Bernard T. Swift, Jr. & Kathy L. Swift

• Court: United States Tax Court
• Decided: February 1, 2024
• Docket: 5354-18
• Status: Unpublished

Opinion

United States Tax Court

T.C. Memo. 2024-13

BERNARD T. SWIFT, JR. AND KATHY L. SWIFT, Petitioners

v.

COMMISSIONER OF INTERNAL REVENUE, Respondent

__________

Docket Nos. 13705-16, 5354-18, Filed February 1, 2024. 11261-19. __________

Jaime Vasquez, A. Leonides Unzeitig, Charles J. Muller III, and Stuart H. Clements, for petitioners in docket Nos. 13705-16 and 5354-18.

Jaime Vasquez, A. Leonides Unzeitig, and Charles J. Muller III, for petitioners in docket No. 11261-19.

Sharmeen Ladhani, David W. Sorensen, Alexander R. Roche, Vivian Bodey, and John Robert Gordon, for respondent in docket No. 13705-16.

Sharmeen Ladhani, Sheila R. Pattinson, Alexander R. Roche, Vivian Bodey, and John Robert Gordon, for respondent in docket Nos. 5354-18 and 11261-19.

MEMORANDUM FINDINGS OF FACT AND OPINION

URDA, Judge: Bernard T. Swift, Jr., is the founder of more than a dozen urgent care centers and physical rehabilitation facilities in and around San Antonio, Texas. From 2004 through 2015 Dr. Swift’s businesses supplemented their traditional insurance by purchasing assorted policies from microcaptive insurance companies that Dr. Swift

Served 02/01/24 2

[*2] also controlled. 1 The premiums paid to the microcaptives dwarfed more traditional insurance premiums, making for healthy deductions for petitioners, Dr. Swift and his wife, Kathy L. Swift. Relying on section 831(b), 2 the microcaptives themselves paid no tax on the premium income received from their sister entities, investing the money as directed by Dr. Swift.

On each of their joint federal income tax returns for 2012 through 2015, the years at issue, the Swifts deducted, inter alia, more than $1 million in premiums paid to the microcaptives and miscellaneous legal fees. The Internal Revenue Service (IRS) examined this arrangement for each of these years and concluded that the microcaptives used the trappings of insurance for purposes of tax avoidance and financial planning. It accordingly issued notices of deficiency that, inter alia, disallowed the claimed deductions and determined accuracy-related penalties.

Consistent with our decisions in Avrahami, 149 T.C. 144, Reserve Mechanical Corp. v. Commissioner, T.C. Memo. 2018-86, aff’d, 34 F.4th 881 (10th Cir. 2022), Syzygy Insurance Co. v. Commissioner, T.C. Memo. 2019-34, Caylor Land, T.C. Memo. 2021-30, and Keating v. Commissioner, T.C. Memo. 2024-2, we will sustain the IRS’s determinations. 3

FINDINGS OF FACT

We held a remote special trial session in these cases via ZoomGov. We incorporate by reference the stipulation of facts, including the jointly

1 “A ‘captive insurance company’ is a corporation whose stock is owned by one

or a small number of companies and which handles all or a part of the insurance needs of its shareholders or their affiliates.” Caylor Land & Dev., Inc. v. Commissioner, T.C. Memo. 2021-30, at *8 n.4; see also Harper Grp. v. Commissioner, 96 T.C. 45, 46 n.3 (1991), aff’d, 979 F.2d 1341 (9th Cir. 1992). “A ‘microcaptive’ is a small captive insurance company,” i.e., one that “take[s] in less than $1.2 million in premiums.” Caylor Land, T.C. Memo. 2021-30, at *8 n.4; see also Avrahami v. Commissioner, 149 T.C. 144, 179 (2017). 2 Unless otherwise indicated, statutory references are to the Internal Revenue

Code, Title 26 U.S.C. (I.R.C. or Code), in effect at all relevant times, regulation references are to the Code of Federal Regulations, Title 26 (Treas. Reg.), in effect at all relevant times, and Rule references are to the Tax Court Rules of Practice and Procedure. All dollar amounts are rounded to the nearest dollar. 3 The Commissioner argues in the alternative that these insurance

transactions lack economic substance. We need not address this argument in light of our conclusion that the captive insurance arrangement did not constitute insurance. 3

[*3] stipulated exhibits contained therein. The Swifts lived in Texas when they timely filed their petitions in these cases.

I. Dr. Swift and His Medical Businesses

Dr. Swift received his doctorate in osteopathy from the College of Osteopathic Medicine in Des Moines, Iowa. He then served in the U.S. Air Force as a flight surgeon assigned to Randolph Air Force Base in San Antonio, Texas. Dr. Swift worked as an emergency physician on the side, ultimately going full time after he left the military in 1980.

A. Texas MedClinic

In 1982 Dr. Swift decided to open the Texas MedClinic (Clinic), an urgent care center, rather than continue as an emergency physician. He operated Clinic as a sole proprietorship during the years at issue (2012 through 2015), with the Swifts reporting its tax information by means of Schedules C, Profit or Loss From Business, attached to their annual returns.

Clinic was successful and grew. It expanded to 13 facilities by 2010. As of 2015, Clinic operated 18 locations in San Antonio, New Braunfels, and Austin, Texas.

Clinic’s practice focused on urgent care and occupational medicine services, as well as minor surgical procedures such as the removals of “lumps and bumps, cysts, . . . [and] skin tags.” The concept of urgent care refers to “the treatment of urgent but non-life-threatening problems.” Clinic thus “see[s] less critically ill patients . . . [with] the usual litany of sprained ankles, sore throats, runny noses, eye injuries, and whatnot.” Occupational medicine encompasses both “caring for injured workers” and “deal[ing] with regulatory issues such as drug testing, regulatory physicals, DOT physicals, [and] asbestos physicals.”

From its founding through 2015, approximately 350 physicians worked at Clinic as independent contractors. During each of the years at issue, approximately 75 independent-contractor physicians worked at Clinic. Clinic averaged gross income of $47,110,423 during the years at issue.

B. Other Businesses

Dr. Swift founded two other businesses relating to medical services. In 2006 he formed an entity focused on sports rehabilitation 4

[*4] (Rehab). 4 As of the end of 2014, Rehab operated eight physical rehabilitation centers, all in Clinic locations and facilities.

During 2012 through 2015 the Swifts filed Schedules C for Rehab as part of their returns. Rehab was a more modest venture with around 12 employees and an average gross income of $1,697,494 during the years at issue.

Dr. Swift opened a separate dermatology practice, Derm Docs, PLLC (Derm Docs), in 2007. Derm Docs had one practicing dermatologist and did not enjoy the success of the other Swift entities, closing its doors in 2012. During its last year Derm Docs brought in $224,073 in gross income.

II. Traditional Insurance for Swift Entities

During all years relevant to these cases, Clinic purchased, separate from any captive policies, both medical malpractice insurance and assorted other lines of general commercial insurance.

A. Commercial Medical Malpractice Coverage

1. Coverage

Clinic bought claims-made medical malpractice insurance policies for all years relevant to these cases. 5 During 2012 through 2015, these policies each had a one-year term and featured no deductible, a $500,000 per-claim limit, and an aggregate limit of $1.5 million.

Clinic designated the same date as both the policies’ effective date and retroactive date, thereby limiting the coverage to those claims that both occurred and were reported during the one-year policy term. The nature of an urgent-care practice meant that Clinic would be aware “fairly quickly” of a catastrophic incident, and during all years relevant to these cases, Clinic’s policies allowed Dr. Swift to trigger coverage by reporting potential claims to carriers himself.

4 In 2011 Rehab was converted from a sole proprietorship into a limited liability

company. 5 A claims-made policy is generally “[a]n agreement to indemnify against all

claims made during a specified period, regardless of when the incidents that gave rise to the claims occurred.” Claims-Made Policy, Black’s Law Dictionary (8th ed. 2004). Such a policy “can also include a retroactive date that limits how far back the incident could have happened.” Avrahami, 149 T.C. at 154. 5

[*5] In 2003 Clinic spent $93,164 on premiums for its commercial medical malpractice insurance, which declined to $66,639 by 2011. The downward trend continued during 2012 through 2015, the years at issue, with Clinic paying $67,059, $34,997, $34,997, and $41,997, respectively, to cover itself and its sister entities. Dr. Swift attributed the decrease to Texas’s efforts to limit medical malpractice liability, which had become law in 2003.

2. Claims History

From its founding until September 1, 2004, Clinic experienced four medical malpractice claims (out of approximately 2,150,000 patient visits), which resulted in settlements ranging from $35,000 to $450,000. Dr. Swift credited Clinic’s success on this front to his proactive approach to risk management, comprising (1) identifying potential weaknesses in doctors, which he would address with “classroom type programs,” (2) reviewing and providing feedback on each doctor’s medical records for the first 90 days of employment, and (3) a peer review process for physician complaints.

From January 1, 2001, through November 1, 2012, Clinic’s insurers paid $1,352,500 relating to medical malpractice claims. No such claims were filed from November 1, 2012, through December 1, 2015.

B. Other Commercial Insurance Coverage

In addition to medical malpractice policies, Clinic bought multiple other lines of commercial insurance. From 2003 through 2011 Clinic paid average annual premiums of $50,738 to obtain such coverage. Clinic paid premiums of $70,030, $80,637, $78,980, and $66,306 during 2012 through 2015, respectively, for these policies.

Clinic purchased a Premier Businessowners Policy to insure buildings it owned up to their replacement cost (valued between $32 million and $45,545,000 during the years at issue), as well as business income and equipment damage. It also bought coverage for doctors’ equipment (averaging $6,645,375 during the years at issue), accounts receivable of $250,000, and fine arts (with an average value of $124,400 during the years at issue). Both of these policies included terrorism coverage at no charge. Rounding out Clinic’s commercial insurance were policies for fiduciary liability, crime coverage, workers’ compensation, and business automobiles, as well as a $10 million umbrella liability policy (which included excess coverage in 2014 and 2015). 6

[*6] III. First Foray into Microcaptive Insurance

A. Understanding the Arrangement

The year 2004 marked a sea change in Dr. Swift’s insurance approach. Before 2004, Clinic complemented its commercial medical malpractice policies with a loss reserve of $500,000 to cover claims falling outside the policy terms. In 2004 Dr. Swift discontinued the loss reserve and set up the first of three microcaptive insurance companies, Castlegate Insurance Co., Ltd. (Castlegate).

Dr. Swift began “explor[ing] the possibility of creating a captive insurance company” because Clinic’s medical malpractice “premiums were rising at a rate that was . . . inappropriate given our attentiveness to managing [its] risks and the insurance company’s risks.” Dr. Swift claimed a desire to “control these policies and . . . control the[se] claims” in a way that he could not with his traditional insurance carriers. He also wanted to pay for “medical malpractice cover[age] with––before tax dollars.”

Dr. Swift’s interest led to Celia Clark, a New York lawyer who specialized in the formation and maintenance of small insurance companies for closely held domestic businesses. Also accompanying Dr. Swift from the start was Tim Schultz, his certified public accountant (CPA). These discussions did not begin with specific insurance offerings that Dr. Swift wanted, but with (1) the financial advantages of operating a microcaptive insurance company and (2) Ms. Clark’s understanding of the legal requirements for a captive to be deemed a true insurance company for tax purposes.

As to the former, the parties discussed section 831(b), which shields from taxation premium income of an insurer with less than $1.2 million in annual premiums (i.e., a microcaptive insurance company). Ms. Clark explained that premiums paid by Clinic to a microcaptive controlled by Dr. Swift would be untaxed and could be used “to purchase [additional] clinics or just land, leased to [Clinic.]”

As to the latter, Ms. Clark emphasized the need for the microcaptive to obtain risk distribution to be considered an insurance company. Relying on her interpretation of our Court’s caselaw and IRS actions, Ms. Clark asserted that 30% of the microcaptive’s total premiums would need to come from unrelated businesses in order for the arrangement to pass muster. Ms. Clark advised that “[t]he 30% 7

[*7] unrelated insurance business is usually achieved through the purchase (subscription) of a piece of reinsurance that relates to a pool.”

During their talks, Ms. Clark explained that if Dr. Swift had “a firm number in mind that [he] desired to use in payment of captive premiums, we can work backward from that to determine appropriate types and levels of coverage.” After deciding to proceed with the arrangement, Dr. Swift and Ms. Clark decided that “tail” medical malpractice coverage was appropriate. As Dr. Swift explained in 2004, this coverage insured claims “that arise from acts committed prior to [the effective date of Clinic’s commercial malpractice policy], but reported after that date.”

