The Wealth Delta Tax: Falsifiable Hypotheses

Keywords

Wealth Delta Tax, falsifiable hypotheses, macroeconomic effects, taxpayer preference, cooperative equilibrium, accumulation-point efficiency, automation resilience, hostile modelling

🚧   THIS PAPER IS UNDER CONSTRUCTION   🚧

Version: 0.1  |  Date: 20 Sep 2026  |  Word count: 11,557 (excl. front matter)

Author Disclosure

Portions of the drafting, editing, literature organisation, and structural review of this paper were assisted by publicly available large language models, including Anthropic’s Claude and OpenAI’s ChatGPT. These tools were used as aids to the author’s research and writing process; the substantive arguments, analysis, interpretations, and conclusions are the author’s own.

This work received no external funding, sponsorship, or other financial support. The author is solely responsible for the content of the paper and for any errors that remain.

Revision History

Revision	Date	Details
0.1	20 September 2026	First full draft: all sections, abstract, appendix

Abstract

The other papers in this series argue for WDT. This one does not.

The WDT project has, across 32 papers, established a mechanism design, a valuation architecture, a governance structure, and a set of revenue projections. That architecture is now sufficiently detailed that the claims embedded in it can be stated precisely enough to be attacked. FAL identifies the five propositions on which WDT most depends, states them in their strongest falsifiable form, and specifies what evidence or modelling would count against each.

The five hypotheses are: that wealthy taxpayers prefer WDT to the existing system when total economic costs rather than tax liability alone are compared (H1); that WDT’s capital allocation, labour displacement, consumer demand, and automatic stabilisation mechanisms combine to produce net positive macroeconomic welfare effects (H2, decomposed into five independently falsifiable sub-hypotheses); that a cooperative institutional equilibrium can be established and sustained through and beyond Phase One’s bootstrapping vulnerability (H3); that taxing changes in wealth produces lower aggregate distortion than taxing intermediate manifestations of the same economic process (H4); and that the WDT tax base is structurally more resilient to automation-driven labour income decline than labour-based fiscal systems (H5).

Two hypotheses carry partial existing support — H2a and H4, grounded in the lock-in welfare results of (WFR) — and H2c has empirical support by analogy from Norway’s 2020 sovereign wealth fund drawdown. The remaining five are hypotheses in search of a test.

For each hypothesis, the paper specifies a precise falsification condition, the minimum requirements a hostile model must satisfy to constitute a genuine test, and the results that would count as serious evidence against WDT. The minimum requirements are set not to make falsification difficult but to ensure that a model that finds a hypothesis false has tested the hypothesis rather than a simplified substitute.

The objective is not to protect WDT from falsification. It is to make falsification possible.

Glossary

Break-even burden. The annual WDT wealth-equivalent burden at which a representative taxpayer at a given wealth level is indifferent between WDT and the existing tax system, accounting for all economic costs of each.

Conditional support. An evidentiary outcome in which a hypothesis holds only within an identifiable parameter range, with the binding constraints explicitly characterised.

Falsification. An evidentiary outcome in which a hypothesis fails across empirically defensible parameter ranges, or survives only under assumptions that are themselves implausible.

Hostile modelling. Modelling conducted by researchers with every incentive and opportunity to demonstrate that a hypothesis fails, rather than by the hypothesis’s proponents.

Positive feedback loop. The chain of mechanisms by which WDT-induced improvements in capital allocation and labour tax displacement are hypothesised to amplify one another, increasing both the WDT tax base and broader welfare over time.

Support. An evidentiary outcome in which a hypothesis holds across a substantial range of empirically defensible parameters. A model that can be made to work by choosing favourable parameters does not meet this standard. The hypothesis must hold across a substantial and realistic range, not merely somewhere within it.

1. Purpose and Methodological Commitment

1.1 What This Paper Is

The other papers in this series argue for WDT. This one does not. Its purpose is to identify the propositions WDT most depends on, state them in falsifiable form, and specify what evidence or modelling would count against them. FAL invites researchers to try to break the architecture the rest of the series establishes.

A research programme that evaluates its own claims only through models designed to support them is advocacy. The WDT project has produced a sufficiently detailed architecture that its embedded claims can be stated precisely enough to be attacked. Another paper defending the design is not what the project needs at this stage.

The author lacks the computational and empirical resources to model all five questions adequately. That is not a weakness to minimise; it is the condition that makes this paper necessary. If the resources existed, the questions would be settled. They are stated here because stating them precisely is what makes hostile modelling possible. A vague objection can always be answered vaguely. A precise hypothesis can be falsified.

1.2 What This Paper Does Not Claim

FAL does not establish that the five hypotheses are true. It does not present new modelling. It does not defend WDT against the falsification conditions it identifies. It takes no position on whether the hypotheses will survive scrutiny.

A paper that presented a list of falsification conditions and then argued each one away would defeat its own purpose. The reader should finish this paper uncertain about whether WDT works as claimed — and equipped to find out.

1.3 Relation to the Existing Project

The WDT series has, across 32 papers, established a mechanism design, a valuation architecture, a governance structure, and a set of revenue projections. Three confirmed literature gaps have been closed by (WFR): the Domar-Musgrave extension to a progressive delta base, a welfare comparison including the delta instrument, and the distributional arithmetic of delta-base concentration under persistent return heterogeneity. Six literature gaps remain open. Ten empirical questions cannot be resolved without a live system (PHASE1 §4).

FAL does not add to that architecture. It selects the five claims that, if shown false, would most seriously damage the case for WDT, and states each with enough precision that a researcher who wants to attack it knows what model to build, what parameters to vary, and what result would constitute a finding.

The relationship to the rest of the series is asymmetric. FAL draws on (WFR), (RATES), (ENV), (BEHAV), (GOV), (CLOSE), (POL), and (PHASE1) for the structural detail behind each hypothesis. Those papers are the source of the claims; this paper is the target painted on them.

2. The Research Question

The question this paper addresses is:

Under what conditions would the behavioural, macroeconomic, institutional, and welfare mechanisms proposed by the Wealth Delta Tax fail to perform as claimed?

Five sub-questions follow, each corresponding to a domain in which WDT’s argument depends on a proposition not yet formally tested.

The first concerns taxpayer behaviour: whether wealthy taxpayers would prefer WDT to existing arrangements once total economic costs, not just tax liability, are compared. The second concerns macroeconomic outcomes: whether the capital allocation, labour displacement, consumer demand, and automatic stabilisation mechanisms combine to produce net positive welfare effects. The third concerns institutional dynamics: whether a cooperative equilibrium between taxpayers and the WDT institution can be established and sustained through the bootstrapping phase before institutional credibility is demonstrated. The fourth concerns the comparative efficiency of the tax base: whether taxing changes in wealth produces less aggregate distortion than taxing the intermediate events — realisations, transactions, income flows — through which the same underlying economic process currently becomes visible to the tax system. The fifth concerns structural resilience: whether the WDT tax base is more robust to automation-driven labour income decline than fiscal systems whose primary revenue base is labour.

These are not the only open questions in the WDT project. The open questions register (WP §9) lists ten Phase One empirical unknowns, six Governing Council calibration parameters, and six jurisdiction-specific preconditions. FAL addresses none of those. It addresses the five propositions whose truth or falsity most determines whether WDT’s core argument stands. The distinction matters: Phase One empirical questions concern how the mechanism would be calibrated and administered; the five hypotheses here concern whether the mechanism works at all.

Each hypothesis is stated below with a precise falsification condition, a specification of the minimum requirements a hostile model must meet to constitute a serious attempt at falsification, and a current status assessment — honest about which hypotheses have theoretical or empirical grounding and which are, at this stage, propositions in search of a test.

3. H1: Wealthy Taxpayer Preference

3.1 Statement

H1. Under plausible parameterisations, wealthy taxpayers will prefer WDT to the existing tax system when their total economic costs are compared rather than their tax liability alone.

3.2 The Comparison That Matters

The intuitive objection to any wealth tax is that the people subject to it will leave. H1 does not dismiss this objection. It reframes the comparison on which the objection depends.

The migration decision is not made by comparing a WDT tax bill against a current tax bill. It is made by comparing the total economic cost of remaining under WDT against the total economic cost of the existing system. Those two quantities are not the same as their tax liabilities, and the gap between them is large enough to matter for the hypothesis.

The WDT total cost has three components: the annual wealth burden at reference parameters, net of expected refund value in loss years, and net of expected institutional participation benefits. (RATES §10) establishes the revenue-weighted annual wealth burden at 0.35% of net worth at reference parameters. The gain-weighted effective lifetime rate is 13.0%, directly comparable to CGT on a materially larger base.

The current system total cost has never been estimated comprehensively for the relevant population. It includes taxes actually paid across income tax, CGT, IHT, and NICs; professional compliance costs; avoidance expenditure; portfolio distortion from the CGT realisation incentive; tax uncertainty; and inheritance planning costs. The NAO’s 2025 review of HMRC’s wealthy individual programme (National Audit Office (2025)) documents that 73% of the roughly 850,000 individuals with assets above £2m or income above £200,000 are represented by a tax agent, that complex tax affairs generate compliance investigations averaging 40 months at the high-value end, and that legal-interpretation disputes account for over half the £1.9bn wealthy tax gap. These are not costs HMRC incurs; they are institutional traces of costs the taxpayer bears. Advani et al. (2020) provides the most rigorous existing UK estimate of the taxpayer-side compliance burden, though it addresses a one-off wealth tax rather than WDT’s ongoing regime. No study has combined these components into a single all-in friction cost estimate for the UK’s high-wealth population. That gap is the primary input FAL cannot supply, and the Appendix Table is intended to sketch the methodology for filling it.

