that might be incurred in the future.
The TAM further concluded, contrary to the assertions of the agents, that the reference in Treas. Reg. section 1.832-4(b) to “the opinion of the district director” does not create an abuse of discretion standard. Rather, taxpayers bear the burden of proving by a preponderance of the evi- dence that the IRS adjustment is incorrect, which may be accomplished by establishing that their reserve estimate was fair and reasonable in accordance with Treas. Reg. section 1.832-4(b). In this regard, the TAM notes (as the court in Physicians Insurance later concluded) that Taxpayers can- not illustrate that their reserves are fair and reasonable merely by showing that such reserves are identical to the reserves shown on the annual statement.
In addition, the TAM commented on the importance of Rev. Proc. 75-56, 1975-2 C.B. 596, which states that “the standard of reasonableness” concerning the computation of unpaid losses will be that set forth in Treas. Reg. sections 1.832-4(a)(5) and -4(b). The IRS observed that the revenue procedure does not set forth a particular testing methodology.
In the final issue involving the “fair and reasonable” standard, the TAM concluded that the use of a hindsight analysis is not appropriate under Treas. Reg. section 1.832-4(b) and Rev. Proc. 75-56. Instead, the TAM concluded that they should test reserves on the basis of a historical development analysis.
The TAM carefully explains what it means by both “hindsight” and “historical development.” This discussion should prove constructive in resolving future controversies about loss reserves. The TAM states:
. . . We understand the term “hindsight” as referring to an analysis of the subsequent development of a particular year’s unpaid loss estimate for the purpose of determining whether the estimate proved sufficiently accurate to be considered fair and reasonable. We understand the term “historical development” as referring to an analysis of the subsequent development of prior years’ unpaid loss estimates for the purpose of determining whether the unpaid loss estimate for the present year is fair and reasonable.
The TAM goes on to set out in detail an approved appropriate historical development test.
For example, if a company’s unpaid loss estimate for Year 10 were at issue, the subsequent development of that estimate would be the amount of losses paid by the company in Year 11 and each year thereafter with respect to losses contemplated by the estimate for Year 10. A hindsight analysis would attempt to determine whether the estimate for Year 10 was fair and reasonable by comparing the original Year 10 estimate with the amount subsequently paid with respect to losses contemplated by the original estimate, i.e., amounts paid in years after Year 10. In contrast, a historical development analysis would attempt to deter- mine whether the estimate for Year 10 was fair and reasonable by comparing the unpaid loss estimates made in Years 1 through 9 with the amounts subsequently paid in Years 1 through 10 with respect to losses contemplated by each of those estimates.
While this simple approach to evaluating reserves may prove favorable to taxpayers, it is justified, as the TAM explains, because it focuses on information available at the time the reserve is set.
. . . Thus, whether a taxpayer’s estimate is fair and reasonable depends upon the information the taxpayer had at the time it made the estimate. In determining whether a taxpayer’s estimate of calculating unpaid losses is fair and reasonable, therefore, the Service should consider the historical development of the taxpayer’s unpaid loss estimates for the years prior to the year in issue.
Furthermore, the TAM almost seems to caution agents against proposing adjustments in isolated instances where reserves may have later proved to be too high.
Use of a historical development analysis is also consistent with the Service’s goal of addressing the potential abuse associated with excessive unpaid loss estimates; since a taxpayer that has overestimated its unpaid loss deduction for a particular taxable year will typically be required to include any excess portion of the estimate in income in subsequent years, only a consistent pattern of overstating estimates of unpaid losses leads to substantial unwarranted tax deferral.9
Furthermore, the IRS also ruled that even though the Taxpayers’ unpaid losses attributable to their A&E liabilities are inherently uncertain, that fact alone does not establish that the reserves are non-deductible contingency reserves. In that regard, the reserve may be deductible if the estimates are fair and reasonable in amount. Finally, the IRS concluded that the Taxpayers’ enhanced reliance on actuarial techniques in Year 5 did not result in a change in method of accounting if Taxpayers were basing their estimates on actual unpaid losses prior to Year 5. However, the IRS noted, if a taxpayer changes the manner in which it estimates actual unpaid losses, the IRS may make an adjustment if the change yields a result that is not fair and reasonable.