To price the coverage, Dr. Swift originally turned to his insurance agent, who reported that “there is no way to obtain an actual quotation covering back to the hire date of the physicians . . . as no carrier will write ‘tail’ coverage only.” “Those carriers who are willing to write ‘tail’ coverage charge approximately 200% of the mature premium for an unlimited extended reporting period.” The insurance agent estimated $976,704 as the cost of the coverage with a $200,000 per-claim limit and a $600,000 aggregate limit.

B. Castlegate

Dr. Swift and Ms. Clark incorporated Castlegate in the British Virgin Islands in October 2004. Castlegate was owned by a limited partnership that, in turn, was controlled by the Swifts through a limited liability company and a family trust in which they were trustees.

Castlegate operated as an insurance company from 2004 through 2009. Each year it reported total premiums just under the $1.2 million cap of section 831(b), as would allow for its premiums to go untaxed. The percentage breakdown of premiums reflected Ms. Clark’s view of risk distribution, with no more than 70% of total premiums attributable to the tail coverage and no less than 30% attributable to Castlegate’s participation in one of three risk distribution programs sponsored by Ms. Clark.

1. Tail Coverage

Castlegate issued its first tail insurance policy for Clinic on November 7, 2004, using the same effective date as its commercial medical malpractice policy. This policy was drafted by Ms. Clark and Dr. Swift by marking up a preexisting commercial medical malpractice 8

[*8] policy. In 2004 the tail insurance covered all current and former Clinic physicians, featuring a $200,000 per-claim limit, a $3 million aggregate limit, and a $15,000 deductible. The policy reflected a premium of $976,700, as estimated by the insurance agent.

Dr. Swift thereafter retained Anthony Bustillo of KPMG LLP (KPMG), to determine premium estimates for the tail coverage. These estimates included (1) a “pure premium” for the coverage derived from industry and internal KPMG client data and (2) what is known in the industry as “expense loads,” which were intended to capture administrative expenses, profit, and contingencies. Dr. Swift ultimately based the premium amount on this information.

From the start, KPMG proved itself flexible in conducting this analysis. KPMG’s first assignment was to provide a premium estimate for Castlegate’s 2004–05 policy year (even though the policy had been written and payments had been made). Mr. Bustillo determined an indicated premium using a 22% load factor to account for expenses. He, however, had failed to take into account the policy’s $15,000 deductible. When alerted to that fact, Mr. Bustillo generated a revised report that incorporated the deductible, raised the load factor to 30%, and produced the same premium, according to Dr. Swift.

For the next policy year, KPMG computed a premium estimate for insuring “current [Clinic] physicians and [Clinic] (the entity)” and not former physicians in the previous year’s policy—because, as Dr. Swift explained in an email, “the calculated premium was going to exceed the $840,000 if [they] included this group.”

From 2004 through 2009, Clinic’s premiums for traditional medical malpractice insurance averaged $100,538 per year, while the tail coverage premiums averaged $825,830. During this time five claims were made against the policy, resulting in total liabilities of $615,000.

2. Risk Distribution Programs

Castlegate participated in three risk distribution programs affiliated with Ms. Clark during its years as an insurance company. As most relevant here, in 2009 Castlegate participated in a risk distribution program involving Pan American Reinsurance Co., Ltd. (Pan American). Under this program, a business insured by a Clark-affiliated microcaptive (like Clinic) would buy terrorism insurance directly from Pan American. Pan American, in turn, would enter into an agreement with the affiliated microcaptive (think Castlegate) under which the 9

[*9] microcaptive would reinsure a portion of the blended risk from the underlying terrorism insurance.

The premium received by the microcaptive for such reinsurance coverage would match the amount that the affiliated business had paid to Pan American for the terrorism insurance. This amount would also represent at least 30% of the microcaptive’s total premiums. We take judicial notice of our holdings in Avrahami that “Pan American was not a bona fide insurance company” and that “we cannot find that the policies it was issuing were insurance.” Avrahami, 149 T.C. at 184–90; see Fed. R. Evid. 201.

3. Investments

Castlegate’s healthy premiums coupled with modest expenses and claims history meant that it had significant resources on hand. By the end of 2008, Castlegate had invested (1) approximately $2 million in three real estate partnerships that bought land for the future use of Clinic and (2) approximately $3 million in mutual funds and equity securities. In 2009 Castlegate stopped issuing policies, and in 2011 it was domesticated and turned into an investment company.

IV. The Second Generation of Swift Microcaptives

A. Formation

In October 2010 Ms. Clark assisted Dr. Swift with the formation of two new microcaptives, Castlerock Insurance Co., Ltd. (Castlerock), and Stonegate Insurance Co., Ltd. (Stonegate) (collectively, Swift captives). These captives were incorporated in the Federation of St. Christopher and Nevis (St. Kitts) and licensed to operate as insurance companies from the St. Kitts Financial Services Regulatory Commission. 6

1. Preliminaries

The record contains no feasibility study verifying the need for two captives. Dr. Swift focused on premiums by stating:

6 In 2006 Ms. Clark assisted in drafting captive insurance legislation for the

island of St. Kitts, which was reviewed and revised by legislators and attorneys from St. Kitts. 10

[*10] The only policies that have been issued to date for Castlegate . . . are medical malpractice, terrorism, and Credit Re policies. . . . In order to get closer to maxing out the premiums, I will be looking for other risks to insure, both from related entities, and also unrelated. Celia has indicated that a reasonable limit for the terrorism insurance . . . is probably not more than $400,000. Actually, we’re thinking in terms of having the captives write a stop loss policy for [Clinic’s] self insured health insurance plan. That premium might come in at $150,000- 200,000 for example. And then there might be others as well. Whatever might come up, and whatever you and Celia can come up with for unrelated party premiums.

A few months later he reaffirmed that “the number of other policies [the Swift captives] write will depend on the total premiums,” estimating that “about $1.5-1.6M in total premiums . . . will probably be all [Dr. Swift] can come up with through [Clinic] this year.” For her part, Ms. Clark verified that “[Clinic] will not be limited to $1,200,000 for the group.”

The business plans of Castlerock and Stonegate set forth concerns about the ability to obtain medical malpractice coverage at a reasonable cost, risks presented by government regulation of the healthcare industry, and threats presented by competition. The plans also indicated that the Swift captives would participate in a “pool with other captive insurance companies . . . [that] will cover business risks relating to terrorist attacks and other hazards . . . [because,] [i]n the United States, it is difficult and expensive to obtain appropriate levels of terrorism risk insurance.”

2. Execution

The Swift captives were incorporated in October 2010, with each owned by a trust for one of the Swifts’ two children. Dr. and Mrs. Swift were the trustees of both trusts, and neither of the children had any role in the operation of the Swift captives. Dr. and Mrs. Swift also served as treasurer and assistant treasurer, respectively, for each of the Swift captives, which gave them authority to open any bank, brokerage, or investment accounts required by the captives.

The directors of the Swift captives were two Kittian companies, Corporate Solutions, Ltd., and Heritor Management Ltd. (Heritor). The 11

[*11] former supplied the principal office, registered office, and registered agent for the captives. The latter provided services including claims management and processing, obtaining insurance licenses, monitoring St. Kitts regulatory compliance, maintaining statutory insurance records, executing insurance policies, and invoicing. Under Heritor’s service agreement with the Swift captives, claims were to be approved and paid out unless Heritor believed coverage was unclear or if the claimed losses were over $50,000, at which point the claim was referred for a second opinion.

Both Swift captives elected to be treated as domestic corporations for United States tax purposes under section 953(d) and elected under section 831(b)(2)(A)(ii) to be taxed solely on investment income. Each of the Swift captives was initially capitalized for $36,500, and no additional capital contributions were made to either captive through the end of 2015.

B. Direct Insurance Offerings

The Swift captives sold multiple lines of insurance to Clinic (covering Clinic, Rehab, and Derm Docs), with each policy featuring the same pricing and an agreement between the two to share any liability. In total the captives received the following premiums between 2010 and 2015: 12

[*12] Policy 2010 2011 2012 2013 2014 2015 Medical $910,562 $946,134 $812,860 $741,644 649,396 $593,522 Malpractice Administrative 101,000 155,936 170,000 102,000 52,000 55,000 Actions Business 116,000 286,136 435,000 134,000 — — Income Business Risk 60,000 73,674 68,000 44,000 — — Indemnity Computer 14,000 22,869 50,000 33,000 — — Operations & Data Employment 31,000 43,256 70,000 37,000 29,000 22,000 Practices Liability Litigation 77,000 150,364 225,000 43,000 18,000 10,000 Expense Cost of — — 14,000 14,000 7,000 7,000 Defense Terrorism 360,000 719,300 540,000 — — — Political — — — 231,000 384,000 384,000 Violence TOTAL $1,669,562 $2,397,669 $2,384,860 $1,379,644 $1,139,396 $1,071,522

As the years at issue are 2012 through 2015, we will provide an overview of the policies as in effect those years.

1. Malpractice Tail Coverage

a. Policy Terms

The tail coverage hewed closely to the model developed by Castlegate, with a $15,000 deductible, $300,000 per-claim limit, and a $6 million annual aggregate limit. The policy covered claims relating to professional services rendered at a Clinic facility between the date the respective physician began work at Clinic and the first day of the policy period. The policy provided coverage “only if” the claim was reported within ten days of receipt of written notice by the insured. Endorsements to the policies included lists of former and current physicians who were covered, including approximately 140 physicians who had left Clinic between 1983 and 2000. 13

[*13] b. Pricing

Like Castlegate, the Swift captives engaged KPMG (again, Mr. Bustillo) to prepare annual actuarial pricing analyses. To provide its estimates, KPMG received information including copies of the commercial medical malpractice policies, a report showing all medical malpractice claims and payments since Clinic’s inception, and a list of all currently employed and formerly contracted physicians and their length of time with Clinic. KPMG’s pricing analyses relied on internal KPMG and industry sources on the ground that the Swift captives’ loss experience was “too sparse to be fully credible on its own.”

To determine the premium for each physician, KPMG stated that it looked to (1) mature claims-made premiums for the policy limits of $300,000 per claim and $6 million in the aggregate, (2) reporting lag factors, (3) physician specialties and (4) the period of exposure. KPMG then offered a range of loads “to contemplate administrative expenses, profit and contingencies,” which would be added to the pure premium to determine the premium. Dr. Swift ultimately chose the load percentage, determined the deductible amount, and drafted the medical malpractice policies.

2. Nonmedical Coverage

At various times during 2012 through 2015 the Swift captives offered nine other lines of insurance:

• Administrative Actions: These policies covered legal expenses, fines, and assessments arising from an administrative action or disciplinary proceeding instituted against the Swift entities. These policies had a per-event limit of $1 million and an aggregate limit of $3 million.

• Business Income: These policies covered business income that the Swift entities lost as the result of reputational damage, new competition, or a legislative or regulatory change. The policies had a per-event limit of $2 million and an aggregate limit of $5 million. These policies were not purchased in 2014 and 2015.

• Business Risk Indemnity: These policies offered excess coverage for business liabilities caused by “construction defects” or events excluded under the policyholder’s commercial policies—for 14

[*14] example, losses from asbestos, climate change, or fungi. The policies did not extend to cover professional liability and had a per-event limit of $1 million in 2012, and $500,000 in 2013, and an aggregate limit of $3 million in 2012, and $500,000 in 2013. These policies were not purchased in 2014 and 2015.

• Computer Operations and Data: These policies indemnified the Swift entities against increased cost of working and reinstatement of data arising out of computer-related malfunctions at covered Clinic locations. The policies had a per- event limit of $1 million and an aggregate limit of $1.5 million. These policies were not purchased in 2014 and 2015.

• Employment Practices Liability: These policies were excess coverage, insuring against expenses incurred in defending against employee claims, including discrimination, sexual harassment, and retaliation. The policies had a per-event and aggregate limit of $2 million.

• Litigation Expense: These policies were excess coverage, insuring any expenses incurred in defending a legal proceeding related to business activities, prosecuting a third party over a matter pertaining to the business, or obtaining any legal consultation pertaining to the business. The policies had a per-event limit and an aggregate limit of $2.5 million in 2012, $250,000 in 2013, and $100,000 in 2014 and 2015.

• Cost of Defense: These policies were excess coverage, offering insurance for “1) the defense [of ] any Claim instituted against an Insured [or its officers and directors] . . . ; 2) any private or governmental Administrative Action . . . ; and 3) any arbitration, mediation, or other alternative dispute resolution.” The policies had a per-event and aggregate limit of $10,000.

