The Rise of Collateralized Attention Derivatives and the Monetization of Temporal Uncertainty

Every era in economic history hides a force so fundamental to its functioning that it remains invisible until the moment it fails.

In 2008, that hidden engine was mortgage-backed securities, arcane financial instruments most people didn’t know existed until they blew up the global economy.

In 1999, it was the dot-com bubble, a frenzy of valuations built on shitty business models and vaporware.

In 1971, it was the dollar’s sudden divorce from gold.

In 2025, it isn’t inflation, interest rates, or even artificial intelligence in the narrow sense.

Those are the usual scapegoats, explanatory variables to the untrained eye.

Convenient, but wrong.

The real fault line runs deeper.

The hidden destabilizer is the monetization of temporal uncertainty through what I call Collateralized Attention Derivatives (CADs)—synthetic financial instruments that package and trade the volatility of human time preference across advertising, labor, and capital markets.

CADs are basically structured bets on human behavior: bundles of predictions about when people will act, sliced into risk tranches, and sold across markets like bonds once were.

The thesis here is deceptively simple.

The most valuable economic resource today is no longer just what we buy, or even what we pay attention to.

It’s the predictive control over the timing of decisions—the ability to slice, bundle, and sell that control or the “when” as an asset.

That “when” has been turned into a financial product.

It’s now collateral. It trades.

Put in simple terms, the economy’s deepest driver is no longer goods or credit, it’s the financialization of human time itself.

And that shift is so profound it has moved the economy’s center of gravity away from producing output and toward allocating reality itself to the highest bidder of time-fragment prediction.

How Time Became Money

Classical economists, most notably those who follow the Austrian tradition, have long argued that time preference is the foundation of everything from savings rates to interest levels.

And that time preference (the degree to which individuals favor present consumption over future consumption) is an exogenous trait of the actor, stable and internal to each person.

Time preference may drift with age, wealth, or circumstance, but it’s not assumed to be directly manufactured by market participants.

This preference, in turn, shapes savings, investment horizons, and the economy’s intertemporal structure of production.

Note: While Austrian theory often treats time preference as given at the instant of choice (i.e., exogenous), economists like Gary Becker (in his 1997 paper) model it as endogenous, shaped by wealth, expected longevity/mortality risk, uncertainty, and deliberate investments in future-oriented (consumption) capital. In his framework, longer horizons and greater resources lower ρ (more patience), while higher mortality risk and uncertainty raise ρ; in short, ρ is responsive to conditions and investment, not purely innate. Similarly, Shoshana Zuboff’s concept of “surveillance capitalism” describes how tech firms extract behavioral surplus from data to predict and modify actions, commodifying attention in ways that echo CADs. What distinguishes the CAD economy is the real-time, algorithmic scale and instrumentability: Platforms don’t just influence preferences slowly via culture—they engineer micro-volatility in ρ in real time and slice that volatility into tradeable risk, turning timing into collateral.

If people can wait, they save and invest, stretching production across time.

If people can’t wait, they consume immediately and capital structures shorten.

Historically, this preference was treated as fixed and internal, shaped slowly by culture, age, or circumstance.

But in 2025 and heading into 2026, that assumption no longer holds up.

These days, time preference isn’t stable at all.

It’s continuously measured, manipulated, and monetized.

Companies track it in real time.

Every scroll on a feed, every delay before clicking “buy,” every subscription renewal or cancellation, and every hesitation in completing a task feeds into live models that estimate both the level of your time preference and its volatility.

They know not just what you want, but how long you’ll wait before clicking “buy.”

Algorithms can—and do—alter these parameters in real time, adjusting your patience or impatience like a dimmer switch, sometimes lengthening your horizon (building anticipation for a future launch) and sometimes collapsing it to the moment, triggering impulsive buying.

The volatility in your time preference—the uncertainty in when you’ll act—has become the new raw material of finance.

The critical innovation is that this volatility (the unpredictability in how your preferences will shift) is now an asset.

