The IPS-based theory of economics

Economics studies how agents, firms, and nations allocate scarce resources through markets and institutions. This essay shows that the major phenomena of economics — market equilibrium and collapse, the role of trust and calibration in price formation, the volatility of financial instruments, and the trade-off between globalisation and redundancy — are all readings on the Fractal Persistence Equation (FPE). There is no separate economic law running beneath the thermodynamics; there is one accounting identity, and economics names the ways collective nodes manage it at the resource-allocation layer.

Formal derivation: information_persisting_systems.md. Vocabulary: glossary.md. State-scale application: ips_geopolitics.md. Trustworthiness token: trustworthiness_token.md.


1 The equation at the economic layer

\[ \mathcal{R}^{(L)} = \Psi(\mathcal{R}^{(L+1)}) \cdot \left[\frac{P_{in}\,\eta(I)}{\omega\,\mathcal{E}_\Sigma\,(1 + \mathcal{D}_{KL} + \Gamma)}\right] \cdot \Phi(\mathcal{R}^{(L-1)}). \]

At the economic layer, the FPE terms substitute as follows:

Why a market or firm is an IPS

Before the substitutions, it is worth being explicit about why the equation applies to a market or a firm at all. Both satisfy the four clauses of Definition 2.1 in the formal paper, with the Markov blanket now drawn around an institution rather than a body.

  1. Boundedness. A market’s Markov blanket \(\partial\Sigma\) is its exchange rules, settlement rails, disclosure law, listing standards, and order book — the interface that separates “inside the market” (cleared positions, posted prices, member obligations) from the rest of the world. A firm’s blanket is its legal personality, employment contracts, and asset boundary. Conditional independence (2.1) holds: given the market’s observations (trades, filings) and actions (clearing, delisting), its internal state (the aggregate belief distribution, the order book state) is separable from external causes.
  2. Drive. A market is a non-equilibrium steady state: it imports free energy as net capital inflows, retained earnings, and the labour of participants, and exports entropy as executed trades, resolved positions, and discarded information (obsolete theses). Stop the fluxes and the market clears its last trade — the \(\tau_d\) bound of Theorem 5.1 is the settlement-and-winddown window. A firm with no payroll or revenue dissolves on the same bound.
  3. Internal model. A market carries an aggregate generative model \(q_\mu\) — the probability distribution over future cash flows and policy paths that prices encode. A firm carries a model in its forecasts, strategy, and capital-allocation rules. Behaviour (capital allocation, trading) is well-described as gradient descent on variational free energy (2.2); this is the formal content of “markets act on what participants expect.”
  4. Identity preservation. A market persists as the same institution (a recognisable exchange with continuous price history) over \(\tau \gg \tau_{\mathrm{relax}}\), where \(\tau_{\mathrm{relax}}\) is the price-update cycle of seconds to days. Participants come and go; the price series and the listing persist. This is what makes the market, not its current traders, the node.

Because the four clauses hold, the market or firm is governed by (4.4). There is no separate economic law running beneath the thermodynamics; economics is the set of substitutions that translate each FPE term into the vocabulary of allocation, price, and capital.

FPE term Economic reading
\(P_{in}\) Usable capital inflow, retained earnings reinvested at positive real return, net equity issuance absorbed by willing holders — not gross turnover
\(\eta(I)\) Total factor productivity, transactions completed without friction cost (no lawyers, no politics), specialisation gains
\(\omega\) Microstructure friction, bureaucratic cost, disclosure mandates, compliance overhead
\(\mathcal{E}_\Sigma\) Environmental volatility: macro shocks, interest-rate regime shifts, geopolitical supply disruption
\(\mathcal{D}_{KL}\) Delusion divergence: gap between aggregate internal model (prices, analyst forecasts, earnings guidance) and observable fundamentals that eventually resolve
\(\Gamma\) Deferred reckoning: rolled debt, pension gaps, regulatory capture, unrepaired scandal, intergenerational transfer that never settled
\(\Phi\) Substrate integrity: productive firms, trust stock, skilled labour, physical infrastructure, natural capital
\(\Psi\) Institutional shelter: rule of law, dollar settlement, central-bank backstop, treaty access, reserve-currency status

An economic system persists not because prices exist but because participants trust that prices aggregate calibrated beliefs — being wrong is costly, insiders cannot systematically harvest uninformed flow without consequence, and tomorrow’s settlement layer will honour today’s ledger entry. When that trust erodes, the system does not merely “slow down.” Its persistence ratio falls toward one and through it.

