The "Release Valve" Hypothesis: A Forensic Map of Wealth-Concentration Mechanics and the Design Space for Keeping Capital Circulating

1Key Findings

1. Billionaire wealth is paper, not cash. Net worth ≠ liquid money. Per Americans for Tax Fairness (January 2024, citing 2022 Federal Reserve Survey of Consumer Finances data), America's billionaires and centi-millionaires held $8.5 trillion in unrealized capital gains in 2022 — a figure that "jumped by more than half — or $3.2 trillion — just since the last Fed survey year of 2019." These ~64,000 households (under 0.05% of the population) hold 18% of all US unrealized gains and realized only $316 billion in 2022, "less than 4% of their $8.5 trillion." ITEP separately estimates ~43% of the wealth of families with over $30M is unrealized gains (~$17 trillion). ProPublica's signature finding: Musk, Bezos and Buffett's combined wealth rose $137 billion from 2014-2018 while their taxable AGI totaled $5.7 billion — 4% of the gain.

2. "Buy-borrow-die" is real but smaller than the slogan suggests. New research by Fox and Liscow (2025) and Yale's Budget Lab finds borrowing is only ~1-2% of the top 1%'s economic income; the dominant strategy is actually "buy, SAVE, die" — the rich earn large taxable income, consume less than they earn, and reinvest the surplus, while the truly untaxed mass is unrealized appreciation shielded by stepped-up basis at death. This matters for your project: the leak isn't mainly the borrowing trick, it's that appreciation is never a taxable event.

3. Bezos's argument (May 20, 2026 CNBC interview with Andrew Ross Sorkin) is the steelman to beat. He argues (a) for-profit companies do more social good than philanthropy ("If I do my job right, the value to society and civilization from my for-profit companies will be much, much larger than the good that I do with my charitable giving"), (b) the bottom half should pay zero income tax, (c) "buy-borrow-die" is a myth ("There's no truth to this 'buy, borrow, die' thing... I'm selling Amazon stock routinely"), (d) "You could double the taxes I pay, and it's not going to help that teacher in Queens," and (e) overspending not under-taxation is the fiscal problem. The counter: Bezos's lifetime giving is ~$4.7B (Forbes), roughly 2% of his ~$283B net worth, versus ex-wife MacKenzie Scott, who "has given away 46 percent of her fortune, summing $26 billion" and donated $7.2B in 2025 alone (Forbes 2025 list). Krugman calls Bezos's "most progressive tax system in the world" claim a "zombie" that ignores payroll and state/local taxes (~80% of Americans pay more in payroll tax than income tax).

4. Velocity and the demand mechanism are the strongest part of your thesis. Mian, Straub and Sufi ("The Saving Glut of the Rich," now in Journal of the European Economic Association vol. 22 no. 3, 2024) document that "Between 1963-1982 and 1983-2019, the top 1%'s saving rate rose from 43% to 54%, while the rest saw theirs fall from 14% to 7%," and that this saving "has not been accompanied by a rise in net domestic investment" — instead it financed dissaving by the government and the bottom 90%. Roughly 30% of the rise in net household debt owed by the bottom 90% (1982-2007) was financed by the top 1%. Their companion paper "Indebted Demand" (QJE vol. 136 no. 4, 2021) formalizes it: "large debt burdens lower aggregate demand, and thus the natural rate of interest," driven by "the simple yet under-appreciated observation that borrowers and savers differ in their marginal propensities to save." MPC evidence (Fagereng, Holm & Natvik, AEJ: Macroeconomics 13(4), 2021): "Low-liquidity winners of the smallest prizes (around US$1,500) are estimated to spend all within the year of winning. The corresponding estimate for high-liquidity winners of large prizes (US$8,300-150,000) is slightly below one-half," with an average MPC around 0.35.

5. The design space is mapped below — and the elective/incentivized quadrant is nearly empty.

2Details

Areas 1-2: Billionaire mechanics and the Bezos argument

A billionaire's wealth is overwhelmingly equity in companies they founded or hold — illiquid, volatile, and untaxed until sold. The "income" the tax system sees is a small fraction of economic income. The Fed's own August 2025 analysis confirms the macro consequence: "the rise in wealth during the past 15 years, much of which accrued to high-income households, did not translate into the same level of consumption as it would have if it had been more evenly distributed," with equity wealth accounting for almost all of the continued rise in concentration since 2009 and a measured MPC out of wealth of only ~3.5 cents per dollar.