3.3 The Break-Even Burden

The most useful object a hostile model could produce for H1 is not a verdict but a break-even burden: the annual WDT wealth-equivalent cost at which a representative taxpayer at a given wealth level becomes indifferent between WDT and the existing system.

Four representative taxpayer brackets are defined by (RATES §5.1), drawing on ONS WAS 2018–20 data with Pareto extrapolation above the top-1% threshold (JUR §2.4):

Bracket	Mean starting wealth	Population
P90–P95	£1,629,190	3,460,000
P95–P99	£2,858,347	2,768,000
P99–P99.9	£7,135,016	622,800
P99.9–P99.99	£19,853,605	62,280

The P90–P95 bracket sits below the canonical £2m threshold at which the WDT rate function produces material liability. H1’s falsification territory therefore lies primarily at P95–P99 and above, where the 0.35% reference burden becomes substantive. If the current system’s all-in friction cost exceeds the break-even burden, remaining is rational on total-cost grounds regardless of the direct tax comparison.

The 0.35% figure is the revenue-weighted average across the full taxpayer distribution, not the burden at any specific bracket. The burden at P99–P99.9 under the logistic rate function is materially above 0.35%; the burden at P90–P95 materially below. H1’s falsification condition should be assessed bracket by bracket. A result showing H1 fails at P99.9 but holds at P99 is informative: it identifies the wealth level at which the break-even flips, which is the primary calibration input for Phase One threshold design.

3.4 The Institutional Participation Component

H1 is not only a cost comparison. WDT taxpayers acquire a continuing relationship with an institution whose capital partly reflects their own contributions. Under the Governing Council architecture (GOV §3), the Taxpayer Chamber holds 25% of formal voting weight, giving the taxable population direct and constitutionally protected representation in rate-setting. The symmetric refund mechanism (WP §3.5) provides contingent downside protection at the same marginal rate as the upside charge. The SWF’s accumulated reserves constitute a systemic claim on the future functioning of the institution.

A taxpayer with a 20–30 year wealth horizon may rationally prefer a lower-friction, more predictable regime with structural downside protection over a higher-friction regime with unlimited avoidance potential but no protection in loss years. Garbinti et al. (2026) provides the most direct evidence on the predictability dimension, finding that the French wealth tax’s compliance architecture generated substantial avoidance expenditure that a more transparent and stable regime would not. Brülhart et al. (2022) finds that Swiss wealth tax responses were driven primarily by avoidance rather than migration, consistent with the framing that current-system avoidance costs are a real and large component of the total-cost comparison.

The expected value of institutional participation is a variable the hostile model should allow to be zero or negative. If wealthy taxpayers do not believe the refund guarantee will be honoured, or do not believe TP Chamber representation will be substantive, the institutional participation component of H1 collapses. Whether those beliefs are rational depends in part on whether H3 holds — the two hypotheses are not independent.

3.5 The Migration Evidence

The existing empirical literature on wealth tax migration constrains but does not determine H1’s falsification condition. Agrawal et al. (2025) estimate that Norway’s 2022 wealth tax increase produced a cross-base fiscal externality of approximately 22 cents of total revenue lost per additional unit of wealth tax raised, once income and consumption tax losses from migrating households are included. Jakobsen et al. (2024) find significant wealth accumulation responses to Danish wealth tax repeal that are difficult to decompose cleanly into migration versus avoidance channels. Iacono & Smedsvik (2024) find behavioural responses to Norwegian reforms that are real but, in aggregate, fiscally modest relative to pre-reform revenues. H. Kleven et al. (2024) surveys the broader taxation and migration literature and finds elasticities that vary substantially by institutional context and tax design.

Three features of the WDT architecture distinguish it structurally from the systems in which these estimates were produced. The delta base produces a revenue-weighted annual burden of 0.35%, structurally below the 1–2% stock levies in the Norwegian, Swedish, and Swiss systems from which most migration elasticity estimates derive (BEHAV §9.2). Route C and Route D hold accumulated basis gaps as departure costs, making exit materially more costly for a taxpayer who has managed declarations strategically (CLOSE §5). The bridging facility (CLOSE §5) decouples physical departure from settlement completion, reducing the practical pressure for aggressive pre-departure restructuring.

None of this means migration under WDT would be zero. Existing elasticity estimates are not directly applicable; a hostile model should treat migration as an endogenous choice at WDT’s specific parameters rather than importing elasticities calibrated to stock-base systems at materially higher effective rates. The direction of the adjustment is not obvious: the lower burden reduces the migration incentive while the departure cost mechanism may increase avoidance-in-place as a substitute.

3.6 Falsification Conditions

H1 is falsified if a heterogeneous taxpayer choice model, meeting the minimum requirements specified in (FAL §10), demonstrates that:

• the current system’s all-in friction cost is robustly below the WDT reference burden across the P95–P99 and higher brackets under empirically defensible compliance cost estimates;
• the expected value of institutional participation is negligible or negative across plausible beliefs about refund credibility and TP Chamber effectiveness;
• avoidance opportunities under WDT are sufficient to make non-cooperation more attractive than participation for a substantial fraction of the taxable population; or
• WDT reduces expected lifetime wealth at reference parameters by enough to make migration rational after accounting for departure costs, bridging facility friction, and re-entry envelope carry-forward.

Current status: Hypothesis only. The migration literature provides constraints on the elasticity parameters relevant to H1, but no model has assessed the full cost comparison at WDT’s specific parameters and institutional features. The break-even burden at each representative bracket has not been estimated.

4. H2: Macroeconomic Effects

4.1 Structure of the Decomposition

The proposition that WDT produces non-negative long-run macroeconomic welfare effects contains five causally distinct claims, each with an independent mechanism, each requiring a different type of model to test, and each independently falsifiable. Collapsing them into one hypothesis would obscure where the chain is most likely to break and would make falsification harder to achieve in practice.

H2a through H2d are therefore stated separately below. H2a concerns capital allocation. H2b concerns labour tax displacement. H2c concerns automatic stabilisation. H2d concerns whether the effects of H2a, H2b, and the consumer demand channel combine to produce a self-reinforcing positive loop. H2e — automation base robustness — is structurally different from the others and is stated as a standalone hypothesis in (FAL §7).

H2d is the only sub-hypothesis that requires the others to hold simultaneously. H2a, H2b, and H2c can each be tested and falsified independently.

4.2 H2a: Capital Allocation Efficiency

H2a. Removing lock-in and debt-preference distortions improves the allocation of capital across investments relative to a realisation-based system, holding revenue constant.

Mechanism. Realisation taxation creates a switching wedge between the gross return differential and the after-tax cost of redeploying capital to a superior asset. A taxpayer holding an appreciated asset faces a CGT liability on sale that does not arise if they hold. This distorts portfolio decisions away from productive redeployment and toward retention of existing positions regardless of their relative merit — the lock-in effect. The WDT eliminates this distortion by construction: tax liability accrues on the change in value whether or not the asset is sold, so the realisation decision is tax-neutral (WFR §4.2).

The welfare cost of lock-in under CGT is established in (WFR §4.2) and (WFR.A §C) as 141–143 basis points in the controlled single-agent comparison. This is a Category 1 result under (WFR §2.3): a cost of an existing operating system, derived without reference to WDT implementation data. The WDT eliminates it by construction, which is a Category 2 result: a property of the mechanism, independent of whether Phase One succeeds. H2a therefore has the strongest existing support of the five sub-hypotheses. The open question is whether the capital allocation improvement survives general equilibrium effects — specifically, whether WDT’s valuation friction and the behavioural responses it induces generate misallocation costs that offset the lock-in gains.

The primary theoretical challenge to H2a comes from Arachi & D’Antoni (2022), who argue that accrual taxation creates intertemporal consumption distortions in loss years that realisation taxation avoids. (WFR §4.2.3) addresses this directly, finding that the symmetric refund restores consumption capacity precisely at the moment the Arachi mechanism would otherwise be most acute, and that the objection applies at the envelope binding boundary — Poor-tier entrants in the first loss year — but not in the general case. H2a’s required model must include this channel and give it every opportunity to dominate.

The broader framework for H2a’s required model is established by Guvenen et al. (2023), whose “Use It or Lose It” model examines wealth taxation and capital allocation in a general equilibrium setting. That model addresses a stock wealth tax rather than a delta instrument; the required adaptation is to replace the stock base with a delta base and introduce the symmetric refund. Gerritsen et al. (2025) and Boadway & Spiritus (2025) provide the heterogeneous returns framework within which the capital allocation comparison should be assessed. Azevedo et al. (2025) contributes recent evidence on the lock-in mechanism in an entrepreneurship context that H2a’s model should also accommodate.

Required model. A portfolio-choice or asset-allocation model holding macro conditions fixed, introducing the delta base and symmetric refund in place of CGT, allowing endogenous portfolio reallocation, and measuring net productive investment change. The model must distinguish productive from unproductive investment and allow valuation friction as a cost on the WDT side.

Falsification condition. H2a is falsified if net capital misallocation under WDT exceeds that under the CGT counterfactual for empirically defensible portfolio elasticity parameters — specifically, if the misallocation cost of WDT valuation friction is larger than the 141–143bp lock-in welfare cost of CGT established by (WFR §4.2).

