INFORMS Journal on Computing, 2023, 35(1): 178-195

We consider the optimal sourcing problem, i.e. how to choose the best suppliers, when the available suppliers may face different problems that affect their ability to deliver goods. This problem is motivated by the increasing severity of supply risks and the difficulty of evaluating common sources of vulnerability in upstream supply chains. These are problems reported by many surveys of goods-producing firms.

We propose a distributionally robust model that (i) accommodates multiple levels of supply disruption, not only full delivery or no delivery, and (ii) uses data-driven estimates of the underlying correlation to develop sourcing strategies in situations where the true correlation structure is ambiguous.

Using this model, we explore a worst-case scenario of supply distribution and show why it is appealing to use such a scenario due to the severe consequences associated with supply chain risks. We also show how our model can help firms decide whether to exert additional effort to better understand the prevailing correlation structure. We test our model using extensive computational experiments to further demonstrate the performance of our distributionally robust approach and show how supplier characteristics and the type of supply uncertainty affect the optimal sourcing decision.

Journal of Management Information Systems, accepted

This research paper explores how language changes affect the analysis of sentiment in financial texts such as annual reports, conference calls, and financial news. This type of analysis, called sentiment analysis, is often used to predict market trends. We propose a new algorithm, Word List Vector for Sentiment (WOLVES), which uses machine learning to understand how the meanings and sentiment strength of words evolve over time—a key feature often overlooked by predefined dictionaries used for sentiment analysis. When the algorithm was tested on different types of financial texts, it revealed that sentiment intensity was relatively stable over time, except for a decrease in negative sentiment in annual reports. This decrease suggests a strategic choice of language by managers to control perceived sentiment. The study further validated these findings using the Sarbanes–Oxley Act and the release of a popular financial sentiment dictionary as key events that could influence sentiment. We found that the improved sentiment measures provided by WOLVES were effective predictors of market reaction, earnings performance, and accounting fraud. The study contributes a new way to analyse sentiment that leverages both human domain knowledge and machine learning. It reconciles some inconsistent findings in the previous literature, and offers insights into strategic language adjustment in corporate communications.

Proceedings of the National Academy of Sciences - PNAS, 2023, 120(47)

Medical errors in hospitals not only lead to staggering costs but, more critically, can result in patient fatalities, with approximately one million deaths annually. While the technical and structural reasons for these errors have been extensively studied, the role of team dynamics within hospitals has received less attention. Our research fills this gap by exploring how 'faultlines' within hospital teams – divisions based on attributes like gender and race – impact patient care and safety. We found that dividing hospital teams into homogeneous subgroups based on these faultlines often leads to a decrease in civility among staff members. This lack of civility is a significant concern, as our study linked it to increased rates of medical errors and patient deaths. Specifically, we observed that a 10% rise in incivility within a unit correlated with up to an 8.87% increase in healthcare-associated infections and a 10.59% increase in patient mortality rates. However, there is a silver lining. In hospital units that foster a strong culture of collaborative conflict management – characterised by mutual respect, active listening, and openness to diverse opinions – the negative effects of these faultlines are significantly mitigated. Patients in such environments experienced fewer adverse events and a lower likelihood of mortality, regardless of the presence of strong faultlines. Our research underscores the importance of training healthcare staff in collaborative conflict resolution and equipping them with tools to understand and improve their own conflict management styles. This approach could be crucial for hospitals aiming to enhance patient care and reduce the incidence of medical errors.

Contemporary Accounting Research, forthcoming

This paper presents new evidence on the economic benefits arising from common institutional ownership. We find a negative and significant effect of common institutional ownership on stock price crash risk. This effect is robust to a battery of robustness checks and is causal according to some identification tests, including difference-in-differences (DiD) analyses on financial institution mergers. We find evidence that the negative effect is attributable to the monitoring role of common institutional owners—a role that is enabled by common owners’ lower information processing cost and greater monitoring incentives owing to governance externalities. We also find that common owners negatively influence crash risk through constraining bad news hoarding and that common owners are more likely to force chief executive officer (CEO) turnover when a firm has higher crash risk. Overall, our results suggest that common institutional shareholders play a unique and effective monitoring role that fends off stock price crashes.

