Driving a More Prosperous Future
The Impact of App Crashes on Consumer Engagement
Journal of Marketing
The authors develop and test a theoretical framework to examine the impact of app crashes on app engagement. The framework predicts that consumers increase engagement after encountering a single crash due to their need-for-closure and curiosity, yet reduce engagement after experiencing repeated and concentrated crashes, primarily because of frustration and perceived task unattainability; the recency of crashes moderates these effects. Field data analysis reveals that while a crash truncates a session and reduces content consumption, it increases page views in the following session. However, this increase in page views does not compensate for the loss during the crashed session. Frequent and more concentrated crashes curtail engagement. Three experiments in which crashes are exogenously manipulated in a different context support the validity and generalizability of these findings, confirm the proposed mediators, and demonstrate how to lessen the negative impact of repeated crashes with post-crash messages. The research adds new dimensions to the task pursuit literature and provides managers with a framework to quantify the economic impact of crashes, analyze content substitution behavior, and assess the bias of a transactional view of crash incidents. Additionally, it offers insights into targeted feature release to more tolerant users and strategic design of post-crash messages.
Savannah Wei Shi, Associate Professor of Marketing & J.C. Penney Research Professor, Leavey School of Business Santa Clara University
Seoungwoo Lee*, Assistant Professor, Yonsei School of Business, Yonsei University Seoul
Kirthi Kalyanam, L.J. Skaggs Distinguished Professor Leavey School of Business, Santa Clara University
Michel Wedel, PepsiCo Chaired Professor of Consumer Science, Robert H. Smith School of Business, University of Maryland
Do Credit Rating Agencies Learn from the Options Market?
Management Science, November 2024
Do credit rating agencies (CRAs) learn from the options market? We examine this question by exploring the relation between options trading activity and credit rating accuracy. We find that as options trading volume increases, credit ratings become more responsive to expected credit risk and exhibit greater ability to predict future defaults. We also find that CRAs rely more on the options market as a source of ratings-related information when firm default risk is higher, options trading is more informative, manager-provided information is of lower quality, and firm uncertainty is higher. Our results are robust to a number of sensitivity tests, including alternative measures of options trading and credit rating accuracy. We reach similar inferences using various approaches to address endogeneity issues, including difference-in-difference analyses and an instrumental variables approach. Overall, our findings are consistent with the view that CRAs incorporate unique information from the options market into their rating decisions which, in turn, improves credit rating accuracy.
Musa Subasi, University of Maryland-College Park
Paul Brockman, Lehigh University
Jeff Wang, San Diego State University
Eliza Zhang, University of Washington-Tacoma
Equity Term Structures without Dividend Strips Data
Journal of Finance
We use a large cross section of equity returns to estimate a rich affine model of equity prices, dividends, returns, and their dynamics. Our model prices dividend strips of the market and equity portfolios without using strips data in the estimation. Yet model-implied equity yields closely match yields on traded strips. Our model extends equity term-structure data over time (to the 1970s) and across maturities, and generates term structures for various equity portfolios. The novel cross section of term structures from our model covers 45 years and includes several recessions, providing a novel set of empirical moments to discipline asset pricing models.
Stefano Giglio, Yale School of Management
Bryan Kelly, Yale School of Management
Serhiy Kozak, R.H. Smith School of Business, University of Maryland
Transforming Products into Platforms: Unearthing New Avenues for Business Innovation
NIM Marketing Intelligence Review, October 2024
It is impossible for brands to ignore digital platform opportunities. Network effects are one of the strongest sources of power and defensibility ever invented and underlie some of the most valuable businesses in the world. Managers and entrepreneurs can leverage the power of platforms by adding some platform elements to their existing products or services, by distributing their brands via existing platforms or by developing their own new platforms. By using one’s own brands as platforms requires creativity but can help businesses unlock new value and build resilient ecosystems around their products. There are three key methods. The first is to invite third-party sellers to enhance existing products. Examples include selling advertising space around products or creating app stores to extend offers. The second is to connect one’s customers by enabling interactions among users to add value. Third, brands might reach out to customers’ customers by enhancing the end-user experience in a way that benefits both themselves and their direct customers. If thoughtfully implemented, any platform strategy will create self-reinforcing feedback loops sparking growth and keeping competitors at bay.
