Inheritance Tax and the UK Tech Cohort

What is being argued, what the disagreement turns on, and what different evidence would mean — for the founders, investors, and operators driving the new economy.

The reformed inheritance tax regime that took effect on 6 April 2026 caps 100 per cent Business Property Relief at £2.5 million of qualifying assets per person (transferable to spouses for £5 million per couple), with 50 per cent relief above. For unlisted UK trading-company shares above the cap — founder equity, venture LP interests, EIS portfolios in unlisted private companies — the result is an effective 20 per cent inheritance tax rate on the excess. The reform is contested in a way most reform debates are not: not over whether the change should have happened, but over whether the death-event valuation mechanism is operationally fit for this particular asset class.

This piece does not argue that any of the positions canvassed below is correct. It sets out what each is, what they share, what their differences turn on, and what different empirical outcomes would imply. The intent is to help a reader think clearly about a question that has been argued at high volume and low resolution.

What is actually in dispute

The design positions on the timing-and-mechanism question have been advanced by knowledgeable people in good faith.

• Position A — Hold the existing mechanism. Keep IHT-at-death, adopt practical fixes, resource HMRC adequately. The principle of the reform requires a death-event tax; the cohort affected is small (HMRC's December 2025 estimate is approximately 1,100 estates affected by the whole APR/BPR reform in 2026-27, of which around 185 include an APR claim, with the unlisted-trading-company-share subset on which this piece focuses being a portion of the remaining BPR-only ~915); mechanism change risks special carve-out for the wealthiest holders. The strongest form of Position A goes further: the operational-mismatch argument is the same argument that advocates of every tax targeting the wealthy have always made — that this tax is uniquely unfit, this asset class uniquely illiquid, this cohort's behavioural response uniquely large. The historical record of these arguments is that they are usually overstated and advanced both by people in the affected cohort and by tax practitioners and academics with no personal stake in the outcome. Position A's strongest reading is to treat the present argument the way comparable arguments have always been treated: with measured scepticism, while still adopting genuinely sensible operational improvements.
• Position B — Switch to Capital Gains Tax on realisation. Replace IHT-at-death with CGT charged on actual proceeds when the heir realises, with no base-cost uplift on death and a long-stop deemed-disposal rule. The death-event mechanism mismatches the asset class; a realisation-based regime collects the same principle against optimised exit values. Note a structural feature Position A's defenders should engage with: if Option B's long-stop is deferrable until actual liquidity, the state effectively becomes a contingent quasi-equity holder across hundreds of UK private companies — economic exposure to outcomes without governance rights. This is structurally a different relationship between state and private business than tax collection, with constitutional and economic-policy implications the public debate has not engaged with.
• Position C — Hybrid, fixes only, mechanism question deferred. Adopt the practical measures both sides accept; defer the mechanism question pending evidence on relocation, dispute caseload, and receipts.
• Position D — Targeted higher threshold for qualifying unlisted trading-company shares. Keep IHT-at-death. Keep the £2.5m / £5m allowance for the general BPR base. Raise the 100% relief threshold (proposals range from £5m to £10m per individual) for a defined sub-class of qualifying unlisted trading-company shares — the cohort the operational mismatch is concentrated in. Combine with the four practical measures. The strongest objection to Position D is the special-pleading-erodes-the-base argument that Position A's strongest form was built to reject: a tech-specific higher threshold is exactly the carve-out for a sympathetic asset class that history shows expands unbounded over time, and the original BPR is itself the carve-out the reform exists to correct. There is also a definitional problem (what is "tech"? SIC codes are gameable; activity-based definitions create case-by-case classification disputes; EIS-style exclusion lists scale up boundary fights with much larger stakes per estate; R&D-intensity tests are clean analytically and worst politically). And a fairness objection: raising the cap for tech founders but not for the family that owns a 200-year-old engineering firm employing 400 people requires a prudential answer (mobility) where readers may want a fairness answer.

The design positions support four practical measures: a collateral-trigger rule, a statutory safe-harbour valuation methodology, targeted tightening of temporary non-residence rules, and structured buy-back guidance for cash-generative companies. These four measures are common ground. The actual disagreement is whether they are sufficient on their own.

One scope note. This piece assumes the principle of the reform — that very large illiquid private holdings should be brought into the IHT base above an allowance. A reader who thinks the reform itself was wrong in principle will find the analysis here operating one level too deep. The disagreement between A, B, and C is a disagreement among people who accept that the reform should have happened.

A note on timing

Before the rest of the discussion: the behavioural response to the announced reforms began before the reforms took effect. The Office for Budget Responsibility's January 2025 costing of the related non-dom reform — a different reform, governing a different population — assumes that approximately 25 per cent of non-doms with excluded property trusts will leave the UK as a result. This is the central UK government assumption for the non-dom reforms, not for BPR. It tells us double-digit departure rates are considered plausible by HMRC for highly mobile wealthy populations; it does not directly forecast the BPR cohort, which has different mobility characteristics (founder equity is anchored in UK companies, customers, teams). The figure is a useful reference point, not an answer.

