Abundant Intelligence and Deficient Demand: A Macro-Financial Stress Test of Rapid AI Adoption

Xupeng Chen

cs.AIMar 10, 2026

Abundant Intelligence and Deficient Demand: A Macro-Financial Stress Test of Rapid AI Adoption

Authors: Xupeng Chen

Organizations: Department of Electrical and Computer Engineering, New York University

Abstract

We formalize a macro-financial stress test for rapid AI adoption. Rather than a productivity bust or existential risk, we identify a distribution-and-contract mismatch: AI-generated abundance coexists with demand deficiency because economic institutions are anchored to human cognitive scarcity. Three mechanisms formalize this channel. First, a displacement spiral with competing reinstatement effects: each firm's rational decision to substitute AI for labor reduces aggregate labor income, which reduces aggregate demand, accelerating further AI adoption. We derive conditions on the AI capability growth rate, diffusion speed, and reinstatement rate under which the net feedback is self-limiting versus explosive. Second, Ghost GDP: when AI-generated output substitutes for labor-generated output, monetary velocity declines monotonically in the labor share absent compensating transfers, creating a wedge between measured output and consumption-relevant income. Third, intermediation collapse: AI agents that reduce information frictions compress intermediary margins toward pure logistics costs, triggering repricing across SaaS, payments, consulting, insurance, and financial advisory. Because top-quintile earners drive 47--65% of U.S.\ consumption and face the highest AI exposure, the transmission into private credit ($2.5 trillion globally) and mortgage markets ($13 trillion) is disproportionate. We derive eleven testable predictions with explicit falsification conditions. Calibrated simulations disciplined by FRED time series and BLS occupation-level data quantify conditions under which stable adjustment transitions to explosive crisis.

Figures & tables

Sector	U.S. Rev. ($B)	Friction Share	Switching Costs	Regulatory Barrier	Net Exposure
SaaS (seat)	$\sim$ 300	High (60–80%)	Moderate	Low	High
Card payments	$\sim$ 120	Moderate (40–60%)	High	High (Durbin, PCI)	Moderate
Insurance brokerage	$\sim$ 50	Moderate (40–50%)	Moderate	High (licensing)	Low–Mod.
Mgmt. consulting	$\sim$ 330	High (50–70%)	Low	Low	High
Financial advisory	$\sim$ 120	Moderate (30–50%)	Moderate	High (RIA, fiduciary)	Moderate
Legal services	$\sim$ 370	Moderate (30–50%)	Moderate	High (bar, UPL)	Low–Mod.

Table 1: Intermediation Exposure by Sector

Figure 1: Macro preconditions for the displacement spiral. (a) Labor share (BLS, FRED series PRS85006173) has declined

\sim

19% from its 1960 peak. (b) M2 velocity (FRED series M2V) has declined 36% from its 1997 peak, consistent with Proposition 2 . (c) Ghost GDP proxy: divergence between real GDP and real personal income growth, widening post-2020. (d) Productivity–compensation gap since the 1970s demonstrates that technological progress need not translate into wage growth. Shaded regions indicate NBER recession dates.

Figure 2: Occupation-level evidence: AI exposure and wage outcomes. (a) Nominal wage growth (2019–2023) vs. AI exposure (

\beta=-6.53

p<0.01

). (b) Real wage growth after adjusting for 18% CPI inflation. (c) Pre-trend placebo (2015–2019): the significant negative slope (

\beta=-3.86

p<0.01

) indicates pre-existing trends, likely reflecting the denominator effect (higher-wage occupations have lower percentage growth). (d) Wage acceleration (DiD): the post-minus-pre difference (

\beta=-2.68

p=0.14

) isolates the AI-specific shift. Bubble size proportional to 2019 employment. Source: BLS OEWS; AI exposure based on Felten et al. (2021) and Eloundou et al. (2023) .

	(1)	(2)	(3)	(4)	(5)	(6)
	Nom. Wage	Real Wage	Employment	Nom. Wage	Pre-Trend	Wage
	Growth	Growth	Growth	Growth	Placebo	Accel.
	2019–23	2019–23	2019–23	2019–23	2015–19	$\Delta\Delta$
AI Exposure	-6.53***	-6.53***	2.22	-4.21**	-3.86***	-2.68
	(2.11)	(2.11)	(4.16)	(2.07)	(1.03)	(1.79)
Log Wage 2019				-1.81

Table 2: Cross-Sectional Regressions: Occupation-Level Outcomes on AI Exposure

Parameter	Symbol	Value	Source
Labor Market
Initial labor share	$s_{L,0}$	0.56	BLS (2024)
MPC, labor income	$\bar{c}$	0.85	Carroll et al. [2017]
MPC, capital income	$1-\bar{c}$	0.15	Model restriction †
Consumption
Top-quintile consump. share	$\chi$	0.47–0.65	Moody’s [2023]; CES/SCF

Table 3: Calibrated Model Parameters

Figure 3: Calibrated displacement dynamics with diffusion and reinstatement (

\rho_{0}=0.002

\bar{d}=0.80

\kappa=2.0

\tau=0

). (a) Labor share under three AI adoption rates. (b) Monetary velocity. (c) Reinstatement comparison under the rapid scenario. (d) Consumption-to-GDP ratio.

Figure 4: Policy response under rapid AI adoption (

g_{A}=0.20

). (a) Labor share trajectories under four policy scenarios. (b) Crisis depth as a function of policy lag

\ell

for three transfer magnitudes.