The mechanism is Theorem 5.1 at this scale: the system’s internal reservoir \(\mathcal{B}\) — the trust stock, capital base, and liquidity that let it absorb shocks — depletes at rate \(\dot{\mathcal B} \le -\delta P_{out}^{req}\) whenever \(P_{in}\eta < \omega\mathcal{E}_\Sigma(1+\mathcal{D}_{KL}+\Gamma)\). Trust erosion is not a mood; it is the denominator (\(\mathcal{D}_{KL}\) of miscalibrated prices, \(\Gamma\) of deferred reckonings, falling \(\Psi\) of weakened enforcement) growing until the numerator can no longer clear it. The “borrowed time” before a crash is the bounded quantity \(\mathcal{B}/(\delta P_{out}^{req})\) in (5.1), and crashes are phase transitions — first-passage events on a depleting reservoir — not smooth slowdowns.

The two operations at the economic layer

Almost every effective market repair — enforcement action, disclosure reform, bankruptcy restructuring, clearing-house redesign — reduces to:

  1. Lower \(\mathcal{D}_{KL}\) — align the aggregate price model with observable fundamentals. Make being wrong costly, make insiders’ edge priced.
  2. Lower \(\Gamma\) — settle, restructure, or write off deferred obligations. Close the unresolved file rather than rolling it.

The thermodynamic reason these two recur is Theorem 5.2: each nat of maintained gap between market \(Q\) and fundamental \(P\) divides the system’s lifetime budget by \(e\). A wrong aggregate model forces every participant to act on predictions that mismatch reality, generating surprise on every price-update cycle; by the Crooks/Jarzynski identity (3.9) each nat of surprise costs at least one nat of excess dissipation per cycle, paid as widened spreads, higher required returns, and capital withdrawal. The honest update (a correct disclosure, a restructured debt, an enforced fine) is a one-time Landauer erasure cost (3.4); suppression — rolling the bad debt, hiding the gap, letting insiders harvest — is that cost charged every cycle, compounded by the \(\Gamma\) it generates. The other denominator terms (\(\omega\mathcal{E}_\Sigma\)) are largely fixed by microstructure and the macro environment; \(P_{in}\) is bounded by capital flows. Only \(\mathcal{D}_{KL}\) and \(\Gamma\) are informational debts the system can actively service. This is why markets that enforce calibration and honest settlement persist, and markets that subsidise delusion and roll debt eventually crash.


2 Markets as information-persisting systems

A stock market — or any coordinated allocation market — is an IPS \(\Sigma^{(L)}\) at the organisational level. It has a Markov blanket (exchange rules, settlement rails, disclosure law, listing standards, order book), a drive (\(P_{in}\): net capital inflows, earnings, index inclusion), and an internal model (\(\mu\): the aggregate probability distribution over future cash flows and policy paths that prices encode).

The market is made of nested IPS

The market-level substrate integrity \(\Phi\) is not the number of tickers; it is the weighted fraction of constituent sub-IPS whose local \(\mathcal{R}_i \ge 1\):

Sub-IPS Primary FPE terms
Listed companies \(\mathcal{D}_{KL}\) between guidance and realised fundamentals; \(\Gamma\) as balance-sheet ossification
Institutional investors Low \(\mathcal{D}_{KL}\) required for fiduciary persistence; their exit raises market \(\Phi\) risk
Market makers Adverse-selection tax when informed flow rises; spreads widen as their \(\mathcal{D}_{KL}\) of uninformed flow grows
Regulators Enforcement capacity = shelter against insider extraction
Political layer External causes in the model — trade policy, sanctions, fiscal path

A market whose companies are deluded but nominally solvent looks healthy on an index print while \(\Phi\) rots underneath. This is the exact firm-mortality pattern: exponential company death rates consistent with a persistence ratio that drifts until it crosses one.

Currency and price as proxies for trustworthiness

Fiat money and equity prices are not trustworthiness tokens — they do not score forecasts against outcomes. They are high-\(\Psi\) shelter artifacts whose value persists while participants believe the settlement layer and price signal remain calibrated.

