In a wide variety of applications, including healthcare, bidding in first price auctions, digital recommendations, and online education, it can be beneficial to learn a policy that assigns treatments to individuals based on their characteristics. The growing policy-learning literature focuses on settings in which policies are learned from historical data in which the treatment assignment rule is fixed throughout the data-collection period. However, adaptive data collection is becoming more common in practice from two primary sources: (1) data collected from adaptive experiments that are designed to improve inferential efficiency and (2) data collected from production systems that progressively evolve an operational policy to improve performance over time (e.g., contextual bandits). Yet adaptivity complicates the problem of learning an optimal policy ex post for two reasons: first, samples are dependent and, second, an adaptive assignment rule may not assign each treatment to each type of individual sufficiently often. In this paper, we address these challenges. We propose an algorithm based on generalized augmented inverse propensity weighted (AIPW) estimators, which nonuniformly reweight the elements of a standard AIPW estimator to control worst case estimation variance. We establish a finite-sample regret upper bound for our algorithm and complement it with a regret lower bound that quantifies the fundamental difficulty of policy learning with adaptive data. When equipped with the best weighting scheme, our algorithm achieves minimax rate-optimal regret guarantees even with diminishing exploration. Finally, we demonstrate our algorithm’s effectiveness using both synthetic data and public benchmark data sets. This paper was accepted by Hamid Nazerzadeh, data science. Funding: This work is supported by the National Science Foundation [Grant CCF-2106508]. R. Zhan was supported by Golub Capital and the Michael Yao and Sara Keying Dai AI and Digital Technology Fund. Z. Ren was supported by the Office of Naval Research [Grant N00014-20-1-2337]. S. Athey was supported by the Office of Naval Research [Grant N00014-19-1-2468]. Z. Zhou is generously supported by the New York University’s 2022–2023 Center for Global Economy and Business faculty research grant and the Digital Twin research grant from Bain & Company. Supplemental Material: The data files are available at https://doi.org/10.1287/mnsc.2023.4921 .

Online platforms such as preprint servers have become an important way to disseminate new scientific knowledge prior to peer review. However, little is known about how attention to preprints may vary across authors from different countries of origin, particularly relative to evaluation in expert-controlled systems such as scientific journals. This study explores how readers allocated attention across preprints in the initial months of the COVID-19 pandemic, a time when there was an increase in demand for new research and a corresponding increase in the use of preprint platforms around the world. We find that, after controlling carefully for article quality and topic, as well as the prominence of the preprint’s ultimate publication outlet, preprints with authors from Chinese institutions receive less attention, and preprints with authors from U.S. institutions receive more attention, than preprints with authors from the rest of the world. In an exploration of potential mechanisms driving the observed effects, we find evidence that when evaluation is more constrained, in terms of lack of knowledge or expertise and increase in time pressure, audiences tend to make greater use of preprint authors’ country of origin as a proxy for quality or relevance. The results suggest that geographic biases may persist or even be exacerbated on platforms designed to promote unfettered access to early research findings. This paper was accepted by Toby Stuart, entrepreneurship and innovation. Supplemental Material: The data files and online appendix are available at https://doi.org/10.1287/mnsc.2023.4936 .

Emotional involvement is known to be necessary but not sufficient for good decision making in the face of uncertainty. It has been conjectured that emotional engagement in anticipation of risky outcomes constitutes “good” emotions. We introduce a new methodology to determine whether anticipatory emotional engagement is beneficial in the context of trading in financial markets. We focus on heart rate changes because they occur at a sufficiently high frequency to discern timing relative to events in the marketplace. After conservatively adjusting for multiple hypothesis testing, we find that participants whose heart rate changes anticipate their order submissions at inflated prices earn significantly more, whereas participants whose heart rate responds to their trades earn significantly less. By investigating cointegration between skin conductance response and the dynamics of individual portfolio values, we confirm the importance of emotional involvement in determining who makes or loses money. This paper was accepted by Bruno Biais, finance. Funding: This work was supported by the Australian Research Council [Grants DP180102284 and LE130100112]. The authors also thank the Monash Business School for funds associated with the Monash Business Behavioural Laboratory facility and equipment used in this study. Supplemental Material: The data files and online appendix are available at https://doi.org/10.1287/mnsc.2023.4883 .