Like the way mortgage repayment risk was bundled into tranches and sold as CDOs, behavioral volatility is packaged into instruments whose value depends on the probability of your future actions.

These instruments, CADs, are now bought, sold, and used as the basis for financing decisions across multiple sectors.

In many cases this is functional, not formal, securitization: cash flows are effectively priced on time-windowed engagement hazards and cohort retention curves—even when nothing is packaged as a registered security.

This is why ad-tech and subscription lenders often key covenants to retention and engagement cohorts: the collateral is behavioral timing, not hard assets.

That said, it can be easy to get caught up in low-level quant work: engagement rates, churn tables, etc.—but that’s not what matters here; most analytics teams already do that stuff.

What truly matters, and what requires real study, is that platforms and financial actors now treat the variance of human behavior itself as collateral.

So what we actually see happening is that time-preference volatility itself is being securitized: hesitation, delay, unpredictability—it can all be bundled into tranches and sold like credit risk once was.

That transformation, from measurement to assetization, is what makes this such a disruptive economic engine.

The Machinery of Collateralized Attention

What I’m calling a collateralized attention derivative (CAD) is, in essence, a forward bet on the probability that a specific behavioral event will occur within a defined time frame.

The raw material is not a mortgage payment or a shipment of commodities but the likelihood of your engagement, your purchase, your labor availability, or your content interaction.

These future behaviors are pooled into cohorts based on how predictable or volatile they are (steady, erratic, or chaotic) and then sliced into tranches, just like mortgage pools were sliced into AAA, BBB, and junk.

At the top are low-volatility “AAA” attention bonds: highly predictable customers who renew subscriptions like clockwork or workers who log in for their shift every Friday without fail.

Lower tranches might include users with more erratic behavior patterns—those who binge one week and disappear the next, or gig workers who respond unpredictably to incentives.

The riskiest tranches, the “equity” layer, contain highly volatile actors whose responses are difficult to predict but who may produce outsized returns if captured at the right moment.

The behavioral data (clicks, churn, availability, etc.) once pooled, is sliced into the following tranches:

• AAA tranches: stable users, highly predictable, loyal subscribers or workers who always renew and are “safe”
• BBB tranches: wobbly users, moderately predictable, sometimes yes, sometimes no
• Equity tranches: volatile, chaotic, high-risk high-reward, can occasionally generate outsized returns

These bundles are then sold to advertisers, lenders, and platforms just like bonds or CDOs once were.

Advertisers buy them to target the right people at the right time.

Banks lend against them, taking engagement forecasts as collateral.

Platforms themselves hedge against volatility by trading on them.

The path from your screen to Wall Street isn’t metaphorical—it looks almost exactly like structured finance.

Platforms capture behavioral traces: clicks, views, watch times, hesitation, churn rates, subscription renewals.

Forecasting models convert these into probabilities of future actions.

In advertising, these instruments manifest as programmatic ad inventory sold not merely on demographic or contextual targeting but on the probability of engagement in precise temporal windows.

In labor markets, they show up as availability guarantees for on-demand workers, backed by the platform’s ability to nudge supply into existence.

In finance, subscription retention curves and engagement forecasts serve as collateral for loans and valuations, allowing companies to raise capital not against current revenue, but against the predicted future behavior of their user base.

The Rise of Control Capital

Traditionally, economic theory has treated capital as the stock of physical tools, machines, and factories used to produce goods—what we can call production capital.

To capture the CAD economy’s shift, however, we must introduce a new variable, control capital (K_c): a distinct, financed, depreciating stock of assets—datasets, algorithms, recommendation engines, user interface surfaces, notification rights, and identity graphs—that does not produce goods directly but shapes preferences and regulates the timing of consumption.

Like production capital, K_c is accumulated and yields returns, but its value derives from altering demand distributions in time, making preferences endogenous and capital-using.

This extends Austrian capital theory, with its emphasis on time preference and the intertemporal structure of production, where ρ was exogenous.

In a CAD economy, ρ now becomes a stochastic process influenced by K_c deployment.

