Matthew Queen
20 January 2023ArticleAccounting & tax analysis

Defining insurance: a critique of the IRS test

The aftermath of the 831(b) tax shelters created bad law and exacerbated the Internal Revenue Service’s (IRS’s) flawed insurance test. In my view three of four of the elements of the IRS insurance test are incoherent and out of step with the practice of insurance.

Practitioners, lawyers in particular, need to remember that the IRS’s insurance test merits challenge on its fundamentals because it is incorrect, arbitrary, and designed to make the job of tax collecting easier.

In addition, defining insurance is a larger issue for the whole industry. With the passage of the 1945 McCarran-Ferguson Act, Congress delegated to the states the exclusive authority to govern the business of insurance. This resulted in many states defining insurance differently, several treatises taking different positions on what constitutes insurance, and the federal government struggling to keep up with when insurance manifests within a contract.

As an example, there is considerable debate in federal courts as to when a warranty product constitutes the unlicensed sale of insurance. See for example Pope v Lake Norman, F. Supp. 2d 309, 2007 WL 2480242. Consequently, the definition of what constitutes insurance is critical for the capital markets.

Genesis of the IRS test

The IRS leverages the following definition of insurance: 1) there must be an insurable interest; 2) there must be risk shifting; 3) there must be risk distribution; and 4) there must be insurance in the commonly accepted sense. Each of these prongs arise out of a case from the 1940s dealing with life insurance and an annuity. The case, Helvering v Le Gierse, involved an annuity where, upon the death of the insured, the residual premiums would be paid as a life insurance payment. This unusual structure was designed to escape the estate tax, which was a more significant tax in the mid-20th century.

The IRS challenged this structure as a tax shelter. The Supreme Court noted that the Internal Revenue Code does not define what constitutes “insurance”, so it made up a definition for the case.

The Supreme Court opined that risk shifting and risk distribution were essential to an insurance contract and that the life insurance payments constituted insurance in its commonly accepted sense (however the annuity payments were not held to constitute premiums).

This definition, developed entirely by the Supreme Court, was revisited in another case questioning what constituted insurance. The matter of Allied Fidelity v Commissioner explored whether a bail bond company could classify its expenses and reserves as tax deductible, as an insurance company. Since surety bonds are insurance, the bail bond company took the position that it was also insurance.

The Supreme Court disagreed, holding that the obligation of a monetary bond is an assurance of performance. The obligation of a bail bond, in particular, is an assurance that the defendant will appear in court. This is not a financial benefit to the government. Consequently, bail bond companies do not constitute insurance.

The Allied Fidelity court analysed the nature of insurance and considered, but effectively ignored, the holding in Helvering v Le Gierse. Rather, the court focused on the fundamental elements of the transaction (in that case, the value of the bond to the government). That wholistic approach was upheld in other cases affirming that surety bonds do constitute insurance. This line of inquiry provides another avenue through which the IRS can assess what constitutes insurance and has largely been lost to the sands of time.

These cases provide two lessons: 1) the IRS’s notion of what constitutes insurance arises out of a life insurance case from the WW2 era which is leveraged against modern property and casualty captive insurance companies; and 2) there are other reasonable methods by which to assess what constitutes insurance that the courts ignore in favour of the current method leveraged by the IRS.

Insurable interest

The first prong of the Helvering test for insurance looks at whether there is an insurable interest. This is a reasonable inquiry. Only fortuitous events can be insured, and all practitioners, treatises, and common sense align with this view. This is a reasonable inquiry.

Risk shifting

Risk shifting is not a reasonable element for what constitutes insurance. First off, the IRS appears to hold to a position that risk shifting occurs only where there is an absolute shifting of the risk from the parent to the captive insurance company. This contradicts common industry practice. For example, mutuals, reciprocals, and risk retention groups never entirely shift risk from the insured to the insurance carrier.

Further, all E&S policies are not guaranteed by the state, which suggests that there remains some risk not transferred to the E&S carrier in the event that reinsurance declines to pay claims.

In addition, there are plenty of admitted carrier policies that fail to absolutely shift risk. Large deductibles and self-insured retentions lower premiums in exchange for the retention of risk.

