831(b) vs 831(a): Who Elects What

Every US captive lives under one of two sentences of the tax code. Section 831(a) is the default: the captive is taxed like any other non-life insurance company. Section 831(b) is an election: a small insurance company can choose to be taxed on its investment income only, with underwriting results left out of the calculation entirely. The choice sounds like a tax question. In practice it is a statement about what kind of insurance company the captive expects to be — and it is far easier to make than to unmake.

What the two sections actually do

Under 831(a), the captive computes taxable income the way the tax code expects of an insurer: premiums earned, plus investment income, less losses incurred, reserve changes, and expenses — taxed at the corporate rate. Underwriting profit is taxed; underwriting losses are deductible and can be carried to other years. The captive is, for tax purposes, a real insurance business with both sides of the ledger recognised.

Under 831(b), a company qualifying as an insurance company may elect to be taxed only on taxable investment income, provided its net written premiums do not exceed the annual cap — $2.9 million for 2026, indexed each year. Underwriting profit is simply not taxed. The election also requires passing one of two diversification tests: broadly, no single policyholder supplies more than 20% of net written premiums, or the captive's ownership does not concentrate in the insured's owner's spouse or lineal descendants in a way the statute treats as estate-planning by another name.

Two features of the election deserve more attention than they usually get. First, it binds: once made, it applies to every later year in which the captive qualifies, and it can be revoked only with the consent of the IRS. Second, it is symmetrical: the same mechanism that exempts underwriting profit makes underwriting losses invisible. A bad year produces no deduction, no carryback, no carryforward. The premium cap, meanwhile, is measured against net written premiums each year — a growing programme does not get to round down.

What the election buys — and what it costs

The arithmetic that makes 831(b) attractive is not subtle. A captive writing $2.5 million in premium against modest expected losses retains most of that premium as surplus, and none of the underwriting margin is taxed on the way in. Compounded over a decade, the difference against a fully taxed insurer is substantial.

The cost is the mirror image. A captive built for volatile, loss-bearing lines — the programme that takes a genuine hammering every few years — gets nothing from the election in the good years it could not have had anyway, and gives up the deductions that would have mattered in the bad ones. And there is a second cost that has nothing to do with arithmetic: the profile that benefits most from the election — high premium, persistently low losses — is precisely the profile the IRS has spent a decade treating as presumptively abusive.

Who elects 831(a)

The default suits captives that behave like insurers in the fullest sense. Programmes writing above the premium cap have no choice, but plenty of captives below it stay on 831(a) deliberately: those retaining loss-heavy lines such as workers' compensation, commercial auto, or medical stop-loss, where deductions for losses and reserve development carry real value; those on a growth path that will cross the cap within a few years, for whom an irrevocable election is a needless constraint; and those for whom the tax treatment of reserves — deductible as they are established, not as they are paid — is itself a material part of the economics. An 831(a) captive files a return that looks like an insurance company's because it is one, and there is a quiet defensive benefit in that: nothing about the structure depends on a tax election to make sense.

Who elects 831(b)

The election fits a narrower profile than its popularity suggests: a captive with premium comfortably under the cap, genuine risk distribution, defensible actuarial pricing, and an expectation of moderate underwriting margins — not a near-total absence of claims. Enterprise-risk captives insuring deductible layers, gaps in commercial cover, and low-frequency/high-severity exposures for a mid-sized operating business are the natural candidates. For that captive, the election converts a real insurance operation's modest expected margin into retained, untaxed surplus, and the diversification tests are passed on the facts rather than engineered.

What the election does not fit is the arrangement designed backwards from the tax result: premiums priced to approach the cap rather than to match the risk, loss ratios in single digits year after year, and captive assets lent back to the owner. That is not a cautionary hypothetical — it is, almost verbatim, the fact pattern the IRS wrote into its disclosure rules.

The 2026 complication

In January 2025 the IRS finalised regulations designating certain micro-captive arrangements as listed transactions — the tests are mechanical, centred on a ten-year loss ratio below 30% and on financing that routes captive assets back to related parties — with a wider band of arrangements, broadly those with loss ratios below 60%, designated as transactions of interest. Both designations carry mandatory disclosure for participants and advisers.

Then 2026 split the map. In March, a federal district court in Tennessee upheld the regulations in full in CIC Services v. IRS. In April, a federal district court in Texas reached the opposite conclusion in Drake Plastics, vacating the listed-transaction designation while leaving the transaction-of-interest rules standing — and appeals are now running in both directions. The practical reading, for now: disclosure obligations under the surviving rules remain in force, an electing captive with a low loss ratio should assume its return will be read with interest, and any structure that relies on the vacatur is making a bet on the circuit courts.

A cleaner way to decide

Strip the promotion away and the decision test is short. Start from expected loss experience, not from year-one tax savings. If the captive is built to pay claims — if a 50–70% loss ratio across the cycle would be a sign of health rather than failure — the election's benefit is modest and its constraints are real; 831(a) keeps the deductions and the flexibility. If the captive writes genuinely low-frequency risk at premium levels an independent actuary would defend, the election is doing what Congress designed it to do — and the documentation that proves it should exist before the first return is filed, not after the first enquiry letter arrives. Either way, the section is a consequence of the programme's design. A captive whose design was a consequence of the section has the whole thing backwards.

Go deeper

Members can take this further in Lesson 6.2: Tax Treatment — 831(a) vs 831(b), which works the deductibility and offshore mechanics in detail; R.9: the 831(b) Micro-Captive Decision Framework, a structured walk through the election decision; and Lesson 6.1: IRS Scrutiny, Risk Transfer Standards, and Listed Transactions. The Captive Regulatory Tracker follows the litigation as it moves. Not a member? Academy Access opens the full curriculum.