Building on this foundation, the following sections quantify how K_c props up macro aggregates like GDP while introducing hidden fragility—resolving the apparent paradox of stable growth amid manipulated demand.

Adding this second category extends the model, placing control capital alongside production capital as a co-equal driver of demand.

Note: To address potential counterarguments, if preferences have always been endogenous (per Becker), why claim a paradigm shift? The difference lies in velocity and monetization—digital control capital operates in seconds, not generations, with US digital ad markets at $317-350B in 2025 (~1.2% GDP but influencing 10-30% retail demand via targeting elasticity). Concentration persists at ~50-55% via Google/Meta in the US—enough for correlated shocks until Web3 fragments it.

Production capital (factories, equipment, skilled labor) still makes things: cars, chips, food, software.

Control capital exists to shape and steer the preferences that determine whether, when, and how those goods are consumed.

Both are funded.

Both generate returns.

But only one—control capital—decides when production capital’s goods get consumed.

Our updated model now shows the two kinds of capital: production (machines, factories, skills) and control (data, algorithms, notifications).

Both feed into what looks like “demand.”

What people actually buy, however, is shaped by both what exists (supply) and how preferences are manipulated (nudges).

Now, in our updated Neuro-Austrian framework, time preference (𝜌) is no longer a constant; it is a stochastic process.

Its drift and volatility can be influenced by the deployment of control capital.

The interest rate, rather than being solely a reflection of real savings and productivity, now also incorporates a volatility premium tied to the instability of time preference.

Equivalently, the observed discount rate can be decomposed as r = r_real +ϕ⋅E[Δρ]+ν⋅Var(ρ) over the relevant horizon.

Profits can accrue not just from producing better goods, but from tightening the predictive net around consumer or worker behavior, extracting “control rents” from the increased certainty of future actions.

Note: I introduce the term ‘control rents’ as a novel concept in this framework. Control rents are the excess profits firms extract, not from making a better product or cutting real costs, but from owning and using control capital to make human time preferences and customer actions more predictable and steerable. Unlike traditional rents from scarce land or resources, these arise from reducing uncertainty in when people act (lowering σ for steadier cash flows) or shifting impatience levels (moving θ to trigger buys/renewals at profitable times). Practically, control rents arise when a company can (i) shift the drift of time preference (push purchases/renewals forward or defer churn) and/or (ii) reduce the volatility of when people act, which raises conversion at a given price and lowers cash-flow timing risk, improving present value and financing terms.

For example: Imagine a big tech platform dumps $1 billion into smarter algorithms to cut the unpredictability in how people decide when to buy or act, basically smoothing out their impatience swings by half. This bumps up the company’s future earnings value today by 10-15% through math models that predict purchase timing. That added profit slice is the “control rent”—it’s extra cash from locking in behavior, baked into the company’s stock price but disguised in GDP stats as real economic growth. But if ad rivals ramp up and eat into those gains (like when Apple’s 2021 privacy update tanked ad returns 20-40%), the rents dry up fast—and many reports are showing US digital profit margins shrinking 5-10% a year.

This equation​ treats time preference ρt as a controlled mean-reverting diffusion.The stock of control capital K_c parametrically shifts two primitives:

1. the long-run target level of time preference θ(K_c (drift), and

2. the residual instability of time preference σ(K_c) (diffusive volatility).

Absent control (K_c at a baseline), ρ_t​ reverts to its natural target at speed κ.

With control, the economy can steer the level of impatience/patience (via θ) and reshape its uncertainty (via σ). In continuous time this delivers a stationary OU-type law with:

• Mean E[ρ_t] → θ(K_c)

• Variance Var⁡(ρ_t) → σ²(K_c)/(2κ)

• Half-life of shocks ln⁡(2)/κ

Operationally, purchase/renewal timing can be written as a hazard λt = f(ρt,X_t); where X_t collects observables like price/discounts, availability/friction, campaign intensity, and seasonality; lowering Var⁡(ρ_t) compresses cash-flow timing risk, which is priced as a volatility premium in discount rates.

