What Makes State Inheritance Tax So Hard for General-Purpose AI to Get Right

Five states impose an inheritance tax - Pennsylvania, Kentucky, New Jersey, Nebraska, and Maryland. For clients with connections to any of those jurisdictions, it’s a real variable in what beneficiaries ultimately receive. And getting it right requires more than knowing the rate schedules.

There are three things that have to work together: the data going into the calculation, the calculation itself, and the ability to actually use the output. 

Starting With the Right Data

Inheritance tax is calculated at the beneficiary level, not the estate level. Unlike estate tax, which applies to the aggregate taxable estate, inheritance tax turns on who is receiving what and what relationship they bear to the decedent. The rate that applies to a sibling in Pennsylvania is different from the rate for a lineal heir. In Kentucky, spouses, lineal heirs, and siblings are exempt, while nieces, nephews, and more distant beneficiaries remain subject to tax. In Nebraska, spouses and charitable organizations are fully exempt, with graduated exemptions applying to others. In Maryland, a broad class of close family members pays nothing, while everyone else is subject to a flat 10%.

And the calculation is only as good as the data behind it. You need each beneficiary's relationship to the decedent, what they're receiving, and whether any exemptions apply. You need to know where assets are sited, and where the decedent was domiciled. All five states tax a resident decedent's full estate but limit their reach to in-state real and tangible personal property when the decedent was a non-resident. Getting those inputs wrong doesn’t produce a wrong answer that looks wrong. It produces a wrong answer that looks plausible. 

Most tools require an advisor to supply that information manually at the time of calculation, which means the result is only as good as what they remember to enter. What's needed instead is a system where inheritance tax is computed as a property of the plan rather than a standalone calculation, where relationship classifications are drawn from ingested trust and estate documents, situs determinations are tied to structured asset data, and beneficiary-level detail comes from the same model driving the rest of the estate. The data doesn’t have to be re-entered. It’s source-verifiable and consistent with everything else in the plan. 

Why the Calculation Is Harder Than It Looks

State inheritance taxes look straightforward at first: a handful of states, class-based rates, defined exemptions. That surface simplicity is misleading, and it's why general-purpose tools, including large language models, don't produce the level of accuracy these calculations require.

The class structures alone vary significantly by state and don't map onto one another. But the bigger challenge is how inheritance tax interacts with other taxes in the stack.

Kentucky allows a deduction for federal estate tax against the inheritance tax base, limited to the portion attributable to Kentucky-situs assets. But federal estate tax is itself affected by the §2058 deduction for state death taxes paid, which includes inheritance tax. So the two calculations are circular, each is an input to the other, and they have to be solved together through iteration rather than sequentially. When the tax is paid from the residue, the default in Kentucky, the tax payment is itself a taxable transfer, which requires a gross-up that iterates until the numbers converge. Nebraska has the same structural dynamic: federal estate tax is deductible against the Nebraska inheritance tax base, which makes the federal and state calculations circular there as well. These aren’t edge cases, but rather a common feature of how inheritance tax interacts with the federal stack in multiple states, and a general-purpose tool won't reliably surface, let alone resolve correctly.

Maryland has a different wrinkle. It is the only state that levies both a state estate tax and an inheritance tax, and the two are interdependent. Inheritance tax actually paid is credited dollar-for-dollar against the Maryland estate tax liability, and where the inheritance tax equals or exceeds the estate tax, no estate tax is due. New Jersey, by contrast, doesn't allow any deduction for federal estate tax against the inheritance tax base, which distinguishes it from Kentucky and Nebraska, and matters in any scenario where both apply to the same transfer. Pennsylvania presents its own layer of complexity. The §9128 deduction allows a reduction in the Pennsylvania taxable base for death taxes paid to other states on the same assets, which matters significantly for clients whose estates span multiple jurisdictions.

Getting it right also means keeping up with it. Maryland's HB17, effective July 1, 2026, moves a non-resident decedent's intangible assets to their home-state situs and out of the Maryland base, a meaningful change for any non-resident client with Maryland-situs property. Rules change, and a calculation built on last year's statute is wrong even if the math is correct. Accurate calculation requires version-stamped rule sets for each state so that scenarios run against the law as it actually applies on the relevant date.

Handling these calculations accurately requires an engine designed to resolve circularity and second-order effects across a complex tax stack, and not a calculator that takes inputs and returns a number. It must model federal estate tax, GST, and state estate taxes for every jurisdiction that levies them, resolving the interdependencies between them through iteration. Inheritance tax fits into that same framework: the state-specific tax coding of beneficiary classes, exemption structures, and residency and situs logic, layered onto an engine that knows how to resolve circular dependencies.

Putting the Output To Work

Getting the math right is necessary, but it's not the end goal. The value is in having that result integrated into something an advisor can actually use with a client to inform comprehensive wealth transfer decision-making.

That means the inheritance tax needs to be part of the estate waterfall, not a separate calculation sitting off to the side. Each beneficiary's projected distribution should already reflect the inheritance tax applicable to their share, based on their relationship to the decedent and the rules of the relevant state. The beneficiary report should show what heirs will actually net. Without that, even an accurate calculation leaves advisors doing manual work to bridge the gap, which is exactly the kind of friction a well-built model should eliminate.

The circular interactions, the state-specific class structures, the interplay between inheritance tax and estate tax, the ongoing maintenance, none of which is visible to the advisor or the client. What’s visible is an estate waterfall that accounts for everything, and answers the question that matters: here is what your beneficiaries are projected to receive. That's where the technical work becomes useful, and it’s what separates a purpose-built planning engine from a tool that approximates the answer. 

‍

Luminary’s domain-specific AI, data model, and deterministic tax calculation engine helps advisors answer client questions with speed, confidence, and clarity. Schedule time with a Luminary expert here to learn more about how we support inheritance tax questions and much more.