AI is already hollowing out the white-collar economy
Economists, researchers, and workers are increasingly asking what happens to an economy built around a premium on human intelligence when that premium vanishes
On a Saturday in February, one of Substack’s most widely read financial newsletters published a provocative thought experiment: What if the AI boom — which has already created enormous wealth and driven corporate capital expenditures to record levels — is not a bullish signal, but a bearish one? What if the same technology boosting white-collar productivity ultimately erodes the broader white-collar economy?
The widely circulated post by Citrini Research opened with a fictional memo dated June 30, 2028: “The unemployment rate printed 10.2% this morning, a 0.3% upside surprise. The market sold off 2% on the number, bringing the cumulative drawdown in the S&P to 38% from its October 2026 highs.”
Though purely hypothetical, the post triggered real-world consequences. That Monday, the Dow dropped 1.7%. Individual companies cited in the piece — Monday.com and DoorDash — fell roughly 7% apiece. IBM declined nearly 13%.
In effect, a speculative Substack essay erased billions in market value. The market’s reaction itself may be as revealing as the argument. If the scenario were implausible, it likely would not have resonated so strongly. Instead, it appeared to tap into latent anxieties about AI’s economic consequences — anxieties that may be more widespread than publicly acknowledged.
The White-Collar Contraction
The central question Citrini raised — what happens to an economy built on the premium value of human intelligence when that premium declines — is increasingly being examined by labor economists and business scholars.
White-collar payrolls have now contracted for 29 consecutive months. According to Aaron Terrazas, former chief economist at Glassdoor, that duration is historically anomalous outside of recessionary periods. The headline unemployment rate, hovering around 4.3%, obscures this dynamic. Terrazas argues that conventional unemployment metrics understate emerging labor slack, which now appears more in underemployment and labor-force exits than in jobless claims.
More revealing indicators — job postings and hiring rates — remain subdued. The signals are diffuse but persistent.
Daniel Keum, a professor at Columbia Business School, characterizes the current environment as a technological shock with two phases. The first phase — already underway — is cost-side displacement. In the United States, where labor is expensive, AI is being deployed primarily to reduce headcount rather than augment human workers.
The second phase — revenue expansion through new AI-enabled products and services — may eventually generate new categories of employment. But that upside remains speculative and likely years away. For now, firms are absorbing labor-saving efficiencies without offsetting job creation.
Not all displacement is direct automation. Some job losses reflect capital reallocation: firms are diverting investment toward AI infrastructure and away from traditional headcount. The substantial capital expenditures announced by Amazon, Microsoft, Google, and Meta are directed largely toward data centers and compute capacity — not payroll expansion.
An instructive proxy for white-collar labor demand is MBA employment outcomes. Elite business school graduates function historically as a leading indicator for high-skill hiring. Recent data is concerning. At Duke University’s Fuqua School of Business, 21% of job-seeking graduates were still unemployed three months after graduation last year, compared with 5% in 2019. Georgetown’s McDonough School reported 25%, up from 8%. Michigan’s Ross School reported 15%, up from 4%. Even Harvard Business School recorded 16% unemployment at the three-month mark — higher than pre-pandemic levels.
AI is unlikely to be the sole factor. Immigration policy changes under President Donald Trump, post-pandemic hiring corrections in technology, elevated interest rates, and trade policy volatility all contribute. Yet even accounting for these variables, weakened demand among the most credentialed labor segments stands out.
The Wage Deflation Question
Beyond employment levels, Citrini also examined the potential for white-collar wage deflation. Traditionally, productivity gains translate — imperfectly but meaningfully — into wage growth. AI complicates that relationship.
As AI increases individual productivity, it simultaneously increases substitutability. When an automated system can replicate aspects of professional work, workers’ bargaining leverage weakens. A junior associate who previously captured 20% of billable value may now capture 10%, even if their output increases, because the credible threat of automation caps compensation.
This dynamic aligns with a longer-term structural trend: labor’s share of GDP in the United States has declined roughly 10 percentage points from its late-1960s peak to approximately 56% in 2024.
Wage compression does not always appear in base salary figures. Compensation adjustments often occur through subtler channels:
Reduced employer contributions to benefits such as health insurance
Lower bonuses or equity grants
Expanded job scope without commensurate pay increases
These shifts reduce total compensation even when nominal salaries remain unchanged. Recent benefits data indicate a multi-year decline in the share of companies fully covering employee-only health premiums, effectively lowering take-home value without explicit salary cuts.
The Broader Economic Risk
Citrini’s thought experiment extrapolates these labor dynamics into a broader macroeconomic cascade: diminished white-collar earnings could weaken consumption, increasing credit risk and reducing demand for housing, automobiles, education, travel, and discretionary goods. Even modest job losses among high earners could generate disproportionate demand contraction.
Yet some economists urge caution. Terrazas notes that current evidence suggests gradual rather than systemic shifts. There is, at present, no definitive causal proof that AI alone is driving these labor trends — only suggestive correlations.
Federal Reserve Governor Christopher Waller publicly pushed back on more alarmist interpretations, emphasizing that AI is a tool rather than an autonomous economic actor. Historically, prior automation waves have ultimately produced net job growth.
The critical distinction this time is whether new categories of work will meaningfully require human labor. Previous technological transitions eliminated tasks but preserved human creativity and judgment as indispensable inputs. The open question is whether advanced AI narrows that remaining domain.
The market reaction to Citrini’s scenario suggests investors recognize the possibility — however uncertain — that this cycle may diverge from historical precedent. Optimists argue that workers will adapt and eventually find improved roles. That outcome remains plausible.
However, current signals imply a near-term redistribution of power away from white-collar labor. If the trajectory continues, the future being constructed may grant these workers less leverage, not more.