The strongest version of Bezos's case rests on a real empirical finding: concentrated capital in the hands of proven allocators does fund a disproportionate share of breakthrough innovation. Venture capital is highly concentrated (Silicon Valley took 39% of US VC in 2019; the top three cities 60%) and there is causal evidence (Howell, Townsend et al., NBER) that VC stimulates innovation. The weakness, established by Mian-Straub-Sufi, is that at the margin today the rich's surplus is not flowing into productive net investment — it is financing debt. Straub's framing (Chicago Booth Review, 2021): productive lending (a road, a factory) "increases productive capacity," but "unproductive lending has an externality that needs to be corrected." That externality framing is your bridge to a Pigouvian design.

Area 3: Velocity of money — what's true and what's loose

US M2 velocity has fallen secularly. The careful causal story is NOT the simplistic "money sits still in a vault." It is: (1) high-MPC households (poor) spend nearly all income, recirculating it immediately; (2) low-MPC households (rich) save 43-54%; (3) when income shifts upward, aggregate consumption demand falls unless the savings are recycled into productive investment; (4) post-2000 they were recycled instead into the debt of everyone else, which services itself by further depressing the borrowers' future demand — Mian-Straub-Sufi's "indebted demand" / "debt trap." Keynesian and post-Keynesian economists (and now mainstream figures like Summers on "secular stagnation," Rachel-Summers on r*) accept that inequality lowers the natural rate of interest. This is the academically respectable core of your intuition.

Area 4: The coercive "release valve" toolkit (everything tried/proposed)

• Pigouvian taxes / carbon-pricing analogy. Pigou (1920): tax an activity by the social cost of its externality to align private and social incentives; cap-and-trade does it via a quantity cap and tradable permits. This is the closest existing template for your "price the clog" intuition. No one has seriously built a Pigouvian tax on capital immobility / hoarding per se — the closest is demurrage (below).
• Wealth taxes. Warren's 2% over $50M + surtax over $1B; Saez-Zucman estimate $3.6T+ over a decade (with a 15% evasion assumption); Smith-Zidar-Zwick estimate far less ($117B/yr vs S-Z's $300B/yr) — the spread reveals how contested behavioral/valuation assumptions are. Switzerland runs a functioning wealth tax; Norway and Spain retain them; most of Europe repealed theirs over evasion and capital-flight concerns.
• Demurrage / Gesell / Freigeld (most direct ancestor of your idea). Money that loses value if held, forcing circulation. Wörgl, Austria (1932-33): stamp scrip depreciating 1%/month circulated ~14x faster than the national shilling, cut local unemployment 16% while Austria's rose 19%, funded public works — then was shut down by the central bank after 13.5 months. The WIR Bank (Basel, 1934-present) is the durable descendant. Caveat: the Mises critique argues Wörgl's success may have come from the spending/public-works injection, not the demurrage feature specifically; US stamp-scrip copies failed partly by taxing transactions rather than holding.
• Negative interest rates are demurrage's modern macro cousin. Denmark, Sweden, Switzerland, ECB, Japan went negative post-2009. Brookings found limited cash hoarding in practice, but the zero-lower-bound (people switch to physical cash) caps how negative you can go — the structural obstacle to using it as a circulation-forcing tool.
• Sovereign wealth funds + citizen dividends. The Alaska Permanent Fund (est. 1976, ~$83 billion) paid every eligible resident a 2025 dividend of $1,000 (set by the Legislature in House Bill 53 — the lowest inflation-adjusted PFD in state history, down from 2024's $1,702), to more than 600,000 Alaskans. Norway's GPFG (~$1T+) funds services rather than cash dividends. The People's Policy Project's "Social Wealth Fund for America" proposes a collectively-owned fund paying a universal dividend. Key finding: among SWFs, Alaska is the only one that directly pays dividends — there's an entrenched "anti-dividend consensus" among fund managers.
• UBI and its funding. Yang's VAT-funded model; LVT-funded; resource/carbon-rent funded; data-dividend funded. The mechanical insight (Citizen Network proposal): a UBI paid into "use-it-or-lose-it" accounts that expire monthly is itself a demurrage-like circulation device.
• Land value tax / Georgism. Tax the unimproved value of land — "the least bad tax" (Milton Friedman) because land is fixed in supply so the tax causes zero deadweight loss; endorsed in principle by Friedman, Stiglitz, Krugman (for city finance). Denmark's grundskyld is a live example. Targets rentier wealth specifically.
• Patronage dividends / cooperatives. Co-ops return surplus to members by patronage (use), not equity ownership — Mondragon (74,000+ workers), grocery co-ops, employee-ownership trusts. A built-in, non-coercive circulation structure, but confined to the co-op sector.
• Chicago Plan / full-reserve banking. Fisher (1935), revived by Benes-Kumhof (IMF 2012): strip banks of money-creation, eliminate bank runs, cut public debt. Monetary-architecture reform — tangential to circulation but relevant to "who controls the money supply."

