Who Actually Manages.

On 12 August 2026 the United States Court of Appeals for the Fifth Circuit did something appellate courts rarely do. It took back an opinion it had issued seven months earlier, withdrew it entirely, and substituted a new one that reached the same result by a materially different road. K Alain, L.L.L.P.; K Alain GP, L.L.C.; Tax Matters Partner v. Commissioner of Internal Revenue, No. 24-60240 (5th Cir. 12 Aug. 2026) (per curiam) (Graves, Engelhardt, and Oldham, JJ.), slip op. at 1 (“The petition for rehearing en banc is DENIED. Treating the petition for rehearing en banc as a petition for rehearing, the petition for rehearing is GRANTED. We withdraw our prior opinion, Sirius Solutions, L.L.L.P. v. Commissioner of Internal Revenue, 165 F.4th 374 (5th Cir. 2026), and substitute the following.”). The case is now captioned K Alain, L.L.L.P. v. Commissioner of Internal Revenue — the same litigation the profession has been calling Sirius Solutions since January. The question was narrow and technical: what does “limited partner” mean in a self-employment tax provision Congress wrote in 1977?

The answer the court gave is not narrow at all. It is this: a limited partner, for these purposes, is “a partner who plays no significant role in managing or running a business.” Id. at 1–2 (“This case concerns the meaning of the term ‘limited partner’ in 26 U.S.C. § 1402(1)(13). Today, we hold its original public meaning is a partner who plays no significant role in managing or running a business. Thus, we VACATE and REMAND so the Commissioner may consider whether the partners at issue fall within that meaning ….”). The subsection is rendered “§ 1402(1)(13)” in the slip opinion; the provision is § 1402(a)(13).

That sentence is a tax holding. It is also, whether the court intended it or not, a statement of the principle this series has been circling for years. The law keeps arriving, from unrelated directions, at the same question about a client’s structure — who actually runs it? — and keeps answering that the paper matters far less than the practice.

Sirius Solutions, L.L.L.P. is a Delaware limited liability limited partnership operating a business-consulting firm out of Houston, with offices in Dallas and London. (It is now called K Alain, L.L.L.P.; the court retained the old names for clarity, and so will we.) For 2014, 2015 and 2016 it allocated essentially all of its ordinary business income to its limited partners — $5,829,402, then $7,242,984, then a loss of $490,291 — and, relying on the limited-partner exception, reported zero net earnings from self-employment. K Alain, slip op. at 3–4.

The Internal Revenue Code taxes the self-employment income of every individual, and “net earnings from self-employment” sweeps in a partner’s distributive share of partnership income. But section 1402(a)(13) carves out “the distributive share of any item of income or loss of a limited partner, as such,” other than guaranteed payments under section 707(c) for services actually rendered. 26 U.S.C. §§ 1401(a), 1402(a), (a)(13), (b); see Social Security Amendments of 1977, Pub. L. No. 95-216, § 313(b), 91 Stat. 1509, 1536; K Alain, slip op. at 2–3. The exception has been on the books, unchanged, since the Social Security Amendments of 1977.

The Service disagreed that Sirius’s partners qualified, issued notices of final partnership administrative adjustment, and moved the self-employment number from zero to $5,915,918 for 2014 and $7,372,756 for 2015. On 20 February 2024 the Tax Court sustained the adjustments, holding itself bound by its own recent decision in Soroban Capital Partners LP v. Commissioner, 161 T.C. 310, 320–21 (2023), which had read “limited partner” to reach only “passive investors.” K Alain, slip op. at 4. Appeal was taken from the United States Tax Court, Nos. 11587-20 and 30118-21.

In January 2026 the Fifth Circuit reversed on a simple ground: a limited partner is a partner in a limited partnership who has limited liability under state law. On rehearing, the panel abandoned that reasoning. Treating a petition for rehearing en banc as a petition for panel rehearing, it granted rehearing, withdrew its prior opinion, and substituted the one now before us. Id. at 1.

The new opinion is an exercise in original public meaning. Tax law, the court said, is federal law, and the Code leaves “limited partner” undefined; the task is to find what the phrase meant to an ordinary informed reader in 1977. Id. at 5, citing United States v. Bess, 357 U.S. 51, 55 (1958), Burnet v. Harmel, 287 U.S. 103, 110 (1932), and Wisconsin Central Ltd. v. United States, 585 U.S. 274, 277 (2018). It found the answer in the contemporaneous dictionaries — Black’s Law Dictionary of 1979 describing limited partners as those who “contribute capital and share in profits but … take no part in running business” — in the Uniform Limited Partnership Act of 1916 and its 1976 revision, both of which turned on whether the partner “takes part in the control of the business,” and in the partnership treatises of the period. Id. at 5–7, citing Limited Partnership, Black’s Law Dictionary (5th ed. 1979); Uniform Limited Partnership Act (1916); Uniform Limited Partnership Act (1976); A.R. Bromberg, Crane and Bromberg on Partnership 141 (1968); and H.G. Reuschlein & W.A. Gregory, Handbook on the Law of Agency and Partnership 435 (1979). It drew as well on the First Circuit’s decision in Plasteel Products Corp. v. Helman, 271 F.2d 354, 356 (1st Cir. 1959), which recognised that a partner who controlled partnership affairs was no longer functioning as a limited partner, while suggesting that lesser involvement did not forfeit the status. Id. at 7.

From that material the court took a distinction it described as managerial versus non-managerial. A limited partner could not manage the partnership, but “perhaps could participate in certain non-managerial aspects of the business.” Id. at 9.