Journal of Finance, forthcoming

We arranged for trained undercover men and women to pose as potential clients and visit all 65 local financial advisory firms in Hong Kong. At financial planning firms, but not at securities firms, women were more likely than men to receive advice to buy only individual or only local securities. Women clients who signaled that they were highly confident, highly risk tolerant or had a domestic outlook, were especially likely to receive this suboptimal advice. Our theoretical model explains these patterns as the result of statistical discrimination interacting with advisors’ incentives. Taste-based discrimination is unlikely to explain the results.

Management Science, advance online publication, 2023

Although prior research suggests strict, fair value–based securities accounting rules cause banks to sell securities into negative liquidity shocks, a value-destroying behavior called “liquidity feedback trading,” the mechanism is uncertain. We find the sooner chief executive officers (CEOs) are permitted to cash out of their stock and option grants, the more prone are their banks to feedback trading. Furthermore, the sooner CEOs can cash out, the more positive their banks’ stock price reaction to news of accounting rule relaxation. We conclude incentives for managerial short-term focus are a mechanism by which stricter accounting rules cause feedback trading. We also find evidence that regulatory compliance concerns play a role.

Journal of Accounting Research, 2023, 61(4), 1263-1312

We investigate how the mandatory adoption of International Financial Reporting Standards (IFRS) by publicly listed firms in the European Union affects peer private firms. We find that private firms’ capital investment decreases significantly after the IFRS mandate, relative to public firms. Private firms also display decreased investment when benchmarked against firms relatively insulated from the impact of the IFRS mandate, but the magnitude of the effect is smaller in this case. These results are consistent with the hypothesis that mandatory IFRS reporting (combined with other reforms), while increasing public firms’ financing and investment, crowds out funding for private firms. The effect is more pronounced for larger private firms and in industries where public peers have greater external financing needs. Our evidence suggests that financial reporting regulations cause shifts in resource allocation in an economy.

The Accounting Review, 2023, 98 (6): 97–124

The larger association between earnings and contemporaneous returns for negative returns than for positive returns is often attributed to conditional conservatism. We reason that this asymmetry may also be driven by the lack of timeliness with which stock price incorporates bad news relative to good news. Consistent with our reasoning, we show that when stock price incorporates bad news with delay, the asymmetry can exist in the absence of conditional conservatism. This suggests the testable hypothesis that the asymmetry decreases (increases) with factors that facilitate (impede) the incorporation of bad news into stock price. Using stock liquidity to test this hypothesis, we find that the earnings-return asymmetry decreases as stock liquidity increases. Our findings support the view that variation in the earnings-return asymmetry also reflects variation in the quality of the return generation process.

Journal of Financial and Quantitative Analysis (JFQA), accepted

We document the causal effects of single-name options trading on the absolute level of information content of prices (stock price informativeness) by exploiting the Penny Pilot Program as an exogenous shock to options trading volume. We find that options trading increases underlying stock price informativeness and information acquisition by both option and stock investors, consistent with the framework of Goldstein and Yang (2015). The findings are driven by firms for which options are more important sources for information and firms with more efficiently priced options. Options market introduction in a sample of 25 other economies also leads to higher price informativeness.

Production and Operations Management, 2023, 32(9), 2825–2845

This study focuses on a broadly adopted but largely underexplored cashback strategy, the praise cashback strategy (PCS), which aims to entice consumers to keep purchases and post positive online reviews. The literature acknowledges the impact of cashback strategies on consumers’ purchase and review behaviors; however, it does not focus on how they affect consumers’ return behaviors. Moreover, there are few studies on the effect of cashback strategies on consumer surplus and social welfare in the presence of strategic consumers, especially in relation to analyzing the proportion of fake reviews. By incorporating consumers’ review and return behaviors into a newsvendor model, we establish a sequential procedure for merchants to determine pricing, inventory, and PCS decisions, and then examine the conditions under which merchants prefer to adopt a PCS and its impact. Our results reveal that the adoption of a PCS is a typical prisoner’s dilemma for merchants in markets where PCSs are prevalent. Only when the PCS phenomenon is pervasive in the market, will merchants adopt a PCS with a small cashback amount. A PCS market negatively impacts consumer surplus and social welfare; however, merchants and consumers may benefit from a small cashback amount. Contrary to the common belief that a PCS undermines the organic evolution of online reviews and reduces return rates, we show that a PCS only elicits a few fake reviews, while merchants may face higher return rates from adopting it. These results benefit future studies on cashback strategies and fake reviews and provide evidence that supports the scientific governance of PCSs.