Print
E-MailS. Sean Cao is Assistant Professor of Accountancy at Georgia State University J. Mack Robinson College of Business; Junbo L. Wang is Assistant Professor of Finance at Louisiana State University E. J. Ourso College of Business; and Baozhong Yang is the H. Talmage Dobbs Jr. Associate Professor of Finance at Georgia State University J. Mack Robinson College of Business. This post is based on a recent paper authored by Mr. Cao, Mr. Wang, Mr. Yang, and Wei Jiang, Arthur F. Burns Professor of Free and Competitive Enterprise at Columbia Business School.
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 by those currently performed by the better-paid and better-educated workers. The existing literature has mostly focused on characterizing the type of jobs that are vulnerable to disruption by, as well as those that could be created due to, AI evolution. In other words, the sentiment of the existent studies mostly involves a theme of “man-versus-machine,” i.e., to characterize the contest between human and AI, to explore ways human adapts, and to predict the resulting job redeployment. There has been relatively little research devoted to prescribing how skilled human workers could tap into a higher potential with enhancement from AI technology, presumably the primary goal for human beings to design and develop AI in the first place. In this study, we aim to connect the contest of “man-versus-machine” (“man v. machine” hereafter) to a potential equilibrium of “man-plus-machine” (“man + machine” hereafter) into the profession of stock analysis. The choice of the setting is primarily motivated by data availability and well-defined performance metrics. However, the inferences from this study apply broadly to many high-skilled professions.
As a first step, we build our “AI analyst” by training a combination of current machine-learning (ML) tool kits using timely publicly available data and information. More specifically, we collect firm-level, industry-level, and macro-economic variables, as well as textual information from firms’ disclosure (updated to right before the time of an analyst forecast) as inputs or predictors, but deliberately exclude information from analyst forecasts (past and current) themselves. We kept training and improving the model until we were confident that our AI analyst is able to beat human analysts as a whole: The AI analyst based on the final “ensemble” model outperforms 53.7% of the target price predictions made by all IBES analysts during the sample period of 2001-2016. Moreover, a monthly rebalanced long-short portfolio based on the differences in the opinions of AI and human analysts is able to generate a monthly risk-adjusted return of 0.84% to 0.92%. Though we do not claim our AI analyst to be the best of the kind, its performance already suggests that the profession of financial analysts is subject to technology disruption as our model is a lower bound of the state-of-the-art.
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 ex-post realization (i.e., the actual year-end stock price). We find human analysts perform better for more illiquid, smaller firms, and firms with asset-light business models (i.e., 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. Analysts affiliated with large brokerage houses also stand a higher chance of beating the machine, thanks to a combination of their abilities and the research resources available to them. Moreover, analysts are more likely to have the upper hand when the associated industry is experiencing distress, suggesting that the AI has yet to catch up on relatively infrequent changes such as an industry recession. This is consistent with the limitation of current machine learning 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 out-smart 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, i.e., an analyst who makes forecasts that combine their own knowledge and the outputs/recommendations from AI models. After we add analyst forecasts to the information set of the machine learning models underlying our AI analyst, the resulting “man + machine” model outperforms 57.3% of the forecasts made by analysts and outperforms the AI-only model in all years. Thus, AI analyst does not displace human analysts yet; and 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. Importantly, the incremental value of human does not decrease as the volume of information (hence the demand for processing capacity) increases, though this constitutes a human disadvantage when alone. Similarly, analysts from small brokerage houses make a similar level of contribution to the man + machine model compared to their counterparts from larger banks, suggesting that AI could potentially help level the disparity in institutional resources.
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 into investment opportunities. An important and popular type of alternative data captures “consumer footprints,” oftentimes in the literal sense, such as satellite images on retail parking lots. Such data, which have to be processed by machine learning models, have been shown to contain incremental information for stock prices. We build on the staggered 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. The underlying premise is that analysts who cover firms that are served by the alternative data could be in the situation of man + machine, as they have the opportunity to use the additional, AI-processed information. Indeed, we find that post alternative data, analysts covering affected firms improve their performance relative to the AI-only forecast model we build. Furthermore, such improvement concentrates in the subset of analysts who are affiliated with brokerage firms with strong AI capabilities, measured by AI-related hiring using the Burning Glass U.S. job posting data. Overall, our results support the hypothesis that analyst capabilities could be augmented by AI, and moreover, analysts’ work possesses incremental value such that they, with the assistance of AI, can still beat a machine model without 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 incoming future.
The complete paper is available for download here.