Economic implications follow immediately: (i) anything priced off when agents act loads on the path of ρ_t; (ii) the discount rate carries a volatility premium increasing in Var⁡(ρ); (iii) firms earn control rents by pushing θ in profitable directions (timing shifts) and/or by lowering σ (timing risk reduction), which raises PV and cuts operational variance.

Labels:

• ρ_t — Time preference at time $t$. Higher values denote stronger present bias (greater weight on immediate consumption/actions). Latent state.

• K_c— Control capital: the asset stock used to shape preference formation and the timing of decisions.

• θ(K_c) — Control-conditioned target (long-run) level of time preference. Comparative statics: ∂θ/∂K_c measures steering power (sign depends on whether control pushes impatience or patience for the use case).

• κ>0 — Speed of mean reversion. Larger κ shortens the half-life ln⁡(2)/κ of shocks/interventions.

• σ(K_c) ≥0 — Residual volatility of time preference under the prevailing control stack. Often decreasing in high-quality K_c (stabilization), but may rise under aggressive short-run campaigns; the sign is an empirical object.

• W_t — Standard Brownian motion (innovation to time-preference not captured by control).

• dt — Infinitesimal time increment (continuous-time limit).

Note: In a baseline OU model without Kc, ρ reverts to θ=0.05 (5% impatience) with σ=0.02 volatility. Add Kc (e.g., $1B ad spend), θ drops to 0.03 (more patience via nudges), σ halves to 0.01—boosting PV by 10-15% via lower discount risk. Real proxy: FRED savings volatility pre/post-stimulus shows similar compression.

Time Preference Under Manipulation

With control capital (K_c) deployed, time preference ρ ceases to be a stable exogenous parameter—it’s a stochastic process, drifting and diffusing based on manipulation intensity.

Mathematically, we can model it as a controlled Ornstein-Uhlenbeck:

where K_c parametrically lowers long-run impatience θ (steering toward profitable timing) and compresses residual volatility σ (stabilizing cash flows).

For clarity, let us consider a baseline without K_c: ρ reverts to θ=0.05 (5% present bias) with σ=0.02.

Ramp K_c (e.g., $1B in algo spend from above), and θ drops to 0.03 while σ halves to 0.01—boosting present values 10-15% via reduced discount risk, per hazard models linking purchase timing to ρ variance.

Empirically, this holds: FRED’s personal saving rate (PSAVERT) proxies ρ shifts, averaging ~8.8% from 1959-2019 with std dev ~2.3% (pre-pandemic stability).

Post-2020, it spiked to 32% in April amid stimulus nudges, pushing std dev to ~6.5% through 2022—clear volatility from exogenous shocks, but platforms amplify it digitally.

In large scale campaigns we’ve run in-house, similar patterns emerge: A/B and synthetic control tests on ad timing showed 15-25% variance in purchase delays post-privacy changes, like iOS 14.5’s ATT framework (2021), which reduced CVRs to 11-22% and dropped ROAS by 17-28% across Meta platforms through 2025.

Platforms counter by dialing K_c to collapse horizons (impulse buys) or extend them (anticipation builds), but efficacy wanes under adaptation—test via cohort retention drops.

In practice, this means platforms can shift your willingness to wait for future goods or collapse it to the present.

Imagine your patience as a line on a chart.

Left alone, it wiggles gently.

Under algorithmic manipulation, however, it spikes and dips—suddenly you feel the urge you must buy today, or suddenly you’re willing to wait.

Mathematically, it drifts and jitters depending on how much control capital is applied and how volatile its effects are.

Empirical evidence supports this volatility: FRED data on the US personal saving rate (PSAVERT) shows dramatic fluctuations post-2020, averaging 8.39% from 1959-2025 but spiking to 32% in April 2020 amid stimulus checks—clear shifts in time preference driven by policy nudges.

Patience isn’t fixed—it spikes when platforms apply interventions.