The Tax Court takes a mature position and generally assumes that so long as the captive is collateralised within the statutory requirements of the domicile that a captive insurance company shifted its risk. This is a good balancing act to avoid a nonsensical element of the test. Since captive insurance companies are extended a licence only if they are sufficiently capitalised within the statutory confines of the domicile, all captive insurance companies should pass this test.

This sleight of hand effectively renders this prong useless and shows that the IRS’s first prong is flawed.

Risk distribution

Actuaries love risk distribution. The actuaries for the IRS state this in the case of Rent-A-Center v Commissioner. The Rent-A-Center court noted that the IRS’s actuaries believe that risk distribution occurs only where there are thousands of points of risk. This definition effectively seals off the value of captive insurance companies from middle-market companies and contains their value to massive corporations since only those companies would have enough internal points of risk to pass this test.

This element is flawed. The issue rests in its implicit assumption that all math is forever discovered. This position drips with arrogance because the IRS assumes that no actuaries or mathematicians will discover how to do more with less data. This assumption is already proven incorrect vis à vis the radical leaps of inference arising out of data science and its applications to the art of rate making (see literally every insurtech in the property and casualty space). Thus, the IRS’s insistence on vast amounts of data necessary to make a rate relegates its mindset to the late 20th century where the most sophisticated economic models leveraged Excel and Matlab.

The value of risk distribution is that thousands of points of risk help actuaries with developing bell curves so that the proper rate per unit of risk can be calculated with some degree of certainty. However, this certainty is not fundamental to insurance and is not an element of what constitutes insurance. Insurance involves risk. Risk taking is fundamental to insurance. A requirement for an absolute quantification of risk contradicts the inherent wager between the insured and the insurer. The insurer is betting on no losses while the insured is not willing to take that chance.

Further, there are innumerable examples of admitted and E&S carriers insuring risks without thousands of points of data. Satellite launches, the first offshore oil rigs, all new vaccines, and literally any new invention of any kind whatsoever lack thousands of points of data. Yet, carriers issue policies for athlete disability coverage, political risk coverage, and other lines of insurance where there are no comparable risks in the market.

Rent-A-Center removes the mask from the IRS and reveals its motive: requiring captive insurance companies to include thousands of points of risk creates a reliable bell curve which IRS actuaries can easily examine in order to assess if it is worth their time to prosecute a captive as a tax shelter. Leveraging smaller chunks of data requires more sophisticated mathematics, training data, and insider knowledge of the current state of the commercial markets which the government lacks. Risk distribution makes the IRS’s job easier.

Fortunately, the court realised that the IRS’s position was egregious. Alas, the Rent-A-Center Court adhered to the necessity of risk distribution. While the court pulled back on the IRS’s position, the IRS maintains that captives are inherently flawed because of their relatively small sample size of points of risk.

The good news for risk distribution is that the Rent-A-Center case leaves open the argument that risk distribution may be secured horizontally via insuring a lot of risks in an organisation, and vertically via reinsurance. This element is generally unexplored by the Tax Court and was avoided in the 831(b) cases. While the Tax Court addressed the nature of risk pools, the opinions in Avrahami, Reserve Mechanical, and Syzygy punted on exploring what amount or types of reinsurance open the door to risk distribution.

Insurance in the commonly accepted sense

When any court issues a test the rules need to be clear, easily applied to various situations, and capable of herding disparate groups of individuals toward similar results. This principle of consistency is the bedrock of justice and case precedence. The fourth prong of the IRS test fails to reach this standard. Insurance is a wide field ranging from warranties to social security to life insurance to health insurance to property and casualty. There are few practices germane to every form of insurance, current and yet to be invented, that constitute insurance in the common sense.

As an extreme example, even the underwriter is not necessary for insurance in the common sense. In parametric coverage the actuaries calculate the per unit rate and then models indicate the premium. For windstorm parametric managing general agents, the whole quoting process frequently consists of inputting a zip code into an Excel spreadsheet and selling the insured on a premium. This shows that something as apparently fundamental as evaluating the risk as an underwriter is not a fundamental element to what constitutes insurance. If we assume that even underwriting is not fundamental to insurance, then inquiring as to insurance in the “commonly accepted sense” is an incoherent standard on which to place a judicial inquiry.