• Terrorism: The policy insured against acts of terrorism, as defined in the Terrorism Risk Insurance Act, as well as assorted acts (such as the dispersion of biological and chemical agents) that result in losses exceeding $100 million, which were committed with the intent to influence or coerce the U.S. government. The policy did “not apply to loss recoverable . . . under other insurance or indemnity,” nor did it cover a loss resulting from acts “occurring in a city with a resident population greater than two million (2,000,000).” The policy had a per-event 15

[*15] and aggregate limit of $6,750,000. This policy was discontinued after 2012.

• Political Violence: The policy insured Clinic for losses to its buildings (including the removal of debris) and its net income against assorted events including acts of terrorism, sabotage, riots, mutiny, civil war, and the use of biological chemical, or nuclear weapons (causing more than $100 million in damage and which were committed with the intent to influence or coerce the United States government). The policy did not apply “to loss recoverable . . . under other insurance or indemnity.” It had a per-event and aggregate limit of $3.3 million and was in effect from 2013 through 2015.

Between 1982 and December 2010 Dr. Swift had not purchased similar coverage for his medical entities from commercial insurance companies. Dr. Swift’s commercial insurance provider was not aware of the existence of the Swift captives, much less these specific lines of coverage, during the years at issue.

These policies featured some unusual terms. Several of the policies identified a claims notification period “as a condition precedent to payment of any benefit.” The policies also provided for payment by promissory note if Clinic “suffers a series of catastrophic loss occurrences that may impair [Clinic’s] solvency.” And the policies each provided for termination “upon the insolvency or bankruptcy of the insured.”

b. Pricing

To price the Swift captives’ nonmedical malpractice policies, Ms. Clark turned to Allen Rosenbach, an actuary for ACR Solutions Group, Inc. Ms. Clark (or someone working for her) would send Dr. Swift’s annual selections and supporting materials to Mr. Rosenbach. This material included information about business metrics, including prior year revenue and expenses, average number of patients, average revenue per patient, payroll, number of employees, prior claims, and changes in business practices.

Mr. Rosenbach prepared a report with premium estimates for each desired line of coverage. The report for each year (and each Swift captive) used stock terms on such estimates: 16

[*16] [T]he base rates and rating factors developed in this report were based on a survey of rating plans obtained from regulatory filings submitted by commercial insurance carriers in the United States of America. . . . Where comparable coverages were unavailable, we used professional judgment to develop reasonable rating guidelines to reflect the expected loss potentials. The Company and/or its representatives supplied key information, both qualitative and quantitative, to help develop the underlying frequency and severity parameters in the base rates. Indicated rates were also reviewed and tempered to reflect historical consistency and rate-on-line ranges.

The reports also included boilerplate regarding coverage comparisons. Specifically, the reports stated that “many of the coverages provided by the Company are either unavailable or unaffordable in the commercial insurance market” and thus “adjustments were made to modify the commercial rates and rating factors to reflect our interpretations of the differences in the captive insurance policies.”

During most of the years at issue Ms. Clark (or one of her employees) provided Mr. Rosenbach with a target for the Swift captives’ premiums. In November 2012, for example, an attorney working with Ms. Clark sent Mr. Rosenbach a memorandum indicating that the captives wished to renew six policies then in effect, which had a $366,118 premium for each Swift captive. The attorney further stated that “the client’s total maximum premium is $1,587,140 to be divided evenly between Castlerock and Stonegate,” including general cost of defense and terrorism risk insurance. Mr. Rosenbach ultimately reached an estimated premium of $1,572,000 for the Swift captives’ 2012 nonmedical policies.

Mr. Rosenbach incorporated feedback from Ms. Clark and her team into his analyses. In November 2013 an attorney working with Ms. Clark wrote Mr. Rosenbach a memorandum detailing the policies selected for 2013, which reflected a premium of $786,000 for each of the Swift captives. The attorney further stated that the “client’s total maximum premium target is $458,356 to be divided evenly between [the Swift captives.]” After Mr. Rosenbach provided his initial calculations, the attorney responded that the Swift captives’ “requested level is extremely low this year” and asked if he would “mind seeing if limits 17

[*17] could be lowered to meet the lower premiums?” Mr. Rosenbach complied. He also worked with the attorney to adjust the premium numbers to hit 30% risk distribution rather than 29.7%.

This collaborative approach to premium pricing was repeated in 2014 and 2015. During those years Mr. Rosenbach made premiums higher or lower as Ms. Clark’s employees instructed, with an eye on 30% risk distribution. In 2015 one of Ms. Clark’s employees noted that the premiums were “very lopsided” and asked whether it would be “possible to adjust the limits upwards and allocate more premium to policies other than [political violence]?” In a later email the lawyer further requested Mr. Rosenbach to “raise the limits on Administrative Actions to a level over $200,000,” noting that the “limit should be raised further if needed to achieve the 30+% risk distributions on the cessions.”

C. Claims

Clinic paid the Swift captives a total of $5,975,422 during 2012 through 2015 (following more than $4 million in 2010 and 2011) to obtain the various lines of insurance coverage. In contrast, Clinic submitted three claims to the Swift captives during the years at issue, resulting in total payments of $339,224 (as of the end of September 2015).

The first of the three claims was made under the 2011–12 medical malpractice tail coverage policies, relating to a Texas state court lawsuit. Clinic received service of this suit on February 1, 2012, although it previously had been sent letters dated November 23, 2010, and July 21, 2011, complaining to Clinic about treatment received on November 22, 2010, and threatening legal action. Although the policy required reporting of the claim “within ten (10) days of receipt by [Clinic] of a written notice of a Claim,” a lawyer working for Ms. Clark sought and received an extension after notifying the Swift captives of the lawsuit on November 6, 2012. Heritor ultimately authorized payment of $13,212 with respect to this claim, which was closed in 2015.

The second claim was made on August 29, 2013, under the Litigation Expense and Administrative Action policies in effect for 2011 through 2012. Clinic sought the payment of legal expenses incurred in defending a wrongful termination lawsuit stemming from a dismissal on October 26, 2012. Although the policies provided that a “Claims Notification Period of 30 days from the date of occurrence of an Insured 18

[*18] Event shall exist as a condition precedent to payment of any benefit hereunder,” Heritor granted Clinic’s extension request.

During the pendency of this claim, the “Cost of Defense policies with both Captives [were] exhausted,” so a member of Ms. Clark’s team requested that excess expenses “be applied instead to the Litigation Expenses policies.” Before mediation in that case, another member of Ms. Clark’s team represented that the fact that the policy requires the Swift captives to consent in writing to a settlement “doesn’t necessarily mean that Dr. Swift has to consult with the [Swift] captive[s] before finalizing the settlement.” According to the lawyer, she suspected that Heritor would provide “a letter approving coverage for the future settlement.”

The Swift captives paid a combined total of $108,012, net a contribution of $3,439 from the risk distribution pool in which they were participating at the time (which will be discussed below). Even after closing, Heritor reopened the claim to approve additional expenses.

Clinic provided notice of the final claim on October 16, 2014, under the Administrative Actions policy issued on December 1, 2013. This notice followed a memorandum from Ms. Clark dated September 12, 2014, in which she stated to her “Clients” that their businesses “may have coverage under an Administrative Actions Insurance Policy issued by your captive for legal and administrative fees relating to a pending IRS audit.” Ms. Clark’s firm reported on Clinic’s behalf that the Department of the Treasury had initiated an audit on February 20, 2014. Again, Heritor approved the untimely claim and ultimately authorized payments totaling $275,793 by the end of 2015.

D. Investment of Premiums

During 2010 through 2015 the Swift captives received just over $10 million in premiums. With the relatively small claims and expenses, the Swift captives had considerable ready money, which Dr. Swift invested in (1) real estate, buying and developing property for three urgent care facilities later leased to Clinic, and (2) the stock market, through investment accounts with Fidelity and the ownership of two limited liability companies organized in 2013. In total the Swift captives invested more than $8 million of the premiums received from 2010 through 2015.

Given the illiquid nature of a large portion of the Swift captives’ investment portfolio, Ms. Clark advised Dr. Swift in April 2013 to 19

[*19] indemnify the Swift captives and protect them “against future capital calls they may be unable to meet because of insurance claims.” In June 2013 Dr. Swift issued a put option to both Swift captives that, if needed, required him to purchase either or both Swift captives’ entire interest in the real estate partnership at a price determined by an appraisal process. The agreements then provided that Dr. Swift could pay via a promissory note to be paid in three equal installments, with the first payment to occur about 30 days after the closing date. Later that year, Ms. Clark’s firm recommended that, going forward “the captive[s] keep[] 30% of the annual premium in cash or cash equivalents.”

E. Risk Distribution and Reinsurance Program

1. Structure

During 2012 through 2015 the Swift captives participated in two risk distribution pools affiliated with Ms. Clark: Jade Reinsurance Group, Inc. (Jade), in 2012 and 2013, followed by Emerald International Reinsurance, Inc. (Emerald), in 2014 and 2015. Jade and Emerald were both Alabama captive insurers, formed to “function as . . . vehicle[s] to pool diverse risks ceded to [them] by” Clark-related microcaptive insurance companies.

As Ms. Clark explained to Mr. Rosenbach when requesting help on “the actuarial end,” “[a]fter three years of using terrorism risk through a reinsurance structure to accomplish risk distribution, [her team was] re-designing the pool to include more types of coverage.” The pools thus represented the next links in the evolutionary chain following the Pan American risk distribution program, in which the Swift captives had participated in 2010 and 2011.

At a high level, Jade (and then Emerald) agreed to reinsure a portion of the risks written by participating Clark-affiliated captive insurance companies, with all of the participating captives, including the Swift captives, paying premiums to the pool for such coverage. In turn, the self-same captives each contracted with the respective pool to reinsure a quota share portion of the pool’s blended liability, with the pool paying for this coverage by releasing a percentage of the total premiums that had been paid to it by all the captives. In Ms. Clark’s view the premium amounts retroceded to the captives pursuant to this arrangement constituted unrelated business premiums for risk distribution purposes. She touted in 2014 that both pools were 20

[*20] “independently managed and have been designed and organized to be in compliance with IRS rulings issued in 2012, and to result, in most cases, in risk distribution well above 30%.”

Turning to specifics, Jade and Emerald both grouped the various insurance policies that could be reinsured as part of their pool as Coverage Part A, B, or C. 7 For policies under Coverage Part A, Jade agreed to insure a net loss above $200,000 in 2012 and $100,000 in 2013, subject to limits of liability of (1) 51% of $800,000 over the $200,000 for 2012 and (2) 55% of $900,000 over $100,000 for 2013.

For its part, Emerald divided Coverage Part A policies on a per- occurrence limit of $100,000. For the policies below this line, Emerald agreed to cover (1) 80% of a loss over 50% of the per-occurrence limit in 2014 and (2) 100% of a loss over $50,000 up to the occurrence limit in 2015. For policies above $100,000 in per-occurrence limits, a participating captive was liable for the first $100,000 and Emerald was liable above that amount subject to (1) a maximum of 80% of the $900,000 over the initial $100,000 for 2014 and (2) 100% of the losses up to $1 million for 2015. The aggregate limits were $720,000 and $950,000 for 2014 and 2015, respectively.

In return for this coverage, Jade received a percentage of the original gross policy premium that varied depending on occurrence limits (with higher occurrence limits being charged a lower percentage). Emerald, on the other hand, charged a flat 30.3% reinsurance premium for Coverage Part A policies during 2014 and 2015.

The arrangements with respect to Coverage Parts B and C, including General Cost of Defense and Terrorism or Political Violence policies, respectively, had many fewer moving parts. For Coverage Part B, Jade agreed to pay $10,000 per claim and in the aggregate, while Emerald agreed to 100% liability of each loss occurrence, capped at $15,000.

7 Coverage Part A policies included the following types of insurance policies:

(1) Administrative Actions; (2) Administrative Actions (Physicians); (3) Business Income and Extra Expense; (4) Business Risk Indemnity; (5) Computer Operations and Data; (6) Contract Cancellation; (7) Cyber Protection; (8) Directors and Officers Liability; (9) Employee Fidelity; (10) Employment Practices Liability; (11) Kidnap, Ransom and Extortion; (12) Litigation Expense; (13) Loss of Key Employee; and (14) Tax Indemnity. Coverage Part B consisted of General Cost of Defense Insurance. Coverage Part C consisted of Terrorism or Political Violence Insurance. 21

[*21] For Coverage Part C, both Jade and Emerald agreed to reinsure a percentage of the underlying premiums, depending on the client’s preference. This approach was consistent with Ms. Clark’s view in 2012 (when setting up Jade) that terrorism and political violence policies could be used as necessary to assure the desired risk distribution.

