Rabbi Trust

A rabbi trust is a trust that an employer establishes and funds to set aside assets for its obligations under a nonqualified deferred compensation arrangement, on terms that keep those assets available to the employer's general creditors. It is often described as informal funding, which is exactly right: money is genuinely segregated from the employer's operating accounts, and just as genuinely still part of the employer's estate.

The name has no official standing, and it is worth saying so up front. Revenue Procedure 92-64 refers to "executive compensation arrangements that are popularly referred to as 'rabbi trust' arrangements" — the quotation marks and the hedge are the IRS conceding this is not a legal term. Neither the Internal Revenue Code nor the regulations contain the phrase. Legally the vehicle is a grantor trust under IRC §§671–679, with the employer as grantor.

The name comes from where the idea started. A 1980 private letter ruling (PLR 8113107) addressed a congregation that had set aside funds in trust for its rabbi's deferred compensation. The IRS ruled the rabbi was not currently taxed: precisely because the assets remained subject to the congregation's creditors. A private letter ruling may not be cited as precedent by other taxpayers under IRC §6110(k)(3), which is why practitioners had no reliable template until the IRS published model trust language twelve years later.

One drafting choice inside that model decides how much the arrangement is actually worth to the employee, and it is easy to miss. The model offers three alternatives: the trust may be irrevocable from the outset, revocable by the employer until a change of control and irrevocable from that point, or irrevocable a set period after a favourable private letter ruling. A revocable version can be unwound by the employer at will, so it secures nothing until whatever event it names actually happens. Everything below about protection against a change of heart assumes the irrevocable form, and the trust agreement is the only place to check which form is in use.

The organising insight: a rabbi trust protects against the employer's change of heart, not against the employer's insolvency. This is the point most competing material states imprecisely, usually by saying a rabbi trust "protects your deferred compensation", which is untrue as to the risk that actually destroys these benefits. What the trust genuinely prevents is the employer deciding, later, to spend the money elsewhere, or new management after an acquisition declining to honour the arrangement: once the trust is irrevocable, the employer has no power to direct the trustee to return the assets or divert them elsewhere until all the benefits have been paid. What it cannot prevent is a bankruptcy court treating the trust assets as part of the employer's estate, because the trust document must say they are available to the employer's general creditors.

That limit is the price of admission, not an oversight. Remove the creditor exposure and the participant has an economic benefit, or constructive receipt, and is taxed on the deferred compensation immediately, in the year it is set aside rather than the year it is paid. Security and deferral trade off one for one, and no drafting removes the trade. A secular trust takes the other side of it: real security for the employee, and immediate taxation. A governmental 457(b) plan is a third case entirely, where the law requires assets to be held in trust for the exclusive benefit of participants: genuine protection under a different regime, not a rabbi trust.

The model trust matters. Revenue Procedure 92-64, as modified by Notice 2000-56, contains the IRS's model rabbi trust language, and three points about it are load-bearing. It is a safe harbor: an arrangement using the model language will not cause constructive receipt or an economic benefit merely because the trust was adopted. The IRS will generally not issue rulings on arrangements that use a materially different trust, except in rare circumstances, so departures from the model carry real uncertainty. And because the employer is the grantor and owner of the trust for tax purposes, the trust's investment income is taxed to the employer each year — the trust is not a tax-sheltered vehicle in its own right. Always cite the guidance as Revenue Procedure 92-64 as modified by Notice 2000-56; the 1992 document alone is incomplete.

Congress deliberately closed the two obvious fixes. IRC §409A(b) added two triggers for immediate income inclusion. Under §409A(b)(1), assets held in a trust located outside the United States cause inclusion regardless of whether creditors could reach them. Under §409A(b)(2), a provision that moves assets beyond creditors' reach on a deterioration of the employer's financial condition: a so-called springing trust, which is exactly the feature that would have made rabbi trusts genuinely safe, also causes inclusion. The market's natural workaround was legislated away.

Two further points. A rabbi trust does not make the underlying plan "funded" for ERISA purposes, so a top-hat arrangement keeps its exemption from ERISA's vesting, funding, and fiduciary rules; participants hold mere unsecured contractual rights with no preferred claim on the trust. And this is not solely an executive curiosity: tax-exempt employers' 457(b) plans, which must be unfunded, commonly use rabbi trusts to hold the deferrals of ordinary nonprofit and hospital employees, so the people bearing this creditor risk are frequently not executives at all.

The employer adopts a nonqualified deferred compensation arrangement, executes a trust using the IRS model language and chooses its revocability alternative, and transfers assets to the trustee. The trustee holds and invests them, pays benefits under the plan's schedule, and, by the trust's own terms, must make the assets available to the employer's general creditors if the employer becomes insolvent.

A hypothetical example showing both halves of the protection. A hospital sets aside $2,000,000 in an irrevocable rabbi trust for the deferred compensation of twelve senior staff, of which Marcus is owed $250,000.

Change of heart: the hospital is acquired, and new management would prefer to redirect that $2,000,000 to a capital project. It cannot. Because the trust is irrevocable, its terms bar reversion for general corporate purposes while benefits remain unpaid. Marcus's money is still there. This is the protection the trust actually delivers, and it is the half a revocable version would not have given him.

Insolvency: instead, the hospital files for bankruptcy owing $30,000,000 to unsecured creditors. The trust's $2,000,000 is available to those creditors, and Marcus's $250,000 claim ranks alongside every other unsecured claim rather than ahead of them. If unsecured creditors ultimately recover 20 cents on the dollar, he receives roughly 20% × $250,000 = $50,000. The trust changed nothing about this outcome. Figures are illustrative.

Pros

• In its irrevocable form, prevents the employer from spending earmarked money on other priorities, and survives a change of ownership or management.
• Uses IRS model language under Revenue Procedure 92-64, so the tax treatment is well settled rather than a matter of argument.
• Preserves the tax deferral, which a genuinely protective trust would destroy.
• Reassures participants enough to make deferral programs viable, and gives them a visible asset pool to monitor.

Cons

• No protection whatsoever against the employer's insolvency — participants remain general unsecured creditors.
• Trust income is taxed to the employer, so the vehicle carries an ongoing tax cost rather than sheltering growth.
• Cannot be drafted to spring protection on financial deterioration, and cannot be held offshore, without triggering immediate taxation under IRC §409A(b).
• The name is reassuring in a way the substance is not, which leads participants to overestimate their security, and a revocable version, which the IRS model permits, delivers even less than they think.
• Adds legal, trustee, and administrative expense to an arrangement that is already documentation-heavy.