The UK Tech IHT Model

An interactive model of the 25-year fiscal effect of three policy options. Move the sliders. Watch the results change.

What the model is doing

The model walks through the same logic as the spreadsheet companion, in three steps.

1. The cohort flow. Each year, a number of UK tech founders enter IHT exposure (their personal equity passes the £2.5m allowance threshold). A fraction of them depart pre-death, with the fraction varying by policy option. The remainder stay and either pay IHT at death (Options A and C) or CGT at exit (Option B).

2. The direct fiscal effect. Each year's stayer cohort produces tax revenue, lagged by the average time to death (Option A and C) or the average time to heir's exit (Option B). The model discounts these future cashflows back to present value at the discount rate. Note: because the model only spans 25 years and the average lag to death is ~20 years, most of Option A's direct revenue stream falls outside the modelling window. The first 25 years capture only the leading edge.

3. The indirect fiscal effect. Each departed founder produces an indirect loss to the UK tax base. This has two parts: the current company's tax base (corporation tax, employment-related tax, VAT) over its remaining UK operating life, partially lost when the founder relocates; and the next company the founder would have built in the UK, which is now built abroad. The next-company effect is probability-weighted by the chance of successful first-company exit and the chance of building a second company.

The model is most useful for understanding which assumptions move the answer most. The answer is not a single number; it is a structure of dependencies.

What the model can and cannot tell you

The model can show you which assumptions matter, what the order-of-magnitude differences between options are under different assumptions, and where the break-points are between options winning and losing. It can be a tool for thinking the question through.

The model cannot tell you what the right answer is. The behavioural response is uncertain. The next-company effect is uncertain. The company tax base is uncertain. The model honestly expresses the dependencies between these inputs and the output. A reader who substitutes more pessimistic behavioural assumptions will get a different answer from a reader who substitutes more optimistic ones, and the model will faithfully report what each set of assumptions implies.

That is the point.

What would change the publication's mind

The model's qualitative finding is that the indirect cost (founder departure, lost subsequent ventures, lost cluster effects) dominates the direct revenue (BPR/APR receipts on the affected estates) by roughly two orders of magnitude in the central case, and by something more modest in the conservative case. The publication uses this finding to argue that mechanism design matters a lot — not just headline-revenue figures. Five things would meaningfully shift this finding:

1. Empirical evidence that pre-death founder relocation is much lower than the model assumes. The model's central case sets pre-death departure-rate-on-incentive at 4–6% per affected founder. If a future Companies House study, an HMRC tracking exercise on the BPR cohort specifically, or an independent academic study found that actual relocation in response to estate-tax-incentive (controlling for other contemporaneous tax changes) is below 1%, the indirect cost shrinks proportionally and the direct vs indirect ratio compresses from ~500–1,000x toward 10–50x. The qualitative finding survives but is much weaker.
2. Empirical evidence that founders who leave continue to pay roughly equivalent UK tax. The model assumes that a relocated founder's UK company-tax base shrinks by ~40-60% over time (operations move, IP migrates, customer base diversifies). If practitioner data showed UK tax base contributions are largely sticky even after personal departure (e.g., founder departs but UK HQ, UK employment, UK supplier base remain), the indirect cost would shrink and the direct vs indirect ratio would compress significantly.
3. Empirical evidence that next-company probability is substantially lower than 30–45%. The model assumes a 30–45% probability that a founder of an affected estate would have started a subsequent significant venture had they remained in the UK. This is the single largest indirect-effect input. If panel data on serial founders showed the next-company rate is below 15%, the indirect cost falls by more than half.
4. Evidence that the network multiplier is below 1.0. The model uses 1.30 in its central case (one founder's exit and continued UK presence increases other UK founders' tax contributions by 30% via mentorship, hiring, investment, and cluster effects). If a serious empirical study showed the multiplier is at or below 1.0 — that one founder's UK presence does not measurably increase the productive activity of others — the indirect-effect calculation would lose a meaningful component.
5. Evidence that heir productivity is high enough to offset the timing mismatch. The publication's principle piece cites Holtz-Eakin et al. (1993) and replications showing heirs of large estates work less and start fewer companies than non-heirs. If new evidence showed that for the specific UK tech-founder-equity heir cohort the productivity gap is small or absent — for example, panel data on UK family-business succession showing heirs match founder generation on entrepreneurial activity — the principle-level argument for taxing the transfer at all weakens, and the mechanism question becomes less consequential.