As artificial intelligence (AI) applications become more pervasive, it is critical to understand how knowledge workers with different levels and types of experience can team with AI for productivity gains. We focus on the influence of two major types of human work experience (narrow experience based on the specific task volume and broad experience based on seniority) on the human-AI team dynamics. We developed an AI solution for medical chart coding in a publicly traded company and conducted a field study among the knowledge workers. Based on a detailed analysis performed at the medical chart level, we find evidence that AI benefits workers with greater task-based experience, but senior workers gain less from AI than their junior colleagues. Further investigation reveals that the relatively lower productivity lift from AI is not a result of seniority per se but lower trust in AI, likely triggered by the senior workers’ broader job responsibilities. This study provides new empirical insights into the differential roles of worker experience in the collaborative dynamics between AI and knowledge workers, which have important societal and business implications. This paper was accepted by Kartik Hosanagar, information systems. Funding: This work was supported by Inovalon [Sponsor of the Health Insights AI Laboratory]. Supplemental Material: The data files and online appendix are available at https://doi.org/10.1287/mnsc.2021.00588 .

When do sellers benefit from the presence of a strong buyer? In a second price auction with independent private values and entry costs for buyers, we allow for one buyer to be stochastically “strong.” We show that, when buyers make positive net profits, seller revenue is higher with a stronger buyer. In contrast, when competition is endogenous so that all buyers other than the strong buyer break even, seller revenue is always lower in the presence of a strong buyer. We also study seller actions such as providing buyer subsidies, setting reserve prices, securing multiple strong buyers, bundling assets, and optimally choosing the time of sale that might alleviate the fall in revenue. This paper was accepted by Will Cong, Finance. Supplemental Material: The online appendix and data are available at https://doi.org/10.1287/mnsc.2023.4938 .

In the high-stakes race for scalability, some blockchains have turned to committee-based consensus (CBC), whereby the chain’s recordkeeping rights are entrusted to a committee of block producers elected via approval voting. Smaller committees boost speed and scalability but can compromise security when voters have limited information. In this environment, voting strategies are naturally nonlinear, and equilibria can become intractable. Despite this, we show that elections converge to optimality asymptotically (in voter numbers), exponentially quickly, and under relatively weak informational requirements. Compared to popular stake-weighted lottery and single-vote protocols used in practice, we find that CBC, when paired with approval voting, can offer meaningful efficiency and robustness gains if enough voters are engaged. This paper has been accepted by Will Cong for the Management Science Special Issue on Blockchains and Crypto Economics. Funding: This work was funded by The Hogeg Blockchain Research Institute. Supplemental Material: The data files and online appendix are available at https://doi.org/10.1287/mnsc.2022.03177 .

Random coefficient multinomial logit models are widely used to estimate customer preferences from sales data. However, these estimation models can only allow for products with positive sales; this selection leads to highly biased estimates in long tail markets, that is, markets where many products have zero or low sales. Such markets are increasingly common in areas such as online retail and other online marketplaces. In this paper, we propose a two-stage estimator that uses machine learning to correct for this bias. Our method first uses deep learning to predict the market shares of all products, where the neural network’s structure mirrors the random coefficient multinomial logit model’s data generating process. In the second stage, we use the predictions of the first stage to reweight the observed shares in a way that corrects for the induced bias and maintains the causal interpretation of the structural model. We show that the estimated parameters are consistent in the number of markets. Our method performs well on simulated and real long tail data, producing accurate estimates of customer behavior. These improved estimates can subsequently be used to provide prescriptive policy recommendations on important managerial decisions such as pricing, assortment, and so on. This paper was accepted by Gabriel Weintraub, revenue management and market analytics. Supplemental Material: The data files are available at https://doi.org/10.1287/mnsc.2023.4893 .