The dotted red lines show campaign spikes where time preference is actively shifted (e.g., sudden ad blitzes, timed discounts).

This volatility is priced like any other financial variable.

A stable consumer base lowers financing costs.

A volatile one raises them.

Platforms invest in K_c to minimize this, converting chaos to revenue—yet overreliance breeds fragility, as shocks reveal manufactured demand’s thin backing.

The Macroeconomic Consequences

This shift explains a lot of the paradoxes you’ve probably been seeing in the macroeconomic data and feeling on the ground.

On paper, things can look fine; in reality, something’s off.

But to see the mechanism clearly and understand what’s truly happening, it helps to revisit a basic point that’s been obscured by political spin and dashboard-watching.

GDP, as currently measured, is an aggregate of spending.

It does not distinguish between productive and unproductive activities, and it’s blind to manipulation intensity.

GDP still “measures” total spending; it simply has no way of measuring how much manipulation it took to sustain that spending.

That’s why headline growth can rise even as authentic willingness to pay decays.

Why?

Because you can goose spending with discounts, nudges, junk impressions, or timing hacks and still post “growth.”

In plain terms: GDP can’t tell the difference between authentic demand (grounded in savings and productivity) and artificially manufactured demand (created by preference control at the moment of purchase).

This disconnect creates an illusion of stability while masking the fragility under the hood: if the machinery of manipulation fails—due to regulatory change, cultural shift, or platform disruption—the propped-up demand will disappear, even if wages and employment haven’t moved yet.

To quantify this disconnect: GDP aggregates total spending without parsing manipulation intensity.

For example, personal consumption expenditures (PCE, ~68% of US GDP per BEA/FRED data in 2024-2025) include digital ad-influenced retail (with US digital ad spend ~$320-350B annually in 2025 per estimates).

Elasticity studies suggest that when PES > 0.3 (i.e., more than 30% of demand at fixed prices comes from nudges), this corresponds to a 10–30% lift in demand attributable to manipulation.

In aggregate, that implies up to 15% of PCE growth could be CAD-propped rather than organic.

Apple’s iOS 14.5 ATT rollout in 2021 exposed the fragility: ad conversion/ROAS fell by 20–40%, yet headline GDP held steady because offsets like SKAdNetwork attribution muted the immediate impact.

However, those offsets are recorded as if they were free.

They don’t account for the unmeasured “K_ccost”—the control-capital expenditure required to manufacture timing.

In 2025, global ATT opt-in rates have slid from ~45% to 25–40%, meaning the hidden control-capital burden is even higher.

If the PES (Preference Endogeneity Score) exceeds 0.3 (i.e., >30% of demand from nudges, proxied by elasticity to platform cues)—then as much as 15–20% of headline growth may be a CAD-propped illusion.

On the surface, this looks stable (for example, the U.S. posted 2.8% GDP rise in 2024 despite high rates).

But underneath, it’s fragile: shocks like Apple’s iOS 14.5 ATT deprecation in 2021 cut ad conversions and ROAS by 20–40%, yet GDP didn’t immediately falter because offsets like organic shifts and SKAdNetwork adaptations masked the impact.

This resolves the paradox—CADs prop spending short-term via Kc (lowering Var(ρ) for predictable cash flows), but amplify busts when efficacy drops, as control rents erode without authentic savings backing.

This may also explain why the usual central-bank manipulation levers haven’t been super effective.

Interest-rate adjustments and balance-sheet operations target things like the output gap, unemployment, and headline inflation.

But none of those tools directly touches the cost of manipulating time preference.

Rates shift financing conditions, not the ρ-control parameters directly.

Firms then adjust K_c and campaign intensity to hit revenue timing targets.

So we get this pattern:

• When rates rise, credit expansion is suppressed. Firms respond by doubling down on control capital manipulation to stabilize cash flow. CAD-driven demand doesn’t necessarily fall; it can intensify as companies squeeze more from their control stack to offset tighter financing.
• When rates fall, financing costs drop and discount rates compress. That raises the present value of CAD cash flows and cheapens investment in control infrastructure. The result is more budget for manipulation, higher control-capital intensity (CCI), and a smoother surface boom—but greater fragility, because a larger share of cash flows is now collateralized on manufactured timing.