None of these five pieces of evidence currently exists in published form for the UK BPR-affected cohort specifically. The publication's central-case parameters are anchored to comparable evidence from adjacent populations (US estate-tax studies, OBR non-dom assumptions, Companies House director data with caveats, IFS commentary on horizontal equity) and the conservative-case parameters compress the indirect-effect ratio toward what the conservative reading of those adjacent studies supports. A specialist working with HMRC microdata, or an independent research institution with access to Companies House panel data, could produce the UK-cohort-specific evidence on these five questions; if and when they do, the model would need to be re-run and the publication's qualitative finding revisited.

What would not change the publication's mind on this question: headline-revenue figures going up or down by a factor of two; the £2.5m threshold being raised or lowered by 50%; the OBR producing a new fiscal forecast that does not separately model the indirect channel. None of these would address the questions the model is actually trying to answer.

An audit path — three places where the same calculation lives

The model exists in three independent representations that should produce the same numbers for the same inputs. A reader wanting to validate the model can compare across all three:

The interactive model on this page. The JavaScript that runs in your browser is visible via View Source. Every input is exposed as a slider with its current value; every output is computed in code you can read. Open the developer console and inspect the calculation as it runs.

The Excel companion. Available here. Every formula is in a cell; every assumption is named; every intermediate value is visible. A specialist modeller can audit the Excel cell-by-cell and check the structural logic.

Hand calculation. The model's structure is simple enough to walk through manually for a single cohort year, then extrapolate. A reader who computes one year's direct revenue (cohort × stay-rate × per-founder tax × discount factor) and one year's indirect cost (departures × tax-base-loss × years × probability-weighting) by hand should land within an order of magnitude of what the JavaScript and the Excel produce. If the three representations diverge, at least one is wrong.

This publication has not done that hand-validation independently. The publication has not had a fiscal economist or modeller perform a cell-by-cell Excel review. The publication has not commissioned an independent recalculation. The audit path exists; the audit has not been performed. Send corrections if you find one.

Sensitivities to test first. Four assumptions move the answer most: pre-death departure rate (the gap between Options A and B), the percentage of company tax base lost on relocation, the next-company probability, and the network multiplier. Setting any of these to its conservative bound compresses the indirect-effect ratio from ~500–1,000x in the central case to ~50–200x in conservative cases. The qualitative finding — that indirect dominates direct — survives every assumption set the publication has run. The precise magnitude does not.

Model. The Excel companion is available here with every formula visible and editable. The interactive model on this page exposes the same calculation logic in JavaScript that runs in your browser — view source on this page or open the Excel to inspect every assumption, formula, and intermediate value. The methodology is documented inside the model and explained in the readable piece. How the model was built: Doug Scott prompted four AI tools (Claude, ChatGPT, Grok, Gemini), the AI tools produced the model structure, the formulas, and the cross-critique, and Doug scanned the output and decided to ship. Doug did not verify the model math against an independent calculation, did not check every formula, and did not have the model reviewed by a fiscal economist or modeller before publication. No human expert reviewed any of this work. AI cross-critique catches some errors and misses others; the errors AI cross-critique misses are exactly the ones a specialist modeller would catch on a careful read. Numbers are estimates anchored to public sources where possible; many are uncertain. The model's output spans roughly two orders of magnitude across plausible assumption sets — do not cite a single number from this model as if it were a forecast. Readers are encouraged to substitute their own assumptions.

Author. Doug Scott, founder and ex-CEO of Redbrain.com. He was born in the UK, lived overseas, and came back. His companies have always been UK-owned, UK-operated, UK-tax-paying. He adapted his own position when the BPR reform was announced; many in his cohort did not. He has invested personal money directly and indirectly into hundreds of very-early-stage UK tech companies and advised many more — the standing the model is built from. The outcome of the policy debate has minimal effect on him personally now. He has been raising the question with government for some time; this model is one of the things AI tools made it possible for him to build to articulate the dependencies systematically. The model is structured to make the dependencies visible, not to argue for a specific outcome.

Found an error? The publication maintains a public corrections log with every dated correction since launch. If you find an error in the model — a formula error, a wrong assumption, an off-by-one, a misnamed cell reference — please send the correction via the email on the about page. The publication will post it on the corrections page with attribution and update the canonical version.