The Partnership That Arrived Too Late.

Fields is an estate-tax case, and readers who do not spend their days in subchapter B may be tempted to file it under “valuation discounts, other people’s problem.” That would be a mistake. The reasoning the Fifth Circuit applied is structurally identical to the reasoning a creditor’s court applies to a hurried transfer, and the failure modes are the same failure modes. Two different bodies of law — transfer tax and voidable transactions — have independently converged on the same test, which is a fairly strong hint that the test is describing something real.

The facts are worth stating precisely, because the precision is the lesson.

Anne Milner Fields was a successful Texas oil-and-gas figure. She was diagnosed with Alzheimer’s disease in 2011. By 2016 she was not managing her own affairs; her great-nephew, Bryan Milner — a career corporate-finance professional whose education she had funded — acted as her attorney-in-fact under a power of attorney.

The certificate of formation for AM Fields, LP was filed on May 20, 2016; Milner executed the partnership agreement on May 25. Fields contributed $16,972,409 in assets for a 99.9941% limited partnership interest. The general partner, holding 0.0059% in exchange for $1,000, was AM Fields Management LLC, of which Milner was the sole member. Milner signed every document both individually and as Fields’s agent. The funding ran in tranches — May 27, June 6, and a final transfer of nearly $10 million from her brokerage account on June 13 — and left Fields with about $2.15 million outside the partnership. Ten days after that last transfer, she died. The estate then valued her limited partnership interest at $10,877,000 — diminishing the gross estate by roughly $6 million against the assets that had gone in.

The Tax Court held in 2024 that section 2036(a)(1) and (a)(2) required the underlying assets, not the discounted partnership interest, to be included in the gross estate, disallowed the discounts, and sustained the penalty. The Fifth Circuit affirmed across the board; the panel later granted the Commissioner’s motion to modify and substituted a revised opinion on July 31, 2026, leaving the disposition unchanged.

Section 2036(a) pulls back into the gross estate property a decedent transferred during life while retaining possession or enjoyment of it, the right to its income, or the right to designate who will possess or enjoy it. It contains one escape hatch: the transfer is respected if it was “a bona fide sale for an adequate and full consideration in money or money’s worth.” I.R.C. § 2036(a) (flush language).

Courts have long read “bona fide sale” in the family-entity context to require a legitimate and significant non-tax purpose for the transaction — and the Fifth Circuit’s formulation in Fields is the sentence estate planners should tape to the wall: the objective evidence must indicate the non-tax reason was a significant factor, and a significant purpose must be an actual motivation, not a theoretical justification. See Strangi v. Commissioner, 417 F.3d 468, 479 (5th Cir. 2005).

The estate offered three non-tax purposes. Each was plausible in the abstract. None survived contact with the record:

• Curing gaps in the power-of-attorney succession. The court was unpersuaded, observing that the existing power of attorney was operative, and that the partnership itself required unanimous consent to change course — hardly an improvement in flexible succession.
• Consolidating asset management. The holdings did not present the kind of pooling or coordination benefit that this rationale assumes, and the court rejected it on the record presented.
• Protecting against fraud and elder abuse. Also rejected as an established significant non-tax motivation on these facts.

The pattern is not that these purposes are unavailable. It is that they were asserted rather than demonstrated. A purpose that first appears in a brief is a theoretical justification. A purpose that appears in board minutes, in correspondence, in a change in how the assets were actually run, and — crucially — in the calendar, is an actual motivation.

Set the transfer-tax vocabulary aside and look at the shape of the transaction. A person with a large, foreseeable claim on the horizon — here, the estate tax, which becomes certain at death and was, by 2016, imminently foreseeable — moves substantially all of her wealth into an entity, weeks before the claim attaches, on terms that change nothing about who really controls or enjoys the property, and then asserts that the property is worth substantially less than what went in.

Read that sentence again with a judgment creditor in the place of the Commissioner. It is the fraudulent-transfer fact pattern, almost badge for badge: transfer to an insider; transfer of substantially all assets; timing that tracks the claim rather than any independent business rhythm; retention of possession and control after the transfer; consideration that is nominal in substance. See Uniform Voidable Transactions Act § 4(b). The UVTA and section 2036(a) were written by different people for different purposes, and they arrive at the same place, because both are asking the same underlying question: did anything real actually change?