Current status: Partial theoretical support from WFR Category 1 results. The lock-in welfare cost of CGT is established; the general equilibrium capital allocation comparison has not been conducted for a delta instrument.

4.3 H2b: Labour Tax Displacement Welfare Effect

H2b. Replacing a portion of labour taxation with WDT revenue, through the constitutionally committed LRR accumulation, produces a net welfare gain relative to the pre-displacement position.

Mechanism. Labour taxation creates deadweight loss by driving a wedge between the private cost of labour supply and its social value. Employer and employee NICs together produce this wedge on both sides of the employment relationship simultaneously. Bilateral NICs removal — funded through LRR drawdown once the long-run reserve is sufficiently capitalised — reduces the cost of labour to supply and to hire at the same time, without requiring either party to absorb the full burden of displacement (ENV §4.2). The welfare gain depends on two conditions holding jointly: first, that the deadweight loss of NICs at the relevant margin exceeds the deadweight loss of WDT including valuation friction and behavioural responses; second, that the LRR accumulates fast enough to deliver meaningful displacement within a policy-relevant horizon.

(ENV §4) identifies bilateral NICs removal as the cleanest implementation case and (RATES §6.2) and (RATES §6.3) establishes that the LRR fills within a 7–29 year range across all 73 historical start years, with a median breakeven of 13 years. Whether 13 years constitutes a politically viable horizon is a question about political economy, not mechanism design.

The UK-specific deadweight loss of NICs is bounded by the literature. H. J. Kleven & Kreiner (2006) produce a marginal cost of public funds of 1.26 for income tax and NICs combined using participation elasticities, the most-cited UK estimate. The literature consistently produces MCPF values of 1.1–1.5 for direct labour taxes, rising when VAT and benefit taper interactions are included. The classical zero-capital-tax result of Chamley (1986) and Judd (1985) has been substantially overturned by Straub & Werning (2020), who establish that positive long-run capital taxation is optimal under realistic finite-horizon conditions — which directly supports the claim that displacing labour tax with WDT revenue does not reduce long-run welfare through capital formation. Bastani & Waldenström (2020) provides the most comprehensive recent survey of the comparative deadweight loss question. Neither the existing literature nor this paper estimates the deadweight loss of WDT versus NICs at revenue equivalence. That comparison is the core input H2b’s required model must produce.

Required model. A labour market or public finance welfare model comparing the deadweight loss of WDT at reference parameters — including valuation friction, liquidity costs, and behavioural responses — against the deadweight loss of the labour taxation it displaces. The LRR timeline should be a variable, not an assumption. The model should produce the deadweight loss comparison as a function of displacement pace, allowing the falsification condition to be assessed at each pace consistent with (ENV §4.1)’s staged displacement schedule.

Falsification condition. H2b is falsified if the deadweight loss of WDT at reference parameters exceeds the deadweight loss of the displaced labour taxation for the relevant population, or if the LRR timeline cannot deliver meaningful fiscal displacement within a politically viable horizon under any plausible Phase One calibration.

Current status: Hypothesis only. The deadweight loss literature provides bounds on the NICs side of the comparison; the WDT side has not been estimated.

4.4 H2c: Automatic Stabilisation

H2c. The symmetric refund mechanism creates a state-contingent fiscal position that dampens asset-market crises more effectively than existing automatic stabilisers, by reaching the leveraged and liquidation-prone population at the moment of maximum systemic risk.

Mechanism. Conventional income tax operates procyclically on the revenue side during recessions: tax receipts fall as incomes fall, and the government’s fiscal position deteriorates simultaneously. Under WDT, a crash produces large negative wealth deltas and triggers automatic refunds to precisely the population most likely to engage in forced asset selling — the mechanism that turns individual portfolio distress into systemic crisis. The refund is mechanical and constitutionally guaranteed rather than subject to discretionary fiscal policy (GOV §4), so it reaches the relevant population without the political and administrative delays that constrain discretionary stabilisation.

The SWF provides a second instrument. Its accumulated reserves are available for countercyclical deployment through the drawdown conditions set out in (GOV.B §E), and the DR chamber’s pre-commitment capacity provides a third instrument without requiring new political authorisation. Two of the three instruments are therefore automatic rather than discretionary.

The opposing effect is real and must be built into the required model. Large refund outflows deteriorate the government’s general fiscal position at the same time as they support household liquidity. The SWF drawdown provides a buffer, but whether that buffer is adequate depends on how much the SRR and LRR have accumulated before the crash. Whether the SWF can plausibly accumulate enough during normal years to absorb the refund outflow in a severe crash without triggering net contractionary effects elsewhere is not answered by the existing architecture papers.

The closest empirical analogue is Norway’s 2020 response to the COVID-19 and oil price collapse. The Government Pension Fund Global deployed a record $37bn drawdown — over four times the previous record — amounting to a fiscal impulse of approximately 5.1% of GDP (Norwegian Ministry of Finance (2020)). The IMF confirmed the GPFG served as a sizeable fiscal buffer operating within a flexible spending rule that permitted countercyclical deviation (International Monetary Fund (2021)). The WDT’s SWF is architecturally similar in its constitutional ring-fencing and countercyclical mandate, though it would operate at an earlier stage of capitalisation in Phase One. Norway’s experience establishes the mechanism’s empirical plausibility without answering the capitalisation-adequacy question for the WDT context.

Required model. A financial stability or crisis dynamics model allowing simultaneous operation of: refund-induced household liquidity injection targeted at the leveraged tail; government general fiscal position deterioration from refund outflows; and SWF drawdown with the capitalisation level as a variable. The model must include asset-market amplification — the forced-selling channel through which individual liquidity problems become systemic — and allow the SWF capitalisation at the moment of crisis to vary between the SRR floor and the LRR full capitalisation level established in (RATES §6.1) and (RATES §6.2).

Falsification condition. H2c is falsified if the combined fiscal effect of refund outflows and SWF drawdown during a crash scenario is net contractionary — either because the SWF is insufficiently capitalised at the moment of crisis, or because government fiscal deterioration offsets the household liquidity benefit across the relevant range of capitalisation scenarios.

Current status: Partial empirical support (analogical). The Norway 2020 GPFG drawdown establishes the SWF countercyclical mechanism’s empirical plausibility. No quantitative model of the WDT-specific stabilisation claim exists.

4.5 H2d: Positive Feedback Loop Integrity

H2d. The capital allocation efficiency gain from H2a, the labour tax displacement welfare gain from H2b, and the consumer demand expansion from the labour tax relief dividend combine to produce a compounding positive effect on the WDT tax base and broader fiscal capacity over time, conditional on the pace of labour tax displacement remaining consistent with maintaining inflation at or near the 2% central bank target.

The positive chain. WDT removes lock-in distortions (H2a), improving capital allocation and raising productivity. Higher productivity raises private returns, expanding the WDT base and increasing fiscal capacity. That fiscal capacity funds labour tax displacement (H2b), raising disposable income for lower and middle earners and expanding consumer spending. Higher consumer demand raises corporate revenues and asset valuations, expanding the WDT base further. The loop closes on itself.

A second amplifier operates through the taxpayer population. Wealthy taxpayers holding demand-correlated assets benefit directly from the consumer demand expansion through higher valuations. The labour tax relief dividend therefore raises the expected value of remaining in the WDT system for the taxpayers who fund it (WFR §4.3). This connects H2d to H3: taxpayers who understand that the positive loop increases the value of their institutional stake have a stronger private incentive to protect the cooperative equilibrium.

The negative chain. WDT reduces after-tax expected returns for the taxable population. Lower expected returns reduce capital formation, investment, and productivity, shrinking the WDT base over time. This chain dominates if the taxable population is sufficiently mobile, avoidance opportunities are sufficiently attractive, or the productivity response to H2a is too small to offset the direct return reduction.

The modeller’s task is to determine which chain dominates, and at what parameter values the positive loop fails to compensate for the negative one. The relevant parameters are capital-flight elasticity, avoidance rate, labour displacement pace, and the productivity response to improved capital allocation. None of these is determined by the design; all are empirical questions.

The inflation condition. The consumer demand channel carries an inflationary risk. If labour tax displacement proceeds faster than the economy’s capacity to absorb the demand stimulus — through productivity growth and supply-side expansion — the result is persistent inflation rather than real output growth. The condition is not that inflation be zero; it is that the displacement pace remain consistent with the central bank’s 2% inflation target under the UK reference framework. The constitutionally-committed LRR pacing in (ENV §4.1) is designed to keep displacement below this bound, but whether any constitutionally-paced displacement schedule achieves this under realistic demand parameters is what H2d’s required model must assess.

The inflation condition is not a background assumption. It is a constraint on the pace at which H2b can operate within H2d, and the model should characterise the displacement pace at which the constraint binds. If that pace is too slow to deliver meaningful fiscal substitution within a politically viable horizon, H2d fails on the inflation condition even if the positive loop otherwise holds.

Required model. A DSGE or heterogeneous-agent general equilibrium model containing WDT, symmetric refunds, SWF, labour tax replacement, capital allocation, heterogeneous returns, investment, consumption, migration, and government expenditure. The positive and negative chains must operate simultaneously. The consumer demand channel must be explicit. The 2% central bank target must serve as the inflation reference. LRR accumulation pace, capital-flight elasticity, and avoidance rate must be inputs rather than assumptions. The primary output is not a verdict but the parameter ranges at which the positive loop fails to dominate the negative one — these are the Governing Council’s calibration targets and Phase One’s primary empirical claims (SWEEPS §3) and (SWEEPS §4).

