Prasanna (Sonny) Tambe

Johann D. Gaebler, Sharad Goel, Aziz Huq, Prasanna Tambe (Forthcoming), Auditing the Use of Language Models to Guide Hiring Decisions.

Abstract: Regulatory efforts to protect against algorithmic bias have taken on increased urgency with rapid advances in large language models (LLMs), which are machine learning models that can achieve performance rivaling human experts on a wide array of tasks. A key theme of these initiatives is algorithmic "auditing," but current regulations -- as well as the scientific literature -- provide little guidance on how to conduct these assessments. Here we propose and investigate one approach for auditing algorithms: correspondence experiments, a widely applied tool for detecting bias in human judgements. In the employment context, correspondence experiments aim to measure the extent to which race and gender impact decisions by experimentally manipulating elements of submitted application materials that suggest an applicant's demographic traits, such as their listed name. We apply this method to audit candidate assessments produced by several state-of-the-art LLMs, using a novel corpus of applications to K-12 teaching positions in a large public school district. We find evidence of moderate race and gender disparities, a pattern largely robust to varying the types of application material input to the models, as well as the framing of the task to the LLMs. We conclude by discussing some important limitations of correspondence experiments for auditing algorithms.

Prasanna Tambe (Forthcoming), Reskilling for the AI Age: How Firms Transform Work for Algorithmic Decision-Making.

Prasanna Tambe and Tiantian Yang (2024), The Hidden Cost of IT Innovation: Access to Emerging Technologies and the Gender Wage Gap, MIS Quarterly.

Abstract: Although an extensive information systems (IS) literature has explored the economic benefits of information technology (IT) investments, how these benefits are distributed between male and female IT workers has not been as closely examined. This paper bridges this gap by addressing two questions: (1) do women and men have similar opportunities to acquire skills related to new IT innovations, and (2) how do these differences affect the gender pay gap? We argue that women are underrepresented in roles that use emerging technologies due to two interrelated processes: demand-side labor market conditions and supply-side job sorting. By analyzing two independent datasets that provide insights into wages, career trajectories, and skill prerequisites for IT roles, we find that women are less likely to apply for jobs requiring expertise with emerging technologies. Such positions often require extended work hours, frequent job mobility, and geographic relocation—which can conflict with the family responsibilities typically assumed by women. Yet, because these positions tend to offer higher wages, women’s underrepresentation in these roles exacerbates the gender pay gap in the IT sector. Our findings stress the importance of creating more flexible job structures and enhancing women’s access to emerging technology roles as critical for achieving gender equity in the IT industry.

David Hsu and Prasanna Tambe (2024), Remote Work and Job Applicant Diversity: Evidence from Technology Startups, Management Science.

Abstract: A significant element of managerial post-COVID job design regards remote work. In an era of renewed recognition of diversity, equity and inclusion, employers may wonder how diverse (gender and race) and experienced job applicants respond to remote job listings, especially for high-skilled technical and managerial positions. Prior work has shown that while remote work allows employee flexibility, it may limit career promotion prospects, so the net effect of designating a job as remote-eligible is not clear from an applicant interest standpoint, particularly when recruiting females and underrepresented minorities (URM). We analyze job applicant data from a leading startup job platform that spans long windows before and after the COVID-19 pandemic-induced shutdowns of March 2020. To address the empirical challenge that remote job designation may be co-determined with unobserved job and employer characteristics, we leverage a matching approach (and an alternative method which leverages the sudden shutdowns) to estimate how applicant characteristics differ for otherwise similar remote and onsite job postings. We find that offering remote work attracts more experienced and diverse (especially URM) job applicants, with larger effects in less diverse geographic areas. A discrete change in job posting to remote status (holding all else constant) is associated with an approximately 15% increase in applicants who are female, 33% increase in applicants with URM status, and 17% increase in applicant experience. Using the application data, we estimate an average estimated compensating wage differential for remote work that is about 7% of posted salaries in this labor market.

Po-Hsuan Hsu, Dokyun Lee, Prasanna Tambe, David Hsu (Working), Deep Learning, Text, and Patent Valuation.