They proxy trustworthiness in one precise sense: a stock price is the present value of a stochastic cash-flow stream under a risk-adjusted measure \(Q\). In prediction-market language, \(Q\) is the aggregate belief distribution over future states. Holding cash or index exposure is holding a share of collective epistemic credit — the society’s accumulated success at pricing uncertainty.

When calibration fails, the same nominal balance is nominal shelter without trustworthiness: strength without calibration. Capital flight, widened spreads, and risk-premium re-rating are not separate phenomena — they are \(\mathcal{R}\) falling toward one expressed in finance vocabulary:

Finance observation FPE reading
Higher required return Participants demand more \(P_{in}\) per unit of exposed \(\mathcal{D}_{KL}\)
Bid-ask spread widening Market makers raise tax on adverse selection
Liquidity withdrawal Critical sub-IPS (market-makers, institutions) exit; \(\Phi\) drops
Volatility spike Epistemic uncertainty over \(Q\) rises
Crash \(\mathcal{R} < 1\) sustained; phase transition, not a single bad draw

Kelly optimality and KL honesty

The Kelly criterion maximises long-run wealth growth by sizing positions proportional to edge. The per-period gain for a binary outcome is:

\[\Delta \log W \propto \log q(y) - \log p_m(y),\]

where \(q\) is the participant’s private belief and \(p_m\) is the market’s aggregate probability. This is the same kernel as the trustworthiness token update and the Proof-of-Trust ledger.

Interpretation: a market that allows Kelly-optimal sizing against a calibrated \(p_m\) rewards lowering \(\mathcal{D}_{KL}\). A market where insiders size on private information without paying the log-score penalty exports \(\mathcal{D}_{KL}\) onto uninformed flow — an adverse-selection tax on market makers and passive investors.

Volatility is uncertainty of the probability

Finance treats volatility as the second moment of price changes. The IPS reading is sharper: volatility is uncertainty over the probability measure itself — how wide, how shifting, how internally inconsistent the implied beliefs are.

Three layers: - Aleatoric: inherent stochasticity of cash flows; contributes to \(\mathcal{E}_\Sigma\). - Epistemic: disagreement and revision as information arrives; shows up in implied vol, forecast dispersion, VIX. - Delusional: systematic gap between market prices and realisable outcomes; \(\mathcal{D}_{KL}\) itself.

Ordinary return volatility mixes all three. VIX is better read as a market-level estimate of epistemic uncertainty over future probabilities. Low volatility is not automatically health — a market can suppress epistemic volatility by suppressing information (thin trading, passive herding, enforcement vacuum) while \(\mathcal{D}_{KL}\) accumulates silently. Fragile, not stable.


3 Trust collapse and adverse selection

The crudest market failure is the adverse selection loop:

  1. Informed participants (insiders, politically connected funds) extract edge from uninformed flow.
  2. Market makers widen spreads → higher \(\omega\) tax on everyone else.
  3. Uninformed liquidity withdraws → \(\Phi\) of the liquidity subgraph drops.
  4. Price impact per unit trade rises → measured epistemic volatility rises.
  5. Institutional fiduciaries reduce participation → \(\Phi\) falls further.

This loop does not require a macro shock. It is a local \(\mathcal{R}_i < 1\) cascade on the trader–market-maker edge. Regulatory \(\Psi\) exists to keep insider \(\mathcal{D}_{KL}\) priced and punished so the loop does not start. The thermodynamic content is Theorem 5.2 applied to the market-maker sub-node: insider \(\mathcal{D}_{KL}\) exported onto uninformed flow is a maintained gap the market-maker must pay each cycle as adverse-selection dissipation (3.9). Each nat of unpriced insider advantage divides the market-maker’s lifetime budget by \(e\) until the node exits — which is exactly step 3 of the loop.

When enforcement weakens, the loop’s gain exceeds its damping. The market can still print record index levels — passive inflows and buybacks maintain \(P_{in}\) — while trustworthiness proxy decouples from persistence. Nominal all-time high, falling \(\mathcal{R}\).

Loss of trust is not a mood. It is rising \(\mathcal{D}_{KL}\) plus falling \(\Psi\) and \(\Phi\). A market that stops being an IPS that prices truth becomes a shelter artifact clearing nominal flow while the persistence ratio erodes — until the next shock proves that trust was the load-bearing variable all along.


