A new study from the National Bureau of Economic Research reveals that recent advances in generative AI are accelerating the divergence in stock market performance between technology-focused and traditional companies.
Firms with high labor exposure to AI saw substantially higher returns after ChatGPT’s release in late 2022 versus those still focused on manual human tasks.
The working paper, published by Andrea L. Eisfeldt, Gregor Schubert and Miao Ben Zhang, analyzes the return premium for portfolios sorted by firm-level generative AI exposure. This novel exposure measure looks at the proportion of occupational categories prone to automation based on each company’s workforce composition.
From November 2022 to March 2023, the study found high AI exposure stocks earned a sizable 0.4% average daily return premium over low exposure stocks.
Many tech and information-based companies already sit at the high end of AI exposure, whereas sectors like retail, transportation, logistics and traditional manufacturing remain on the lower end.
Generative AI, powered by models like OpenAI’s GPT LLM, appear poised to materially widen the return gap between both sets of companies going forward.
Within industries, smaller high-growth firms with more focus on intangible assets and R&D also tend to have greater exposure, meaning the technology disproportionately benefits companies already attracting investor attention.
The launch of ChatGPT itself, from OpenAI, represented a major shock to expectations about AI’s potential, driving a 0.43% return premium during its release window for high versus low exposure portfolios. Investors rapidly incorporated updated views on the transformative capabilities of generative AI into valuations across the market.
On average, the study shows exposed firms saw over 100% higher annualized returns than unexposed firms after ChatGPT’s release. This underscores how rapidly expectations are shifting.
The researchers say generative AI promises to automate many tasks involving the production and synthesis of text, images, code and other content. This will have a broad impact on firms whose workers focus on non-routine cognitive tasks. The technology was frequently discussed in earnings calls for exposed industries.
While AI could boost productivity for some firms, the study highlights it may also disrupt industries like transportation, finance, and real estate that involve significant manual work. The exposure measure developed provides a valuable tool for tracking the differential impact of AI across sectors.
Read the full article here