Despite the multitude of steps under each program, Ms. Clark assured participants that “[i]n all cases, the total premiums ceded to Jade [or Emerald] by a [captive] will be at least 30% and will usually be substantially above that.”

The reinsurance by Jade and Emerald was only one side of the coin, however. Jade and later Emerald also entered into quota share retrocession agreements with the participating captives to retrocede a quota share of the pool’s blended risk from each Coverage Part to each participating captive.

In return for effectively reinsuring its reinsurer, each participating captive, including the Swift captives, received premiums corresponding to the quota share of risk retroceded. These amounts would “not be released to the captives immediately” but held in a trust account and released to the participating captives in tranches throughout the year. Pursuant to the Jade pool, “half of the funds [would] be released” from the trust account after 90 days, with the remainder released after 180 days, “less a . . . holdback as a continuing loss reserve until the end of the policy period (one year plus any extended reporting period).”

Emerald tweaked Jade’s approach somewhat. In 2014 Emerald built in “a 5% holdback as a continuing loss reserve until all obligations of the pool [had] been settled and paid,” anticipating that the trustees might be directed to reserve greater amounts. In 2015 the funds were to remain in the trust account for 180 days, “at which time half of the funds not used to pay losses or reserved for expected losses [would] be released.” After 280 days 25% of the original amount would be released (less the funds used to pay losses or which were reserved). The remaining funds would be released “when all obligations of the pool [had] been settled and paid.” If at any time Emerald held amounts “less than those required to pay [l]osses,” the participating captives were required to, “within thirty (30) days of notice[,] provide additional funds . . . for amounts equal to such difference.” 22

[*22] If the respective pool did not have enough money in its accounts to cover a filed claim, each captive in the pool would be asked to pay its fair share of the claim to the pool. In planning conversations with Mr. Rosenbach about Jade, Ms. Clark identified “meaningful deterrents to claims against the pool.” Among other things, she noted that the pool excluded high severity and high frequency lines of insurance, that each captive would need to pay up to its retained limit before making a claim, and that the pool would have the authority to exclude an insured making excessive claims from future pools.

Despite skin-deep differences between Jade and Emerald, their general structure was the same and can be seen from a diagram included in Ms. Clark’s 2015 memorandum overviewing Emerald:

To participate in the programs, a captive was required to submit an application for reinsurance to the relevant pool, which requested a limited range of information: entity name, business organization (e.g., sole proprietorship), location, business activity, gross revenue, value of property insured, and whether the reinsurance company had previously received any claims. Ms. Clark then circulated the applications to all other pool participants, who had less than a week to determine whether to exclude fellow participants from their respective pool. Dr. Swift, for example, chose to exclude three applications in 2015 because of different property value and revenue. The captive thereafter would enter into a trust agreement, a reinsurance agreement, and a quota share retrocession agreement, as necessary to participate in both risk distribution programs. 23

[*23] The fees for participation in Jade and Emerald were $5,000 in 2012, $6,000 in 2013, and $6,325 in 2014 and 2015. All told, 94 captives participated in Jade with a total gross volume of $22,700,000 in 2012 and $33,100,000 in 2013. With respect to Emerald, 150 captives participated with total gross volume of $36,600,000 in 2014, and 159 captives with total gross volume of $36,200,000 in 2015.

2. The Swift Captives’ Participation in Jade and Emerald

a. 2012 and 2013

The Swift captives participated in the Jade reinsurance pool in 2012 and 2013. In the first year each of the Swift captives had premiums of $1,192,430, and Jade reinsured $360,050, which represented 30.2% of the total. The amounts retroceded to the Swift captives under the quota share agreement matched the amount paid to it. The pattern of matching premiums was repeated in 2013, with Jade receiving reinsurance premiums of $207,450 (30.1% of total premiums of $689,822) from each of the Swift captives, and the Swift captives receiving the same amounts in their roles as retrocessionaires. All told, the Swift captives received back 99.59% and 98.74% of the reinsurance premiums they paid to Jade in 2012 and 2013, respectively. 8

b. 2014 and 2015

The Swift captives’ experience with Emerald was more of the same. Specifically in 2014 each captive had total premiums of $569,698, resulting in $171,450 in premiums reinsured by Emerald (30.1% of the whole) and retroceded from Emerald to each captive. Likewise, in 2015, each captive had total premiums of $535,761 and reinsurance premiums and premiums retroceded of $170,800, which constituted 31.9% of the whole. Ultimately, the Swift captives each received back 94.98% and 98.99% (before the final distribution) of the reinsurance premiums paid to Emerald in 2014 and 2015, respectively.

8 Under the 2013 pool, two claims to captive insurance companies reinsured by

Jade were approved, and Jade was responsible for $360,853 of loss, payable from trust funds. In January 2014 Jade also paid the Swift captives $1,719 for a claim payment related to pool coverage under Coverage B. 24

[*24] V. IRS Examination and Notices of Deficiency

The Swifts’ tax returns were prepared by their CPA, Mr. Schultz, who had done so since 2004. For tax years 2012 through 2015, the Swifts reported gross income, total expenses, and insurance expense (other than health insurance) as follows:

Year Gross Income Total Expenses Insurance Expense 2012 $51,939,335 $50,037,253 $2,518,374 2013 45,778,832 45,194,057 1,495,278 2014 45,214,460 43,535,206 1,253,373 2015 45,509,064 44,273,720 1,181,184

Clinic was the only Swift entity that claimed insurance deductions for the captive insurance premiums.

A. Examination

The IRS conducted an examination into the Swifts’ tax returns for each of the years at issue. Revenue Agent Allen Sohrt conducted the examination with respect to the Swifts’ 2012 through 2014 tax years and Revenue Agent Elia Maglaya conducted the examination into their 2015 tax year.

1. 2012 and 2013

On December 31, 2015, Revenue Agent Sohrt notified the Swifts via letter that he had completed his review for their 2012 and 2013 tax years and recommended disallowance of the applicable captive premium payments, although the final decision rested with IRS District Counsel. In his letter he also pointed out that he recommended the imposition of 20% accuracy-related penalties in his revenue agent report. About a week later, on January 4, 2016, Group Manager Cynthia Tam signed a Civil Penalty Approval Form approving the assertion of 20% accuracy- related penalties under section 6662(c) and (d) for the Swifts’ 2012 and 2013 tax years.

2. 2014

Two years later, Revenue Agent Sohrt completed his examination for the Swifts’ 2014 tax year and, on October 2, 2017, sent the Swifts Form 4549, Report of Income Tax Examination Changes, for that year. The report reflected a 40% accuracy-related penalty but did not include 25

[*25] the 20% penalty. On October 19, 2017, Group Manager Tam signed Civil Penalty Approval Forms for the Swifts’ 2014 tax year. In addition to the 40% penalty shown on the revenue agent’s report, Group Manager Tam approved, in the alternative, a 20% accuracy-related penalty for both negligence and substantial understatement.

3. 2015

As to 2015, Revenue Agent Maglaya likewise recommended the accuracy-related penalty for negligence and substantial understatement. Group Manager Arturo Velasquez signed a Civil Penalty Approval form on April 9, 2019, approving the penalty. By letter dated April 9, 2019, Group Manager Velasquez sent to the Swifts an examination report that, inter alia, took the position that the 20% accuracy-related penalty applied.

B. Notices of Deficiency

The IRS issued three notices of deficiency to the Swifts, disallowing the amounts deducted as insurance premiums and related legal expenses. 9 The IRS first issued to the Swifts a notice of deficiency that determined for 2012 and 2013 deficiencies of $893,809 and $596,855 (stemming from the disallowance of deductions for the insurance premiums and related legal and professional expenses), as well as alternative 20% accuracy-related penalties under section 6662(a). The IRS later issued a notice of deficiency determining a deficiency of $494,259 for their 2014 tax year, as well as an alternative 20% accuracy-related penalty. The IRS finally issued to the Swifts a notice of deficiency for their 2015 tax year determining a deficiency of $461,524 and a 20% accuracy-related penalty as an alternative position.

OPINION

I. Burden of Proof

In general the Commissioner’s determinations in a notice of deficiency are presumed correct, and the taxpayer bears the burden of proving that the determinations are in error. Rule 142(a); Welch v. Helvering, 290 U.S. 111, 115 (1933). The taxpayer bears the burden of proving entitlement to any deduction claimed. INDOPCO, Inc. v.

9 The notice of deficiency for each year at issue also determined a 40% accuracy-

related penalty. The Commissioner has since conceded that penalty, and we accordingly will not address it. 26

[*26] Commissioner, 503 U.S. 79, 84 (1992). Thus, a taxpayer claiming a deduction on a federal income tax return must demonstrate that the deduction is provided for by statute and must maintain records sufficient to enable the Commissioner to determine the correct tax liability. See I.R.C. § 6001; Hradesky v. Commissioner, 65 T.C. 87, 89–90 (1975), aff’d per curiam, 540 F.2d 821 (5th Cir. 1976); Treas. Reg. § 1.6001-1(a).

If, in any court proceeding, the taxpayer puts forth credible evidence with respect to any factual issue relevant to ascertaining the liability of the taxpayer and meets certain other requirements, the burden of proof shifts to the Commissioner. I.R.C. § 7491(a)(1) and (2). 10 When each party has satisfied its burden of production, then the party supported by the weight of the evidence will prevail, and thus a shift in the burden of proof has real significance only in the event of an evidentiary tie. See Knudsen v. Commissioner, 131 T.C. 185, 189 (2008), supplementing T.C. Memo. 2007-340.

We do not perceive an evidentiary tie in these cases and are able to decide the issues on the preponderance of the evidence. See, e.g., Bordelon v. Commissioner, T.C. Memo. 2020-26, at *11.

II. Insurance

A. General Principles

Section 162(a) allows the deduction of “all the ordinary and necessary expenses paid or incurred during the taxable year in carrying on any trade or business.” Insurance premiums are typically deductible under section 162(a) as ordinary and necessary expenses if paid or incurred in connection with a trade or business. Treas. Reg. § 1.162- 1(a).

Insurance companies are generally taxed on taxable income, including premium income and investment income, in the same manner as other corporations. See I.R.C. § 831(a); see also Syzygy, T.C. Memo. 2019-34, at *27. Section 831(b), however, provides an alternative taxing structure for certain small insurance companies. See Avrahami,

10 The U.S. Court of Appeals for the Fifth Circuit, to which an appeal in these

cases would ordinarily lie, see I.R.C. § 7482(b)(1), has likewise held that, if an “assessment is arbitrary and erroneous, the burden shifts to the government to prove the correct amount of any taxes owed,” Portillo v. Commissioner, 932 F.2d 1128, 1133 (5th Cir. 1991), aff’g in part, rev’g and remanding in part T.C. Memo. 1990-68. 27

[*27] 149 T.C. at 175. Specifically, an insurance company with written premiums not over $1.2 million in its tax year that makes a valid section 831(b) election is subject to tax only on its investment income (and thus not its premium income). See I.R.C. § 831(b)(1) and (2). 11 To make a valid section 831(b) election, a captive must be an insurance company, however. See I.R.C. § 831(c); Syzygy, T.C. Memo. 2019-34, at *28.

Neither the Code nor the Treasury Regulations define insurance. See R.V.I. Guar. Co., Ltd. & Subs. v. Commissioner, 145 T.C. 209, 224 (2015); Securitas Holdings, Inc., & Subs. v. Commissioner, T.C. Memo. 2014-225, at *18. The categorization nonetheless has profound effects: “[W]hile insurance is deductible, amounts set aside in a loss reserve as a form of self-insurance are not.” Caylor Land, T.C. Memo. 2021-30, at *31; see also Harper Grp., 96 T.C. at 46. When the insurer and the insured are related (including in the case of captive or microcaptive insurers), the line between insurance and self-insurance blurs. See Avrahami, 149 T.C. at 176–77.