This is the most technically demanding model in the series. Guvenen et al. (2023) provides the closest existing framework; Fagereng et al. (2020) provides the heterogeneous returns calibration; Straub & Werning (2020) provides the theoretical foundation for the claim that WDT does not collapse long-run capital formation. None of these addresses the full positive-loop structure H2d requires.

Falsification condition. H2d is falsified if the negative loop dominates for empirically defensible parameter ranges, or if constitutionally-paced LRR accumulation produces demand stimulus that persistently breaches the 2% central bank target under realistic behavioural parameters, or if the pace of displacement required to remain within the 2% bound is insufficient to deliver meaningful fiscal substitution within a politically viable horizon.

Current status: Hypothesis only. (ENV §9.1) explicitly assigns the general equilibrium model to future work. (WFR §7.2) names general equilibrium as a limitation of the welfare comparison. No quantitative model of the positive loop exists.

5. H3: Cooperative Institutional Equilibrium

5.1 Statement

H3. A WDT jurisdiction can reach and maintain a stable cooperative equilibrium in which wealthy taxpayers have sufficient long-term incentives to preserve the credibility of the WDT/SWF institution rather than collectively attempting to undermine, capture, or exit it.

5.2 What the Hypothesis Is Not

H3 does not claim that wealthy taxpayers will behave altruistically, or that they will comply because they believe taxation is just. It claims something more limited: that the repeated-game value of preserving a functioning WDT institution can exceed the short-term gain from individually exploiting or capturing it.

The taxpayer’s question is not “do I like paying taxes?” It is “what institutional arrangement maximises the expected net value of my wealth over a 20–30 year horizon?” WDT offers lower compliance friction, symmetric downside protection, formal governance representation through the TP Chamber (GOV §3), a contingent claim on SWF returns, and a more predictable fiscal regime. Whether these benefits are worth more than the expected gain from defection, capture, or exit is a question about private incentives under a specific institutional structure — the kind of question a hostile game theorist is well positioned to answer.

H3 is also structurally linked to H1. If H1 holds — if wealthy taxpayers prefer WDT on total cost grounds — then the institutional stake they acquire by participating gives them a positive incentive to protect the equilibrium. If H1 fails, the cooperative stake logic weakens substantially. The two hypotheses share a common empirical foundation and a common vulnerability.

5.3 The Repeated-Game Structure

The equilibrium H3 requires is not a one-shot compliance decision. It is a repeated interaction in which each taxpayer’s decision to cooperate, defect, capture, or exit is conditioned on what they expect others to do, what they expect the institution to be worth in future periods, and what they expect the consequences of defection to be for the institution they depend on.

The tension operates across three levels simultaneously.

At the individual level, the incentive to defect is real: a single taxpayer who successfully manipulates the TP Chamber to reduce their own rate, or who extracts a large refund through strategic declaration, captures a private gain at the institution’s expense. This is the standard free-rider problem, and the standard analysis says it should dominate.

At the collective level, the free-rider analysis breaks down if the taxable population is small and sophisticated enough to understand the systemic consequence of generalised defection: if everyone manipulates the institution, it loses credibility, the refund guarantee becomes worthless, and the regime collapses into the adversarial dynamic that existing wealth taxes have historically produced (POL §3). The taxable population under WDT Phase One is small — the P99.9–P99.99 bracket is approximately 62,000 individuals — and the systemic consequence of generalised defection is visible to each of them.

At the long-term level, the cooperative incentive is strengthened by the accumulation of envelope depth. Each year a taxpayer remains in the system, their lifetime contribution envelope deepens and their accumulated refund entitlement grows. A taxpayer who has been in the system for fifteen years and experienced two significant loss-year refunds has a materially larger contingent claim on the institution’s future functioning than a new entrant. The cooperative stake compounds over time, which means the incentive to defect is highest at entry and lowest at maturity. Early Phase One is therefore the most acute vulnerability window.

The foundational analysis of cooperative equilibrium in multi-party resource governance is Ostrom (1990). The specific collective action problem — concentrated costs, diffuse benefits — is Olson (1965). The path-dependence mechanism by which early cooperative behaviour compounds into durable institutional norms is Pierson (2000). None of these addresses the specific structure H3 requires: a repeated game in which the governed population also holds governance stakes and contingent financial claims on the institution’s reserves. (DAgostino2026?) comes closest by making institutional credibility endogenous to compliance behaviour, but does not include taxpayer governance representation or contingent reserve claims. The required model is novel.

5.4 The Governance Architecture as Anti-Capture Design

H3 does not require the cooperative equilibrium to survive without structural support. The WDT governance architecture is designed to make capture costly.

The ten enumerated structural clauses (GOV §5.2) define the properties the system must preserve to remain the same kind of tax. The DR chamber — filled by monthly lottery from the general population at 50% vote share — provides a standing majority that cannot be captured by the taxable population alone; TP and FS combined hold only 50% and must persuade a substantial fraction of DR to pass any proposal (GOV §3). The anti-collusion guarantee is structural: DR’s unanimous opposition must independently be sufficient to defeat any joint TP/FS proposal. The rebalancing mechanism (GOV.B §D) imposes automatic vote share reduction on any chamber that misses votes, removing the option of strategic non-participation.

Three of the ten structural clauses rest partly on judgment rather than foundational axioms: the Route D auction mechanism, DR’s lottery constitution, and mandatory permanent public transparency (GOV §5.2). These are the clauses most vulnerable to a sustained political campaign for modification. The repeated-game model should test whether the structural safeguards are sufficient to prevent gradual erosion of these three — the failure mode Pierson (1994) identifies as “institutional brittleness” in the welfare state context.

The governance design is a constraint on capture, not a guarantee against it. The question is whether a small, concentrated, well-resourced taxable population can find paths around it — through regulatory capture of the Valuation Bodies, through strategic constituency dissolution of the DR chamber, through accumulated vote-share rebalancing that shifts power toward FS, or through political capture of the Administrator. The failure-mode taxonomy in (GOV.A §A) identifies three specific degeneration paths; the required model must test all three.

5.5 The Bootstrapping Phase

H3’s most acute vulnerability is not the mature equilibrium but the period before it is established. The cooperative equilibrium requires a credible institution with a demonstrated refund track record and a meaningful SWF reserve. Phase One is precisely the period when none of these exist.

The bootstrapping problem is primarily a political design choice rather than a modelling problem. Its length is determined by the Phase One entry threshold. A sufficiently high threshold compresses the bootstrapping window substantially. The SRR fills faster from a small high-wealth population contributing at high rates than from a large lower-wealth population contributing at low rates; in the limit a handful of participants at P99.9+ wealth levels can capitalise the SRR within one or two assessment cycles. The trade-off is a smaller initial institutional base and a longer path to Phase Two. The threshold is a political variable, set by whoever establishes the legislation; the bootstrapping vulnerability window shrinks as the threshold rises.

A further mechanism may partially close the window that remains. A taxpayer who has calculated that institutional survival is in their long-run interest has a rational private incentive to voluntarily prefund the SRR beyond their mechanical contribution. This is not altruism; it is the same logic that leads sovereign debt market-makers to support the credibility of institutions they depend on. Voluntary prefunding is not assumed to occur; it is a mechanism that can occur if the cooperative equilibrium logic holds strongly enough, and it gives the bootstrapping phase a partial self-correcting property. The taxpayers most exposed to institutional failure in Phase One have a private incentive to prevent it. The repeated-game model should include voluntary prefunding as an available action and characterise whether it emerges as an equilibrium behaviour, and if so under what conditions.

The two paths for Phase One are:

Path A. Threshold chosen to compress bootstrapping window → SRR capitalises rapidly → first refund demonstrated → envelope deepening → credibility accumulates → voluntary prefunding reinforces → cooperative norm strengthens into Phase Two.

Path B. Threshold too low, or voluntary prefunding insufficient → extended vulnerability window → exit or avoidance → weak SWF → lower credibility → institutional failure before the cooperative equilibrium is established.

H3 claims Path A is attainable. Whether it is attained depends on the threshold choice and on whether the cooperative equilibrium logic is strong enough to generate voluntary prefunding during the window that remains. Both are empirical questions the existing architecture papers cannot settle (POL §6, ADD §10).

5.6 Required Model

A repeated game or institutional model containing:

• individual and collective capture, including regulatory capture of executive bodies
• lobbying and political intervention capacity
• strategic exit and strategic compliance as endogenous choices
• institutional credibility as an endogenous variable, determined by refund track record and SWF capitalisation
• SWF returns and accumulated reserves, with capitalisation level varying across Phase One scenarios
• future tax liabilities and refund claims as elements of the cooperative stake
• voluntary prefunding as an available action — not an assumed behaviour
• Phase One entry threshold as a parameter controlling bootstrapping window length
• the DR lottery mechanism and constituency dissolution trigger as structural safeguards to be tested

The model must be given every opportunity to make cooperation collapse. If a stable cooperative equilibrium exists, the model should characterise the parameter range within which it holds. If it does not, that is a serious result against WDT and should be reported as such — not adjusted until it disappears. The model should also test whether voluntary prefunding emerges as an equilibrium behaviour, and at what cooperative stake level and under what conditions it breaks down.

