COMPUTATIONAL SOCIAL SCIENCE WORKSHOP
Abstract: In this talk, I will discuss three papers that apply the tools of natural language processing to develop novel, time-varying measures of organizational culture and demonstrate their relationship to consequential outcomes. The first paper develops measures of cultural heterogeneity based on employee reviews of nearly 500 publicly traded firms on a leading company review website and shows how they relate to firms’ innovation success, profitability, and market valuation. The second paper uses a word embedding model applied to a corpus of corporate quarterly earnings calls to develop measures of performative atypicality–the extent to which organizational leaders’ identity performances for external audiences deviate from normative expectations. Performative atypicality breeds ambiguity, which increases the standard deviation of financial analysts’ forecasts of future performance. Through the mechanism of motivated reasoning, this ambiguity can in turn lead to a positive bias in analyst forecasts and portend a negative earnings surprise. The third paper uses Bidirectional Encoder Representations from Transformers (BERT) to develop a novel approach to identifying the visionaries in a strategic field. We apply BERT to the corpus of quarterly earnings calls to derive a measure of prescience–the tendency of firms to communicate in ways that are novel at a given point in time but that become commonplace in the future. As one might expect, prescience is positively related to future stock returns. Yet, contrary to theories of incumbent advantage, small firms are more likely to exhibit prescience than are large firms.
Sameer B. Srivastava is Associate Professor and Harold Furst Chair in Management Philosophy and Values at UC Berkeley’s Haas School of Business and is also affiliated with UC Berkeley Sociology. His research unpacks the complex interrelationships between the culture of social groups, the cognition of individuals within these groups, and the connections that people forge within and across groups. Much of his work is set in organizational contexts, where he uses computational methods to examine how culture, cognition, and networks relate to career outcomes and organizational success. His work has been published in journals such as American Journal of Sociology, American Sociological Review, Administrative Science Quarterly, Management Science, and Organization Science. Sameer co-directs the Berkeley-Stanford Computational Culture Lab and the Berkeley Culture Initiative. He holds a PhD in Sociology and Organizational Behavior from Harvard University.
This virtual workshop is open to the Yale community. To receive Zoom information, you must subscribe to the Computational Social Science Workshop at this link: https://csap.yale.edu/computational-social-science-workshop.
This workshop is cosponsored by the Center for the Study of American Politics (CSAP) and the Yale School of Management (SOM) with support from the Initiative for Leadership and Organization at SOM.