Given the lack of a statutory definition, the meaning of insurance “has thus been developed chiefly through a process of common-law adjudication.” R.V.I., 145 T.C. at 224–25; see also, e.g., Caylor Land, T.C. Memo. 2021-30, at *31. The U.S. Supreme Court long ago explained that “[h]istorically and commonly insurance involves risk-shifting and risk-distributing.” Helvering v. Le Gierse, 312 U.S. 531, 539 (1941). Building on this foundation, to determine whether an arrangement constitutes insurance, we “look[] to four nonexclusive but rarely supplemented criteria: [1] risk-shifting; [2] risk-distribution; [3] insurance risk; and [4] whether an arrangement looks like commonly accepted notions of insurance.” Caylor Land, T.C. Memo. 2021-30, at *32; see also Avrahami, 149 T.C. at 181; Rent-A-Center, Inc. v. Commissioner, 142 T.C. 1, 13 (2014); Rsrv. Mech., T.C. Memo. 2018-86, at *33.

“In our [five] prior microcaptive cases, we have focused on the elements of risk distribution and ‘commonly accepted notions of insurance.’” Caylor Land, T.C. Memo. 2021-30, at *32; see also Avrahami, 149 T.C. at 181–97; Keating, T.C. Memo. 2024-2, at *51–52; Syzygy, T.C. Memo. 2019-34, at *29; Rsrv. Mech., T.C. Memo. 2018-86,

11 For tax years after December 31, 2016, Congress raised the premium ceiling

to $2,200,000 and added certain diversification requirements to make a section 831(b) election. See Consolidated Appropriations Act, 2016, Pub. L. No. 114-113, § 333(b), 129 Stat. 2242, 3108 (2015). These changes do not have any bearing on the years at issue. 28

[*28] at *33–34. We will do so again, and we again reach the conclusion that the microcaptive arrangement before us does not constitute insurance.

B. Risk Distribution

Risk distribution occurs when the insurer pools a large enough collection of unrelated risks, or risks that are “generally unaffected by the same event or circumstance.” Rent-A-Center, 142 T.C. at 24; see also Avrahami, 149 T.C. at 181. “The idea is based on the law of large numbers—a statistical concept that theorizes that the average of a large number of independent losses will be close to the expected loss.” Avrahami, 149 T.C. at 181; see also R.V.I., 145 T.C. at 228; Securitas, T.C. Memo. 2014-225, at *25–26. Thus, “[b]y assuming numerous relatively small, independent risks that occur randomly over time, the insurer smoothes out losses to match more closely its receipt of premiums.” Rent-A-Center, 142 T.C. at 24 (quoting Clougherty Packing Co. v. Commissioner, 811 F.2d 1297, 1300 (9th Cir. 1987), aff’g 84 T.C. 948 (1985)); see also Securitas, T.C. Memo. 2014-225, at *25–26 (“As the size of the pool increases, the chance that the loss per policy during any given period will deviate from the expected loss by a given amount (or proportion) declines.” (quoting AMERCO v. Commissioner, 96 T.C. 18, 33 n.14 (1991), aff’d, 979 F.2d 162 (9th Cir. 1992))). Distributing risk also “allows the insurer to reduce the possibility that a single costly claim will exceed the amount taken in as a premium.” Securitas, T.C. Memo. 2014-225, at *25 (quoting Clougherty Packing Co. v. Commissioner, 811 F.2d at 1300).

In analyzing risk distribution, we look to the actions of the insurer as it is the insurer’s risk, not the insured’s, that is reduced by risk distribution. See Rent-A-Center, 142 T.C. at 24; see also Humana Inc. v. Commissioner, 881 F.2d 247, 251 (6th Cir. 1989), aff’g in part, rev’g and remanding in part 88 T.C. 197 (1987); Clougherty, 811 F.2d at 1300. We have concluded on two occasions that a captive insurer had established risk distribution solely by insuring commonly owned brother-sister entities.

Each involved coverage at an impressive scale. The captive in the first instance offered workers’ compensation, automobile, and general liability insurance covering “between 2,623 and 3,081 stores; . . . between 14,300 and 19,740 employees; and . . . between 7,143 and 8,027 insured vehicles,” with operations in all 50 states, the District of Columbia, Puerto Rico, and Canada. Rent-A-Center, 142 T.C. at 24. The 29

[*29] captive in the second case offered workers’ compensation, automobile, employment practice, general, and fidelity liability insurance to 25 to 45 separate entities in more than 20 countries with more than 200,000 employees and 2,250 vehicles. Securitas, T.C. Memo. 2014-225, at *26.

Microcaptive insurers have not fared as well with respect to showing risk distribution; all of our previous cases have found compliance with this requirement lacking. See Avrahami, 149 T.C. at 182–90; see also Caylor Land, T.C. Memo. 2021-30, at *33–39; Syzygy, T.C. Memo. 2019-34, at *29–37; Rsrv. Mech., T.C. Memo. 2018-86, at *34–48. In those cases the microcaptives have attempted to demonstrate risk distribution in two different ways: (1) direct policies to brother-and-sister entities and (2) participation in a risk pool. See Avrahami, 149 T.C. at 182–90; see also Caylor Land, T.C. Memo. 2021- 30, at *33–39; Syzygy, T.C. Memo. 2019-34, at *29–37; Rsrv. Mech., T.C. Memo. 2018-86, at *34–48. The Swifts assay a similar climb and take the same tumble.

1. Direct Written Policies

The Swifts first argue that the Swift captives policies themselves establish risk distribution. This argument appears to have been a somewhat belated revelation, as the Swift captives’ business plans indicate that they were participating in Jade and Emerald risk pools (and the Pan American pool before them) for risk-distribution purposes.

In our previous microcaptive cases, we “have focused on both the number of insureds and the total number of independent risk exposures” when assessing risk distribution. Rsrv. Mech., T.C. Memo. 2018-86, at *34. In each of those cases “we found there wasn’t a large enough pool of unrelated risk from the policies issued to the related entities.” Caylor Land, T.C. Memo. 2021-30, at *34.

We reach the same conclusion here. The Swift captives insured, at most, three entities (in 2012), which dropped to two in 2013 when Derm Docs closed its doors. 12 The Swift captives issued only nine lines of coverage in 2012 and 2013 and six lines in 2014 and 2015. These numbers are comparable to those we have found wanting in several of our previous microcaptive cases. See Avrahami, 149 T.C. at 184 (finding

12 The Commissioner argues that the Swift captives insured only one entity.

We need not address this argument because even with three insured entities, we agree with the Commissioner’s contention that the Swift captives failed to distribute risk. 30

[*30] seven types of policies to four entities fell short); Caylor Land, T.C. Memo. 2021-30, at *19, *21, *35–37 (finding 11–12 policies concentrated in two entities fell short); Rsrv. Mech., T.C. Memo. 2018-86, at *35–36 (between 11 and 13 policies for three entities fell short).

Moreover, Clinic and Rehab did not insure “a sufficient number of unrelated risks to allow the law of large numbers to predict losses.” See Caylor Land, T.C. Memo. 2021-30, at *36; see also Rsrv. Mech., T.C. Memo. 2018-86, at *35–36. The Swift captives’ six or nine policies covered an operation spanning approximately 28 locations (as of 2015) and a workforce that ranged between 530 (2012) and 341 (2015) workers during the years at issue, including its independent-contractor physicians. The Swift captives’ risk exposure pales in comparison with that we have deemed satisfactory for the law of large numbers to apply. See R.V.I., 145 T.C. at 214 (finding insurance company issued 951 policies covering 714 different insured parties with 754,532 passenger vehicles, 2,097 real properties, and 1,387,281 commercial-equipment assets); Rent-A-Center, 142 T.C. at 2 (finding over time, captive insured 14,300 to 19,740 employees, 7,143 to 8,027 vehicles, and 2,623 to 3,081 stores); Harper Grp., 96 T.C. at 51 (finding captive insured 7,500 customers covering more than 30,000 different shipments and 6,722 policies); cf. Caylor Land, T.C. Memo. 2021-30, at *36 (“[T]his is called the law of large numbers—not small numbers or some numbers.”).

The Swifts argue, however, that the law of large numbers applies here considering the millions of doctor-patient interactions covered by the medical malpractice tail policies. The Swifts are using the wrong metric to evaluate the risk: Michael Angelina, the more persuasive of the Swifts’ experts, and the Commissioner’s experts looked to the number of doctors, which is standard in the industry when evaluating risk. 13 Consistent with the majority of the experts, KPMG likewise considered doctors, not patient interactions, when offering its price estimates. We will follow their lead.

Nor do we think the number of physicians sufficient for risk- distribution purposes. We first disagree with the Swifts that all of the independent-contractor physicians that worked at a Clinic location stretching back to 1982 represented live risk exposures. During the years at issue Texas had a two-year statute of limitations on medical

13 Indeed, it strikes us that using the doctor-patient interaction as the

appropriate unit of measurement for risk exposure would be tantamount to treating as the correct unit of measurement for risk exposure in the automobile insurance context every time a car is put into gear. 31

[*31] malpractice claims and a ten-year statute of repose. See Tex. Civ. Prac. & Rem. Code Ann. § 74.251 (West 2003). Given this legal regime, doctors who left Clinic’s service before 2002 should not be considered in the risk-distribution analysis, leaving approximately 199 current or former physicians. We do not believe that this is an adequate number of risk exposures, concentrated in one line of insurance, for the operation of the law of large numbers. In short, the captives “face[d] a number of independent risks that are at least a couple orders of magnitude smaller than the captives in cases where we’ve found sufficient distribution of risk.” Caylor Land, T.C. Memo. 2021-30, at *37.

In addition to failing to show sufficient risk exposures, we hold in the alternative that the Swifts have not demonstrated that the Swift captives faced independent risks, also necessary for risk distribution. Several of the policies insured overlapping risks, as illustrated by the seamless switch in coverage between the cost of the defense policy and the litigation expenses policy in connection with the 2013 wrongful termination claim. This 2013 wrongful termination claim does not represent an isolated incident, with the Commissioner’s expert David Russell pointing out that one event might trigger multiple policies, including Administrative Actions, Business Income, and Litigation Expenses.

Moreover, “there was no geographic diversity . . . in the entities that [the Swift captives] insured.” See id. at *38. Clinic’s locations were concentrated in the San Antonio-Austin area, within a 100-mile radius, particularly significant given the heavy investment in terrorism and political violence insurance.

Nor was there industry diversity, with both Clinic and Rehab operating in well-defined slices of the medical field. The doctors did not introduce such diversity even in the circumscribed context of the tail coverage. KPMG grouped in the same risk category all of the physicians that contracted with Clinic, assigning no value to individual claim history, specialty, or full-time or part-time status. And Dr. Swift touted that he tightly controlled physician practice in this setting to achieve uniformity of performance and desirable outcomes, which belies diversity.

In summary, we conclude that the Swift captives failed on multiple levels to establish risk distribution through the direct policies in effect during the years at issue. 32

[*32] 2. Reinsurance Pools

The Swifts argue in the alternative that the Swift captives distributed risk by their participation in the Jade and Emerald reinsurance pools. We are thus called to determine “whether [each] quota-share arrangement was a true insurance arrangement for the distribution of risk.” Rsrv. Mech. v. Commissioner, 34 F.4th at 912.

In our previous cases we have analyzed this question by reference to the factors used to determine whether a company is a bona fide insurer:

(1) whether the company was created for legitimate nontax reasons;

(2) whether there was a circular flow of funds;

(3) whether the entity faced actual and insurable risk;

(4) whether the policies were arm’s-length contracts;

(5) whether the entity charged actuarially determined premiums;

(6) whether comparable coverage was more expensive or even available;

(7) whether it was subject to regulatory control and met minimum statutory requirements;

(8) whether it was adequately capitalized; and

(9) whether it paid claims from a separately maintained account.

See, e.g., Reserve, T.C. Memo. 2018-86, at *38–39. To be clear, we do not consult these factors to determine whether Jade or Emerald “meet the formal definition of an insurance company” but to decide whether their products constituted insurance as necessary for Castlerock and Stonegate to distribute risk. See Rsrv. Mech. v. Commissioner, 34 F.4th at 912. Several factors convince us that Jade’s and Emerald’s risk pools did not suffice on this score. 33

[*33] a. Circular Flow of Funds

Under the reinsurance agreements, the Swift captives paid reinsurance premiums to Jade and to Emerald to reinsure a portion of their risk. Pursuant to trust agreements and quota share retrocession agreements, Jade and Emerald returned to the Swift captives 99.59% and 98.74% of the reinsurance premiums paid to Jade in 2012 and 2013, respectively, and 94.98% and 98.99% of the reinsurance premiums paid to Emerald in 2014 and 2015, respectively. “While not quite a complete loop, this arrangement looks suspiciously like a circular flow of funds.” Avrahami, 149 T.C. at 186; see also Syzygy, T.C. Memo. 2019-34, at *30–31; Rsrv. Mech., T.C. Memo. 2018-86, at *40–41.

