Since its inception and as it rises, artificial intelligence (AI) constantly makes human beings rethink their own roles. Concerns abound that AI could replace human tasks – and increasingly skilled ones – and thus displace jobs currently performed by the better-paid and better-educated workers. This concern becomes more significant with the advent of newer and better AI technologies, including those from Google and Open AI.
But there has been relatively little research devoted to prescribing how skilled human workers could tap into a higher potential with enhancement from AI technology. In a recent study, we aim to connect the contest of ‘man versus machine’ to a potential equilibrium of ‘man plus machine’ regarding the profession of stock analysis. The choice of the setting is primarily motivated by data availability and well-defined performance metrics but the inferences from this study apply broadly to many highly skilled professions.
Building the AI analyst
As a first step, we build our AI analyst by training a combination of current machine-learning (ML) toolkits using timely, publicly available data and information. More specifically, we collect company-level, industry-level and macroeconomic variables, as well as textual information from company disclosures, social media and news as inputs or predictors, but deliberately exclude information from analyst forecasts (past and current) themselves.
The AI analyst trained with such data is able to beat human analysts as a whole: it outperforms 54.5 percent of the target price predictions made by all Institutional Brokers’ Estimate System analysts during the sample period of 2001-2018. Moreover, a monthly rebalanced long-short portfolio based on the differences in the opinions of AI and human analysts is able to generate an annual abnormal return of 8.08 percent to 14.19 percent.
Next, we examine the circumstances under which human analysts retain their advantage, in that a forecast made by an analyst beats the concurrent AI forecast in terms of lower absolute forecast error relative to the actual year-end stock price. We find human analysts perform better for more illiquid, smaller companies and firms with asset-light business models (such as higher intangible assets), consistent with the notion that such firms are subject to higher information asymmetry and require better institutional knowledge or industry experience to decipher.
Moreover, analysts are more likely to have the upper hand when the firm itself or its associated industry is experiencing distress, suggesting that the AI has yet to catch up on relatively infrequent changes such as company distress or industry recession. This is consistent with the limitation of current ML and AI models, which lack reasoning functions and thus cannot learn effectively from infrequent events. As expected, AI enjoys a clear advantage in its capacity to process information and is more likely to outsmart analysts when the volume of public information is larger.
The superior performance of an AI analyst does not rule out the value of human inputs. If human and machine have different relative advantages in information processing and decision-making, then human analysts may still contribute critically to a ‘hybrid’ analyst: one who makes forecasts that combine human knowledge with the outputs/recommendations from AI models.
Attaining the best performance
After we add analyst forecasts to the information set of the ML models underlying our AI analyst, the resulting man-plus-machine model outperforms 57.8 percent of the forecasts made by analysts and outperforms the AI-only model most years. Thus, the AI analyst does not displace human analysts yet; in fact, an investor or analyst who combines AI’s computational power and the human art of understanding soft information can attain the best performance.
Finally, we resort to an event study to sharpen the inference of the impact of integrating man and machine in stock analyses. In recent years, the infrastructure of Big Data has created a new class of information about companies that is collected and published outside of the firms, and such information provides unique and timely clues about investment opportunities.
An important and popular type of alternative data captures ‘consumer footprints’, often in the literal sense, such as satellite images of retail parking lots. Such data, which has to be processed by ML models, has been shown to contain incremental information for stock prices. We build on the introduction of several important alternative databases and conduct a test of analysts’ performance versus our own AI model before and after the availability of the alternative data. We find that post-alternative data availability, analysts covering affected companies improve their performance relative to the AI-only forecast model we build.
Further, such improvement is concentrated in the analysts affiliated with brokerage firms that have strong AI capabilities, measured by AI-related hiring by these firms. Overall, our results support the hypothesis that analyst capabilities could be augmented by AI and, moreover, that analysts’ work possesses incremental value such that they, with the assistance of AI, can still beat a machine model that doesn’t have human inputs.
If there is some external validity from stock analysis to skilled workers in general, the inference from our study is encouraging news for humans in the age of AI. The complementarity between humans and machines documented in this study also provides guidance about how humans can adapt to survive and thrive in the age of machines. For example, reforming education and professional training to strengthen soft skills and creativity can help human professionals to better prepare for the future.
Sean Cao
Wei Jiang
Junbo Wang