The directly observable evidence is suggestive but contaminated. Financial Times analysis of Companies House records shows 3,790 UK company directors changed their primary residence to abroad between October 2024 and July 2025 — a 40 per cent increase on the same period a year earlier, with director departures in April 2025 alone running 79 per cent above April 2024. Approximately 150 directors moved specifically to the UAE in the second quarter of 2025. But this figure is much larger than the BPR-affected cohort (~1,100 estates across the whole APR/BPR reform per HMRC December 2025 estimates, with the BPR-only unlisted-share subset in the low hundreds), and the period coincides with multiple tax-policy changes (non-dom abolition, residence-based IHT for non-doms, CGT changes, carried-interest changes). The April 2025 spike in director departures correlates more directly with the non-dom reform's effective date than with the BPR reform, which took effect a year later. The data shows that wealthy UK directors were relocating in higher numbers; it cannot isolate the BPR-driven component.

The named tech-founder cases (Nik Storonsky of Revolut, who has registered UAE residency; Herman Narula of Improbable, who has been reported as preparing to emigrate, citing a mooted exit-tax rather than the BPR reform specifically) and adviser-survey reporting from Sifted (15–20 per cent of new business enquiries at one specialist firm now concern UAE relocation) are illustrative rather than evidential — three named cases do not establish a rate, not all of those cases are responses to the BPR reform, and adviser-survey evidence is qualitative. The widely-cited Henley & Partners figure of 16,500 millionaire departures in 2025 has been forensically critiqued by Tax Policy Associates and Tax Justice Network and should not be cited as evidence. The OBR figure and the Companies House data are the more credible sources, with the caveats above about what each can and cannot tell us about the specific cohort this analysis addresses.

What this means for the rest of this piece: any analytical framework calibrated against post-April-2026 data will measure a population that has already partly responded. The early-mover cohort has already exited the dataset. So measured outcomes will understate true behavioural response. The reader should hold this in mind through the analysis below.

What the disagreement turns on

Five questions, none of which has been answered with rigour:

1. How large is the relocation channel under the existing mechanism? Empirical, answerable through HMRC dynamic modelling. Currently relying on adviser surveys and trade-press reporting, neither sufficient.
2. How quickly does SAV dispute caseload grow as the cohort grows? Empirical. Depends partly on the rate of cohort growth, which industrial strategy is trying to accelerate.
3. How effective is the collateral trigger against sophisticated structuring? Technical. Depends on drafting quality.
4. Can a 10- or 15-year long-stop be administered without recreating the original problem? Empirical. Depends on the realistic distribution of realisation events across the cohort.
5. What is the regime for? Normative. Revenue, fairness across asset classes, or industrial-strategy alignment? When they conflict, which wins? Ministers, not analysts, must answer this.

Reasonable readers reach different conclusions on these questions, and that is most of what produces the disagreement between the positions. A reader with a clear view on the questions has, by implication, a clear view on the policy.

What each of the design positions would look like in operational form

The four practical measures (collateral-trigger rule; statutory SAV safe-harbour methodology; targeted tightening of temporary non-residence rules; structured buy-back guidance) are common ground. The design positions endorse them. They should be adopted on their own merits, separately from the harder design question.

Position A in operational form. Adopt the four measures. Resource SAV. Run the regime as enacted. Routine post-implementation review at two to three years. No pre-committed triggers. Position A's defenders argue this is the right posture under uncertainty: observed evidence should drive any subsequent change, not thresholds set ex ante.

Position B in operational form. Adopt the four measures. Legislate CGT-on-realisation for unlisted trading-company shares above the £2.5m / £5m allowance, with a long-stop deemed disposal at year 10 to 15 and a payment-on-realisation deferral provision. Position B's defenders argue the switch should happen on the available evidence, without waiting for confirmation that may arrive too late.

Position C in operational form. Adopt the four measures. Commission HMRC dynamic modelling. Defer the mechanism question. Two variants: without triggers — review afresh when evidence accumulates, with no pre-commitment; with triggers — pre-commit to a Position B switch if defined evidence thresholds are breached at a published twelve-month review point, with parallel legislation drafted so the trigger is time-deliverable. Position C's defenders are split between the two variants. The trigger thresholds in any such framework face a calibration problem the analysis cannot solve from outside government — without HMRC microdata, internal HMT working papers, and AEOI exchange information, any threshold an analyst proposes is an illustration of the kind of trigger that would be needed, not a number a government could act on. The full version of this paper sets out illustrative thresholds (around 60 departures, 350 active SAV files, receipts more than 20 per cent below forecast, 35 per cent of advised estates planning relocation, average SAV settlement at less than 65 per cent of opening assessment) but these are illustrative, not actionable.