The Swifts argue that under section 832(b)(4)(A), the Swift captives never received as income the portion of Clinic’s premium payments that were later paid to Jade and Emerald for reinsurance. Ms. Clark refutes this argument, explaining in her annual memoranda describing the structure of Jade and Emerald: “Your business will pay all insurance premiums, initially to your [captives.]”

Even assuming that Ms. Clark was incorrect, the end result was the transfer of nearly all of the reinsurance premium amounts from Clinic, which was 100% owned by Dr. Swift, to the Swift captives, which were owned by trusts for the benefit of the Swifts’ adult children, with the Swifts acting as trustees. Cf. Avrahami, 149 T.C. at 186 (“The end result of two years in the Pan American program was the transfer of $720,000 from an entity owned 100% by the Avrahamis to one owned 100% by Mrs. Avrahami.”); Rsrv. Mech., T.C. Memo. 2018-86, at *41.

b. Arm’s-Length Contracts

Nor do we believe that the Swift captives entered into arm’s- length contracts with Jade and Emerald, for the same reasons that we laid out in Reserve Mechanical, T.C. Memo. 2018-86, at *42, and Avrahami, 149 T.C. at 188–89. Jade, and later Emerald, agreed to reinsure a portion of the risks insured under the direct policies written by the Swift captives, with premiums purportedly related to the specific risks that the reinsurer assumes. At the same time, the respective reinsurer retroceded to the Swift captives a share of the total reinsurance premiums received from approximately 100 captives that insured diverse companies in various lines of business.

The fact that the premium amounts for the two different types of insurance agreements matched belies the idea that the parties entered 34

[*34] into these contracts at arm’s length. It beggars belief that for each of the four years at issue the premiums paid to Jade and Emerald to reinsure a portion of the Swift captives’ risk equaled the premiums paid to the Swift captives for assuming a quota share portion of Jade and Emerald’s blended risk from approximately 100 different captives. See Rsrv. Mech., T.C. Memo. 2018-86, at *41–42; see also Rsrv. Mech. v. Commissioner, 34 F.4th at 906, 912.

Although the Swifts argue that “reinsurance premiums are normally priced as a percentage of the original premium, this contention is a red herring. As in Reserve Mechanical, T.C. Memo. 2018-86, at *41–43, the Swifts failed to show why each percentage was reasonable and why they perfectly aligned in light of the different risks being assumed (some of which were wholly subject to client discretion). From our perspective, it appears that the percentages were reverse engineered by Ms. Clark and Mr. Rosenbach to ensure that the reinsurance and retrocessions premiums both equaled at least 30% of a participating captive’s total premiums as Ms. Clark believed necessary.

The chance that a qualifying loss would not have been paid under either the Jade or Emerald pool also raises questions whether a reasonable business would enter into these contracts absent tax motivations. See Avrahami, 149 T.C. at 188. Both were thinly capitalized, with Jade ceding 97.5% of its premiums within the first six months of each year in which the Swift captives participated and Emerald releasing 95% on the same timeline in 2014. Although Emerald slowed this cession rate in 2015, this alteration does not change our view that the Swift captives entered into reinsurance contracts with companies that would have difficulties making good on claims, precisely because of the promise, i.e., the premiums would be returned, at the heart of the arrangement. As in Avrahami, 149 T.C. at 189, Jade and Emerald would be required to go hat in hand to the participating captives to cover cash shortfalls, despite their inability to force any of the participating captives to pay more money into the pool to cover the claim.

And we cannot ignore that the risk distribution pools actively sought to block reinsurance coverage. As Ms. Clark explained to Mr. Rosenbach, she had erected “meaningful deterrents to claims against the pool.” Specifically, with respect to Coverage Part A, each captive would need to pay up to a retained limit before making a claim against the pool, and the pool would have the authority to exclude an insured making excessive claims from future pools. In short, the 35

[*35] contracts were set up to dissuade participants from using the pools as reinsurance.

c. Actuarially Determined Premiums

Both Jade and Emerald charged premiums as a percentage of the participating captives’ direct written policies, assigning different percentages to various types and amounts of coverage. “We have held that premiums were not actuarially determined where there has been no evidence to support the calculation of premiums and when the purpose of premium pricing has been to fit squarely within the limits of section 831(b).” Syzygy, T.C. Memo. 2019-34, at *34; see also Avrahami, 149 T.C. at 196; Rsrv. Mech., T.C. Memo. 2018-86, at *43.

We begin by observing that the Jade and the Emerald premiums produced loss ratios that deviated significantly from the industry standard. The loss ratio generally represents the “[p]ercentage of each premium dollar an insurer spends on claims.” Loss Ratio, Insurance Information Institute, https://www.iii.org/resource-center/iii-glossary/L (last visited Dec. 7, 2023). As we have said before, “[a]s the size of the pool increases, the chance that the loss per policy during any given period will deviate from the expected loss by a given amount (or proportion) declines.” Securitas, T.C. Memo. 2014-225, at *25–26; see also Rent-A-Center, 142 T.C. at 24.

In his report the Commissioner’s expert Dr. Russell stated that the industry loss ratios for reinsurance companies averaged 66.1%, 56.4%, 69.6%, and 66.3% in 2012, 2013, 2014, and 2015, respectively. Jade’s and Emerald’s loss ratios, on the other hand, ranged between 0.13% in 2012 and 7.91% in 2015. Although we do not contest the Swifts’ representation that Jade and Emerald together paid out millions of dollars in claims, this point is of no moment when seen in the context of the loss ratios. The tiny loss ratios suggest that the premiums were priced much higher than what the risks called for, which calls into question whether these were actual insurance arrangements intended to distribute risk. See Rsrv. Mech. v. Commissioner, 34 F.4th at 912.

Moreover, while this arrangement represents a slight variation on the theme in Reserve Mechanical, the fundamental defects remain: Both pools were designed to give the gloss of risk distribution, working backwards from predetermined premiums. See id. at 906, 912. We start with Coverage Part A. Charging a uniform reinsurance premium percentage to all captives participating in the pool based on general lines 36

[*36] of coverage and amount of underlying premium plainly fails to account for the specific risks presented by each of the agreements being reinsured through the pools.

And the reinsurance premium for Coverage Part C, which encompasses terrorism and political violence coverage, fluctuated as necessary to achieve 30% risk distribution. Jade and Emerald agreed to reinsure such coverage “depending on the client’s preference,” handing Ms. Clark and Mr. Rosenbach a flexible tool to adjust the reinsurance premiums to whatever level necessary to hit 30% risk distribution overall. The Swift captives here provide a practical illustration, with their Coverage Part C reinsurance premium percentages jumping from 82% in 2012 to 100% in 2013, then down to 80% in 2014 before trending up to 85% in 2015. The (tiny) risk did not change, the percentages were altered to fit the needs of the moment. This is unsurprising considering that, at the beginning of these arrangements, Ms. Clark informed Mr. Rosenbach of the plan to “add terrorism risk to the pool at whatever split we need to get to or above 50%.”

It was incumbent on the Swifts to show how Mr. Rosenbach derived these premiums in light of the various risks purportedly being reinsured. See Rsrv. Mech., T.C. Memo. 2018-86 at *43; see also Rsrv. Mech. v. Commissioner, 34 F.4th at 906, 912. The Swifts and their experts have failed to do so, with the evidence before us instead showing that Ms. Clark and Mr. Rosenbach (and their helpers) were simply manipulating numbers to design a system where 30% of total premiums would be allocated to reinsurance before being retroceded back.

d. Conclusion

Based on the factors discussed above, we find that Jade’s and Emerald’s policies were not bona fide insurance arrangements. See Avrahami, 149 T.C. at 190; Syzygy, T.C. Memo. 2019-34, at *36–37; Rsrv. Mech., T.C. Memo. 2018-86, at *47; see also Rsrv. Mech. v. Commissioner, 34 F.4th at 911–12 (observing that the “heart of the problem” is that the “product was not actual insurance” and the company sponsoring the pool, “as a matter of substance, . . . did not perform the functions of an insurance company—regardless of label— vis-à-vis the quota share arrangement”). Accordingly, the Swift captives could not use their reinsurance through the quota-share agreement to achieve the risk distribution that they lacked. 37

[*37] C. Insurance in the Commonly Accepted Sense

The absence of risk distribution alone is enough to conclude that the arrangements between the Swift captives and their insureds were not insurance. See Avrahami, 149 T.C. at 190–91. We also conclude, in the alternative, that the arrangements did not constitute insurance in the commonly accepted sense. See id. at 191; Caylor Land, T.C. Memo. 2021-30, at *39–49; Rsrv. Mech., T.C. Memo. 2018-86, at *48.

In making this evaluation, we look at numerous factors, “including whether the company was organized, operated, and regulated as an insurance company; whether the insurer was adequately capitalized; whether the policies were valid and binding; whether the premiums were reasonable and the result of an arm’s-length transaction; and whether claims were paid.” Avrahami, 149 T.C. at 191. We have also considered whether “the policies covered typical insurance risks and whether there was a legitimate business reason for acquiring insurance from the captive.” Id.; see also Caylor Land, T.C. Memo. 2021- 30, at *40.

1. Organized, Operated, and Regulated as an Insurance Company

The Swift captives were incorporated in St. Kitts, subject to regulation under its laws, and licensed to operate as insurance companies by its Financial Services Regulatory Commission. They each kept their own books and records, maintained separate bank accounts, prepared financial statements, and held meetings of their boards of directors.

“Apart from observing these formalities, however, the facts demonstrate that [the Swift captives were] not operated as . . . insurance compan[ies].” See Rsrv. Mech., T.C. Memo. 2018-86, at *50; see also Keating, T.C. Memo. 2024-2, at *53; Caylor Land, T.C. Memo. 2021-30, at *42 (“[W]e ‘must look beyond the formalities and consider the realities of the purported insurance transaction.’” (quoting Hosp. Corp. of Am. v. Commissioner, T.C. Memo. 1997-482, 1997 WL 663283, at *24)).

As an initial matter, Clinic conducted no due diligence into the need for two microcaptive insurance companies offering these lines of insurance. This omission would seem bizarre if these were actual insurance companies, a point easily seen in the context of tail insurance. For decades before setting up Castlegate, Clinic effectively addressed risks not covered by its commercial medical malpractice policy with a 38

[*38] loss-reserve fund of $500,000. In 2004 Dr. Swift began paying more than $800,000 per year to provide similar protection as his loss reserve. With a total of $615,000 in claims over the five years of Castlegate’s existence, we struggle to see the business reason for one microcaptive insurance company, much less a second. As to the other lines, Dr. Swift plainly communicated that he wanted to add coverage in the hopes of “maxing out” premiums, not for any real business need. That does not strike us as how the insurance industry normally operates. See Keating, T.C. Memo. 2024-2, at *60 (“A much more detailed explanation of the need for such expensive policies was warranted than the ones provided by [the taxpayer].”).

The Swift captives “also made investment choices only an unthinking insurance company would make.” See Avrahami, 149 T.C. at 193. Specifically, the Swift captives invested millions of dollars in premiums in a real estate limited partnership that owned and developed three of Clinic’s urgent care facilities, among other real estate projects. The Swift captives’ holdings were so illiquid that Dr. Swift issued a put option to both in June 2013 that, if needed, required him to purchase either or both captives’ entire interests in the limited partnership at a price determined by an appraisal process. Most of the premiums not tied up in real estate were invested in the stock market through brokerage accounts. Like the Swifts’ own expert, Mr. Angelina, “[w]e do not think that an insurance company in the commonly accepted sense would invest” so heavily in assets that could not be accessed to pay claims. Avrahami, 149 T.C. at 193; Syzygy, T.C. Memo. 2019-34, at *40; Avrahami, 149 T.C. at 193.

The handling of claims also seems off. Only three claims were made under 30 lines of insurance during the four years at issue, with one of them relating to the audit that resulted in this litigation. Despite stern warnings regarding notification periods, all three claims were approved despite being filed months after the expiration of the relevant periods. We find unusual the idea (articulated by one of Ms. Clark’s team members) that Dr. Swift did not need to consult with the captive before finalizing the settlement of one of the claims and that Heritor would provide “a letter approving coverage for the future settlement.” See also Keating, T.C. Memo. 2024-2, at *63 (“[The captives] paid claims. Nonetheless, the process by which those claims were handled was abnormal.”).