4 Globalisation as a redundancy trade-off

Late-20th-century economic integration was not only “more trade.” It was a substrate-topology experiment: move critical supply paths from parallel (domestic fallback, high tariffs) toward series (long chains, just-in-time, single-source nodes) in exchange for higher coupling efficiency \(\eta\) and lower per-unit \(\omega\) in calm conditions.

Term Pre-globalisation Hyper-globalisation (≈1990–2019)
\(\Phi\) topology More parallel capacity on food, energy, industry More series paths; critical cut-sets abroad
Middle bracket Lower \(\eta\), higher local \(\omega\) High \(\eta\), scale gains, inventory stripped
\(\Psi\) Thinner global shelter Thick \(\Psi_{\text{eff}}\): WTO, dollar rails, US security umbrella
Tail risk Local shocks absorbed domestically Cut-set failure → discontinuous \(\Phi\) collapse

States and firms chose collectively to spend less on substrate redundancy and more on productivity, betting that global \(\Psi\) would stay high.

What failed in the 2020s stress test

Shocks mapped cleanly onto FPE channels:

Event Channel Mechanism
COVID PPE / pharma shortage \(\Phi\) substrate Single-region production was the critical cut-set; series graph snapped
Suez Canal blockage \(\Phi\) Maritime cut-set, not mean shipping health
Semiconductor shortage \(\Phi\) One fab geography dominated critical path
Russia–Ukraine energy shock \(\Psi\) + \(\Phi\) Shelter edge removed; gas substrate for EU industry
US–China decoupling \(\Psi\) Trade and tech shelter channels deliberately constrained

Fingerprint: supply shocks were discontinuous on critical edges, not smooth decline in average national GDP. This is the substrate composition law (Theorem 5.3 and Section 2.5 of the formal paper): \(\Phi\) is a composition operator over sub-IPS, and structurally critical parts enter as a bottleneck (\(\min\)), redundant parts as a softer pool. World-system persistence is gated by the weakest critical cut-set, not by the mean of national accounts — a single failed fab or blocked canal zeros a critical factor and, because \(\Phi\) is a multiplicative postfactor, the whole product falls regardless of how healthy the average sub-node is. This is the precise sense in which the late-20th-century bet “mispriced which nodes were critical”: it optimised the mean \(\mathcal{R}_i\) while leaving critical-cut-set \(\mathcal{R}_i\) exposed.

Repricing: nearshoring, friend-shoring, strategic stockpiles

Current policy is re-purchasing \(\Phi\)-redundancy — duplicate fabs, strategic reserves, regional trade blocs — which raises \(\omega\) and lowers calm-environment \(\eta\). This is the insurance premium that the late-20th-century bet avoided paying. The bet won on medium-term \(\mathcal{R}\) for decades; it mispriced which nodes were critical.

What this is not: a claim that autarky maximises \(\mathcal{R}\). Pure series autarky is also fragile. The optimum is a mixed graph with identified critical cut-sets. The goal is not maximum redundancy — it is minimum critical-path exposure to single-source failure for the variables the survival of the node depends upon.


5 Classical economic theories, FPE read

Every major school of economics is a partial reading of the FPE. The table below locates each school’s core claim inside the accounting identity.