Using breadth of corporate bond fund ownership to proxy for the combination of short-sale constraints and divergence of opinions in the corporate bond market, this paper investigates whether and to what extent Miller’s overpricing theory holds for corporate bonds. Consistent with the theory, we document a significant and positive relation between changes in breadth and the cross-section of future corporate bond returns. Because of the bounded upside of debt, the ability of changes in breadth to predict returns appears to be less pronounced for corporate bonds relative to stocks. Furthermore, this predictability is stronger among bonds with higher credit risk, indicating that the effect of short-sale constraints and divergence of opinions on bond mispricing tends to be larger for riskier bonds with more upside potential and, thus, more scope for overvaluation. This paper was accepted by Lukas Schmid, finance. Supplemental Material: The internet appendix and data are available at https://doi.org/10.1287/mnsc.2023.4950 .

Some nonprofit organizations (NPOs) manage a complex workforce composed of a mix of volunteers, part-time workers, and full-time workers. We study the NPO’s finite-horizon staffing problem to determine the optimal initial staff planning decisions and per period optimal hiring and assignment decisions given a budget, capacity constraints, and an uncertain supply of volunteers and part-time workers. Our main goal is to solve this problem in a way that is effective and easy to implement while obtaining interesting managerial insights. To this end, we first demonstrate that the optimal staffing policies are computationally challenging to identify in general. However, we demonstrate that a prioritization assignment policy and a hire-up-to policy for part-time workers can be conveniently applied and are close to optimal. These policies are, in fact, optimal under staff scarcity and staff sufficiency. In our numerical analysis, we study the value and impact of the general optimal solution that considers flexibility and turnover of part-time workers versus the prioritization assignment policy and a constant hire-up-to policy that omit flexibility and turnover behaviors. We further suggest two easy-to-implement heuristics and theoretically analyze them and run a numerical performance study. We observe that both heuristics have low relative optimality gaps. Finally, we extend our analysis by studying how the optimal policy varies under three different practical considerations: a concave social value objective, nonzero volunteer costs, and dynamic volunteer behaviors. This paper was accepted by Jayashankar Swaminathan, operations management. Funding: This work was supported by the Comunidad de Madrid [Grant EPUC3M12], the Ministerio de Ciencia e Innovación [Grant PID2021-127657NA-I00], and the Ramon y Cajal Fellowship [RYC2020-029303-I]. Supplemental Material: The data files and online appendices are available at https://doi.org/10.1287/mnsc.2023.4923 .

Exploiting a unique high-frequency, individual-level database, we (1) construct individual-level, incentive-compatible proxies of impulsivity based on video gaming behavior and (2) use these proxies to evaluate predictions concerning how impulsivity shapes individuals’ responses to a relaxation of credit constraints as captured by receiving a credit card. We discover that precard gaming intensity—as measured by the frequency and amount of game expenditures—is strongly and positively associated with (a) the probability of defaulting on credit card debt in the future, (b) postcard expenditures on luxury and addictive items, (c) surges in consumption spending immediately after receiving the credit card, and (d) rapid debt accumulation after obtaining the card. Differences in financial literacy, income, income variability, education, and demographics do not drive the results. The results are consistent with (1) neurological and psychological studies stressing that excessive gaming is associated with impulse control deficiencies and (2) behavioral theories stressing that impulsivity, i.e., time-inconsistent preferences for immediate gratification and ineffective strategies for avoiding myopic cues and temptations, substantially influence individual expenditure patterns and borrowing decisions when liquidity constraints are relaxed. This paper was accepted by Kay Giesecke, finance. Funding: C. Lin acknowledges the financial support from the National Natural Science Foundation of China [Project 72192841] and the Research Grant Council of the Hong Kong SAR, China [Project T35-710/20-R]. Supplemental Material: The data files and online appendix are available at https://doi.org/10.1287/mnsc.2023.4931 .