The policies also displayed various oddities. The provision for payment by promissory note if Clinic “suffer[ed] a series of catastrophic 39

[*39] loss occurrences that [might] impair [Clinic’s] solvency” would hamper Clinic’s recovery from a serious loss. Most of the policies deviate from industry standard by not providing a refund of unearned premiums in the event of cancellation and by tying cancellation to Clinic’s insolvency. And the General Cost of Defense policy featured combined premiums of $14,000, which exceeded the per-claim and aggregate limit of $10,000 for 2012, and nearly did the same in 2013 (limits set at $15,000). As the Commissioner’s expert Donald Bendure opined in his report, “[t]his policy is in effect a deposit account for legal fees with a limit so low as to be of minimal use from a risk management standpoint.”

Although the Swift captives were organized and regulated as insurance companies, they were not operated as such.

2. Adequate Capitalization

A captive is adequately capitalized as long as it meets the minimum capitalization requirements of its regulators. See Avrahami, 149 T.C. at 193; R.V.I., 145 T.C. at 231; Harper Grp., 96 T.C. at 50, 60; Syzygy, T.C. Memo. 2019-34, at *41. Although the Swift captives were thinly capitalized, they complied with St. Kitts law.

3. Valid and Binding Policies

“To be valid and binding an insurance policy should, at a minimum, identify the insured, define an effective period for the policy, specify what is covered by the policy, state the premium amount, and be signed by authorized representatives of the parties.” Rsrv. Mech., T.C. Memo. 2018-86, at *54; see also Avrahami, 149 T.C. at 194; R.V.I., 145 T.C. at 231. Here, the parties do not dispute that the policies issued by the Swift captives identified the insured, stated the premium amount, and were signed by Heritor as the authorized insurance manager.

The Commissioner argues, however, that the policies contain conflicting terms, pointing to the exact phrasing we critiqued in Avrahami, 149 T.C. at 194. We note that this is hardly the only example of questionable draftsmanship, with several of the policies acting effectively as excess coverage masquerading as primary. We have previously decided that this factor weighed against a taxpayer where policies combined ambiguities and contradictions with late issuance of the policies. Syzygy, T.C. Memo. 2019-34, at *42. The problems with these policies strike us as venial, not mortal, and we will treat this factor as neutral. 40

[*40] 4. Reasonableness of Premiums

We next consider whether the Swift captives’ premiums were reasonable and the result of an arm’s-length transaction. See, e.g., Avrahami, 149 T.C. 194–95.

As a general matter, we have serious reservations about the reasonableness of premiums developed to hit a preordained target for tax purposes, as here. See Caylor Land, T.C. Memo. 2021-30, at *45–46; Syzygy, T.C. Memo. 2019-34, at *34; see also Keating, T.C. Memo. 2024- 2, at *59 (finding premiums to be unreasonable where the client “provided . . . an amount he was willing to pay or a target premium for all policies,” which “played an outsized role in . . . underwriting”). “It is fair to assume that a purchaser of insurance would want the most coverage for the lowest premiums[, and that] [i]n an arm’s-length negotiation, an insurance purchaser would want to negotiate lower premiums instead of higher premiums.” Syzygy, T.C. Memo. 2019-34, at *33–34; see also Keating, T.C. Memo. 2024-2, at *59. “Seemingly, the main advantage of paying higher premiums is to increase deductions.” Syzygy, T.C. Memo. 2019-34, at *34. “We have held that premiums were not actuarially determined when there has been no evidence to support the calculation of premiums and when the purpose of premium pricing has been to fit squarely within the limits of section 831(b).” Id.

In these cases, Dr. Swift had a long history of playing the microcaptive insurance version of the “Showcase Showdown” from the Price Is Right: obtaining premiums close to, but not over, the limit imposed by section 831(b) or a pre-set target. The voluminous record before us leaves the firm impression that premium amounts were engineered to suit the tax needs of the moment, not to account for any risk.

a. Malpractice Tail Coverage

Looking at the derivation of the premiums more closely confirms us in our view. We begin, as did Dr. Swift, with tail coverage. Again, this coverage seemed an unusual choice in 2004 given (1) little loss history, (2) a practice unlikely to produce considerable tail risk, (3) proactive measures to further minimize risk, and (4) a moderate reserve that had proved itself fully up to the task of meeting losses that had arisen. By the years at issue Clinic had spent over $4,600,000 for such coverage, while paying out less than $400,000 in claims. See Keating, T.C. Memo. 2024-2, at *63. We believe that, if this were 41

[*41] intended to be insurance, Dr. Swift would have stopped paying premiums at this level long before the years at issue.

Also counting against the reasonableness of the premiums was the design of the coverage during the years at issue. The tail policy had a per-occurrence limit of $300,000 and an aggregate limit of $6 million despite a total of 42 medical malpractice claims from 1982 through 2015. Only four of those claims exceeded $300,000, lending support to the notion that the limits were set unreasonably high to further goose premiums for a coverage that “no carrier will write.” 14

The rate on line, which measures insurance cost per unit by dividing the premium paid by the occurrence limit, casts further doubt on the reasonableness of these premiums. “A higher rate-on-line means that insurance coverage is more expensive per dollar of coverage,” which “leads to a greater deduction for premiums.” See Syzygy, T.C. Memo. 2019-34, at *31. For the years at issue the commercial medical malpractice policies purchased by Clinic had a rate on line of 9.299% with respect to the occurrence limits and 3.1% with respect to the aggregate limit. The captives on the other hand had a rate on line of 233.12% on the occurrence limits and 11.656% in the aggregate basis. See Keating, T.C. Memo. 2024-2, at *61 (finding premiums to be “patently unreasonable” when “the average rate-on-line for . . . captive policies during the years at issue was more than ten times greater than the average rate-on-line for comparable commercial insurance policies”).

KPMG’s premium estimate analyses fail to change our mind. These estimates rely on general industry and internal KPMG data to derive a pure premium, which was then adjusted for various factors including the expense load. At trial, however, Dr. Swift and the KPMG representative who testified struggled to explain these calculations and the data relied upon. We have particular concerns considering the strong likelihood that the underlying data used to derive the premiums involved entities in the health care field, but in different states and with different risk factors and practices.

The Commissioner’s experts (Evelyn Toni Mulder and Daniel Lupton) showed numerous weaknesses in the KPMG analyses, which

14 We also bear in mind the contrast between the tail premium and the

commercial malpractice premiums that Clinic paid during the years at issue. Clinic paid approximately $700,000 per year for its tail coverage and $44,763 for its commercial medical malpractice coverage, which featured no deductible, a $500,000 occurrence limit, and a $1.5 million aggregate limit. 42

[*42] suggested that the premiums had been significantly overstated. Among other things, the KPMG analyses during the years at issue (1) inexplicably excluded low loss years from the calculation of pure premiums, (2) failed to tailor general industry data regarding loss reporting to reflect the significantly shorter lag time experienced by urgent care centers such as Clinic, and (3) did not accurately account for differences in risk associated with full- and part-time physicians.

In making our determination, “we consider more than whether the premiums chosen can be arrived at by actuarial means.” Rsrv. Mech., T.C. Memo. 2018-86, at *60. “Without a comprehensible explanation we can’t find these premium amounts justified.” Avrahami, 149 T.C. at 196. The Swifts have failed to demonstrate that the data used by KPMG accurately reflected risks Clinic faced or resulted in reasonable and actuarially determined premiums.

b. Nonmedical Malpractice Coverage

We reach the same conclusion with respect to the nonmedical malpractice lines of insurance, which were part of Dr. Swift’s avowed effort “to get closer to maxing out the premiums” to the Swift captives. During the years at issue Clinic paid an average of $73,968 in premiums to maintain their longstanding, expansive lines of commercial insurance. At the same time, Clinic paid the Swift captives an average of $794,500 for various niche lines of insurance, many of which were excessive. As documented by Dr. Russell, the annual rates-on-line for the Swift captives’ policies were 50 (or more) times greater than the commercial policies for the same period.

The premium analyses of Mr. Rosenbach, who did not testify at trial, fail to persuade us that these astounding numbers are reasonable. From the record before us we understand that, aside from the terrorism and political violence lines, Mr. Rosenbach generally relied on a 2005 filing by the Chubb Group of Insurance Companies (Chubb) with the Florida Department of Financial Services to determine a base rate for most of the lines of insurance, which he then adjusted to take into account various factors that he found relevant.

Neither the Swifts nor their experts have provided a persuasive explanation as to how Mr. Rosenbach exercised his judgment to 43

[*43] determine the base rates, factors, and ultimately, the premiums. 15 See Avrahami, 149 T.C. at 195 (“Rosenbach also made adjustments based on his professional judgment—most without a coherent explanation.”). Absent such explanation, we are left with the impression left by Mr. Rosenbach’s emails with Ms. Clark and her team, i.e., that he reverse-engineered premiums with a patina of actuarial methods. Keating, T.C. Memo. 2024-2, at *60 (“The premiums were . . . inflated by numerous subjective, judgment-driven factors, each of which could modify the premiums significantly; and there is very little documentation to support how [these factors were applied].”).

The premiums for terrorism and political violence insurance likewise were not reasonable. This coverage supplemented Clinic’s terrorism coverage under its commercial policies, which cost nothing and covered the replacement cost of Clinic’s buildings and equipment (valued between $33 million and $45 million) among other things. The Swift captives’ terrorism and political violence coverage operated as excess coverage of $6,750,000, except in narrow conditions not covered by the commercial insurance such as a nuclear, biological, or chemical attack in a city with a population of less than 2 million people. For such coverage, the Swift captives charged premiums of $540,000, $231,000, $384,000, and $384,000 for 2012 through 2015, respectively.

The Commissioner’s experts persuasively demonstrate that similar commercial coverage would be a fraction of the premium charged. Dr. Russell explained that the Swift captives’ terrorism insurance premiums were approximately 1,400 times the highest commercial rates. Ms. Taylor and Mr. Lupton agreed, explaining that a high-end estimate for coverage for the years at issue would be $5,430 rather than the $1.5 million paid by Clinic. The record before us demonstrates that the premiums for this coverage were not reasonable but merely a mechanism so that the Swift captives could hit the risk- distribution target set by Ms. Clark.

In summary, we agree with Dr. Russell’s observation that “[w]hile it is not unlikely for an insured to file few or no claims over an extended period, it is not economically justifiable that the Swift entities would

15 Even in the isolated instances where there was a thin ligament connecting

Mr. Rosenbach’s work to the Chubb filing, the end result was shaky. For example, he changed the deductible factor from year to year even though the deductibles for the policies remained constant. And he failed to take into account differences between Chubb and the Swift captives when determining expenses, which obviously would be quite stark. 44

[*44] rationally continue to pay premiums at the levels Castlerock and Stonegate . . . charged.” The premiums were neither reasonable nor actuarially determined. This factor thus weighs against the Swift captives’ being insurance in the commonly accepted sense.

5. Payment of Claims

Clinic submitted three claims to the Swift captives during the years at issue. The Swift captives paid these three claims, but as discussed above, there were problems with the way they were handled. While this factor weighs slightly in favor of the Swifts, “we do not regard this as overwhelming evidence that the arrangement constituted insurance in the commonly accepted sense.” See Syzygy, T.C. Memo. 2019-34, at *45; see also Rsrv. Mech., T.C. Memo. 2018-86, at *61.

6. Conclusion

Although the Swift captives displayed some attributes of insurance companies, they failed to operate as insurance companies and their premiums were nonsense. We therefore conclude that the Swift captives did not provide insurance in the commonly accepted sense.

7. Effect on the Swift Captives

Our holding in this regard has two major consequences. First, because the Swift captives’ policies were not contracts for insurance, they do not fall within the meaning of insurance company in section 831(c), which is defined in section 816(a) as “any company more than half of the business of which during the taxable year is the issuing of insurance or annuity contracts or the reinsuring of risks underwritten by insurance companies.” This makes the Swift captives ineligible to make an election under section 831(b) for the tax years at issue. Likewise, the Swift captives must meet this definition of “insurance company” to elect to be treated as domestic corporations under section 953(d)(1)(B). See also Avrahami, 149 T.C. at 198. Therefore, the Swift captives’ section 953(d) election is likewise invalid for the tax years at issue. We sustain the Commissioner’s determinations with respect to the Swift captives, so the Swift captives must recognize the premiums they received as income for the years at issue.