The closest structural analogues are Ostrom (1990) on commons governance with stake-holding participants, Tsebelis (2002) on veto player architecture as a constraint on institutional erosion, and Levi (1988) on the political economy of revenue bargaining. The compliance psychology literature — Tyler (1990) on procedural justice, Gangl et al. (2015) on power-trust dynamics, H. Kleven et al. (2011) on third-party reporting — provides the behavioural foundations for the individual compliance decisions within the game. None of these provides the full required structure. The game-theoretic model needed for H3 is the most novel modelling challenge among all five hypotheses.

5.7 Falsification Conditions

H3 is falsified if the model, meeting the minimum requirements in (FAL §10), demonstrates that:

• individual capture is always privately optimal for a significant fraction of the taxable population regardless of institutional credibility or cooperative stake depth;
• cooperative behaviour cannot be sustained across any bootstrapping window length consistent with politically plausible Phase One threshold choices;
• wealthy taxpayers have stronger incentives to exit before institutional benefits materialise than to remain and accumulate cooperative stake, even at threshold levels that minimise the bootstrapping window;
• voluntary prefunding does not emerge as an equilibrium behaviour at any plausible cooperative stake level; or
• a small group can profitably destroy institutional credibility faster than it accumulates through voluntary prefunding and envelope deepening combined.

Current status: Hypothesis only. The governance architecture is designed to resist capture, but no formal model has tested whether it succeeds. The game-theoretic structure required — repeated game with taxpayer governance stakes, contingent reserve claims, and capture as an available strategy — has no direct antecedent in the literature.

6. H4: Accumulation-Point Efficiency

6.1 Statement

H4. Conditional on a comparable comprehensive economic base and appropriately specified redistribution, taxing changes in wealth produces lower aggregate distortion than taxing intermediate manifestations of the same economic process — particularly realised gains and existing wealth stocks.

6.2 The Mechanism

Standard tax systems do not tax wealth accumulation. They tax events that occur within the process of accumulation: a gain realised on sale, an income flow earned during a period, a stock of assets held at assessment date. The interventions occur at intermediate points:

accumulation → asset holding → realisation / transaction / income event → tax

Each intervention point creates a class of behavioural decisions that are tax-motivated rather than economically motivated: when to sell, what to hold, how to structure an asset, which legal form to use, when to realise income, how to time a bequest. The cumulative distortion from these decisions — measured as deadweight loss, misallocation, and avoidance expenditure — is the tax system’s compliance cost in the broadest sense.

WDT intervenes at the underlying change in economic capacity:

accumulation → wealth increase → tax

The tax point is the thing being taxed, not an event triggered by what the taxpayer does with it. The realisation decision, the holding decision, the structuring decision, and the timing decision are all tax-neutral under a pure delta base. The hypothesis is that this single change — moving the tax point to the underlying process rather than its intermediate manifestations — produces lower aggregate distortion than any system that retains intermediate-point interventions, holding revenue constant.

The intellectual lineage of this claim runs through the Haig-Simons comprehensive income tradition (Haig (1921), Simons (1938)) and the prior accrual proposals that reached the delta concept and stopped at valuation difficulty (LR.B §3). Shakow (1986) and Auerbach (1991) are the most direct prior proposals; both establish the distortion-reduction logic without implementing the self-balancing mechanism that makes WDT’s valuation architecture viable. The portfolio theory framework within which H4’s comparative claim should be assessed is Sandmo (1977).

The 141–143 basis point lock-in welfare cost established in (WFR §4.2) and (WFR.A §C) is a Category 1 result: a cost of an existing operating system, established without reference to WDT implementation data. It directly supports H4 for the CGT comparison. H4 extends this to a general claim across all six tax systems considered in (WFR §3) and (WFR §4).

H4 and H2a are related but distinct. H2a is the applied capital allocation instantiation of H4’s theoretical claim: it tests whether removing lock-in improves capital allocation in practice. H4 is the broader comparative proposition: whether the delta base is less distortionary in aggregate than any system that taxes intermediate events. H4 can hold even if H2a produces only modest capital allocation gains, provided the sum of all distortion reductions exceeds the sum of all new costs introduced by accrual.

6.3 The Costs WDT Introduces

H4 is not established by the WFR welfare comparison alone. (WFR) operates under controlled conditions that suppress several costs WDT introduces. H4 requires the comparative claim to survive when those costs are included.

The costs that must be modelled against H4’s distortion reduction are: valuation friction — the administrative and professional cost of operating a four-route valuation architecture across the full taxpayer population; liquidity constraints — the cash-flow cost of a tax liability that accrues on paper gains before they are realised; migration and restructuring — the avoidance responses that reduce the effective base and shift costs onto the remaining population; and the enforcement cost of operating the Route D auction mechanism as a deterrent.

(WFR §7.3) names implementation cost quantification as a named limitation — the modelling gap closest to what H4’s hostile comparison requires. Arachi & D’Antoni (2022) provides the most direct challenge on the liquidity side; (WFR §4.2.3) addresses it directly. Guvenen et al. (2023) provides the GE framework within which valuation friction and avoidance responses should be modelled. Gerritsen et al. (2025) and Boadway & Spiritus (2025) provide the heterogeneous returns framework that determines whether the distortion comparison changes under return heterogeneity. Dalle Luche et al. (2026) is the most recent empirical entry into this dispute using Italian data.

The consumption tax tradition — Kaldor (1955), Bradford (1986) — makes the strongest alternative claim to H4: that taxing consumption rather than accumulation produces even less distortion than a delta base, because consumption taxation removes the intertemporal investment decision from the tax base entirely. Hebous et al. (2024) and European Commission (2026) represent the most recent policy-level assessments of this comparison. H4 does not claim to definitively resolve it; it claims that the delta base is less distortionary than the systems currently in operation.

6.4 What a Hostile Model Should Measure

For equivalent revenue, a comparative welfare model should report, across flat WDT, progressive WDT, CGT, income tax, stock wealth tax, and consumption tax — the six systems compared in (WFR §3) and (WFR §4):

• asset allocation distortion and lock-in magnitude
• valuation friction, administrative complexity, and compliance costs
• avoidance expenditure and tax-base elasticity
• liquidity costs and intertemporal consumption distortion
• migration responses and base erosion
• deadweight loss and welfare at comparable revenue

The (WFR) controlled comparison is the starting framework; it establishes the baseline under suppressed distortions. H4 asks whether the WDT advantage survives when the suppressed costs are restored. The model should vary the magnitude of each cost component systematically and report the parameter range within which H4 holds.

6.5 Falsification Conditions

H4 is falsified if a comparative welfare model, meeting the minimum requirements in (FAL §10), demonstrates that after accounting for valuation costs, liquidity constraints, migration, avoidance, administrative costs, and behavioural responses, WDT generates greater total distortion than the alternative systems across empirically defensible parameter ranges.

The falsification criterion is clean and does not require the model to show that WDT is catastrophically bad. It requires only that the reduction in realisation and intermediate-point distortion is offset by valuation friction, liquidity costs, and avoidance costs sufficient to leave total distortion higher than the counterfactual.

Current status: Partial theoretical support from WFR Category 1 results. The lock-in welfare cost of CGT is established as 141–143 basis points under controlled conditions. The general comparative claim including all WDT-specific costs has not been quantified.

7. H5: Automation Base Robustness

7.1 Statement

H5. The WDT tax base is structurally more resilient to automation-driven labour income decline than fiscal systems whose primary revenue base is labour income, because the delta base captures capital appreciation regardless of whether that appreciation was generated by human or automated production.

7.2 The Mechanism

As automation increases the productive value of capital relative to labour, most existing fiscal systems face a structural squeeze: labour income declines, labour tax revenues decline, and fiscal pressure to maintain welfare expenditure increases. The revenue base shrinks at the same time as the cost base grows.

WDT does not share this structure. Tax liability accrues on the change in net worth regardless of whether the underlying appreciation was produced by human labour, automated production, or compounding capital returns. A firm whose value grows because it deployed robotics rather than workers generates exactly the same WDT liability for its shareholders as a firm whose value grew through human capital investment. The automation of production is fiscally invisible to the mechanism; what matters is the change in wealth, not its source (ENV §4.3) and (ENV §4.5).

This differs categorically from robot taxes, which require identifying a taxable agent and a taxable event neither of which automation naturally produces; from capital income taxes, which require a realised income flow to trigger liability; and from stock wealth taxes, which tax the accumulated stock rather than the marginal change and therefore do not automatically scale with automation-driven capital appreciation.

H5 is therefore a comparative structural claim, not a macroeconomic prediction. It does not assert that WDT revenue will grow under automation. It asserts that WDT revenue will be more resilient to automation-driven labour income decline than systems whose primary base is labour — because the delta base tracks capital appreciation wherever it occurs, while labour-base systems track a component of production that automation structurally displaces.

7.3 The Dependency on H2d

H5 is not unconditional. Two dependencies on H2d must be acknowledged.

The first is direct. H5’s base resilience claim depends on automation-driven capital appreciation flowing through the WDT base without being captured at the upper tail through avoidance structures inaccessible to routine assessment. Ring et al. (2026) establishes that personal holding companies significantly reduce effective tax progressivity in Scandinavian systems; a similar dynamic under WDT — automation-concentrated capital flowing into Route D structures with sufficiently remote realisation horizons to defer settlement indefinitely — would undermine base capture without falsifying the mechanism design.

