Knowledge that Transforms

The COVID-19 pandemic forced organizations, including higher education institutions, to rapidly adjust their operations. In the face of the pandemic, most higher education institutions shut down their campuses and transitioned to emergency remote teaching mode. This study examines digital resilience in higher education institutions through the conceptual lens of disaster response management, by assessing the role played by the centralized governance of information technology (IT) investments. We posit that centralized IT helps organizations maintain customer satisfaction with services during a crisis (e.g., student satisfaction with classes during COVID-19) by facilitating the organization-wide transition to an emergency operational mode and supporting its service operations. Consolidating data on IT investment, governance, and course evaluations from 463 U.S. higher education institutions from 2017-2020, we show that centralized IT helped organizations adapt better to the pandemic in terms of maintaining student satisfaction. Moreover, we found that centralized IT investments geared toward facilitating organizational coordination and providing instructional and technical support played a pivotal role in enabling ERT and improving student ratings during the crisis. These results are corroborated by interviews with CIOs of U.S. higher education institutions. Additional analyses also suggest that the effectiveness of centralized IT governance is contingent upon organizational size, dissimilarity of local units, and the strategic role of the CIO. We also discuss theoretical extensions toward digital resilience as well as practical implications.

Unstructured multimedia data (text and audio) provides unprecedented opportunities to derive actionable decision-making in the financial industry, in areas such as portfolio and risk management. However, due to formidable methodological challenges, the promise of business value from unstructured multimedia data has not materialized. In this study, we use a design science approach to develop DeepVoice, a novel nonverbal predictive analysis system for financial risk prediction, in the setting of quarterly earnings conference calls. DeepVoice forecasts financial risk by leveraging not only what managers say (verbal linguistic cues) but also how managers say it (vocal cues) during the earnings conference calls. The design of DeepVoice addresses several challenges associated with the analysis of nonverbal communication. We also propose a two-stage deep learning model to effectively integrate managers’ sequential vocal and verbal cues. Using a unique dataset of 6,047 earnings call samples (audio recordings and textual transcripts) of S&P 500 firms across four years, we show that DeepVoice yields remarkably lower risk forecast errors than that achieved by previous efforts. The improvement can also translate into nontrivial economic gains in options trading. The theoretical and practical implications of analyzing vocal cues are discussed.

Data breaches can severely damage a firm’s reputation and its customers’ confidence. Firms must therefore continuously invest in security measures to prevent such breaches. However, the effectiveness of security investment has been questioned by both practitioners and academics. We illustrate the bidirectional dynamic relationship between information technology (IT) investment and data breaches moderated by threat and countermeasure security awareness using an eight-year panel of 311 U.S.-listed firms to provide empirical evidence that threat awareness broadens firms’ scope for addressing data-breach issues by investing more in IT than in security. Countermeasure awareness equips firms with sufficient knowledge and experience to ensure effective implementation of IT, which provides more comprehensive protection than security investment alone. Our results suggest that firms should evolve beyond the reactive mindset of solely upgrading security and begin nurturing both threat awareness and countermeasure awareness to address the underlying IT system issues that are the cause of data breaches.

Emotion artificial intelligence, the algorithm that recognizes and interprets various human emotions beyond valence (positive and negative polarity), is still in its infancy but has attracted attention from industry and academia. Based on discrete emotion theory and statistical language modeling, this work proposes an algorithm to enable automatic domain-adaptive emotion lexicon construction and multidimensional emotion detection in texts. Using a large-scale dataset of China’s movie market from 2012 to 2018, we constructed and validated a domain-specific emotion lexicon and demonstrated the predictive power of eight discrete emotions (i.e., surprise, joy, anticipation, love, anxiety, sadness, anger, and disgust) in online reviews on box office sales. We found that representing overall emotions through discrete emotions yields higher prediction accuracy than valence or latent emotion variables generated by topic modeling. To understand the source of the predictive power from a theoretical perspective and to test the cross-culture generalizability of our prediction study, we further conducted an experiment in the U.S. movie market based on theories on emotion, judgment, and decision-making. We found that discrete emotions, mediated by perceived processing fluency, significantly affect the perceived review helpfulness, which further influences purchase intention. Our work shows the economic value of emotions in online reviews, generates insight into the mechanism of their effects, and has managerial implications for online review platform design, movie marketing, and cinema operations.

Charity fundraising is becoming increasingly reliant on online platforms such as crowdfunding platforms. However, overwhelmingly, crowdfunding campaigns are not meeting their goals. Therefore, it is imperative to examine how the success of charity fundraising campaigns can be improved. In this paper, we focus on the design of project images on a crowdfunding website, which portray the themes and content of the projects. Employing the stimulus-organism-response (S-O-R) model, we investigate the relationships between image attributes (S) and image emotions (O), and between image emotions (O) and campaign outcomes (R). We developed and trained a deep neural network model to identify the emotions conveyed in the images, and then implemented it to analyze project images from a popular crowdfunding platform. We applied the obtained image emotions together with the objective image attributes and the project outcome metrics to explore from a design perspective, what image attributes evoke image emotions, and how image emotions are related to the success of charity fundraising projects. Our results confirm these relationships and further suggest that the roles of image emotions on the success of crowdfunding campaigns vary with project characteristics such as the project budget and category. In addition, the image emotions of competing projects on the crowdfunding platform were found to reduce the project’s performance. In an extended study, we conducted an online randomized controlled experiment by manipulating image attributes to reexamine the causal relationships and verify the mediating roles of positive and negative empathies between image emotions and campaign outcomes. This research contributes to the charity fundraising literature from a novel perspective of emotions in project images. It presents new and unique findings regarding the mediation roles of positive and negative empathies and the limitations of the emotion of sadness in certain types of charity fundraising. In addition, our findings provide useful insights for practitioners seeking to design successful online charity campaigns.