As stated previously, the key to sustaining determinations of unpaid loss reserves is contemporaneous actuarial support for the amounts reported on the annual statement and then used to compute taxable income. Both Minnesota Lawyers and Physicians Insurance illustrate that unsupportable and identifiable additions to actuarially determined reserves are likely to be challenged by the IRS and, if no actuarial basis is provided for such additional amounts, the IRS’s determination is likely to be sustained. In certain cases, companies can obtain a range of estimated amounts from the actuaries in order to offer management the ability to choose a conservative reserve within the given range. Both Utah Medical and TAM 200115002 provide support for the use of ranges.10 If only a point estimate is available under actuarial guidelines, any additional reserves must be actuarially determined in order to satisfy the fair and reasonable standard.
TAM 200115002 is also significant in its prohibition of IRS reliance on hindsight in disallowing reserves. While the Internal Revenue Manual has for many years proscribed hindsight analysis by agents, that Manual provision has been disregarded by agents on occasion. IRM, Handbook 4232.1, Techniques Handbook for Specialized Industries — Insurance, section 366. The TAM’s principal explanation of the reason hindsight is inappropriate should minimize or eliminate the use of hindsight analysis by agents in the future.
22 Insurance Tax Review 53
1 In Minnesota Lawyers, the court adopted the government expert’s point estimate which was slightly higher than the point estimate recommended by the taxpayer’s actuary for one of the years in question.
2 Similarly, and more importantly, the IRS in TAM 200115002 (described herein) concludes that, “A taxpayer’s estimate of unpaid losses will typically be considered fair and reasonable for tax purposes if the taxpayer estimates its unpaid losses on the basis of a recognized methodology that is appropriate for its particular line of business, calculates the estimate in accordance with actuarial standards, and properly takes into account its prior experience.”
3 In Utah Medical, Kilbourne testified that Tillinghast’s estimates were not fair and reasonable because Tillinghast’s (i) lookback approach was flawed; (ii) method ignored point estimate results; and (iii) conclusion resulted in an overstated range of estimates. Kilbourne then computed his best estimates of the unpaid loss reserves which were lower than the lowest end of the range computed by the taxpayer’s actuaries. In Utah Medical, the court rejected Kilbourne’s argument that the lookback approach was flawed, and, ultimately, sustained the taxpayer’s determination of unpaid loss reserves.
4 The three criticisms were: (1) the use of incorrect premium data in AMI’s application of the Bornhuetter-Ferguson actuarial method; (2) inappropriate interpolation of loss development factors in AMI’s application of the paid loss development actuarial method; and (3) inappropriate selection of factors generally in the AMI analysis. Id. at 926.
5 The list notes the following qualitative factors: (i) a trend toward increased claims against corporations; (ii) possible liability to the Wisconsin Patients Compensation Fund for settlements or bad faith claims; (iii) turnover in clients; (iv) pending tort reform legislation; (v) greater uncertainty with “new states” and other lines of business recently offered; and (vi) increased litigation resulting from petitioner’s aggressive claims defense. Id. at 927, n. 16.
6 For 1993, respondent’s determination was $46,508,000 and respondent’s experts determination was $50 million. For 1994, respondent’s determination of $45,549,000 and the expert’s determination was $39 million.
7 For example, the court noted that Otto’s report stated that Tillinghast’s estimates reflected a conscious decision to overstate Physician’s reserves, a statement which he could not support during his testimony.
8 Under the current regulations, Treas. Reg. section 1.832-4(a)(14) contains the definition of “losses incurred.”
9 At this point in footnote 11, the TAM explained further —Therefore, where a taxpayer’s unpaid losses in years prior to the year in issue have consistently proven to be greater than necessary, the taxpayer should be aware that its methodology of calculating unpaid loss estimates may be flawed, or that the data underlying such estimates may be unreliable.
10 In TAM 200115002, the IRS stated, "Furthermore, an estimate chosen from a range of reasonable estimates may, in certain cases, satisfy the requirements set forth in the regulations. See Utah Medical Ass’n v. Commissioner, T.C. Memo 1998-458. Since different actuar- ies or different actuarial techniques may [end up] at different results in at- tempting to arrive at a “most accurate” estimate, the language “as nearly as it is possible to ascertain them” does not preclude the use of ranges. Whether a range is appropriate will depend on the facts of each case; for example, a range may be warranted with regard to more volatile lines of insurance. Simi- larly, the breadth of such a range will also depend on the facts of each case."