Charitable Remainder Trusts and CLATs — modeling split-interest trusts in wealth-tech

Charitable Remainder Trusts (CRTs) and Charitable Lead Annuity Trusts (CLATs) are the workhorse split-interest trusts of high-net-worth philanthropic planning. Each splits a property's interest between a charitable beneficiary and a non-charitable beneficiary across time, with prescribed actuarial valuations driving the upfront tax deduction and the ongoing tax treatment.

The instruments are not exotic — well-resourced family offices use them routinely, and their administration spans decades. But the data model required to advise on them is more demanding than a typical wealth-tech schema, and the math is sensitive in ways that often get hidden behind "consult your estate planning attorney." This article is the working note on what a wealth-tech platform actually needs to surface useful advice on these structures.

The two structures, mechanically

Charitable Remainder Trust (CRT): a donor contributes appreciated property to an irrevocable trust. The trust pays an annuity (Charitable Remainder Annuity Trust, CRAT) or a fixed-percentage of trust assets (Charitable Remainder Unitrust, CRUT) to a non-charitable beneficiary for a term (up to 20 years) or for the beneficiary's life. At the end of the term, the remaining trust assets pass to the named charitable beneficiary.

The donor receives an immediate income-tax deduction equal to the present value of the charitable remainder interest, computed using IRS-prescribed actuarial methodology and the §7520 rate (a monthly Treasury-rate-derived discount rate). The trust itself is tax-exempt and can sell the contributed appreciated property without recognizing gain — the gain is recognized only as it flows out through the annuity payments to the non-charitable beneficiary.

Charitable Lead Annuity Trust (CLAT): the inverse structure. A donor contributes property to an irrevocable trust. The trust pays a fixed annuity (CLAT) or fixed percentage (CLUT) to a charitable beneficiary for a term. At the end of the term, the remaining trust assets pass to the non-charitable beneficiary (typically the donor's heirs).

CLATs are gift-tax-leveraging tools. The donor's gift-tax cost is the present value of the remainder interest, which can be engineered close to zero ("zero-out CLAT") by structuring the annuity payments to deplete the trust's expected return entirely. If the trust's actual investment returns exceed the §7520 rate at funding, the excess passes to the non-charitable beneficiary tax-free.

The interplay of these structures with the §7520 rate, life expectancy assumptions, and projected investment returns is what makes them sophisticated and what makes them require a real data model.

The §7520 rate dependency

§7520 of the Code mandates the discount rate for valuing actuarial interests in property — it's set monthly at 120% of the federal mid-term rate. As of mid-2026, the rate has fluctuated between roughly 4% and 6% in recent years.

The §7520 rate dramatically affects the upfront tax mathematics:

• For a CRT, a higher §7520 rate increases the present value of the charitable remainder (because future cash flows discount more heavily), increasing the donor's deduction.
• For a CLAT, a higher §7520 rate decreases the present value of the charitable lead (the annuity stream is worth less in PV terms), increasing the present value of the remainder — which is the donor's gift-tax cost. Higher rates make CLATs less efficient.

For high-rate environments (when §7520 rate is above expected investment returns), CRTs are favored. For low-rate environments (when §7520 rate is below expected investment returns), CLATs can produce dramatic gift-tax leverage.

Wealth-tech platforms that surface CRT/CLAT advice without surfacing the §7520-rate sensitivity are giving incomplete advice. The same trust structure that's optimal at one rate is suboptimal at another, and the donor's funding decisions should be timed to the rate environment.

interface SplitInterestTrustModel {
  trust_type: 'CRAT' | 'CRUT' | 'CLAT' | 'CLUT';
  funding_date: ISODate;
  funding_§7520_rate: number; // The rate at funding date (locked)
  
  // For CRTs / CLATs both
  funding_assets: AssetContribution[];
  funding_fmv: number;
  funding_basis: number; // Carried over for income recognition
  
  // Annuity / unitrust spec
  payment_type: 'annuity' | 'unitrust';
  payment_amount?: number; // For annuity
  payment_pct?: number;   // For unitrust
  payment_frequency: 'monthly' | 'quarterly' | 'annual';
  
  // Term spec
  term_type: 'fixed_years' | 'life' | 'lives' | 'lesser_of';
  term_years?: number;
  measuring_lives?: Person[];
  
  // Beneficiary spec
  non_charitable_beneficiary: Person | Trust;
  charitable_beneficiary: PublicCharity | DAF;
  
  // Computed at funding (locked)
  charitable_deduction_at_funding?: number; // For CRT
  taxable_gift_at_funding?: number;          // For CLAT (close to 0 in zero-out)
  
  // Ongoing administration
  current_trust_value: number;
  cumulative_distributions: number;
  remaining_term_years: number;
  
  // Tier accounting (for CRTs — required by trust accounting rules)
  tier_1_ordinary_income: number;
  tier_2_capital_gain: number;
  tier_3_other_income: number;
  tier_4_corpus: number;
}

The "tier accounting" fields are CRT-specific and represent one of the most-mishandled requirements. The IRS prescribes that distributions from a CRT carry character based on a four-tier ordering — distributions are deemed first to come from the highest-character income, then the next, etc. The trust must track the historical and current tier balances; the beneficiary's tax treatment of each distribution depends on the tier from which it's deemed to come.

