Incorporator The Journal of U.S. Formation
Start Free Quiz
Editorial 7 MIN READ

Micro-captives in 2023: the IRS keeps winning, and the rules are about to harden

Why the § 831(b) election is not dead, why the promoters are, and what a clean captive looks like going into 2024

The Editors
Staff, Incorporator.org
7 min read Updated
Contents 6 sections
  1. What Notice 2016-66 did, and what the proposed regs change
  2. Avrahami, Reserve Mechanical, Caylor: the doctrinal spine
  3. The 2023 enforcement posture
  4. The § 831(b) election is not dead
  5. What a clean 2024 formation looks like
  6. Sources

he IRS has spent eight years turning the § 831(b) micro-captive insurance market into a case study in what happens when a tax shelter attracts a listed transaction designation. In April 2023 the agency moved again, proposing regulations (REG-109309-22) that would replace Notice 2016-66 with a formal listing under Treas. Reg. § 1.6011-4, and the Large Business & International Division opened a new compliance campaign the same quarter.

If you are forming a captive this year, the facts on the ground matter more than the headlines. The § 831(b) election is alive. The promoters selling it as a premium-deduction machine are not.

What Notice 2016-66 did, and what the proposed regs change

Notice 2016-66, issued November 1, 2016, identified certain § 831(b) micro-captive transactions as "transactions of interest," a tier below a listed transaction but carrying the same disclosure machinery under Treas. Reg. § 1.6011-4. Participants and material advisors were required to file Form 8886 and Form 8918 respectively, with penalties under IRC § 6707A for non-disclosure and a six-year statute of limitations. The Notice targeted captives where loss-ratios ran below 70% over a computed period, or where the captive financed related parties, two fact patterns the Service had identified in promoter audits as common to abusive structures.

CIC Services, LLC challenged the Notice under the Administrative Procedure Act and the Anti-Injunction Act, arguing the IRS had bypassed notice-and-comment. The Supreme Court held for CIC Services in 2021 on the Anti-Injunction Act question, letting the APA claim proceed, and the Eastern District of Tennessee ultimately set Notice 2016-66 aside as to the plaintiffs. That procedural loss is the direct cause of what came next.

REG-109309-22, published at 88 Fed. Reg. 21547 on April 10, 2023, would redo the listing through full notice-and-comment rulemaking. The proposed regulations designate § 831(b) micro-captive transactions that meet certain loss-ratio or financing tests as listed transactions for § 6011 purposes, rather than transactions of interest. The comment period closed July 10, 2023; Treasury has signaled final regulations for 2024. Two substantive changes are worth noting. The loss-ratio computation window shortens and the loss-ratio trigger rises from 70% to 65%, and a separate financing test captures captives that lend, transfer, or otherwise make premiums available to related parties. If your captive trips either, you file.

Avrahami, Reserve Mechanical, Caylor: the doctrinal spine

The Tax Court has held for the Commissioner in every fully litigated micro-captive case of consequence since 2017. That record, more than any Notice, is what reshaped the market.

Avrahami v. Commissioner, 149 T.C. 144 (2017), was the opening decision. Judge Holmes worked through the Harper factors (risk-shifting, risk-distribution, insurance in the commonly accepted sense) and concluded the Avrahamis' Saint Kitts captive insured too few independent risks to distribute them and that the premiums, set to hit the § 831(b) ceiling exactly, were not arm's length. The captive was held not an insurance company for federal tax purposes. The premiums were not deductible under § 162, and the § 831(b) election was unavailable because there was no insurance to elect into.

Reserve Mechanical Corp. v. Commissioner, T.C. Memo 2018-86, took the same analytical frame to an Anguilla captive and reached the same result, with the Tenth Circuit affirming in 2022 at 34 F.4th 881. The appellate opinion is the cleanest recent statement of the risk-distribution requirement: a captive insuring the owner's operating company plus a thin "risk pool" of similarly situated captives does not distribute risk when the pool is structured so each participant bears only its own losses in substance.