Andrei Hagiu, Associate Professor of Information System, Boston University; Bobby Zhou, Associate Professor of Marketing, University of Maryland
The Theory-Based View and Strategic Pivots: The Effects of Theorization and Experimentation on the Type and Nature of Pivots
Strategy Science
We examine how formalization in cognitive processes (theorization) and evidence evaluation (experimentation) influence the type (frequency and radicalness) and nature (impetus, clarity, and coherence) of entrepreneurial pivots. We use a mixed-method research design to analyze rich data from over 1,600 interviews with 261 entrepreneurs within a randomized control trial in London. A quantitative analysis that complements human-coded and machine learning-coded measures reveals that conditional on pivoting, theorization and experimentation are complementary in their association with making single radical pivots. The extensive qualitative-case comparison further elucidates interactions between theorization and experimentation that generate differences in the nature of pivots that range from purposeful (clear and coherent rationale deriving from articulated theory and experimentation), postulatory (informed by articulated theory but not incorporating nuances or surprises generated from experimentation), and remedial (stemming from adjustments to preformed theories that drew on prior experiences) to reactive (driven by environmental stimuli absent a clear theory of value). These insights contribute to the theory-driven strategic decision-making literature and offer practical insights for entrepreneurs, incubators, and policymakers on the benefits of a scientific approach to entrepreneurship.
Valentine, Jacob (Doctoral Candidate, University of Maryland); Novelli, Elena (Professor, Bayes Business School); Agarwal, Rajshree (Lamone Professor of Strategy and Entrepreneurship, University of Maryland)
Improved LISA Analysis for Zero-Heavy Crack Cocaine Seizure Data
INFORMS Journal of Data Science
Local Indicators of Spatial Association (LISA) analysis is a useful tool for analyzing and extracting meaningful insights from geographic data. It provides informative statistical analysis that highlights areas of high and low activity. However, LISA analysis methods may not be appropriate for zero-heavy data, as without the correct mathematical context the meaning of the patterns identified by the analysis may be distorted. We demonstrate these issues through statistical analysis and provide the appropriate context for interpreting LISA results for zero-heavy data. We then propose an improved LISA analysis method for spatial data with a majority of zero values. This work constitutes a possible path to a more appropriate understanding of the underlying spatial relationships. Applying our proposed methodology to crack cocaine seizure data in the U.S., we show how our improved methods identify different spatial patterns, which in our context could lead to different real-world law enforcement strategies. As LISA analysis is a popular statistical approach that supports policy analysis and design, and as zero-heavy data is common in these scenarios, we provide a framework that is tailored to zero-heavy contexts, improving interpretations and providing finer categorization of observed data, ultimately leading to better decisions in multiple fields where spatial data is foundational.
Eunseong Jang, The Robert H. Smith School of Business, University of Maryland
Margret Bjarnadottir, The Robert H. Smith School of Business, University of Maryland
Marcus Boyd, National Consortium for the Study of Terrorism and Responses to Terrorism, University of Maryland
S. Raghavan, The Robert H. Smith School of Business & Institute for Systems Research, University of Maryland
How to Talk When a Machine Is Listening: Corporate Disclosure in the Age of AI
The Review of Financial Studies, March 2023
Growing AI readership (proxied for by machine downloads and ownership by AI-equipped investors) motivates firms to prepare filings friendlier to machine processing and to mitigate linguistic tones that are unfavorably perceived by algorithms. Loughran and McDonald (2011) and BERT available since 2018 serve as event studies supporting attribution of the decrease in the measured negative sentiment to increased machine readership. This relationship is stronger among firms with higher benefits to (e.g., external financing needs) or lower cost (e.g., litigation risk) of sentiment management. This is the first study exploring the feedback effect on corporate disclosure in response to technology.
