If AI disappoints? The transmission of US big-tech earnings information – Financial institution Underground

Daniel Ostry, Roger Vicquéry and Emilio Zaratiegui

There’s rising concern amongst policymakers, worldwide organisations, and even big-tech Chief Government Officers (reveals I, II and III) that the present synthetic intelligence (AI) growth options valuations more and more indifferent from fundamentals. The Financial institution’s February 2026 Financial Coverage Report famous that an asset value correction is a key danger to the worldwide economic system, whereas the Financial institution’s July 2026 Monetary Stability Report introduced a state of affairs for the way an AI correction may unfold. On this publish, we examine how unfavorable big-tech earnings information transmits to international markets, which knowledgeable discussions round this state of affairs. We discover that the results ripple far past tech: fairness indices decline, credit score spreads widen and the US greenback depreciates. This final consequence, along with the restricted response of Treasury yields, suggests muted flight-to-safety dynamics, not like different monetary stress episodes.

Why would an AI crash transmit in a different way from a typical monetary shock? One cause could also be that the present AI growth has been linked to expectations of higher economy-wide productiveness, for the reason that tech-sector is more and more a main engine of progress within the US. This view has been a key driver of non-public sector capital inflows into the US whilst reserve managers have shied away from Treasuries since 2015. Ought to anticipated AI-driven productiveness features disappoint, traders could retreat from each US debt and fairness markets. That may be a pointy departure from typical stress situations such because the 2008 international monetary disaster by which heightened international danger aversion precipitated a flight-to-safety alongside a big correction in (US) fairness markets.

Empirical setup

To check how big-tech information transmits throughout markets, we assemble a every day shock sequence across the earnings bulletins of the Magnificent 7 US tech firms. These tech giants have grown to signify over one third of the S&P 500 index, reflecting, partly, their dominance in growing and scaling revolutionary applied sciences – most lately associated to AI – which can render information disclosed at their earnings related for combination productiveness in addition to the broader index’s profitability. By specializing in slim home windows round their earnings bulletins, this method follows the commonplace logic of occasion research. Our contribution is to assemble a brand new sequence of Magazine-7 earnings shocks capturing information intrinsic to those corporations. This process helps to partial out the results of combination shocks (eg, financial coverage) that will collectively affect big-tech corporations’ earnings and asset markets, though information relating to how Magazine-7 corporations’ earnings load on these combination shocks may nonetheless be current.

Formally, our ‘Magazine-7’ fairness value shocks, outlined in Equation (1) beneath, are constructed as the proportion change within the inventory value of a given Magazine-7 agency round its earnings announcement. The load  on these modifications is given by the agency’s share within the S&P 500 to manage for the truth that their collective market capitalisations have risen from 3% to 35% of the S&P 500 over our pattern. Since these corporations report earnings after markets have closed, we measure the change between the closing value on earnings day and the opening value on the next day. If a couple of agency broadcasts earnings on the identical day, we add them up.

Chart 1: Magazine-7 fairness value shocks

Supply: Workers calculations.

Chart 1 plots our Magazine-7 fairness value shocks, which we assemble from 2000 to the top of 2025. The shocks largely lie between -0.5% and 0.5% and are bigger after 2020, reflecting the Magazine-7’s bigger share in S&P 500 lately. By the use of instance, the most important unfavorable shock in our pattern occurred on 28 April 2022, when Amazon introduced 2022 Q1 earnings that have been considerably beneath market expectations. In addition they materially revised down their anticipated 2022 Q2 working revenue. Collectively, these bulletins led Amazon’s inventory value to tumble by 10% in after-hours buying and selling. This occasion triggered a big market response. Regardless of Amazon constituting solely about 3% of the S&P 500 on the finish of 2021, the index fell by near 4% inside a day. The market response was not contained solely to US shares: over the approaching days, UK shares declined as properly, credit score spreads in each jurisdictions widened and the US greenback depreciated. Under, we set up that many of those dynamics signify systematic patterns following Magazine-7 fairness value shocks, which we argue could present a believable base case for the way an AI crash state of affairs may transmit to monetary markets.

Specification and outcomes

Armed with our shock sequence, we now examine how big-tech fairness value surprises have an effect on international asset costs utilizing a parsimonious native projection framework.

the place  corresponds to an asset value of curiosity in nation , together with fairness indices, efficient trade charges, credit score spreads, nominal authorities bond yields and break-even inflation charges. Controls in  embrace lags of the dependent variable in addition to different asset value modifications. Our coefficients of curiosity, , measure the marginal results of a unfavorable 1% Magazine-7 fairness value shock on   enterprise days after the earnings announcement.

Focusing first on the fairness market, Chart 2 traces the response () of the S&P 500 and the FTSE 100 to a -1% Magazine-7 fairness value shock. Importantly, for the reason that underlying Magazine-7 inventory value modifications are weighted by every agency’s measurement within the S&P 500, a -1% shock mechanically implies a 1% fall within the S&P 500 index on affect. In Panel (a) of Chart 2, we see that the S&P 500 declines by practically 2% to the shock, implying vital spillovers to different corporations within the index as properly. These spillovers are of comparable magnitude to the mechanical impact. Relatedly, inside two days, the FTSE 100 declines by 1% as properly regardless of no mechanical impact, showcasing that fairness indices exterior the US, on this case the UK, are considerably affected by information intrinsic to main US tech corporations. Each results are comparatively persistent as properly.