Position D in operational form. Adopt the four measures. Legislate a higher BPR threshold for a defined sub-class of qualifying unlisted trading-company shares (proposals range from £5m to £10m per individual). The general £2.5m / £5m threshold remains for farms, mature family businesses, and the bulk of the affected cohort. Position D's defenders argue this is one of several designs proposed in response to the cohort-heterogeneity observation; advocates argue it is the most directly responsive while critics argue it creates definitional problems around what counts as a qualifying unlisted trading company the article identifies. Position D's critics argue this is the special-pleading carve-out for a sympathetic asset class that has historically been how tax bases erode, that the definitional problem (what is "tech"?) has no clean statutory solution, and that the fairness objection is harder to defend than supporters typically acknowledge.

The publication does not adjudicate between them. Each is internally coherent; each rests on different assumptions about behavioural response, policy stability, and the normative question of what the regime is for. The reader weighing the four can see what each would actually involve and form their own view.

What different evidence would mean

Four scenarios, ordered by what HMRC modelling and observed receipts data could plausibly show. The reader should hold the timing point above in mind: each scenario describes what the data could measure, but the data will measure a population that has already partly responded. A small measured response could mean either a small underlying response or a large response already absorbed into the pre-reform departures.

Worth noting before the scenarios: the companion policy options paper finds Option B's central-case revenue impact in the range of approximately −£100 million per year over five years, with a wider range of −£200m to +£600m depending on which behavioural assumption is correct. Option B is not obviously revenue-positive on the central case. B's stronger argument is on cohort retention and behavioural elasticity, not on direct receipts. Framings of the choice as "which option collects more tax" are operating on a question the evidence does not clearly answer.

If the relocation channel is small and SAV caseload manageable — annual departures below ~30, caseload growth tracking estate count, receipts within forecast — Position A is broadly vindicated. The four practical measures are sufficient. Mechanism change is unnecessary. Treasury collects what was forecast; the affected cohort finds the regime more navigable than the announced response suggested; the principle of fairness across asset classes is preserved. The venture-stage cohort still faces residual operational difficulty, but too small a group to drive policy change.

If the relocation channel is meaningful but bounded — annual departures in the 30–80 range, caseload somewhat faster than estate count, receipts 5–15% below forecast — the picture is mixed. Holding the regime delivers most of the revenue but loses some; switching may collect more but at the political cost of being seen to capitulate. The regime is workable for most but pressure persists for the segment most exposed to mechanism mismatch. Whether this is acceptable depends on Question 5: revenue, fairness, or industrial-strategy alignment as primary.

If there is substantial capital flight and operational pressure — annual departures above 80, caseload superlinear with multi-year disputes, receipts more than 20% below forecast — Position B's case strengthens substantially. The regime in its current form is not collecting what was forecast and is producing a measurable behavioural response. Mechanism change recovers some receipts going forward, but the cohort that left does not return. The principle of fairness is sustained nominally but undermined operationally. The damage to industrial strategy is significant and persistent — affecting not just current founders but the next cohort considering the calculation.

If the data is contested or delayed — the likeliest real-world outcome, on the historical record of UK tax policy reviews — any conditional framework's credibility depends on the trigger firing under conditions advocates will agree have been met. If the data is genuinely contested, the trigger becomes another consultation rather than a forcing function. The regime drifts in its current form for several years, regardless of the underlying empirics. This is the scenario that makes both Position A's and Position B's defenders most uncomfortable with conditional frameworks, in different ways.

The limits of this analysis

Three limits are worth naming. First, the analysis is system-internal by design. It asks how the UK regime should work, not which tax system the cohort should sit inside. An AI cross-critique session put the boundary more cleanly than the article's drafts had: the article operates in a "given the system, fix it" model, while sophisticated actors operate in a "given multiple systems, choose one" model. That is the fundamental mismatch. The OBR's 25 per cent figure is for a different population governed by a different reform; the Companies House director-departure data is contaminated by other tax changes happening at the same time. Both are suggestive of a broader behavioural environment in which mobile wealth is moving more than usual, but neither directly establishes that the BPR cohort is responding at scale. The article's fixes may work within the UK system while high-value activity relocates or restructures outside it. This boundary is intentional — system-selection responses are not legislatively addressable in the same way as system-internal ones — but it matters, and the system-selection question deserves its own treatment, which this piece does not provide. Second, the analysis is asymmetric in its quantification — friction is measurable, outcomes are not, and the reader should weight accordingly. Third, the normative question (what the regime is for) is a political choice, not an empirical finding, and no analysis can settle it.

Closing

The reformed inheritance tax regime is the right reform in principle in the view of all four serious positions. The mechanism for one specific asset class is contested on grounds that turn on five questions whose answers are not fully knowable today. Four positions and several conditional frameworks have been proposed; each rests on different empirical assumptions and different normative priorities; each has identifiable consequences under each plausible empirical outcome.

This piece has tried to set out the question, the open evidence, and what different outcomes would imply, without arguing for any of them. Readers will reach different conclusions depending on which empirical outcomes they consider most likely and which normative framing they treat as primary. That is the nature of policy questions answered under genuine uncertainty — not a failure of analysis, but what the question actually looks like.