This creates a new type of boom-bust cycle: neuro-malinvestment.

Companies build and time production off false signals—not because money is cheap per se, but because manipulated time preference tricked them into thinking demand looked durable when it wasn’t.

The nudges keep the line flat, until they don’t.

And when manipulation efficacy drops—say a tracking identifier is deprecated, distribution is lost, or consumers’ preferences change—the manufactured demand collapses, and the misallocation is revealed.

You can already see echoes of this pattern from the post-2022 Bidenflation era: a policy-induced inflation spike, aggressive hikes meant to ‘control’ it, followed by firms compensating through control spending to hold revenue timing steady.

The causal chain on the ground isn’t ‘rates → real demand’ so much as ‘rates → financing conditions → control intensity → perceived demand stability.’

Here’s a simple way to tell the difference between mainstream narrative and what’s actually happening:

• Rate-driven malinvestment is the intertemporal misallocation caused by credit expansion and artificially low rates that falsify the interest signal, leading entrepreneurs to mistake cheap money for savings-backed demand and to launch production plans that later prove unsustainable. You’ll see it in credit/term-structure anomalies, cheap financing of long projects, and the eventual unwind when rates normalize.
• Neuro-malinvestment (a new term we introduce in this paper) is the intertemporal misallocation caused by engineered shifts in time preference. This is the same error as rate-driven malinvestment under a different false signal: engineered shifts in time preference—via control capital and CADs—manufacture near-term purchase hazards and apparent demand stability that firms misread as sustainable, savings-backed demand. You’ll see it in rising TPVI (time-preference volatility), elevated CCI, higher ACR (attention collateralization), and outsized sensitivity to platform/privacy shocks relative to rate moves.

Both are forms of intertemporal discoordination—one rate-driven, the other preference-driven—and they can coexist and amplify each other. Diagnose the former via credit/term-structure anomalies, and the latter via TPVI, CCI, and sensitivity to platform/privacy shocks.

Put simply: the economy no longer runs primarily on savings and productivity; it runs on algorithmic nudges, manipulative preference shaping, and volatility management.

Monetary policy has only indirect influence over that substrate.

Raise rates, and firms compensate by turning the control dial up.

Cut rates, and you subsidize the control dial.

Cheap credit fuels investment not in productive capacity, but in manipulation infrastructure.

In both cases, without separate constraints on control capital, the structure drifts toward greater dependence on manufactured timing—and the eventual crash becomes a matter not of interest-rate normalization, but of control efficacy breaking.

Systemic Risk

Like mortgage-backed securities, CADs look diversified on paper but are tied to the same fragile foundation.

Manipulation efficacy is not independent across tranches; it’s concentrated in a handful of control surfaces and distribution pipes (e.g., Apple’s tracking identifiers, Google’s ad infrastructure, Meta’s feeds).

We’ll call this the CAD correlation factor—common exposure to a small set of preference-control surfaces (OS identifiers, ad pipes, recommendation feeds).

When any one of these is impaired (privacy law with teeth, OS policy change, mass audience migration), tranche performance co-moves across portfolios as the common control surfaces fail simultaneously.

When that happens, PES—the preference endogeneity score (how much a market’s demand depends on control capital)—plummets across portfolios as cues stop working.

This type of crash won’t look like a traditional financial crisis.

It will feel like a sudden cultural/economic shift: ads stop clearing, subscriptions churn, gig workers disappear, and coverage commitments miss—as if the economy’s ‘mood’ changed overnight.

The structure is opaque.

Investors, advertisers, and even platforms often don’t see how correlated their exposures to preference volatility really are.

That opacity encourages over-leverage on the assumption that behavioral control tomorrow will be as effective as today.

The unwind, when it hits, won’t be a bank run or a tape crash—it will be a synchronized drop in engagement, retention, and responsiveness: a collective demand-timing snap instead of a major credit event.