Sean Cao, Associate Professor (with tenure), Robert H. Smith School of Business, University of Maryland, United States of America
Large language models and synthetic health data: progress and prospects
JAMIA Online, October 2024
There is growing interest in the application of machine learning models and advanced analytics to various healthcare processes and operations, including the generation of new clinical discoveries, development of high-quality predictions, and optimization of administrative processes. Machine learning models for prediction and classification rely on extensive and robust datasets, particularly for deep learning models common in health, creating an urgent need for large health datasets. Yet datasets can be insufficiently large due to the rapid evolution of diseases, such as coronavirus disease 2019 (COVID-19), rarity of disease, or the myriad obstacles to sharing and acquiring existing health data, including ethical, legal, political, economic, cultural, and technical barriers. Synthetic data provide a unique opportunity for health dataset expansion or creation by addressing privacy concerns and other barriers. In this paper, we review prior literature and discuss the landscape of machine learning models used for synthetic health data generation (SHDG), outlining challenges and limitations. We build on existing research on the state of the art in SHDG and prior broad explorations of the potential risks and opportunities for large language models (LLMs) in healthcare. We contribute to the literature with a focused assessment of LLMs for SHDG, including a review of early research in the area and recommendations for future research directions. Six promising research directions are identified for further investigation of LLMs for SHDG: evaluation metrics, LLM adoption, data efficiency, generalization, health equity, and regulatory challenges
Daniel Smolyak, Department of Computer Science, University of Maryland
Margret V. Bjarnadottir, Robert H. Smith School of Business, University of Maryland
Kenyon Crowley, Accenture Federal Services
Ritu Agarwal, Center for Digital Health and Artificial Intelligence, Carey Business School
Seductive Language for Narcissists in Job Postings
Management Science
Prior research indicates that narcissistic executives engage in earnings management and other negative organizational behaviors, and many studies ponder why firms hire such individuals, especially into corporate accounting positions. Utilizing a selection of terms from real-world job postings that we characterize as either describing a “Rule-Bender” or “Rule-Follower" candidate, we first conduct several validation studies which reveal that these terms vary predictably across types of job postings, that people generally agree with our categorization of these terms, and that Rule-Benders are viewed as possessing worse managerial skills but a higher proclivity for unethical behavior. We then demonstrate that narcissistic job seekers are more attracted to job postings that describe the ideal candidate using Rule-Bender terms for both general positions (Experiment 1) and senior accounting positions (Experiment 2). Finally, we examine firm characteristics that might lead professional recruiters to incorporate Rule-Bender language into Chief Accounting Officer job postings and find that Rule-Bender terms are preferred for higher-growth, higher-innovation firms (Experiment 3), and when more aggressive reporting would benefit the firm (Experiment 4). Our results suggest that recruiters’ language choices can attract Rule-Bending narcissists to firms, perhaps even in unintended circumstances.
Jonathan Gay (University of Mississippi), Scott Jackson (University of South Carolina), Nick Seybert (University of Maryland)
Bayesian Ensembles of Exponentially Smoothed Life-Cycle Forecasts
Manufacturing and Servoce Operations Management
We study the problem of forecasting an entire demand distribution for a new product before and after its launch. Firms need accurate distributional forecasts of demand to make operational decisions about capacity, inventory and marketing expenditures. We introduce a unified, robust, and interpretable approach to producing these pre- and post-launch distributional forecasts. Our approach is inspired by Bayesian model averaging. Each candidate model in our ensemble is a life-cycle model fitted to the completed life cycle of a comparable product. A pre-launch forecast is an ensemble with equal weights on the candidate models’ forecasts, while a post-launch forecast is an ensemble with weights that evolve according to Bayesian updating. Our approach is part frequentist and part Bayesian, resulting in a novel form of regularization tailored to the demand forecasting challenge. We also introduce a new type of life-cycle or product diffusion model with states that can be updated using exponential smoothing. The trend in this model follows the density of an exponentially tilted Gompertz random variable. For post-launch forecasting, this model is attractive because it can adapt itself to the most recent changes in a product’s life cycle. We provide closed-form distributional forecasts from our model. In two empirical studies, we show that when the ensemble’s candidate models are all in our new type of exponential smoothing model, this version of the ensemble outperforms several leading approaches in both point and quantile forecasting. In a data-driven operations environment, our model can produce accurate fore- casts frequently and at scale. When quantile forecasts are needed, our model has the potential to provide meaningful economic benefits. In addition, our model’s interpretability should be attractive to managers who already use exponential smoothing and ensemble methods for other forecasting purposes.
Xiaojia Guo (Assistant professor, Robert H. Smith School of Business, UMD), Casey Lichtendahl (Google), Yael Grushka-Cockayne (Professor, Darden school of business, University of Virginia)