The second is structural. If automation destroys the consumer demand channel on which H2d’s positive feedback loop depends — by reducing labour income and household spending broadly enough to suppress corporate revenues and asset valuations — then H5’s base resilience and H2d’s positive loop weaken simultaneously. A hostile model that stresses both simultaneously would be more informative than testing each independently, and the automation scenarios within which H2d fails should be identified first, since those are the scenarios where H5’s resilience is most tested.

7.4 Required Model

An automation scenario model containing:

• multiple automation pace trajectories — gradual, rapid, concentrated in capital-intensive sectors
• heterogeneous asset concentration — tracking whether automation-driven appreciation flows to a broad wealth distribution or concentrates at the upper tail
• avoidance capacity as a variable — specifically, whether Route D holding period extension and corporate restructuring absorb automation-driven appreciation faster than the WDT base captures it
• the H2d consumer demand channel — operating in the same simulation so that the common vulnerability can be tested simultaneously
• a labour-tax revenue counterfactual — explicitly tracking whether WDT base growth exceeds labour-tax revenue decline across each automation scenario

Fagereng et al. (2020)’s heterogeneous returns calibration provides the baseline for the return distribution across wealth tiers; Jordà et al. (2019)’s long-run return series provides the historical context for capital return trajectories under structural economic shifts. Guvenen et al. (2023) provides the GE framework closest to the required model.

7.5 Falsification Conditions

H5 is falsified if an automation scenario model, meeting the minimum requirements in (FAL §10), demonstrates that:

• WDT revenue growth does not exceed labour-tax revenue decline across plausible automation pace and capital concentration scenarios;
• automation-driven capital appreciation concentrates in a population with sufficient avoidance capacity — through Route D deferral, corporate restructuring, or holding company structures — to prevent base capture tracking the appreciation;
• the avoidance capacity of the automated-capital-holding population grows faster under automation than the WDT base it is avoiding; or
• the destruction of the consumer demand channel under aggressive automation severs the H2d positive loop at the same time as it undermines H5’s resilience claim, leaving both hypotheses without supporting mechanism.

Current status: Hypothesis only. Open questions register item 12 (automation and tax-base migration) is assigned to MACRO and requires Phase One data before modelling can be calibrated (ENV §9.2, PHASE1 §6.2).

8. Interactions Between the Five Hypotheses

The five hypotheses are not fully independent. Several share common mechanisms, common supporting evidence, and common vulnerabilities. A researcher planning a modelling programme should understand these dependencies before choosing which hypothesis to attack first.

H1 and H3 share the cooperative stake logic. H1 argues that wealthy taxpayers prefer WDT on total economic cost grounds, partly because of the institutional participation component — governance representation, downside protection, predictability. H3 argues that the cooperative equilibrium is maintained because wealthy taxpayers have long-run incentives to protect the institution they depend on. If the cooperative equilibrium fails, the credibility of the refund guarantee and the substance of TP Chamber representation both deteriorate, and the institutional participation component of H1 shrinks toward zero. The two hypotheses must be assessed together; a model that tests H1 while holding H3’s equilibrium fixed is not testing the full claim.

H2d and H3 share the demand-correlated asset channel. The positive feedback loop in H2d generates higher asset valuations through the consumer demand channel, and wealthy taxpayers holding demand-correlated assets benefit from those higher valuations. Taxpayers who believe H2d will operate have a stronger private incentive to protect the institution that delivers it. A hostile model that assumes H2d fails should not then expect H3 to hold on the same cooperative stake logic.

H2d and H5 share the consumer demand vulnerability. H2d’s positive loop relies on labour tax relief raising disposable income and sustaining demand for the assets wealthy taxpayers hold. H5’s base resilience relies on automation-driven capital appreciation not being offset by a collapse in the demand for goods and services that corporate revenues depend on. Aggressive automation scenarios that destroy the labour income share broadly enough to suppress consumer demand constitute a simultaneous stress on both hypotheses. A model that tests H5 under aggressive automation should include the H2d consumer demand channel, so that the interaction is explicit rather than assumed away.

H2a is the most structurally independent. Capital allocation efficiency does not require H2b, H2c, H2d, H3, or H5 to hold. The lock-in distortion and its elimination by the delta base are mechanical properties of the tax instrument, testable in isolation with a portfolio-choice model at fixed macro conditions. H2a already has the strongest existing support of all five hypotheses through the WFR Category 1 results. A researcher wishing to maximise the value of a single modelling effort should start here: a hostile GE model of WDT capital allocation effects calibrated to the delta instrument would immediately establish whether the 141–143 basis point lock-in gain survives general equilibrium dynamics, and that result would constrain every subsequent hypothesis test.

H4 is the broadest and deepest claim. It does not depend on WDT being implemented, on Phase One succeeding, or on any of the other four hypotheses holding. It is a claim about where in the accumulation process taxation is least distortionary. If H4 is falsified — if a full comparative welfare model including all WDT-specific costs shows that the delta base generates more aggregate distortion than intermediate-point systems — the case for WDT is weakened at the level of mechanism design, not just implementation. If H4 holds, it provides the theoretical grounding that makes each of the other hypotheses worth testing.

9. What Counts as a Serious Result

This section must appear in the paper to prevent goalpost movement — the tendency, common in contested empirical literature, to accept only results that confirm the prior and dismiss adverse findings as “not quite right” on scope grounds.

Three outcome categories are defined here. Any modelling that engages with one of the five hypotheses should report its result using one of these categories. If the result does not fit cleanly into one, the paper should explain why.

9.1 Support

The hostile model finds the hypothesis holds across a substantial range of empirically defensible parameters, with the supporting parameter range characterised. The characterisation matters: “the hypothesis holds” without specifying the parameter range is not a finding. A finding tells the Governing Council and Phase One evaluators where the parameters need to sit.

9.2 Conditional Support

The hypothesis holds only within an identifiable parameter range, with the binding constraints explicitly stated. Examples: H1 holds if and only if migration elasticity is below a specified value; H2b holds if and only if the labour displacement pace remains within a specified range; H3 holds if and only if Phase One SRR capitalisation exceeds a specified threshold; H2d holds if and only if the constitutionally-paced LRR accumulation remains consistent with the 2% central bank inflation target under the relevant demand parameters.

Conditional support is a substantive result. It tells the Governing Council where the live parameters need to be calibrated and gives Phase One a set of empirical targets to monitor. A conditional support finding for H2d, for example, that identifies the capital-flight elasticity above which the negative loop dominates the positive one, is exactly the kind of input (SWEEPS §3), (SWEEPS §4) and (PHASE1 §5) are designed to use.

9.3 Falsification

The hypothesis fails across empirically defensible parameter ranges, or survives only under assumptions that are themselves implausible.

A model that can be made to work does not constitute support. Any hypothesis can be defended by choosing favourable parameters. The standard is that the hypothesis holds across a substantial and realistic range, not merely somewhere. A result that shows H3 holds if and only if the bootstrapping phase lasts zero days and every taxpayer is fully rational with infinite time horizons is not support; it is a near-falsification with a theoretical escape hatch.

10. Minimum Requirements for Hostile Modelling

This section specifies the minimum model requirements for each hypothesis. A model that does not meet these requirements may still be interesting, but it does not constitute a serious attempt at falsification and should not be reported as one. The requirements are set not to make falsification difficult but to ensure that a model that finds a hypothesis false has actually tested the hypothesis rather than a simplified substitute.

Hypothesis	Model type	Minimum requirements
H1	Heterogeneous taxpayer choice	Four brackets: P90–95 (£1.6m mean), P95–99 (£2.9m), P99–99.9 (£7.1m), P99.9–99.99 (£19.9m) per (RATES §5.1); current-system all-in friction cost modelled across all five components including compliance, avoidance, lock-in, and uncertainty; migration as endogenous choice; institutional participation value allowed to vary including to zero
H2a	Portfolio-choice / asset-allocation	Revenue-equivalent CGT baseline; endogenous portfolio reallocation; productive vs unproductive investment distinguished; valuation friction included as a WDT cost; Arachi & D’Antoni (2022) intertemporal consumption channel included
H2b	Labour market / public finance welfare	Bilateral NICs removal modelled; WDT deadweight loss including valuation friction and behavioural responses; LRR timeline as a variable not an assumption; H. J. Kleven & Kreiner (2006) MCPF of 1.26 as the NICs deadweight loss reference
H2c	Financial stability / crisis dynamics	Simultaneous household liquidity injection and government fiscal position deterioration; SWF capitalisation as a variable from SRR floor to LRR full capitalisation; asset-market forced-selling amplification mechanism included; Norway 2020 GPFG drawdown as empirical calibration reference
H2d	DSGE or heterogeneous-agent GE	Positive and negative loops operating simultaneously; consumer demand channel explicit; 2% central bank target as inflation reference; LRR accumulation pace as a variable; capital-flight elasticity and avoidance rate as inputs not assumptions; primary output is parameter ranges not a verdict
H3	Repeated game / institutional	Individual and collective capture; strategic exit and compliance as endogenous choices; institutional credibility endogenous to refund track record and SWF capitalisation; voluntary prefunding as an available action; Phase One entry threshold as a parameter; DR lottery and constituency dissolution mechanisms included as structural safeguards to be tested
H4	Comparative welfare	All six tax systems (flat WDT, progressive WDT, CGT, income tax, stock wealth tax, consumption tax) on a common revenue-equivalent basis per (WFR §3–§4); valuation friction, liquidity, migration, avoidance, and administrative costs all included; (WFR) controlled comparison as the baseline to be extended
H5	Automation scenario	Multiple automation pace and capital concentration scenarios; avoidance capacity as a variable including Route D deferral and corporate restructuring; H2d consumer demand channel included in the same simulation; labour-tax revenue counterfactual explicitly tracked

11. What Would Constitute Evidence Against WDT

The following results, if produced by a model meeting the minimum requirements in (FAL §10), would be serious evidence against WDT. They are listed here so they cannot later be dismissed as unimportant or off-scope.