Early warning: watch for ROAS and LTV cohort curves to degrade in parallel across unrelated verticals right after a privacy/platform change (ID deprecation, cookie kill, algorithm tweak).

If those same curves react far less to a rate move but swing hard on platform or privacy news, your exposure is to preference control, not credit.

That’s systemic CAD risk, not monetary sensitivity.

You’re just seeing the CAD correlation factor revealed in real time.

Making the Invisible Visible

There are a lot of false signals in the data right now, but this ‘hidden’ economy can be measured if you know what to look for.

• A Time-Preference Volatility Index (TPVI), inferred from churn patterns, micro-savings delays, or impulse elasticity, tracking how stable or unstable people’s patience is.

• Control Capital Intensity (CCI), showing how much companies spend on manipulation vs. production.

• An Attention Collateralization Ratio (ACR), measuring how much debt is based on future behavior forecasts.

• A Preference Endogeneity Score (PES), revealing how much of demand comes from nudges rather than intrinsic desire.

Together, these four measures form an early-warning dashboard for the CAD economy.

Test this framework: If CAD risk is real, expect synchronized ROAS/LTV drops across sectors post-privacy shocks, outpacing rate-hike effects.

Industry reports show that after Apple’s iOS 14.5 rollout in 2021, Meta’s global opt-in rates fell to around 45% (sliding further to 25–40% by 2025). That drop tanked ROAS 20–40%, even though interest rates were steady—an impact far sharper than the demand response to the 2022–2023 rate hikes.

Forecast: A 2026 EU privacy crackdown could spike TPVI 20%, revealing neuro-malinvestment.

If TPVI and ACR are rising, CCI is accelerating, and PES stays elevated, you’re not looking at organic, savings-backed demand, you’re looking at manufactured timing.

For readers who want the quantitative backbone, here’s how each metric can be operationalized in practice:

• TPVI (Time-Preference Volatility Index): infer from renewal-lag dispersion, BNPL adoption elasticity to tiny friction changes, and response latency distributions; normalize by segment.
• PES (Preference Endogeneity Score): estimate elasticity of purchase probability to platform-controlled cues (timing/placement) at fixed price; PES = 1 − (cue-elasticity adjusted for off-platform anchors).
• ACR (Attention Collateralization Ratio): share of financing whose covenants/valuation multiples reference engagement/retention forecasts rather than realized cash flow.
• CCI (Control-Capital Intensity): capex + opex on data/models/content/placement rights divided by total capex + opex.

Taken together, these metrics act like seismographs for systemic fragility, giving you an early-warning dashboard.

The same way credit spreads once betrayed hidden leverage before 2008, synchronized spikes in TPVI, ACR, and CCI with persistently high PES tell you that demand is no longer self-sustaining.

That’s when systemic CAD risk is building beneath the surface, waiting for a cultural, technical, or regulatory shock to trigger the unwind.

This mix reliably produces the pattern Figure 4 illustrates below: short-term conversion lift now, margin compression next period, and rising fragility to any privacy/platform shock that lowers manipulation efficacy.

In the short run, investing heavily in control capital (data, targeting, nudging tools) boosts sales.

But over time, the more companies spend on manipulation, the more their future margins shrink—because everyone is competing on the same lever, bidding up ‘control rent.’

The CAD economy doesn’t just change finance; it redirects talent and capital away from building real things.

The same engineering skill that could advance space technology, biotechnology, or energy is instead consumed by the race to perfect micro-manipulations of human time preference.

Some of the brightest engineers in the world are not designing reactors, rockets, or new materials—they are designing systems to optimize when you click.

They are working on optimizing push notifications.

They are working on evolutionary algorithms that show you exactly what you want and at the perfect time, almost as if your phone is listening to everything you say and think.

The opportunity cost is immense but invisible in GDP.

GDP and stock indices treat all revenue equally, whether it is earned by curing disease or by perfectly timing an in-app purchase prompt.

