AI Boosts Market Efficiency But Raises 'Tail Risk,' Bernstein Says

Artificial intelligence is making financial markets more efficient but is also introducing new vulnerabilities that could lead to more severe volatility, according to a research note from Bernstein. The analysis suggests a fundamental trade-off for investors: lower average market inefficiency in exchange for greater tail risk.

How AI Improves Market Efficiency

At the company level, AI is accelerating the processing of vast amounts of data, including earnings reports, regulatory filings, and alternative datasets, the note explains. This allows for faster analysis than humanly possible, which helps narrow information gaps, improve price discovery, and reduce the size of earnings surprises as analyst forecasts converge more quickly with actual results.

Beyond faster processing, AI is also reshaping the research process. Bernstein highlights that automated workflows are cutting the time needed for earnings reviews and financial model updates. This allows analysts to expand their coverage, with the note pointing to a sharp rise in research on emerging market small-cap stocks over the past year, a trend that could erode the valuation premium often associated with under-researched companies.

New Vulnerabilities and Systemic Risks

Despite the benefits, Bernstein warns that the widespread adoption of similar AI models creates new dangers. As more investors rely on comparable tools trained on overlapping data, trading signals could become increasingly synchronized, leading to crowded positions and making market reversals more severe during periods of stress.

Ad

The note pointed to recent market events as illustrations of these risks, including the August 2024 unwind of the yen carry trade and instances of AI-generated misinformation that briefly moved U.S. equities. These examples show how automated strategies can accelerate volatility before markets have time to verify new information.

The 'Reflexivity Problem'

Analysts also describe a potential "reflexivity problem," where AI-generated recommendations increasingly influence investor behavior, pushing prices in ways that reinforce the models' future outputs. This feedback loop could strengthen momentum trades, increase market concentration in a handful of assets, and produce larger valuation extremes.

Taken together, Bernstein's findings suggest AI is likely to reduce average market inefficiencies through better research and execution. However, this comes at the cost of increasing the likelihood of larger, though less frequent, market dislocations during periods of high stress.