Caylor Land & Development, Inc. v. Commissioner, T.C. Memo 2021-30, added a fourth taxpayer loss on materially similar facts, and piled on accuracy-related penalties under IRC § 6662 for the reasonable-cause failure. Taken together, the cases define a negative space. A § 831(b) captive that is a reinsurance-pool participant, has claims-to-premiums running near zero, sets premiums at or just under the statutory ceiling, and lends surplus back to the owner, is going to lose in the Tax Court and pay the 20% penalty on the way out.

The 2023 enforcement posture

The LB&I compliance campaign announced in March 2023 formalized what had been a case-by-case examination strategy. The Service's public statements frame it as a continuation of audits already in-flight and a widening to newly disclosed 2016-66 participants. The 2019 and 2020 settlement initiatives had resolved roughly 80% of then-docketed micro-captive cases on IRS terms, which left a smaller but more contentious remainder and a backlog of un-audited returns flagged by Form 8886 filings.

The terms of the 2019 initiative, broadly, required full inclusion of the previously deducted premiums, forfeiture of the § 831(b) election, and a 10% or 5% § 6662 penalty depending on the taxpayer's posture. Counsel who worked those files report that captives which had any real third-party underwriting record fared better, and captives that existed mostly on paper did not.

For advisors, § 6662A is the provision to keep in view. It imposes a 20% penalty on understatements attributable to reportable transactions (30% if undisclosed), with no reasonable-cause exception for undisclosed listed transactions. If REG-109309-22 is finalized and your captive was not disclosed on Form 8886, the math gets worse.

The § 831(b) election is not dead

This is the part the captive-selling industry will not say plainly, and the IRS has said only glancingly. Nothing in the cases or the proposed regulations calls the § 831(b) election itself abusive. The election allows a small insurance company (premium ceiling of $2.8 million for tax year 2023, indexed annually per Rev. Proc. 2022-38) to be taxed only on investment income rather than on underwriting profit. Congress re-wrote the provision in the PATH Act of 2015 to add diversification requirements specifically aimed at closing estate-planning exploits, and those requirements now appear at IRC § 831(b)(2)(B). Treasury has not proposed repealing any of it.

What has been ruled abusive, repeatedly, is the particular construction in which a closely held operating business forms a captive, has it write policies covering risks the operating business mostly self-insured anyway, joins a "pool" whose arithmetic leaves each captive with its own loss experience in substance, sets premiums at the deductibility- maximizing number, and treats the captive's surplus as an estate- planning vehicle.

A captive that insures a genuine risk the market either does not cover or prices poorly (cyber for certain regulated industries, specific professional liability tails, business-interruption exposures with long contingent periods), distributes that risk across enough independent insureds to satisfy Harper, charges an arm's length premium an unaffiliated actuary will sign, pays claims, and does not lend surplus back to the parent, is a captive the Service has not successfully attacked. Those captives exist. They are less common than the promoter population of 2016 suggested.

What a clean 2024 formation looks like

The through-line of the case law is that the IRS cares about substance in four places: risk-distribution, arm's length pricing, claims handling, and the captive's financial relationship with the parent.

Risk-distribution is best addressed by insuring a meaningful roster of operating sister entities plus genuine third-party risk, rather than a pooling arrangement whose mechanics reduce back to each participant bearing its own loss. Pricing should come from an independent actuary whose methodology would survive a deposition, not a tax-optimization spreadsheet. Claims should be adjusted by a third party at arm's length and paid without negotiation when triggered. The captive's assets should be invested in liquid instruments held by an unrelated custodian, and loans back to the parent should not happen.

Domicile choice (Vermont, Delaware, Hawaii, a long list of offshore options) is secondary to those four. A well-run Vermont captive will beat a poorly run Nevis captive every time, in examination and in the Tax Court. Formation counsel who has lived through a § 831(b) exam knows which promoters' templates to throw out on sight, and that is usually the single highest-value input a new captive buyer can get.

For founders reading this because someone pitched a captive at a year-end tax meeting, the question to ask is not whether the structure "works" in the abstract. It is whether the risk you would insure is real, and whether the person selling the captive is also the person who will pool it, manage it, price it, and adjust its claims. If the answer to the second is yes, walk.

Sources

Keep reading

More from the journal.