GDP accounting treats both as equal contributions, but society does not reap equal benefits.

A dollar earned from nudging a purchase is treated the same as a dollar earned from curing disease.

The system pushes toward manipulation because the returns are immediate and predictable.

This shift also erodes consumer sovereignty, the philosophical core of the Austrian framework.

If preferences are shaped by the same entities that profit from satisfying them, the notion of voluntary exchange becomes muddied.

The market process is no longer merely coordinating plans among independent actors; it’s recursively engineering the actors themselves.

But like every hidden engine that came before it, the CAD economy will eventually reveal itself.

It will arrive when a major platform’s manipulation efficacy drops sharply—whether from regulation, technological change, or cultural shift that makes certain manipulations ineffective overnight.

When that happens, entire CAD structures and their web of contracts—valuations, loans, ad markets—will collapse together.

The contagion will spread not through the banking system, but through the synchronized failure of ad campaigns, labor supply chains, and subscription revenue projections.

The recognition will feel like déjà vu: the discovery that what looked diversified was, in fact, one fragile thing sliced into many pieces.

But this time the collateral won’t be houses—it will be our collective willingness to act or wait, packaged into securities we never knew existed.

This event will force an uncomfortable reckoning: much of what has been counted as robust, organic demand will be revealed as dependent on an artificially maintained preference architecture.

Without it, entire sectors will find their markets smaller, more volatile, and less predictable.

Note: Decentralized tech (e.g., Web3 IDs) could fragment control surfaces, reducing CAD correlation. Yet, data shows concentration persists, +50% ad spend via Google/Meta, making shocks systemic until diversified.

Final Thoughts

The U.S. economy in 2025 doesn’t just run on production, or even on attention.

It runs on the volatility of human choice.

Collateralized Attention Derivatives take the most personal thing we have—our sense of when to act—and turn it into a marketable security.

This is why GDP can look strong while the economy feels fake, why monetary policy seems practically useless, and why culture and regulation now move markets more than interest rates.

The central reality is that the volatility of time preference, once a background parameter, has become the raw material of a sprawling, opaque, and systemically important set of markets.

These collateralized attention derivatives are not a side effect of digital ads or gig work; they are the structural core of how value is extracted, priced, and traded.

And this transformation forces us to rethink both theory and policy.

In my humble opinion, Austrian capital theory must now be extended to treat preferences as endogenous and capital-using.

And macroeconomic policy must recognize preference volatility as a driver of cycles and fragility.

As such, we are extending Austrian capital theory by making time preference a state variable with capital-using production; it connects to asset pricing by putting a volatility premium on ρ.

At stake isn’t just allocation efficiency; it’s whether we’re comfortable collateralizing the variance of human agency.

At the end of the day, however, society must ultimately decide whether the engineering of decision-timing is a sane basis for economic prosperity.

Rule for Analysts: Treat CAD-linked revenue as a contingent liability; covenant to improving PES and stable-to-lower TPVI, cap CCI growth in cheap-credit windows, and stress-test ROAS/LTV to platform/privacy shocks rather than rate moves.

We’re already seeing the machinery reveal itself in the data.

The system is showing signs of stress.

TPVI spikes show rising instability of patience; CCI ratios reveal firms pouring more into manipulation than production; ACR levels tell us financing is leaning heavily on forecasted engagement; and PES scores confirm demand is more engineered than organic.

Those are the footprints of CAD fragility.

But most people still do not recognize the danger we’re in right now.

They are still focused on interest rates, which are just low-tech distractions for enthusiasts and nerds.

Until that recognition arrives, the CAD engine will hum quietly beneath the surface—allocating not just capital, but choice itself, to those best positioned to predict and manipulate the shadows of what we might do next.

If you like The Unconquered Mind, sign up for our email list and we’ll send you new posts when they come out.

If you liked this post, these are for you too: 

Tariffs, Triffin, and the Fall of America

America’s Hidden Economic Civil War

Invisible Economics & Post Scarcity Societies