AI and Economic Transition: consumer cohort effects Report Interpretation

Overview

This macro report examines how AI adoption may affect US consumer cohorts through labor income, financial wealth and inflation. Morgan Stanley’s central conclusion is that college-educated, high-income, urban households—especially established workers—may be safer than conventional concerns about white-collar displacement imply, although younger professionals and a rapid-adoption scenario remain vulnerable.

Core views

Morgan Stanley frames AI’s consumer effects through three connected channels: labor markets, wealth and prices. Its prior general-equilibrium work suggests the outcome depends primarily on the pace of AI diffusion, the speed and scale of new task creation, and feedback effects from asset markets. The base case assumes AI diffuses over about a decade, roughly twice the speed of the internet era, with conservative wealth effects. In that scenario, temporary displacement is offset by productivity-led real-wage gains, new jobs, higher profits and disinflation, raising output and income. Peak unemployment rises by around 0.2-0.5pp in the base case, versus 0.6pp-4.1pp in a rapid-diffusion bear case in which full adoption takes six years and skill mismatches and weak demand overwhelm job creation. The report challenges the idea that white-collar consumers are automatically the principal AI casualties. Its task-based occupational-exposure analysis finds that high-exposure jobs are concentrated among college-educated, higher-income and metropolitan households—collectively termed CHIC households. About 26% of workers are in high-exposure occupations, compared with 56% in medium-exposure and 18% in low-exposure roles. Median income is $97k in the high-exposure group versus $46k in the low-exposure group; nearly 70% of high-exposure workers have at least a bachelor’s degree, compared with about 10% of low-exposure workers. Yet exposure may mean either automation or augmentation. Unemployment in highly exposed occupations is about 0.5pp above the level implied by cyclical trends, but the report sees effects so far as relatively small. Younger workers are more exposed to automation of entry-level tasks, whereas experienced workers may capture augmentation, productivity and wage gains. New-job creation is an important offset in the base case. AI-related postings are concentrated in professional and technical services, information and finance, tend to carry higher estimated salaries than non-AI postings, and tilt toward workers with experience: 32% target zero-to-two years of experience, compared with 55% for all postings with stated numerical experience requirements. AI-related jobs increased from 2.1% to 4.7% of all postings over the past year. The report therefore concludes that the same CHIC households with high exposure in existing occupations, particularly workers with experience, currently appear better positioned for AI-created jobs. By contrast, embodied AI and robotics may affect manual occupations more directly, but Morgan Stanley expects slower diffusion because of hardware costs, safety requirements, capital intensity and insufficient physical-world training data. Its Global Robot Model projects US robot stock rising to 35 million by 2030 from 15 million in 2026, while humanoid robots do not exceed 1 million until 2035. Wealth effects make the distributional result especially important. The top income quintile accounts for 87% of direct household holdings of corporate equities and mutual funds; college-educated households account for 85%, and households aged 55 and above account for 79%. Stocks represent around 30% of net worth for the top 20% income cohort but only 10% for the bottom 40%. Household wealth increased by roughly $21tn from 1Q24 to 1Q26, with direct equities contributing half of the rise. Morgan Stanley estimates consumers typically spend 3-5 cents of each dollar of permanent wealth gains. With top-cohort equity wealth roughly six times annual labor income, a 4% permanent rise in equity wealth can offset the spending impact of a 1% labor-income decline under an assumed 90% marginal propensity to consume from labor income and 4% from financial wealth; the bottom 40% would require a 24% wealth increase. This means relatively modest sustained asset gains can support affluent consumers’ spending even during temporary labor-income weakness. Inflation produces a different near- versus longer-term distribution. AI infrastructure demand could lift prices for electricity, software and chip-related goods, although most such categories have small CPI and PCE weights. Electricity is the main near-term concern: it is 4% of spending for lower-income cohorts versus 2% for upper-income cohorts, while vehicle spending is 8% for upper-income consumers versus 5% for lower-income consumers. The report finds more evidence of data-center-related pressure in electricity prices than of higher chip costs passing through to vehicle prices; in areas with greater data-center concentration, electricity prices have risen more. Low-income, non-college-educated and Southern households are consequently more exposed to near-term price pressure. Over the medium to longer run, Morgan Stanley expects productivity gains to be disinflationary, first in high-adoption industries such as information technology, financials and communications. High-income households spend about 11% on financial services and insurance versus 4% for the bottom quintile, and 11% on other services including professional and communications services versus 5% for the lowest quintile, positioning them to benefit earlier from service-sector disinflation. Bringing the channels together, the base case is positive for CHIC households overall: wealth effects, productivity-led wages, new jobs and eventual disinflation outweigh displacement, though younger highly educated workers may experience short-term weakness and inequality widens. Morgan Stanley expects affluent households’ large share of discretionary spending—including furniture, vehicles, luxury goods and recreation—to support those categories. Lower-income households may still gain in absolute terms from broader wage gains, disinflation or potential monetary easing, but receive less of the initial upside and face more near-term inflation pressure. In the bear case, rapid adoption produces an initial CHIC employment shock; if asset prices also fall by 25-30% or more for a sustained period, affluent consumers reduce discretionary spending and their large spending share turns the shock into broader demand weakness across all cohorts. Monetary and fiscal policy responses become critical in that outcome.