Area 5: The elective/incentivized angle — where the gap is

This is the quadrant you care about and it is conspicuously underbuilt. What exists:

• Giving Pledge — the flagship voluntary mechanism, and it has largely failed. The Institute for Policy Studies' "Giving Pledge at 15" (2025): original signers who are still billionaires got 283% richer (166% inflation-adjusted) since signing; only 8 of 22 deceased pledgers fulfilled their pledge; ~80% of gifts went to their own foundations/DAFs, not working charities. New signatories collapsed from 113 in the first five years to 4 in 2024; Thiel calls it a "fake Boomer club." Lesson: pure reputation incentives without a reward structure or enforcement don't move concentrated capital.
• Opportunity Zones — the flagship tax-incentive-for-deployment mechanism, and it's been captured. $100B+ deployed, but 93% went to metro areas, much was not additive (would have happened anyway), and it "structurally favors projects that generate high returns, rather than the greatest social impact" (luxury apartments over need). Scheduled for 2026 expiration; the Center for American Progress and Urban Institute call for retooling or sunset.
• Social impact bonds / pay-for-success, patient capital, impact investing — exist but small, with weak/voluntary impact measurement.

The structural lesson across all three: voluntary or lightly-incentivized capital release reliably gets gamed toward the highest private return unless (a) the reward is tied to a verified circulation/impact metric and (b) there's a default cost to inaction. Carbon markets work because the cap creates scarcity and the externality has a price; the elective part is how you comply, not whether. No existing instrument applies that exact architecture to capital concentration.

Area 6: The thermodynamic / econophysics framing

This directly addresses your "scientific formula" intuition — and it is a real, published field, but read it carefully because it cuts both ways.

• Dragulescu and Yakovenko (2000, European Physical Journal B) showed that if money is conserved in random pairwise exchange (like energy in particle collisions), the equilibrium distribution is the exponential Boltzmann-Gibbs law, P(m) ∝ e^(−m/T), where the "temperature" T = M/N is average money per agent. Entropy maximization drives the system there.
• Empirically, Yakovenko's two-class finding is robust across countries: the bottom ~97% fit the exponential ("thermal") distribution; the top ~3% follow a Pareto power law ("superthermal") that swells and collapses with asset markets. The Gini of the pure exponential portion is exactly 0.5.
• The crucial caveat for your project: money is NOT actually conserved — banks create it via lending (the Chicago Plan literature), and the time-reversal symmetry that produces Boltzmann-Gibbs breaks down with debt and proportional (percentage) returns, which generate the Pareto tail. So there's no single "optimal" distribution the system "wants" — rather, free exchange has an attractor (exponential + Pareto tail) that already embeds substantial inequality. Yakovenko/Semieniuk's later work even argues globalization drives the world toward maximum-entropy mixing (global Gini saturating ~0.5), which they link to stagnation.
• Implication: your intuition that there's a "natural" amount of circulation has a real analogue (velocity ~ 1/(money held)), but the physics says concentration is the spontaneous tendency, not a fixable one-time clog. That argues for a continuous, self-renewing circulation incentive (a "thermostat"), not a one-shot release.

Area 7: Canadian context

• Concentration: Statistics Canada — the top 1% (~292,560 people) generate $140B+ in annual market income; the top 0.1% (~29,260 people) $62.8B. The Parliamentary Budget Officer (PBO) maintains a high-net-worth family database precisely because survey data undercounts the top. In Q1 2025 the net worth of the bottom 20% of households was ~10% lower than in Q1 2022, while the top 20%'s wealth rose 12.5% over 2022-2025.
• Capital gains treatment — the live drama: Budget 2024 (Freeland) raised the inclusion rate from 50% to 66.67% on gains over $250K/individual and all corporate/trust gains; PBO costed it at ~$17.4B over five years. It was deferred (Jan 31, 2025), then cancelled outright by PM Mark Carney on March 21, 2025, explicitly to "catalyze investment across our communities and incentivize builders, innovators, and entrepreneurs." The Lifetime Capital Gains Exemption increase to $1.25M was kept. This is the central tell about the Carney fiscal frame: pro-investment, supply-side, explicitly rejecting a capital-gains hike, while pursuing ~15% departmental spending cuts.
• Wealth-tax proposals: NDP 2025 platform — 1% over $10M, 2% over $50M, 3% over $100M; PBO costed the wealth tax at ~$94.5B over five years (using a 35% behavioral wealth-reduction assumption). Greens: 1% over $20M. Carney Liberals rejected a wealth tax; favor a luxury tax and CRA enforcement (an extra $1B/yr to CRA). The Diefenbaker-era Carter Commission ("a buck is a buck is a buck") is the Canadian intellectual anchor for taxing capital like labor.
• A Canadian "release valve" project has a natural opening: Carney's pro-investment frame is hostile to wealth taxes but rhetorically receptive to incentivized capital deployment — which is exactly your elective/Pigouvian quadrant.

