Fitch Stress Test Models 1.5% US Contraction After 35% Equity Drop
The stress reaches recession scale only after falling household wealth is combined with a direct business-investment shock; tighter financial conditions make the contraction deeper. Fitch is describing a conditional path, not forecasting an AI crash.
Fitch Ratings' September 9 stress scenario starts with a 35% fall in US equities over six months, a 15% decline in equities elsewhere and a separate confidence shock to US business investment. Run through Oxford Economics' Global Economic Model, the combination drives year-on-year US GDP growth to negative 1.5% in the second quarter of 2027 and cuts private capital expenditure by more than 6%. Fitch does not use that path as its base case.
For macro portfolios, the useful result is the interaction among three exposures. Falling equity wealth weakens household demand; the confidence shock hits investment directly; financing conditions determine how much monetary easing can cushion both. Fitch's world-growth measure falls below 1% as the US shock spills abroad.
The monetary-policy assumption materially changes the outcome. In Fitch's main stress, the Federal Reserve eases sharply as demand and inflation weaken. A second version overlays higher sovereign yields and wider credit spreads, producing a deeper contraction because lower policy rates transmit less effectively to households and companies.
Fitch's public summary does not allocate the GDP loss among wealth, investment, credit and confidence channels, so it cannot support a precise claim about which one contributes most. It does show that the recession result requires more than an isolated repricing of technology shares: the market decline has to damage spending and investment while financial conditions fail to offset enough of the shock.
Bank of America challenges the premise that today's AI trade already has the hallmarks of a classic bubble. Its global equity-derivatives work points to earnings growth that has outpaced share-price gains in parts of the sector and a proprietary bubble-risk indicator that remains outside bubble territory. That is counterevidence to calling a crash imminent; it is not a probability estimate for Fitch's 35% scenario.
Fitch's contribution is therefore useful as a conditional macro stress, not as a market-timing call. A technology-led drawdown becomes recessionary in the model when it is accompanied by falling household demand, a direct capex retrenchment and impaired financial transmission. Portfolio stress tests that shock AI equities without those macro channels will not reproduce the recession Fitch is warning about.
Sources
Fitch Ratings, Severe Equity Price Shock Scenario Would Entail US Recession: https://www.fitchratings.com/research/sovereigns/severe-equity-price-shock-scenario-would-entail-us-recession-09-09-2026
MarketWatch, Bank of America AI bubble-risk analysis: https://www.marketwatch.com/story/history-shows-the-bar-to-disrupt-ai-is-surprisingly-high-says-bank-of-america-ddf543d4