Many wealth-tech platforms model the trust at the asset-allocation level without modeling the tier accounting. For an active CRT, this means the platform can't actually advise on the beneficiary's tax treatment of distributions — a gap the beneficiary's accountant has to fill manually each year.

The actuarial mathematics

The valuation of the split interests requires actuarial computation. For a CRT, the charitable deduction is the present value of the remainder interest, which is the funding amount minus the present value of the annuity / unitrust stream.

For an annuity (CRAT), the math is a standard annuity-PV calculation:

PV(annuity) = annual_payment × annuity_factor(7520_rate, term_or_life_table)

For a unitrust (CRUT), the math is more complex because the payment varies with the trust value, which itself depends on investment performance. The IRS-prescribed methodology (Treas. Reg. §1.664-4) uses a unitrust valuation table that produces the PV factor as a function of the §7520 rate and the term.

For lives (rather than fixed terms), the relevant mortality table is IRS Table 2010CM (current) for trusts created after May 1, 2009. The table varies; new mortality tables are published periodically and the rate at funding date is locked.

A wealth-tech platform that wants to advise on CRT structuring needs the actuarial library to perform these computations. Several open-source libraries exist; the IRS publishes the tables. The platform doesn't need to recreate the math but does need to consume it correctly.

The CRT 5%-50% guardrails

For a CRT to be qualified under §664, the annuity / unitrust payment must be at least 5% and not more than 50% of the trust's initial value. Additionally, the present value of the charitable remainder must be at least 10% of the funding value.

These guardrails interact: a 50% CRUT for a young beneficiary's life will fail the 10% remainder test; a 5% CRUT for a short term will fail the same test. The valid design space is bounded.

A platform advising on CRT structuring should validate that the proposed structure passes both guardrails before presenting it to the donor. This is one of the most-common professional errors that platforms can prevent — design a structure, model the tax savings, and only at the funding step discover the structure doesn't qualify.

The CLAT zero-out engineering

CLATs are typically engineered to "zero out" — to produce a remainder PV close to zero, eliminating gift-tax cost. The structure: choose a §7520 rate, an annuity payment, and a term such that the present value of the annuity payments equals the funding amount. Any actual investment returns above the §7520 rate at funding pass to the remainder beneficiary as gift-tax-free wealth transfer.

In a high-rate environment, zero-out CLATs are difficult — the annuity payment must be steep enough that the PV calculation works, which means the trust may have insufficient assets to fund the annuity if returns underperform. In a low-rate environment, zero-out CLATs are highly attractive — the annuity payment is modest and the trust has substantial buffer.

The platform should surface a structuring recommendation that takes the current rate environment into account: "At the current §7520 rate of 5.4%, a zero-out CLAT requires an annuity of $X per year for Y years, expected to leave $Z to the remainder beneficiary at expected market returns."

What the test corpus needs

A test corpus for split-interest-trust modeling needs:

• Donors at the funding stage (structuring decisions)
• Active trusts with their first 1-3 years of distributions
• Mature trusts (10+ years in) with substantial tier-accounting history
• Trusts approaching termination (transition to charitable beneficiary)
• Failed structures (test data violating the 5/50/10 guardrails)
• Mixed-asset trusts (the funding included multiple asset classes with different bases)
• Trusts with NIMCRUT (net-income-with-makeup CRUT) variants
• CLATs that performed above projection (excess passes to remainder)
• CLATs that performed below projection (annuity may be unfunded; constructive distributions)

Each scenario produces structurally different tax treatment and structurally different advice. A platform that exercises only "newly funded" trusts misses the administration-stage scenarios where most of the actual planning value lives.

The synthetic-corpus value

Realistic split-interest-trust data is rare in production datasets — these structures are concentrated in HNW populations and tracking systems are usually accountant-managed spreadsheets. A purpose-built synthetic corpus with documented trust structures, tier-accounting state, and beneficiary mappings is required for the testing to exercise the math.

The corpus also serves as a template for the data the platform should be capturing. If the production schema doesn't include tier-accounting fields, the platform can't advise on distribution tax treatment. If the schema doesn't include the §7520 rate at funding (as a locked field), the platform can't recompute the deduction.

Split-interest trusts are sophisticated instruments, but their complexity is structurally enumerable. A wealth-tech platform that models them correctly differentiates itself in the HNW segment where these structures are common and the planning value per recommendation is six or seven figures. Without the model, the platform falls back to generic "consult your estate planning attorney" — which is true but doesn't help the client decide which structure to use, when to fund, or how to administer the ongoing trust.