Recession	Name	Real PCE	Real GDP	Primary
		Decline (%)	Decline (%)	Mechanism
1973Q4–1975Q1	Oil shock	-2.4	-1.8	Stagflation; supply-driven
1980Q1–1980Q3	Volcker I	-2.2	-1.3	Monetary tightening
1981Q3–1982Q4	Volcker II	-1.5	-2.7	Monetary tightening
1990Q3–1991Q1	S&L crisis	-0.8	-1.4	Financial; real estate
2001Q1–2001Q4	Dot-com	-0.3	0.2	Investment-led; mild

Table 4: Historical U.S. Recession Comparison: Consumption Declines

	Predicted Pattern	Observable Indicator	Data Source	Falsification Condition
H1	White-collar unemployment rises or wages stagnate in AI-exposed occupations	Information-sector unemployment; BLS wage percentiles for AI-exposed vs. control occupations	BLS CPS, OEWS; JOLTS; Indeed Hiring Lab	AI-exposed occupations show faster wage growth or lower unemployment than controls through 2028
H2	SaaS revenue compression: seat-based pricing collapses, ARR growth decelerates	Median SaaS net revenue retention; new ARR per public SaaS company	Public SaaS earnings; KeyBanc SaaS survey; Bessemer Cloud Index	SaaS net retention remains above 110% and seat-based pricing persists
H3	Credential inflation in AI-exposed roles	Credential requirements for stable-task roles	Lightcast/ Burning Glass	Requirements decline in AI-exposed roles
H4	Consumption decline concentrated in upper-income cohorts	Consumer spending by income quintile; retail sales in high-income ZIP codes	BEA PCE; credit card spending data (Facteus); Census retail	Consumption declines are uniform across quintiles or concentrated in lower quintiles
H5	Private credit stress in software/tech portfolio; public–private mark gap widens	Default rates in PE-backed software loans; CLO spreads; public–private NAV gap	Moody’s; Preqin; SEC Form PF; BIS	PE-backed software defaults below historical base rates and gap within 200 bps through 2028
H6	Monetary velocity declines despite positive GDP growth	$V=\text{GDP}/M2$ ; divergence between GDP growth and personal income growth	FRED; BEA NIPA	Velocity increases or remains stable while GDP grows

Table 5: Testable Predictions and Early Warning Indicators

Mechanism	Established	Directional	Speculative
AI productivity & slowdowns	14–56% task-level gains; experienced devs slower with AI in field	Gains extend to complex tasks; overhead limits net effect	Economy-wide acceleration
Displacement spiral & reinstatement	Labor share $\downarrow$ 19%; AI-exposed wages compressed; new tasks historically offset automation	Agentic tools perform with limited oversight; AI-complementary roles emerging	Self-reinforcing macro feedback; reinstatement dominates
Ghost GDP & intermediation	Velocity $\downarrow$ 36% since 1997; friction reduction is core AI capability	GDP–income gap widening; SaaS pricing pressure	Velocity acceleration; sector-wide margin collapse
Consumption & deflation	Top 20% = 47–65% of spending; inference costs $\downarrow$ 30%/yr	High-income workers most AI-exposed; cloud prices falling	Concentrated shock; deflation offsets compression
Credit & fragility	Private credit $2.5T; mortgages$ 13T; leverage/opacity flagged by BIS	Software LBOs assume growth; multiple stressors could trigger repricing	Systemic cascading defaults; AI as unique trigger

Table 6: Evidence Assessment Matrix

AI and Labor

AI and labor studies how artificial intelligence changes jobs, wages, skills, hiring, and the organization and quality of work.

See all 94 papers

Cognitive Debt: AI as Intellectual Leverage and the Dynamics of Systemic Fragility
Jun 13, 2026Shuchen MengAI Risk Management Cognitive Offloading
New York University
Generative AI and the Reorganization of Labor Demand
May 22, 2026Fangyan Wang, Zaiyan Wei, Yang WangAI and Labor
Mitch Daniels School of Business, Purdue University; West Lafayette, IN 47907
The Open Source Economic Index of AI Adoption and Capability
May 23, 2026Seamus Somerstep, Aritra Guha, Divesh Srivastava +1AI Agent Evaluation AI Agent Benchmarks
University of Michigan · AT&T Chief Data Office · IFM-MBZUAI

Jun 13, 2026J/K move · Enter open · S save

Cognitive Debt: AI as Intellectual Leverage and the Dynamics of Systemic Fragility

Shuchen Meng

New York University

AI Risk Management Cognitive Offloading

Open paper AskPDF arXiv

We develop a formal theory of cognitive debt: the stock of unverified reasoning obligations that accumulates when individuals use AI as a substitute rather than a complement for first-principles cognition. The model features two state variables per agent, cognitive capital and cognitive debt, and a multiplicative production technology in which cognitive capital functions as collateral that determines the return to AI adoption. We establish six propositions. Rational agents incur positive cognitive debt because the costs are deferred, partially external, and masked by short-run productivity gains. Tranquil periods lower subjective risk assessments, raise AI substitution intensity, and compound leverage, generating a cognitive Minsky moment in which subjective risk falls while true systemic fragility rises. Expected crisis losses are convex in aggregate leverage. Post-crisis, output-target pressure can produce a false-correction loop in which agents patch AI failures with more AI. The decentralised equilibrium over-adopts substitutive AI relative to the social optimum because of systemic risk, cognitive public goods, and arms-race externalities. In a two-type heterogeneous-agent economy, high-cognitive-capital agents adopt AI more intensively and may eventually erode their unaided cognitive capital below that of initially lower-skilled agents.

Abstract

AI and Labor

Explore similar work

Cognitive Debt: AI as Intellectual Leverage and the Dynamics of Systemic Fragility

Generative AI and the Reorganization of Labor Demand

The Open Source Economic Index of AI Adoption and Capability