• The current system’s all-in friction cost for the UK wealthy population is robustly below the WDT reference burden across the P95–P99 and higher brackets under empirically defensible compliance cost estimates. (Falsifies H1)
• The break-even WDT burden is below the current system’s all-in friction cost at any of the four representative wealth brackets. (Falsifies H1 at that bracket)
• The capital misallocation cost of WDT valuation friction exceeds the 141–143 basis point lock-in welfare cost of CGT across realistic portfolio elasticity ranges. (Falsifies H2a and weakens H4)
• The deadweight loss of WDT at reference parameters exceeds the deadweight loss of the NICs it displaces, or the LRR timeline cannot deliver meaningful fiscal displacement within a politically viable horizon at any Phase One threshold. (Falsifies H2b)
• SWF capitalisation cannot plausibly absorb refund outflows in a severe crash without producing net contractionary effects on the broader economy across the relevant range of capitalisation scenarios. (Falsifies H2c)
• The negative macroeconomic loop dominates the positive feedback loop for empirically defensible capital-flight and avoidance parameters, or constitutionally-paced LRR accumulation persistently breaches the 2% central bank target, or the displacement pace required to remain within that bound is insufficient to deliver meaningful fiscal substitution within a politically viable horizon. (Falsifies H2d)
• Individual capture of the Governing Council is always privately optimal for a significant fraction of the taxable population regardless of institutional credibility, or the bootstrapping phase cannot reach Path A under any plausible Phase One threshold choice. (Falsifies H3)
• After accounting for valuation, liquidity, migration, avoidance, and administrative costs, WDT generates greater total distortion than CGT or an income tax across realistic parameters. (Falsifies H4)
• Under plausible automation scenarios, WDT base growth does not exceed labour-tax revenue decline because automation-driven capital appreciation concentrates in a population with sufficient avoidance capacity to prevent base capture. (Falsifies H5)

12. Research Challenge

The WDT research programme has developed a theoretical and institutional architecture across 32 papers. It has established a mechanism, a valuation architecture, a governance structure, and a welfare comparison. It has closed three confirmed literature gaps. It has documented ten empirical questions that can only be resolved by a live system, six calibration parameters that await Governing Council decision, and five hypotheses — set out above — that require modelling the existing papers cannot provide.

The author does not have the computational or empirical resources to model all five adequately. That limitation is not an argument for WDT.

It defines a research opportunity.

Researchers who believe that WDT would produce capital flight that offsets the lock-in welfare gain, or that the positive feedback loop is dominated by the investment reduction effect, or that the Governing Council would be captured before the cooperative equilibrium is established, or that valuation friction makes the delta base more distortionary than CGT in practice, or that automation-driven capital appreciation would be absorbed by avoidance structures faster than the WDT base can capture it — all of these objections can now be formalised and tested against hypotheses stated with enough precision to be falsified. The minimum requirements in (FAL §10) are set to ensure that a test is a test.

The most valuable contribution to the WDT research programme at this stage would not be another paper defending the design. It would be a model demonstrating that one of these hypotheses fails under realistic assumptions. A conditional support finding — identifying the parameter range within which a hypothesis holds — would be almost as valuable, because it tells the Governing Council and Phase One evaluators what the binding constraints are and where measurement effort should concentrate.

The objective is not to protect WDT from falsification. It is to make falsification possible.

13. Conclusion

Five propositions underlie WDT’s core argument. They have been stated above with enough precision to be falsified, and the falsification conditions have been stated without softening.

The first concerns what wealthy taxpayers actually face when they compare WDT against the existing system. The comparison is not between tax bills; it is between total economic costs. At a revenue-weighted annual burden of 0.35%, the WDT reference figure is low relative to what can plausibly be estimated as the current system’s all-in friction cost for the relevant population. Whether H1 holds depends on that estimation being done properly — at each representative wealth bracket, across all five cost components, with migration treated as an endogenous choice at WDT’s specific parameters rather than imported from stock-base systems at materially higher effective rates. That estimation has not been done. The break-even burden at each bracket is the number that matters, and it has not been calculated.

The second concerns whether WDT’s macroeconomic mechanisms work as claimed. H2 has been decomposed into five sub-hypotheses because they have different causal structures, require different models, and can be independently falsified. H2a — the capital allocation efficiency claim — already has the strongest support: the 141–143 basis point lock-in welfare cost of CGT is a Category 1 result from (WFR), established without reference to WDT implementation data. H2b, H2c, and H2d have no quantitative foundation at all. H2d is the most demanding because it requires H2a, H2b, the consumer demand channel, and the inflation condition to hold simultaneously. The positive feedback loop either compounds or unravels depending on parameters — capital-flight elasticity, avoidance rate, displacement pace, productivity response — that are empirical questions no existing model has answered for the WDT instrument.

The third concerns whether a cooperative institutional equilibrium can be established and sustained. H3 is the most unusual of the five hypotheses. It is not a macroeconomic or welfare claim; it is a claim about the strategic behaviour of a small, concentrated, well-resourced population in a repeated game against an institution they also govern. The literature does not contain a model with this structure. The bootstrapping phase is H3’s most acute vulnerability, and its length is a political choice rather than a fixed parameter — a sufficiently high Phase One threshold compresses it substantially, and voluntary prefunding by taxpayers who understand the institution’s vulnerability may compress it further. Whether either mechanism is sufficient is exactly what a hostile game theorist should attempt to break.

The fourth concerns the comparative efficiency of the tax base. H4 is the broadest claim and the one with the deepest theoretical roots. It holds that moving the tax point to the underlying change in economic capacity produces less aggregate distortion than any system that retains intermediate-point interventions, once revenue equivalence is imposed. The WFR Category 1 results support the mechanism for CGT specifically. The general comparative claim, including all WDT-specific costs, has not been quantified.

The fifth concerns structural resilience to automation. H5 does not predict that WDT revenue will grow under automation. It predicts that it will be more robust to labour income decline than systems whose primary base is labour — because the delta base tracks capital appreciation wherever it occurs and regardless of how it was generated. The claim is structural and comparative. Its common vulnerability with H2d is that both depend on the consumer demand channel, and aggressive automation scenarios that destroy labour income broadly enough to suppress demand constitute a simultaneous stress on both hypotheses.

Two hypotheses have existing partial support: H2a and H4, both grounded in WFR’s Category 1 results on CGT lock-in. H2c has empirical support by analogy from Norway’s 2020 GPFG drawdown. H1, H2b, H2d, H3, and H5 are hypotheses in search of a test. H2d is the most demanding, requiring the most technically complex model and the largest number of simultaneously operating mechanisms. H3 is the most novel, requiring a game-theoretic structure that has no direct antecedent in the existing literature.

The falsification conditions are stated as stated. The minimum modelling requirements are set at levels that constitute a genuine test. The list of results that would count as evidence against WDT is meant to be taken seriously.

The objective is not to protect WDT from falsification. It is to make falsification possible.

A. Hypothesis Summary Table

The table below summarises each hypothesis, its core claim, its primary threat, its falsification condition, the model type required, and its current evidential status. It is intended for use by researchers selecting an entry point into the modelling programme.

#	Hypothesis	Core claim	Primary threat	Falsification condition	Model type	Current status
H1	Wealthy taxpayer preference	Total economic cost of WDT ≤ total economic cost of current system for the relevant population	Current-system friction costs below 0.35%; migration rational after departure cost accounting	Current-system all-in cost robustly below WDT reference burden at P95+ brackets	Heterogeneous taxpayer choice	Hypothesis only
H2a	Capital allocation efficiency	Removing lock-in and debt-preference distortions improves capital allocation, holding revenue constant	WDT valuation friction generates misallocation exceeding the 141–143bp lock-in gain	Net misallocation under WDT exceeds CGT counterfactual for defensible elasticity parameters	Portfolio-choice / asset-allocation	Partial theoretical support (WFR Category 1)
H2b	Labour tax displacement welfare	Replacing NICs with WDT revenue produces net welfare gain via bilateral removal	WDT deadweight loss exceeds NICs deadweight loss; LRR timeline too slow	WDT DWL > NICs DWL, or LRR cannot deliver displacement in a viable political horizon	Labour market / public finance welfare	Hypothesis only
H2c	Automatic stabilisation	Symmetric refund reaches leveraged population at moment of systemic risk, dampening crisis	SWF insufficiently capitalised; net fiscal effect contractionary	Combined refund outflow and SWF drawdown is net contractionary across capitalisation scenarios	Financial stability / crisis dynamics	Partial empirical support (analogical — Norway 2020)
H2d	Positive feedback loop	Capital allocation gain + labour displacement + consumer demand combine to compound WDT base	Negative investment loop dominates; inflation breach under LRR pacing	Negative loop dominates at defensible parameters, or LRR pacing breaches 2% target	DSGE / heterogeneous-agent GE	Hypothesis only
H3	Cooperative institutional equilibrium	Wealthy taxpayers have sufficient long-run incentives to preserve rather than capture the institution	Individual capture always privately optimal; bootstrapping phase fails	Capture optimal regardless of institutional credibility, or Path A unattainable at any threshold	Repeated game / institutional	Hypothesis only
H4	Accumulation-point efficiency	Taxing wealth change produces less aggregate distortion than taxing intermediate events, at revenue equivalence	Valuation friction + liquidity costs + avoidance exceed realisation distortion eliminated	WDT total distortion > alternative systems after all WDT-specific costs included	Comparative welfare	Partial theoretical support (WFR Category 1)
H5	Automation base robustness	WDT base more resilient to automation-driven labour income decline than labour-base fiscal systems	Automation-concentrated capital absorbed by avoidance structures faster than base captures it	WDT revenue growth ≤ labour-tax revenue decline across plausible automation scenarios	Automation scenario	Hypothesis only