Three cases focus on this prong of the test.

In the case of Harper Group v Commissioner, the court grappled with whether a captive conducted insurance in the commonly accepted sense. The court noted that:

  • The company was regulated by an insurance regulator;
  • The company was adequately capitalised;
  • Premiums were the results of arm’s-length transactions; and
  • Policies were valid and binding.

These factors were formalised in Securitas Holdings v Commissioner. In that case, insurance in the commonly accepted sense constituted:

  • The insurer is organised, operated, and regulated as an insurance company;
  • The insurer is adequately capitalised;
  • The insurance policies are valid and binding;
  • The premiums are reasonable; and
  • The premiums were paid and losses satisfied.

These factors were revisited in the 831(b) case involving Reserve Mechanical Corp v Commissioner. There, the court noted that insurance in the commonly accepted sense covered:

  • Whether it was created for non-tax purposes;
  • Whether there is a circular flow of funds;
  • Whether the entity faced actual and insurable risk;
  • Whether the policies were arm’s-length contracts;
  • Whether the entity charged actuarially determined premiums;
  • Whether comparable coverage is more expensive or available;
  • Whether it is subject to regulatory control and meets minimum statutory requirements;
  • Whether it is adequately capitalised; and
  • Whether it paid claims from a separately maintained account.

Some of these factors are reasonable inquiries. For example, a circular flow of funds or an intent to form an insurance company for some other purpose than insurance is a reasonable inquiry. However, these factors all stem from the case of Helvering v Le Gierse.

The Helvering Court stated that its criteria were: “The proper framework. They are not independent or exclusive for federal tax purposes. They inform each other and, to the extent not fully consistent, confining each other’s potential excesses.” The court realised that Congress delegated to the states the exclusive authority to regulate the business of insurance but the mere existence of an insurance licence is not “dispositive” regarding the existence of an insurance carrier.

Thus, the question becomes whether this prong has any meaning. All of the factors employed by the courts to visit this prong generally assess whether the company is doing business as an insurance company. That is not a real standard. There are too many types of insurance companies for this standard to hold water. For almost every standard (eg, whether comparable coverage is more expensive or available) there are arguments against the standard (eg, comparable coverage is irrelevant for uninsurable risks, hard markets, or innovative products and services).

In summary, the “commonly accepted sense” prong amounts to a sniff test. These types of tests open the door to bias, prejudice, and inconsistency. This is a betrayal of the basic function of the court system and demands reconsideration.

A better solution

State law, industry, and treatises all depart from the IRS’s definition of insurance. These definitions vary considerably but generally aggregate around a handful of concepts that are fundamental to insurance:

  • The insured possesses an interest in the insured thing;
  • The insured is subject to a risk of loss by losing the thing or an impairment of interest in the thing;
  • The insurer assumes at least some risk of the loss;
  • The insurer’s financials indicate the ability to pay that loss;
  • In consideration for this service of assuming the risk of loss, the insurer accepts premium;
  • Insurance for the risk of loss is the primary motivation of the transaction; and
  • This type of transaction constitutes something that should be regulated for the benefit of the public interest.

Insurance manifests in contracts containing these elements. Each of these elements are objective or reasonably determined without complicated tests or expensive discovery. These elements apply to every form of insurance. Also, these elements align with the common practice of insurance without reducing the transaction to a series of boxes to check in order to appease fickle regulatory bodies.


The wake of the 831(b) tax shelter captive insurance companies led to bad law. Bad law takes decades to resolve, if ever. Therefore it is vital to remember that the IRS’s position with regard to what constitutes insurance is arbitrary, uninformed, and designed to make its job easier with respect to prosecuting taxpayers. Its fundamental concern is about tax collections—not insurance. Thus, the blind spots in its definitions are fundamental and merit challenge in appropriate cases.

Matthew Queen is chief executive of Sherbrooke Captive Insurance and chief risk officer of parent company Goldner Capital Management. He can be contacted at