8. Effect on the Swifts

The second major consequence is that, if Clinic’s payments are not for insurance, “then they are not ordinary and necessary business 45

[*45] expenses and may not be deducted under section 162(a).” See Avrahami, 149 T.C. at 199; see also Syzygy, T.C. Memo. 2019-34, at *46. We therefore sustain the Commissioner’s determination to adjust the Swifts’ income by disallowing these deductions.

III. Legal Expense Deductions

In the notices of deficiency, the Commissioner also disallowed deductions for certain legal and professional fees paid to Ms. Clark, which were claimed on Clinic’s Schedules C. The Swifts assert in a conclusory statement in their brief that these were ordinary and necessary business expenses because Ms. Clark’s firm “provided valuable legal services during the years at issue by advising Dr. Swift regarding the proper formation and operation of . . . section 831(b) insurance companies.” The Swifts fail to develop this argument and have forfeited this issue. See, e.g., Estate of Spizzirri v. Commissioner, T.C. Memo. 2023-25, at *17 n.9.

Even if we were to overlook this forfeiture, the Swifts would not prevail on this point. The deductibility of legal expenses under section 162(a) depends on the origin and character of the claim for which the expenses were incurred and whether the claim bears a sufficient nexus to the taxpayer’s business or income-producing activities. See United States v. Gilmore, 372 U.S. 39, 48–49 (1963); Mylan, Inc. & Subs. v. Commissioner, 156 T.C. 137, 152 (2021), aff’d, 76 F.4th 230 (3d Cir. 2023). For these legal fees to be deductible, “the origin of those legal services must have been rooted in [their] Schedule C business.” Test v. Commissioner, T.C. Memo. 2000-362, 2000 WL 1738858, at *4. The legal fees in our cases related to the formation and operation of wholly independent business entities, i.e., the Swift captives, and the Swifts have failed to establish that the payment of microcaptive formation and operation expenses bears a sufficient nexus to Clinic’s business of providing urgent care and occupational medicine services.

We therefore find the Commissioner correctly disallowed the legal expense deductions the Swifts claimed in the years at issue.

IV. Penalties

In each of the notices of deficiency, the Commissioner determined a 20% accuracy-related penalty against the Swifts, premised on an underpayment attributable to negligence and a substantial understatement of income tax. See I.R.C. § 6662(a) and (b)(1) and (2). 46

[*46] Section 7491(c) generally provides that “the Secretary shall have the burden of production in any court proceeding with respect to the liability of any individual for any penalty.” This burden requires the Commissioner to come forward with sufficient evidence indicating that the imposition of the penalty is appropriate. See Higbee v. Commissioner, 116 T.C. 438, 446 (2001). Once he meets his burden of production, the burden of proof is on the taxpayer to “come forward with evidence sufficient to persuade a Court that the Commissioner’s determination is incorrect.” Id. at 447.

A. Supervisory Approval Requirement

The Commissioner’s burden of production under section 7491(c) includes establishing compliance with section 6751(b)(1), which provides that “[n]o penalty . . . shall be assessed unless the initial determination of such assessment is personally approved (in writing) by the immediate supervisor of the individual making such determination.” See Graev v. Commissioner, 149 T.C. 485, 493 (2017), supplementing and overruling in part 147 T.C. 460 (2016); see also Chai v. Commissioner, 851 F.3d 190, 217, 221–22 (2d Cir. 2017), aff’g in part, rev’g in part T.C. Memo. 2015- 42. In Belair Woods, LLC v. Commissioner, 154 T.C. 1, 14–15 (2020), we explained that the “initial determination” of a penalty assessment is typically embodied in a letter “by which the IRS formally notifie[s] [the taxpayer] that [it] ha[s] completed its work and . . . ha[s] made a definite decision to assert penalties.” Once the Commissioner introduces evidence sufficient to show written supervisory approval, the burden shifts to the taxpayer to show that the approval was untimely, viz, “that there was a formal communication of the penalty [to the taxpayer] before the proffered approval” was secured. Frost v. Commissioner, 154 T.C. 23, 35 (2020); Thompson v. Commissioner, T.C. Memo. 2022-80, at *6.

“The word ‘determination’ has ‘an established meaning in the tax context and denotes a communication with a high degree of concreteness and formality.” Oxbow Bend, LLC v. Commissioner, T.C. Memo. 2022- 23, at *5 (quoting Belair Woods, 154 T.C. at 15); accord Beland v. Commissioner, 156 T.C. 80, 85 (2021). “[T]he ‘initial determination’ of a penalty assessment will be embodied in a formal written communication to the taxpayer, notifying him that the Examination Division has completed its work and has made a definite decision to assert penalties.” Belair Woods, 154 T.C. at 10. A “mere suggestion, proposal, or initial informal mention” of penalties does not, we have held, constitute an 47

[*47] initial determination under section 6751(b)(1). Tribune Media Co. v. Commissioner, T.C. Memo. 2020-2, at *19. 16

The Swifts make two arguments regarding supervisory approval. First, they argue that the Commissioner did not establish that the group managers who signed the penalty approval forms were the respective examining agents’ immediate supervisors. “We have repeatedly held that a manager’s signature on a penalty approval form, without more, is sufficient to satisfy the statutory requirements [of section 6751].” Nassau River Stone, LLC v. Commissioner, T.C. Memo. 2023-36, at *11.

The Swifts also raise the vague contention that “there are instances where the revenue agent indicated penalties to petitioners, but had not yet sought managerial approval.” We understand that the Swifts are challenging the supervisory approval of the 2012 and 2013 penalties on the ground that Revenue Agent Sohrt sent a letter with a revenue agent report “includ[ing] the 20% accuracy related penalty” on December 31, 2015, before obtaining his supervisor’s written approval on January 4, 2016. We do not believe that this communication possessed the high degree of concreteness and formality that we associate with a determination for purposes of section 6751. The letter enclosing the revenue agent report noted that the revenue agent was “recommending a disallowance of the applicable captive premium payments,” but made clear that no final decision had been made:

Because the statute of limitations will be expiring [in four months], it will be necessary to move the case forward for the possible issuance of a Statutory Notice of Deficiency.

16 We recognize that there is a split among circuits as to whether written

supervisory approval must be obtained before the IRS issues a notice of deficiency, Chai v. Commissioner, 851 F.3d at 221, or merely before the assessment, Kroner v. Commissioner, 48 F.4th 1272, 1278–79 (11th Cir. 2022), rev’g in part T.C. Memo. 2020- 73; see also Laidlaw’s Harley Davidson Sales, Inc. v. Commissioner, 29 F.4th 1066, 1074 (9th Cir. 2022) (holding that section 6751(b)(1) “requires written supervisory approval before the assessment of the penalty or, if earlier, before the relevant supervisor loses discretion whether to approve the penalty assessment”), rev’g and remanding 154 T.C. 68 (2020). As stated previously, appeal of these cases would presumably lie in the Fifth Circuit. I.R.C. § 7482(b)(1)(A); Golsen v. Commissioner, 54 T.C. 742 (1970), aff’d, 445 F.2d 985 (10th Cir. 1971). Golsen stands for the proposition that this Court will apply the decision of the court of appeals that is “squarely in point where appeal from our decision lies to that Court of Appeals and to that court alone” and, as a corollary, that this Court’s own views will be given effect to the extent the relevant court of appeals has not expressed one. See Golsen, 54 T.C. at 757. The Fifth Circuit does not appear to have taken a clear stance on the section 6751(b)(1) issue. 48

[*48] Internal Revenue Service District Counsel will review the issue and will make a determination as to whether a Statutory Notice of Deficiency will be issued.

We accordingly conclude that this was not an initial determination for purposes of section 6751.

B. Section 6662

The Code imposes a 20% penalty on the portion of the underpayment of tax attributable to a substantial understatement of income tax. 17 See I.R.C. § 6662(a), (b)(2). An understatement of income tax is substantial if it exceeds the greater of $5,000 or “10 percent of the tax required to be shown on the return.” I.R.C. § 6662(d)(1)(A). The Commissioner has met his prima facie burden, as each of the understatements at issue plainly exceeds $5,000 and is greater than 10% of the tax required to be shown on the return:

Year Reported Tax Liability Corrected Tax Liability Understatement

2012 $1,520,783 $2,414,592 $893,809

2013 1,106,436 1,703,291 596,855

2014 2,046,545 2,540,804 494,259

2015 1,484,612 1,946,136 461,524

The accuracy-related penalty does not apply to any part of an underpayment of tax if it is shown that the taxpayer acted with reasonable cause and in good faith with respect to that portion. I.R.C. § 6664(c)(1); Rogers v. Commissioner, T.C. Memo. 2019-61, at *31, aff’d, 9 F.4th 576 (7th Cir. 2021). The Swifts bear the burden of proving that they had reasonable cause and acted in good faith with respect to the underpayments. See Higbee, 116 T.C. at 449.

17 “Only one accuracy-related penalty may be applied with respect to any given

portion of an underpayment, even if that portion is subject to the penalty on more than one of the grounds set forth in section 6662(b).” Sampson v. Commissioner, T.C. Memo. 2013-212, at *7–8 (citing New Phoenix Sunrise Corp. & Subs. v. Commissioner, 132 T.C. 161, 187 (2009), aff’d, 408 F. App’x 908 (6th Cir. 2010)). Consequently, we will not determine whether the Swifts are liable for penalties for negligence. 49

[*49] “Reasonable cause requires that the taxpayer have exercised ordinary business care and prudence as to the disputed item.” Neonatology Assocs., P.A. v. Commissioner, 115 T.C. 43, 98 (2000), aff’d, 299 F.3d 221 (3d Cir. 2002). The determination of whether a taxpayer acted in good faith is made on a case-by-case basis, considering all the pertinent facts and circumstances. Treas. Reg. § 1.6664-4(b)(1). “A taxpayer’s knowledge, education, and experience are relevant factors to indicate reasonable cause and good faith.” Rogers, T.C. Memo. 2019- 61, at *31. For underpayments related to passthrough items we look at all pertinent facts and circumstances, including the taxpayer’s own actions, as well as the actions of the passthrough entity. See Treas. Reg. § 1.6664-4(e).

Reliance on a tax professional may constitute reasonable cause if that professional advises the taxpayer on a substantive tax issue. See United States v. Boyle, 469 U.S. 241, 250–51 (1985); Treas. Reg. § 1.6664-4(b). For the reliance to be reasonable, a taxpayer must prove that “(1) [t]he adviser was a competent professional who had sufficient expertise to justify reliance, (2) the taxpayer provided necessary and accurate information to the adviser, and (3) the taxpayer actually relied in good faith on the adviser’s judgment.” Neonatology, 115 T.C. at 99.

The Swifts argue that they had reasonable cause for their reporting positions because they reasonably relied on the advice of Ms. Clark and their CPA, Mr. Schultz, at a time when the law surrounding microcaptive insurance companies was novel. Ms. Clark was the primary promoter of the transaction, however, so the Swifts could not reasonably rely on any advice she offered. See, e.g., Avrahami, 149 T.C. at 206; 106 Ltd. v. Commissioner, 136 T.C. 67, 79 (2011), aff’d, 684 F.3d 84 (D.C. Cir. 2012).

Neither could the Swifts rely on the advice of Mr. Schultz, who had been involved with Dr. Swift’s microcaptive insurance scheme since 2004 and who “participated in structuring the transaction.” 106 Ltd., 136 T.C. at 79 (quoting Tigers Eye Trading, LLC v. Commissioner, T.C. Memo. 2009-121, 2009 WL 1475159, at *19). Even if Mr. Schultz avoided the promoter label, the Swifts fail to establish the content of any substantive tax advice provided by Mr. Schultz on which they relied. 18

18 The Swifts also argue that their penalties should be reduced because they

relied on “substantial authority” in accordance with Treasury Regulation § 1.6662- 4(d)(3). The Swifts, however, did not raise this argument until their answering brief, and we thus decline to consider it. See, e.g., Clay v. Commissioner, 152 T.C. 223, 236 50

[*50] While the Swifts insist that Mr. Schultz was “very involved” and “asked many insightful and thorough questions . . . in the form of detailed correspondence,” these questions do not constitute advice nor do they indicate to the Court any advice Mr. Schultz ultimately relayed to the Swifts, upon which they supposedly relied.

V. Conclusion

For the reasons set forth above, we will sustain the deficiency determinations by the IRS. We further find that the Swifts are liable for the alternative 20% accuracy-related penalties under section 6662.

To reflect the foregoing,

Appropriate decisions will be entered.

(2019) (deeming “an issue raised for the first time in a party’s answering brief to be abandoned and conceded”).