The Impact of Artificial Intelligence on Firm Growth, Product Innovation, and Operating Efficiency

Tania Babina, Anastassia Fedyk, Alex He, and James Hodson, authors of the study “Artificial Intelligence, Firm Growth, and Product Innovation,” published in the January 2024 issue of the Journal of Financial Economics, developed a new measure of firm-level AI investments to determine whether AI can transform economies and spur economic growth. They used a combination of datasets that capture both the stock of and the demand for AI-skilled employees among U.S. firms: resume data from Cognism Inc., which offer job histories for 535 million individuals globally (covers approximately 64% of the entire U.S. workforce as of 2018 with a representative breakdown across industries), and job postings data from Burning Glass, which capture 180 million job vacancies. Their AI measure allowed them to analyze the patterns of AI adoption and examine its potential benefits for the adopting firms and industries.

Explaining their process: “Our algorithm learns the AI-relatedness of each job empirically from the detailed skills of the job postings. First, we measure the AI-relatedness of each skill in the job postings data, based on that skill's co-occurrence with the core AI skills—machine learning, computer vision, and natural language processing. Second, we obtain a measure of AI-relatedness of each job posting by averaging the AI-relatedness of all skills required by the job posting. Finally, we leverage the most AI-related skills identified from the job postings data to classify AI workers in the less structured resume data. For each employee, we consider whether skills with the highest AI-relatedness (e.g., “deep learning”) appear either in the job title, in the job description, or in any publications, patents, or awards received during that job. This gives us a classification of each employee of each firm at each point in time. We aggregate both the resume data and the job postings data to the firm level and match to public firms in the Compustat database. Encouragingly, the two measures of AI investments, although based on two independent datasets, are highly correlated and yield consistent results.” Their data sample covered the period 2010-2018. Following is a summary of their key findings:

· In both employee resume and job postings datasets, the fraction of AI jobs has increased dramatically over time, growing more than seven-fold from 2010 to 2018, from 0.04% in 2007 to 0.29% in 2018.

· The share of AI jobs is highest in the technology sector, but the rate of increase in AI investments over time is similar across sectors.

· AI-investing firms experience higher growth in sales, employment, and market valuations.

· A one-standard-deviation in AI investment was associated with a 19.5% increase in sales, an 18.1% increase in employment and a 22.3% increase in market valuation. The results were ubiquitous across major industry sectors (e.g., manufacturing, finance, and retail), supporting the idea that AI is a general-purpose technology.

· Faster growth comes primarily through increased product innovation, both new and improved products. Firms that adopted AI showed more product patents, trademarks, and more updates to their product portfolios than those that do not—AI can reduce the costs of product innovation, which improves the quality of existing products and allows firms to create new products.

· Growth was not due to operational efficiencies. There was no evidence that operating costs can be lowered by either improving the production process or replacing human labor with AI methods—there was no evidence that AI was used to improve firm efficiency by automating tasks.

· Firms that were connected to universities that produce a high number of AI graduates were more likely to adopt AI technologies—firms with connections to AI-universities are at an advantage.

· The effects of AI adoption materialized after 2-3 years—patience is required as the learning curve is of a gradual nature.

· As industry sales and employment increased, so did industry concentration. The result is that it is more than likely that the benefits of AI will disproportionately accrue to larger firms reinforcing a “winner takes all, or almost all” competitive environment.

Their findings led Babina, Fedyk, He, and Hodson to conclude: “Our results highlight that new technologies like AI can contribute to growth and superstar firms through product innovation.”

Their findings raise the question of whether investors can exploit this information now that they are public. David McLean and Jeffrey Pontiff, authors of the 2016 study “Does Academic Research Destroy Stock Return Predictability?” found that anomaly returns declined by about 60% post-publication.

As Andrew Berkin and I explained in our book, The Incredible Shrinking Alpha, another important issue is that while IBM’s Watson can outwit individuals (even champions in their fields, e.g., chess), individuals are not the competition when it comes to investing. Instead, Watson competes with the collective wisdom of the millions of individuals trading in stock markets each day – a much tougher competitor. In addition, as soon as new information (such as the publication of a paper revealing a profitable anomaly) is obtained, the process of acting on that information gets incorporated into market prices very quickly. Today, the competition is the collective decision making of not just humans but also machines, algorithms, and algorithms predicting what other algorithms will do next. Thus, any successful AI strategy is likely to be short lived. A perfect example of this type of evolution is that in their December 2023 paper “ Textual Analysis by Hedge Funds,” authors Sipeng Zeng and Kuo Zhou found that after the 2018 release of Google’s BERT model, machine-download funds immediately began to adjust their positions based on the sentiment index constructed by BERT.

Providing further insights into the issue, Dimensional explained in a research note: “Active investors have long attempted to get an informational edge on markets by using artificial intelligence (AI) processes to retrieve and process data. For example, tools that gauge sentiment from social media or scrape text from company financial reports predate ChatGPT by many years. Material information gleaned from running AI processes is very likely a subset of the vast information set known by the market in aggregate and reflected in market prices. If new information is obtained, the process of acting on that information incorporates it into market prices. Another reason to question AI’s role in helping with market timing is limitations with its predictions. AI’s forecasting ability fares well when assessing patterns that are relatively stable. The market is fantastically complex. So much so that no one knows exactly how much a particular piece of information impacts a price, because there are so many other simultaneous inputs. AI trying to predict market prices is like self-piloting cars trying to read stop signs with words, shapes, and colors that differ every day.”

Before summarizing, there is a major concern we need to address related to the use of AI tools such as ChatGPT: Look-ahead bias. These tools are trained on a corpus (a collection of writings or texts) by feeding in past writing. Thus, when research shows great historical results using ChatGPT, it could be that it simply knows what will happen since that was in its training set, rather than because of some great pattern recognition ability.

Investor takeaways

An interesting, and surprising takeaway from the research is that AI technologies have benefited firms through product innovations rather than through reductions in operating expenses or improvements in productivity. Babina, Fedyk, He, and Hodson did note that their results contrasted with previous literature finding that “IT investments were associated with economically large productivity increases but mixed results on firm growth measures such as market share.” With that said, the authors cautioned: “Our results speak to the early wave of AI adoption, and efficiency gains, if present, may be more backloaded.”

There is no doubt that artificial intelligence has changed the way financial institutions construct portfolios and trade. With that said, the collective wisdom of the market is a powerful force that ensures that the price quoted is the best estimate of the value of a security given its risk. The market also quickly adapts to new strategies (the adaptive markets hypothesis) with competition minimizing, if not eliminating, excess profit opportunities. Thus, there’s no reason to think that the use of AI should lead to persistent investment outperformance, with any advantages gained likely being short lived.

Larry Swedroe is the author or co-author of 18 books on investing, including his latest Enrich Your Future.