References

Advani, A., Chamberlain, E., & Summers, A. (2020). A wealth tax for the UK: Final report. Wealth Tax Commission. https://www.wealthandpolicy.com/wp/WealthTaxFinalReport.pdf

Agrawal, D. R., Foremny, D., & Martínez-Toledano, C. (2025). Wealth tax mobility and tax coordination. American Economic Journal: Applied Economics, 17(1), 402–430. https://doi.org/10.1257/app.20220615

Arachi, G., & D’Antoni, M. (2022). Taxation of capital gains upon accrual: Is it really more efficient than realisation? Fiscal Studies, 43(1), 39–61. https://doi.org/10.1111/1475-5890.12255

Auerbach, A. J. (1991). Retrospective capital gains taxation. American Economic Review, 81(1), 167–178. https://www.nber.org/papers/w2792

Azevedo, E. M., Scheuer, F., Smetters, K., & Yang, M. (2025). Dilution vs. Risk taking: Capital gains taxes and entrepreneurship (Working Paper No. 34512). National Bureau of Economic Research. https://doi.org/10.3386/w34512

Bastani, S., & Waldenström, D. (2020). How should capital be taxed? Journal of Economic Surveys, 34(4), 812–846. https://doi.org/10.1111/joes.12380

Boadway, R., & Spiritus, K. (2025). Optimal taxation of normal and excess returns to risky assets. Scandinavian Journal of Economics, 127(2), 366–389. https://doi.org/10.1111/sjoe.12566

Bradford, D. F. (1986). Untangling the income tax. Harvard University Press.

Brülhart, M., Gruber, J., Krapf, M., & Schmidheiny, K. (2022). Behavioral responses to wealth taxes: Evidence from Switzerland. American Economic Journal: Economic Policy, 14(4), 111–150. https://doi.org/10.1257/pol.20200258

Chamley, C. (1986). Optimal taxation of capital income in general equilibrium with infinite lives. Econometrica, 54(3), 607–622. https://www.jstor.org/stable/1911310

Dalle Luche, M., Guzzardi, D., Palagi, E., Roventini, A., & Santoro, A. (2026). Tackling the regressivity at the top of the Italian tax system: An optimal taxation framework with heterogeneous returns to capital. Review of Income and Wealth. https://doi.org/10.1111/roiw.70086

European Commission. (2026). Wealth taxation, including net wealth, capital and exit taxes: A study for the European Commission. Publications Office of the European Union. https://op.europa.eu/en/publication-detail/-/publication/fd44e869-3870-11f1-be39-01aa75ed71a1/language-en

Fagereng, A., Guiso, L., Malacrino, D., & Pistaferri, L. (2020). Heterogeneity and persistence in returns to wealth. Econometrica, 88(1), 115–170. https://doi.org/10.3982/ECTA14835

Gangl, K., Hofmann, E., & Kirchler, E. (2015). Tax authorities’ interaction with taxpayers: A conception of compliance in social dilemmas by power and trust. New Ideas in Psychology, 37, 13–23. https://doi.org/10.1016/j.newideapsych.2014.12.001

Garbinti, B., Goupille-Lebret, J., Muñoz, M., Stantcheva, S., & Zucman, G. (2026). Tax design, information, and elasticities: Evidence from the French wealth tax (Working Paper No. 31333). National Bureau of Economic Research. https://doi.org/10.3386/w31333

Gerritsen, A., Jacobs, B., Spiritus, K., & Rusu, A. V. (2025). Optimal taxation of capital income with heterogeneous rates of return. The Economic Journal, 135(665), 180–211. https://doi.org/10.1093/ej/ueae083

Guvenen, F., Kambourov, G., Kuruscu, B., Ocampo, S., & Chen, D. (2023). Use it or lose it: Efficiency and redistributional effects of wealth taxation. Quarterly Journal of Economics, 138(2), 835–894. https://doi.org/10.1093/qje/qjac047

Haig, R. M. (1921). The concept of income: Economic and legal aspects. In R. M. Haig (Ed.), The federal income tax (pp. 1–28). Columbia University Press.

Hebous, S., Klemm, A., Michielse, G., & Osorio-Buitron, C. (2024). How to tax wealth (How to Note 2024/001). International Monetary Fund. https://www.elibrary.imf.org/view/journals/061/2024/001/061.2024.issue-001-en.xml

Iacono, R., & Smedsvik, B. (2024). Behavioral responses to wealth taxes: Evidence from a Norwegian reform (Working Paper No. 11335). CESifo. https://doi.org/10.2139/ssrn.4991833

International Monetary Fund. (2021). Norway: 2021 Article IV consultation — Press release; staff report; and staff statement (IMF Country Report 21/104). International Monetary Fund. https://www.imf.org/en/Publications/CR/Issues/2021/06/09/Norway-2021-Article-IV-Consultation-Press-Release-Staff-Report-and-Staff-Statement-460986

Jakobsen, K., Kleven, H., Kolsrud, J., Landais, C., & Muñoz, M. (2024). Taxing top wealth: Migration responses and their aggregate economic implications (Working Paper No. 32153). National Bureau of Economic Research. https://doi.org/10.3386/w32153

Jordà, Ò., Knoll, K., Kuvshinov, D., Schularick, M., & Taylor, A. M. (2019). The rate of return on everything, 1870–2015. Quarterly Journal of Economics, 134(3), 1225–1298. https://doi.org/10.1093/qje/qjy032

Judd, K. L. (1985). Redistributive taxation in a simple perfect foresight model. Journal of Public Economics, 28(1), 59–83. https://doi.org/10.1016/0047-2727(85)90020-9

Kaldor, N. (1955). An expenditure tax. George Allen & Unwin.

Kleven, H. J., & Kreiner, C. T. (2006). The marginal cost of public funds: Hours of work versus labor force participation. Journal of Public Economics, 90(10–11), 1955–1973. https://doi.org/10.1016/j.jpubeco.2006.03.006

Kleven, H., Knudsen, M. B., Kreiner, C. T., Pedersen, S., & Saez, E. (2011). Unwilling or unable to cheat? Evidence from a tax audit experiment in Denmark. Econometrica, 79(3), 651–692. https://doi.org/10.3982/ECTA9113

Kleven, H., Landais, C., Muñoz, M., & Stantcheva, S. (2024). Taxation and migration: Evidence and policy implications. Journal of Economic Perspectives, 38(2), 3–30. https://doi.org/10.1257/jep.38.2.3

Levi, M. (1988). Of rule and revenue. University of California Press.

National Audit Office. (2025). Collecting the right tax from wealthy individuals. National Audit Office. https://www.nao.org.uk/reports/collecting-the-right-tax-from-wealthy-individuals/

Norwegian Ministry of Finance. (2020). Revised national budget 2020. Norwegian Ministry of Finance. https://www.regjeringen.no/en/aktuelt/key-figures-in-the-revised-national-budget-2020/id2701786

Olson, M. (1965). The logic of collective action: Public goods and the theory of groups. Harvard University Press.

Ostrom, E. (1990). Governing the commons: The evolution of institutions for collective action. Cambridge University Press.

Pierson, P. (1994). Dismantling the welfare state? Reagan, Thatcher and the politics of retrenchment. Cambridge University Press.

Pierson, P. (2000). Increasing returns, path dependence, and the study of politics. American Political Science Review, 94(2), 251–267. https://doi.org/10.2307/2586011

Ring, M. A. K., Seim, D. G., & Zucman, G. (2026). Personal holding companies, tax progressivity, and inequality (Working Paper No. 35534). National Bureau of Economic Research. https://doi.org/10.3386/w35534

Sandmo, A. (1977). Portfolio theory, asset demand and taxation: Comparative statics with many assets. Review of Economic Studies, 44(2), 369–379. https://doi.org/10.2307/2297074

Shakow, D. J. (1986). Taxation without realization: A proposal for accrual taxation. University of Pennsylvania Law Review, 134(5), 1111–1205. https://doi.org/10.2307/3311978

Simons, H. C. (1938). Personal income taxation. University of Chicago Press.

Straub, L., & Werning, I. (2020). Positive long-run capital taxation: Chamley–Judd revisited. American Economic Review, 110(1), 86–119. https://doi.org/10.1257/aer.20150210

Tsebelis, G. (2002). Veto players: How political institutions work. Princeton University Press.

Tyler, T. R. (1990). Why people obey the law. Yale University Press.