School / theorist Core claim FPE location What IPS adds
Keynesian (Keynes, Hicks) Aggregate demand shortfall causes recessions; fiscal stimulus restores output \(P_{in}\) contraction; \(\Gamma\) (debt from prior period) blocks numerator recovery Explains why demand shortfalls are sticky: \(\Phi\) (trust stock) has fallen; \(\Gamma\) deferred from bubble phase compounds denominator; stimulus only works if \(\mathcal{D}_{KL}\) (false models from boom) is cleared
Austrian (Mises, Hayek) Capital structure and time preference determine long-run prosperity; forced credit expansion creates unsustainable booms \(\omega\) from mis-specified capital allocation; \(\Gamma\) from lending that cannot be repaid; \(\Phi\) sub-nodes built on false price signals IPS formalises “malinvestment” as \(\mathcal{D}_{KL}\) baked into capital stock: clearing costs are structural, not just demand-side
Efficient Market Hypothesis (Fama) Prices fully reflect available information; no systematic edge without private information Near-zero \(\mathcal{D}_{KL}\) between aggregate price model \(Q\) and fundamental \(P\) EMH holds in the limiting case of saturated honesty incentives; IPS explains when it fails — when insider \(\mathcal{D}_{KL}\) advantages are not priced, triggering the adverse-selection loop in §3
Financial instability (Minsky) Stability breeds instability: calm periods encourage leverage until a “Minsky moment” \(\Psi\) rises during calm → \(\mathcal{D}_{KL}\) accumulates (models diverge from risk-adjusted \(P\)) → sudden \(\Psi\) removal + \(\Phi\) collapse Minsky’s cycle is the signature of a system where \(\mathcal{D}_{KL}\) compounds silently (vol suppressed) until \(\mathcal{R}\) passes one discontinuously
Institutional economics (North, Acemoglu) Institutions determine incentives and long-run growth \(\Psi\) as the formal shelter; \(\omega\) as enforcement overhead; \(\Gamma\) as legacy rule-layers Formalises institutional quality as shelter coefficient product — each independent institution multiplies \(\Psi_{\text{eff}}\)
Modern Monetary Theory (Kelton, Wray) A currency-issuing state cannot involuntarily default; fiscal space is the real constraint State \(P_{in}\) ≠ limited by taxes alone; sovereign \(\Psi\) is its own shelter channel True in the short run (no involuntary \(\Psi\) failure for own-currency issuer); IPS adds the constraint: \(\mathcal{D}_{KL}\) between stated fiscal promise and realised value of currency is the real bill — paid by inflation and eventual trust collapse

Synthesising statement. Each school correctly identifies one channel in which the FPE denominator rises or the numerator falls. They disagree primarily because they hold the other channels fixed — the classic partial-equilibrium error. The FPE holds all channels simultaneously and shows that any single-channel intervention (stimulus, monetary rule, deregulation) can improve \(\mathcal{R}\) short-run while worsening a hidden channel long-run. The question is always the full accounting: where does the bill go?


6 Falsifiable economic predictions from IPS

  1. Critical-path monotonicity: global industrial stress should track the weakest critical cut-set, not mean sector health. When critical-path nodes (TSMC, Rotterdam port, Saudi Aramco capacity) are constrained, the effect should be discontinuous in a way that mean-field economic models cannot capture.

  2. Shelter removal scaling: losing one trade or alliance channel should degrade \(\mathcal{R}\) roughly as \(-\log\Psi_j\) on that channel (from the composition law), not as \(1/N\) of total shelter. The asymmetric effect of Russian gas removal on EU industry (vs. equivalent share of other energy) is a testable instance.

  3. Delusion cost is not metaphor: each additional nat of market \(\mathcal{D}_{KL}\) (earnings guidance systematically missing, insider advantage pricing out calibrated participants) incurs excess thermodynamic dissipation per update cycle. The accumulated toll shows in widened spreads, higher required returns, and eventual crash amplitude.

  4. Trust is the load-bearing variable: across market crashes, the timing of the phase transition (not the economic fundamentals) should correlate with measurable trust indicators — enforcement-action frequency, insider-trading exposure events, settlement failures — rather than with the first arrival of macro bad news.


7 Diagnostic questions (any market or economic system)

  1. \(\mathcal{D}_{KL}\): Where does the aggregate price signal diverge from observable fundamentals? Earnings guidance vs. realised? Official growth statistics vs. satellite data, electricity, freight?
  2. \(\Gamma\): What debt, pension, or regulatory obligation has been deferred? What scandal or structural failure has been left unresolved?
  3. \(\Phi\): Is the substrate of critical firms, skilled labour, physical infrastructure, and natural capital intact? Are sub-IPS (\(\mathcal{R}_i\)) passing or rotting under the index headline?
  4. \(\Psi\): Is the institutional shelter — rule of law, enforcement, settlement, reserve access — real or eroding? Is the enforcement layer funded and independent?
  5. Critical-path topology: Which supply chains are series rather than parallel? Where are single-source nodes on the critical path to continued operation?
  6. Trust gradient: Is the system rewarding calibration or narrative performance? Does being wrong cost anything?

See also