Despite a large body of literature on online reviews, none have considered the nuanced impacts of how multidimensional reviews affect product sales differently for mainstream vs. niche products. This study seeks to fill this knowledge gap by conducting complementary studies in two product categories (i.e., automobiles and laptops) with different methods (a field study and three lab experiments). Our paper reveals three key insights into the emerging literature and phenomenon on multidimensional review systems: (1) the interdimensional rating variance is more negatively related to product sales for mainstream products than for niche products in the same category, (2) the intradimensional rating valence on the dominant dimension of a product is more positively related to product sales for niche products than for mainstream products, and (3) the intradimensional rating variance on the dominant dimension of a product is more negatively related to product sales for niche products than for mainstream products. Our research provides important managerial implications for both product providers and review platforms.

Online health communities (OHCs) play an important role in enabling patients to exchange information and obtain social support from each other. However, do OHC interactions always benefit patients? In this research, we investigate different mechanisms by which OHC content may affect patients’ emotions. Specifically, we notice users can read not only emotional support intended to help them but also emotional support targeting other persons or posts that are not intended to generate any emotional support (auxiliary content). Drawing from emotional contagion theories, we argue that even though emotional support may benefit targeted support seekers, it could have a negative impact on the emotions of other support seekers. Our empirical study on an OHC for depression patients supports these arguments. Our findings are new to the literature and have critical practical implications since they suggest that we should carefully manage OHC-based interventions for depression patients to avoid unintended consequences. We design a novel deep learning model to differentiate emotional support from auxiliary content. Such differentiation is critical for identifying the negative effect of emotional support on unintended recipients. We also discuss options to alter the intervention volume, length, and frequency to tackle the challenge of the negative effect.

With the rapid development of fintech, the need for dynamic credit risk evaluation is becoming increasingly important. While previous studies on credit scoring have mostly focused on single-period loan default prediction, we call for a new avenue—multiperiod default prediction (MPDP)—to depict risk profiles over time. To address the challenges raised by MPDP, such as monotonic default probability prediction and complex relationship accommodation, we propose a novel approach, hybrid and collective scoring (HACS). We design a hybrid modeling strategy to predict whether and when a borrower will default separately through a default discrimination model and a default time estimation model, respectively, and synthesize them through a probabilistic framework. To accommodate various possible patterns of default time and measure the distribution of default probability over successive time intervals, we propose a joint default modeling method to train the default time estimation model. Empirical evaluations at the model (time-to-default prediction performance and discrimination performance) and mechanism (identifiability and discriminability) levels, as well as impact analyses at the application (granting performance and profitability performance) level, show that HACS outperforms the benchmarked survival analysis and multilabel learning methods on all fronts. It can more accurately predict time-to-default and provide financial institutions and investors better decision-support in granting loans and selecting loan portfolios.

User engagement, a key factor in the success of social media platforms, has long been based on permanent content. A recent paradigm shift in platform design has led large social media providers to implement ephemerality features that by default make shared content disappear after a certain amount of time. However, very little is known about how ephemerality features affect user engagement and behavior in social media. Drawing upon the technology affordance perspective, we conducted a qualitative multimethod study involving individual interviews and focus groups. Our findings show that the affordances arising from features with varying degrees of ephemerality (i.e., snaps and stories) differ from those of permanent content features in terms of self-presentation, browsing others’ content, and communication. Adopting a multidimensional conceptualization of user engagement, we show the positive (e.g., more content sharing) and negative (e.g., cognitive burden from context loss) effects for snaps and stories that should be cautiously considered by social media platforms aiming to introduce such features. Finally, we reveal new user behaviors that relate to sharing snapshots of fleeting value as snaps or experiences of transient value as stories.

We contribute to the emerging literature on quality certification by digital platforms by studying the launch of the Airbnb Plus service, wherein the platform inspects properties and provides a badge that presumably signals the quality of the property and the reliability of the host. Our identification strategy relies on the fact that the Airbnb Plus service was launched in different cities at different times, and listings within the cities received the certification at different times. Using a staggered difference-in-differences estimation strategy in conjunction with suitable matching methods, we found that the Airbnb Plus certification increased the weekly booking rate of Plus listings by about 6.8% on average (direct effect). We also found some evidence that non-Plus listings saw a temporary decline in booking rate when one or more nearby properties received a Plus certification (externality effect). The net impact of the Airbnb Plus service on the platform itself was an annual increase in revenue of about $37,500 for the average 2-kilometer zone in a U.S. city that included one or more Plus listings, as compared to matched zones without any Plus listings (local platform effect). We performed additional analyses, including a randomized experiment, to demonstrate the robustness of our findings. Overall, our results suggest that platform-endorsed quality certification has significant economic impacts—not just on the listings that receive the certification but on other listings on the platform as well as on the platform itself.