Analysis framework

The report starts with base and bear general-equilibrium AI-adoption scenarios, then applies microdata on occupations, household balance sheets and spending patterns across income, education, geography and age. It evaluates labor displacement, productivity-led wage growth and job creation; translates equity holdings into consumption effects using marginal-propensity-to-consume assumptions; and compares cohort spending baskets to assess near-term inflation and longer-run disinflation exposure.

Methodology notes

• Macroeconomics

  General-equilibrium AI adoption scenarios

  Morgan Stanley uses prior general-equilibrium modeling to compare a moderate diffusion base case with a rapid-diffusion bear case, incorporating task creation, wealth feedback and labor-market adjustment.

• Industry AnalysisVolume-price decomposition

  Three-channel consumer transmission analysis

  The report separates AI’s consumer impact into labor income, financial wealth and inflation, then compares how each channel affects demographic cohorts.

• Quantitative, Factor, and Portfolio Theory

  Task-based occupational AI exposure index

  Using the Felten et al. framework, occupations are scored by how their tasks and abilities align with current and anticipated AI capabilities, distinguishing exposure from a guaranteed negative outcome.

• Corporate Fundamentals and FinanceFree cash flow analysis

  Marginal propensity to consume from labor income and wealth

  The report estimates how permanent equity-wealth changes could offset lost labor income in consumer spending, with results differing sharply by cohort wealth levels.

Key data

• Base-case AI diffusionAbout 10 years; roughly 2x the internet’s paceAssumes productivity gains without a downturn.
• Base-case peak unemployment increase+0.2-0.5ppTemporary rise offset by wages, task creation, disinflation and wealth effects.
• Bear-case peak unemployment increase+0.6pp to +4.1ppRapid six-year adoption scenario; outcome depends on indirect wealth and demand effects.
• Workers in high AI-exposure occupations26%Compared with 18% low exposure and 56% medium exposure.
• Median income by exposure group$97k high exposure; $46k low exposureBased on occupation median annual salary data as of 2024.
• AI-related job-posting share4.7%, up from 2.1% over the past yearEvidence of expanding AI-related hiring.
• Top 20% share of direct equity and mutual-fund holdings87%Illustrates concentration of AI-related wealth effects.
• Household wealth changeAbout $21tn from 1Q24 to 1Q26Direct equities accounted for half of the increase.
• Asset gain needed to offset a 1% labor-income decline4% for top 20%; 24% for bottom 40%Assumes 90% MPC from labor income and 4% from financial wealth.

Impact & implications

Morgan Stanley argues that consumer outcomes from AI cannot be inferred from occupational exposure alone. Under its base case, affluent, educated urban households’ asset ownership and positioning for productivity and new-job gains support aggregate discretionary consumption, while distributional inequality increases. A rapid deployment path combined with a meaningful asset-market decline would reverse that logic and spread weaker demand across consumer cohorts.

Risks

• Rapid AI diffusion could cause job displacement to outpace task creation, producing recession-like outcomes.
• A sustained asset-market decline of 25-30% or more could cause affluent households to reduce spending and amplify a broader demand shock.
• Skills mismatches may prevent displaced workers from moving quickly into newly created AI-related roles.
• Software or other AI-related price pressures could exceed expectations, reducing consumer purchasing power.
• Distributional results depend on the pace of adoption, asset-market performance, and monetary and fiscal policy responses.

What to watch

• Monthly and quarterly cohort-level employment, unemployment, wage growth and labor-income measures.
• Morgan Stanley’s AI labor tracker and AI productivity tracker.
• AI-related job postings, including industry, compensation and experience requirements.
• Changes in net worth and equity wealth by income, age and education cohort.
• Cohort inflation measures, especially electricity prices in data-center-concentrated regions.
• Spending trends in discretionary categories and across income, age, geography and education groups.