In Fields, nothing real changed. Fields put in 99.9941% and got back a claim on 99.9941%. The 0.0059% general partner was an entity wholly owned by her own attorney-in-fact — the same person who had been managing her affairs before the partnership existed and who continued managing them after. There was no independent administration, no genuine relinquishment, no third party whose judgment could override the family’s. The discount asked the law to price an economic separation that had not occurred.

The Fifth Circuit sustained the twenty percent accuracy-related penalty under section 6662, rejecting the estate’s good-faith and reliance-on-professional-advice defenses and characterizing the roughly $6 million discount as too good to be true. Citing Sun v. Commissioner, 880 F.3d 173, 181 (5th Cir. 2018). That characterization is the part practitioners should sit with.

A structure can fail on the merits and still leave the client no worse off than if nothing had been attempted. That is the comforting assumption behind a great deal of late planning: we may as well try; the downside is that it does not work. Fields is a direct rebuttal. The downside was that it did not work and cost an additional penalty on the deficiency, plus years of litigation through the Tax Court and a court of appeals. Failed planning is not free. In the creditor context the analogue is worse still, because an avoided transfer typically hands the creditor an additional cause of action, additional discovery, and a narrative of concealment to argue from.

The recurring Lighthouse position is that protection must be seasoned, irrevocable and discretionary, and independently administered. Fields illustrates all three by their absence, but it is worth being concrete about what the first one is doing.

Seasoning is not superstition, and it is not a waiting period for its own sake. Time performs a specific evidentiary function: it makes the non-tax, non-creditor purpose observable. A partnership that has operated for eight years has eight years of distributions, meetings, tax returns, disagreements, and decisions that a court can inspect. Its purpose is not asserted; it is documented in the ordinary course. A partnership that has operated for twenty-nine days has nothing to show but the intentions of the people who formed it — and when the only intention a court can objectively verify is the one visible from the timing, that is the intention the court will find.

The same logic explains why independent administration matters. A trustee or manager who is genuinely independent will, over time, do things the settlor did not want. Those moments are the proof. A general partner wholly owned by the family member already running the assets will never generate that proof, because there is no friction to record.

And it explains the irrevocability point. Fields retained, in every practical sense, everything she had before. A transfer that leaves the transferor’s economic position unchanged is not a transfer that the law will price as one — whether the law asking the question is section 2036 or section 4 of the UVTA.

For clients and their advisers, we would draw four points from Fields, offered as general observations rather than advice on any particular structure:

1. Build on a clear day. The single variable most predictive of outcome in both the transfer-tax and creditor lines of cases is the interval between the structure and the claim. It is also the one variable that cannot be improved retroactively.

2. Make the non-tax purpose do visible work. If the stated purpose is consolidated management, the management should actually change and the change should be documented. If it is succession, the structure should improve on the alternative that already exists, not merely restate it.

3. Give up something real. Retained possession, retained enjoyment, retained power to designate — these are the express triggers of section 2036(a), and their creditor-law cousins are the badges. A structure that costs the client nothing in control will generally be worth nothing in protection.

4. Put someone genuinely independent in the chair. Not a relative wearing a general-partner hat. Not the attorney-in-fact who was already in charge.

Whether any of this changes a given family’s analysis turns entirely on that family’s facts, its jurisdiction, and its timing. What Fields settles is narrower and firmer: a structure assembled at the bedside will be read as what the calendar says it is.

The estate in Fields was not defeated by a clever argument or an unfavorable circuit. It was defeated by a date. The partnership agreement and the death certificate were one month apart, and no amount of drafting quality could put distance between them after the fact.

There is a version of this plan that works, and it is not exotic. It is the same plan, executed in 2008 instead of May 2016, with a general partner nobody in the family controlled, with a decade of ordinary operation behind it, and with a settlor who had genuinely let go of something. That version is not available in June. It was available for years, and it was not taken.

That is the whole of the lesson, and it is not a new one. It is only newly expensive.

This note is general commentary for informational purposes and is not legal or tax advice for any specific matter; outcomes turn on the facts, the timing, and the jurisdiction, and should be assessed with qualified counsel.