Abstract: This paper uses deep learning and natural language processing (NLP) methods on the US patent corpus to evaluate their predictive power in estimating two measures of patent value: (i) investor reaction to patent announcements as measured in Kogan et al., 2017 and (ii) forward citations. While forward citations have traditionally been used as measures of economic value in the literature, their utility is mainly retrospective. Contemporaneous predictions of patent value, as embodied in investor reactions to patent grants, can be important for managers and policy-makers for prospective decision making. We compare the prediction performance of models using the structured features of the patent (number of citations, technology class, etc.) to deep learning and NLP methods. Relative to linear regression models using the same features, deep learning models reduce mean absolute error (MAE) by approximately 32%. Incorporating patent text further lowers the MAE by 13%.

Prasanna Tambe and John Horton (2020), The Death of a Technical Skill, Information Systems Research.

Abstract: In 2010, Steve Jobs announced that Apple would no longer support Adobe Flash—a popular set of tools for creating Internet applications. After the announcement, the use of Flash declined precipitously. We show there was no reduction in Flash hourly wages because of a rapid supply response: Flash specialists, especially those who were younger, were less specialized, or had good “fall back” skills quickly transitioned away from Flash; new market entrants also avoided Flash, leaving the effective supply per Flash job opening unchanged. As such, there was no factor market reason for firms to stay with Flash longer.

Prasanna Tambe, Lorin M. Hitt, Daniel Rock, Erik Brynjolfsson (Under Review), Digital Capital and Superstar Firms.

Abstract: General purpose technologies like information technology typically require complementary firm-specific human and organizational investments to create value. These complementary investments produce a form of capital, which we call Information technology-related intangible capital ("ITIC''). An understanding of how the accumulation of ITIC contributes to economic growth and differences among firms has been hindered by the lack of measures of the stock of ITIC. We use a new, extended firm-level panel on IT labor investments along with Hall’s Quantity Revelation Theorem to construct measures of both the prices and quantities of ITIC over the last thirty years. We find that 1) prices vary significantly for ITIC, 2) significant quantities of ITIC have been  accumulating since the 1990s, with ITIC accounting for at least 25% of firms’ assets by the end of our panel, 3) that it has disproportionately accumulated in small subset of high-value, superstar firms, and 4) that the accumulation of ITIC predicts future productivity.

Prasanna Tambe, Xuan Ye, Peter Cappelli (2020), Paying to Program? Engineering Brand and High-Tech Wages, Management Science.

Abstract: We test the hypothesis that information technology (IT) workers accept a compensating differential to work with emerging IT systems and that employers that invest in these systems can, in turn, capture greater value from the wages they pay. We show that much of the utility IT workers derive from these systems is from skills acquired on the job. This is principally true for younger workers at employers where skill development is encouraged, and the effects are stronger in thicker markets where workers with newer skills have more outside options. An analysis of the text in online employer reviews supports the notion that IT workers value access to interesting IT systems above most other employer attributes. These findings are important because (1) they provide evidence of how worker preferences can influence corporate IT investment decisions, (2) they shed light on factors influencing IT skill development, and (3) they point to a potentially important explanation for returns from IT investments.

Prasanna Tambe, Peter Cappelli, Valery Yakubovich (2019), Artificial Intelligence in Human Resources Management: Challenges and a Path Forward, California Management Review.

Abstract: There is a substantial gap between the promise and reality of artificial intelligence in human resource (HR) management. This article identifies four challenges in using data science techniques for HR tasks: complexity of HR phenomena, constraints imposed by small data sets, accountability questions associated with fairness and other ethical and legal constraints, and possible adverse employee reactions to management decisions via data-based algorithms. It then proposes practical responses to these challenges based on three overlapping principles—causal reasoning, randomization and experiments, and employee contribution—that would be both economically efficient and socially appropriate for using data science in the management of employees.

Ariel C. Avgar, Prasanna Tambe, Lorin M. Hitt (2018), Built to Learn: How Work Practices Affect Employee Learning During Healthcare Information Technology Implementation, MIS Quarterly.

Abstract: We test the hypothesis that work practices complement IT investment, in part, by accelerating how rapidly employees acquire the skills needed to use new IT systems. We combine support request data from an EMR vendor with survey responses on work practices from 962 employees from 15 client nursing homes. Nursing homes using work practices that prior studies have shown to complement IT investment—those promoting discretion, teamwork, training, high staffing levels, and communication—experienced more rapid declines in requests for technical support. We then show that these benefits are due, in part, because the use of these work practices facilitates learning for workers in frontline occupations who otherwise may not have the freedom to experiment with and adapt the new technology systems. For many frontline workers, discretion was more important than training in explaining IT learning. Implications for the healthcare industry are discussed.