Area 8: Where the genuinely novel gaps are

1. **A Pigouvian price on capital immobility, not capital itself.** Every existing tool taxes the stock (wealth tax), the realization (capital gains), or the land. None prices the externality you've identified — capital sitting in low-velocity, low-MPC holdings — while leaving an elective escape hatch (deploy it productively and avoid the charge). This is demurrage generalized from cash to all concentrated capital, with a carbon-market compliance structure.
2. A verified-circulation credit market. Carbon markets created a tradable instrument out of an externality. No equivalent exists for "demonstrated recirculation of capital into high-MPC hands or productive net investment." A credit earned for verified deployment (wages, productive capex, local investment) that offsets a default holding charge would replicate the carbon-market architecture in the wealth domain.
3. Use-it-or-lose-it dividend accounts at national scale. The demurrage principle is proven (Wörgl) but was killed institutionally; combining it with a sovereign-wealth-funded citizen dividend paid into expiring accounts is technically trivial today (CBDC/digital rails) and untried at scale.
4. **Reputation/gamified giving with a reward rather than a pledge.** The Giving Pledge failed because it's pure reputation with no reward and no default cost. A structure that confers tradable status, returns, or matching for verified release — and imposes a quiet cost for hoarding — is the unbuilt version.

3Recommendations

1. Anchor the project on the demand mechanism, not a conservation metaphor. Lead with Mian-Straub-Sufi "indebted demand" and "saving glut of the rich" — this is peer-reviewed, mainstream-adjacent, and says exactly what you intuit (concentration depresses demand and the natural rate) without the vulnerable "money is conserved" claim. Drop or heavily qualify the literal thermodynamic conservation framing; use econophysics only as the "concentration is the spontaneous attractor, so you need a continuous thermostat" argument.
2. Stake your claim in the empty quadrant: elective-but-incentivized capital release on a Pigouvian/cap-and-trade template. Concretely prototype "a price on capital immobility with a deployment escape hatch." Benchmark against carbon markets for the compliance architecture and against Opportunity Zones for the failure modes to avoid (require verified additionality; tie the reward to a circulation metric, not to private return).
3. Use Canada as the testbed and frame it to survive the Carney filter. A wealth tax is dead under Carney; an incentivized deployment mechanism that rewards builders is rhetorically alive. Design the pitch as pro-investment ("capital that moves is rewarded; capital that clogs pays") rather than redistributive.
4. Build the impact-verification layer first. Every voluntary mechanism (Giving Pledge, OZ, impact bonds) failed on measurement and additionality. The defensible novel contribution is a credible, gaming-resistant metric for "this capital actually recirculated." Without it, you rebuild the Giving Pledge.

Thresholds that change the plan: If credible additionality/verification proves impossible, abandon the elective approach and fall back to the highest-efficiency coercive tool (land value tax, which has the strongest economic pedigree and least deadweight loss). If the political window for any capital-side instrument closes, pivot to the distribution side (sovereign-wealth-funded expiring-dividend accounts), which needs no behavioral cooperation from the wealthy.

4Caveats

• The "clog" metaphor is partly misleading. Money is created and destroyed by banks; it is not a conserved fluid. The real mechanism is differential MPC and the recycling of savings into debt rather than investment — interpret all "circulation" language through that lens.
• "Buy-borrow-die" is contested. Fox-Liscow and the Yale Budget Lab show borrowing is only 1-2% of top-1% income; the Tax Policy Center and a Washington Post editorial call it a "myth." Stepped-up basis and high saving ("buy-save-die") are the bigger leaks. Don't over-rely on the borrowing trick rhetorically.
• Wealth-tax revenue numbers are highly contested (Saez-Zucman vs Smith-Zidar-Zwick differ 2-3x on behavioral response and valuation). Present ranges, not point estimates.
• Wörgl is celebrated but disputed — the boom may owe more to the spending injection than the demurrage feature, and it ran only 13.5 months before suppression.
• Econophysics is a minority/heterodox program; mainstream economists dispute that physical analogies capture institutional reality. Use it as suggestive framing, not proof of an "optimal distribution."
• Several sources are advocacy outlets (IPS, Jacobin, NDP, World Socialist Web Site, Mises) — their facts are usable but their framing is directional; the most load-bearing ones are flagged.
• The Bezos interview details come from a May 20, 2026 CNBC interview reported across CNBC, Yahoo/Moneywise and others; some secondary recaps were used only for corroboration.
• The 43%→54% top-1% saving-rate figures are from the July 2025 revision of the Saving Glut paper; the earlier "3 percentage points of national income" framing is from the 2020-2021 NBER drafts. Both are the authors' own findings; cite the version you rely on.