It then rejected the Tax Court’s contrary rule in terms that will be quoted for a long time: Soroban had “selected a rule divorced from statutory text and that appears to prohibit even the most minor involvement in corporate affairs,” resting on “just a few sentences of operative analysis — citing no contemporary textual authority,” and making “no attempt to ground its rule in the original public meaning of ‘limited partner’ in 1977.” Id. at 9 and n.3. The court noted, pointedly, that the Service’s own instructions had for four decades defined a limited partner by liability rather than control — Package X in 1978, and the Form 1065 instructions from 1976 through 2017 — so that “parties were led to believe that what mattered was liability, not control.” Id. at 10, citing Internal Revenue Service, Package X: Informational Copies of Federal Income Tax Forms 137 (1978) and the Instructions for Form 1065 for 1976, 1977, and 2015–2017. Citing Loper Bright Enterprises v. Raimondo, 603 U.S. 369, 398 (2024), it observed that federal courts, not the Tax Court and not the Commissioner, decide the relevant questions of law. Id. at 8.

The judgment was vacated and the case remanded so that the Commissioner may consider whether these particular partners fall within the meaning the court had just supplied. Id. at 2, 11 (“We therefore VACATE its judgment and REMAND the case for further proceedings consistent with this opinion.”). Judge Graves dissented, reading the text and structure of section 1402(a)(13) to exempt only those “functioning as passive investors,” and would have affirmed. Id. at 12 (Graves, J., dissenting).

It would be easy to file this away as a self-employment tax development for partnerships in Texas, Louisiana and Mississippi. That reading is too small.

Notice what the court made dispositive. Not the entity’s state-law label. Not the certificate filed with the Secretary of State. Not the limited-liability shield. The partner’s actual role in running the business. A partner who manages is not a limited partner, whatever the partnership agreement calls him. A partner who does not manage may be one, even if he performs real services.

That is precisely — word for word, almost — the test that creditor law has applied to structures for two centuries. A trust in which the settlor still decides who gets what, and when, is a trust that a court will read as his own pocket, and it will say so using the language of sham or alter ego. A holding company that its owner controls in every meaningful respect is, as we have put it before, a coat of paint. A family limited partnership in which the “limited” partners in fact direct the investments, sign the cheques, and overrule the general partner is a partnership that a determined creditor will attack with real prospects of success. The fraudulent-transfer badges make the point from the other direction: retention of control over transferred property is one of the classic circumstantial indicia from which courts infer that the transfer was never real. See Uniform Voidable Transactions Act § 4(b) (badges of fraud, including whether “the debtor retained possession or control of the property transferred after the transfer”), as adopted in varying form across the states.

So the client who has been told that his LLLP interest is “limited” because the document says so is now exposed on two independent fronts at once, and the exposure has a single cause. If he in fact runs the business, the Fifth Circuit’s test denies him the tax exception. And if he in fact runs the business, the alter-ego and sham arguments that a judgment creditor will make are already half-written. There is no version of that fact pattern in which the label saves him.

The converse is the more useful lesson, and it is the reason we have always designed the way we do. A structure in which management genuinely sits elsewhere — a real general partner or manager with real authority, an independent trustee with genuine discretion, decisions actually made and minuted by people who are not the client — is the structure that survives both inquiries. Not because it was engineered to win a self-employment tax argument, but because it was engineered to be true. Truth is durable across doctrines. Labels are not.

We would be doing our readers no service if we let the good news stand unqualified.

This is one circuit. K Alain binds the Fifth Circuit. Soroban remains the Tax Court’s own position for cases appealable elsewhere, and the split between a “no significant management role” test and a “passive investor” test is now open and obvious. The Service may seek further review; Congress has been asked repeatedly to clarify section 1402(a)(13) and may finally do so. Nobody should build a structure on the assumption that this is settled, because it plainly is not.

Nothing was decided in the taxpayer’s favour on the facts. The court vacated and remanded. Whether Sirius’s own partners played “no significant role in managing or running” a consulting firm whose income they received is a question that has not yet been answered. A consulting partnership is not an obvious candidate for a finding of non-management.

The new test is harder to apply, not easier. A bright-line status test was predictable. A fact-intensive inquiry into who does what is not. Judge Graves’s dissent identifies the practical hazard from the other side, and even practitioners who welcome the result concede that outcomes on audit will now turn on evidence — organisational charts, meeting minutes, actual signature authority, who was on which call — rather than on a filing.

And most importantly: do not restructure to chase a tax result. A client who strips himself of management authority on paper while continuing to run the enterprise in fact has achieved nothing on either front. He has merely created a documentary record that contradicts his own conduct, which is the worst possible position from which to defend anything. If the management is to sit elsewhere, it must actually sit elsewhere — and it must have sat there long enough, and evidently enough, that the arrangement is a description of reality rather than an argument about it.

We have written before that protection must be seasoned, irrevocable and discretionary, and independently administered. K Alain is not an asset-protection case, and it will not be cited in one. But it validates the third of those principles from an unexpected quarter, and it does so in the plainest possible terms: the law is increasingly unwilling to accept a label as a substitute for a fact.

For clients, the practical question is not “what does my partnership agreement call me?” It is “if a court examined the last five years of how this enterprise was actually run, what would it conclude?” A structure that answers that question comfortably is a structure that works — for tax, for creditors, and for the long-horizon reasons a family builds one in the first place. A structure that answers it badly will keep losing, in one forum after another, for the same reason each time.

Build it so that the answer is comfortable. Build it early, so that there are five years to examine.

This note is general commentary for planning discussion and is not legal advice or a tax opinion for any particular person, entity, or transaction. Self-employment tax treatment of partnership interests is unsettled and varies by circuit; charging-order and alter-ego standards vary by state and by jurisdiction of formation. Every conclusion described here depends on specific facts, documentation, and timing, and should be assessed with qualified counsel.