Chart 2: Fairness-market response to a -1% Magazine-7 fairness value shock

Notes: The chart plots the response of the S&P 500 (left) and FTSE 100 (proper) to a -1% Magazine-7 fairness value shock, as described in Equation (1) and estimated in keeping with Equation (2). 68% (blue) and 90% (gentle blue) confidence intervals constructed with Newey-West commonplace errors.

Supply: Workers estimates.

Turning to trade charges, Chart 3 plots the responses of the greenback and the pound nominal efficient trade charges (NEER), which transfer reverse to 1 one other. We estimate {that a} -1% Magazine-7 fairness value shock depreciates the greenback on affect, with a peak efficient depreciation of 0.3% per week after the earnings announcement. Sterling, then again, appreciates in efficient phrases by about 0.5% over the same timeframe.

Total, together with the sooner fairness value responses, this USD depreciation means that unfavorable US tech-sector earnings information leads traders to pivot away from US equities, depreciating the greenback by outweighing another flight-to-safety dynamics into US authorities bonds. Whereas our identification technique imposes no restrictions on the underlying mixture of structural shocks driving our Magazine-7 fairness surprises, these outcomes are in line with traders decoding unfavorable US big-tech earnings information as downward revisions to future US productiveness (Chahrour et al (2024)).

Chart 3: Change price response to a -1% Magazine-7 fairness value shock

Notes: The chart plots the response of the USD NEER (left) and GBP NEER (proper) to a -1% Magazine-7 fairness value shock, as described in Equation (1) and estimated in keeping with Equation (2). 68% (blue) and 90% (gentle blue) confidence intervals constructed with Newey-West commonplace errors.

Supply: Workers estimates.

Searching for these flight-to-safety dynamics, we examine the responses of US and UK 10-year authorities bond yields. Panels (a) and (b) of Chart 4 present that yields in each jurisdictions fall by a few foundation factors on affect, in step with a muted flight-to-safety or a decline in combination demand, earlier than rising, though this rebound shouldn’t be statistically vital. An increase in yields following a unfavorable Magazine-7 earnings information is in line with new proof by Andrews and Farboodi (2026) on the discharge of latest AI fashions, which is known to be excellent news about future productiveness. Lustig et al (2026) argue that that is the results of US fiscal sustainability growing in productiveness, which means that US authorities bond holders are successfully lengthy AI. These offsetting channels could clarify the general muted response on bond markets.

Curiously, Panel (c) of Chart 4 reveals that 10-year breakeven inflation charges meaningfully decline within the US following unfavorable Magazine-7 earnings information, doubtlessly indicating vital demand-side results to the shock. Because of this, US 10-year actual yields rise. The response within the UK, nevertheless, is extra attenuated on affect, however grows over time.

Chart 4: Authorities-bond market response to a -1% Magazine-7 fairness value shock

Notes: The chart plots the response of US 10-year sovereign nominal yields (prime left), UK 10-year sovereign yields (prime proper), US 10-year breakeven inflation (backside left) and UK 10-year breakeven inflation (backside proper) to a -1% Magazine-7 fairness value shock, as described in Equation (1) and estimated in keeping with Equation (2). 68% (blue) and 90% (gentle blue) confidence intervals constructed with Newey-West commonplace errors.

Supply: Workers estimates.

Lastly, Chart 5 plots the responses of US and UK credit score unfold indices to the unfavorable Magazine-7 fairness value shock. US credit score spreads rise on affect, with the impact rising to between 5 and 10 foundation factors after three weeks. The outcomes are related for the UK spreads, though the on-impact response is extra muted. Total, these findings showcase that big-tech fairness value shocks spillover to company credit score borrowing charges as properly.

Chart 5: Company credit score market response to a -1% Magazine-7 fairness value shock

Notes: The chart plots the response of US (left) and UK credit score spreads to a -1% Magazine-7 fairness value shock, as described in Equation (1) and estimated in keeping with Equation (2). 68% (blue) and 90% (gentle blue) confidence intervals constructed with Newey-West commonplace errors.

Supply: Workers estimates.

Conclusion

The present AI fairness growth has been characterised by expectations of higher US productiveness progress. On this publish, we have now explored how an AI correction may unfold, leveraging a novel high-frequency fairness value shock across the earnings bulletins of the magnificent-7 US tech firms.

These shocks present proof that what originates in Massive Tech doesn’t keep confined to Massive Tech. On unfavorable Magazine-7 earnings information, fairness markets decline, credit score spreads widen, the US greenback depreciates, and bond yields are muted, suggesting {that a} downward revision to US productiveness outweighs flight-to-safety dynamics. These findings have two coverage implications. First, vital cross-border spillovers to fairness and credit score markets indicate that policymakers shouldn’t depend on corrections to elevated US tech-sector valuations remaining confined to the US. Second, a USD depreciation may exacerbate the macroeconomic results of those spillovers, growing the draw back dangers to an AI crash globally. Specifically, international economies have traditionally benefitted from a USD appreciation in instances of stress when it comes to export competitiveness and a internet wealth switch that helps offset the valuation losses on their USD asset holdings. Accounting for a possible USD depreciation is subsequently paramount for policymakers when contemplating AI crash situations.

Daniel Ostry, Roger Vicquéry and Emilio Zaratiegui work within the Financial institution’s International Evaluation Division.

If you wish to get in contact, please e mail us at bankunderground@bankofengland.co.uk or depart a remark beneath.

Feedback will solely seem as soon as accepted by a moderator, and are solely revealed the place a full title is equipped. Financial institution Underground is a weblog for Financial institution of England workers to share views that problem – or help – prevailing coverage orthodoxies. The views expressed listed below are these of the authors, and will not be essentially these of the Financial institution of England, or its coverage committees.