Mack Institute Research Archives

New Report: Generative AI Adoption in the U.S. Military

Stylized depiction of the American flag with glowing red stripes and star patterns against a dark, mountainous landscape.

The U.S. Department of Defense is one of the most complex organizations in the world — so how is it approaching generative AI? A new report from Wharton’s Prof. Serguei Netessine and WG’25 Andrew Stiles examines the DoD’s early steps toward integrating GenAI and highlights lessons that industry leaders can…Read More

The Six Dimensions of Strong Theory

One of the most important features on which to judge the merit of any academic paper is the strength of its theory. Although commentary about what constitutes strong theory is widespread, there is no holistic account of the full range of existing perspectives. To address this oversight, I construct a typology composed of six dimensions of strong theory: importance, interestingness, actionability, generality, simplicity, and accuracy.…Read More

Outlier Neglect: A Decision-Making Bias with Implications for Hiring, Investment, and Consumer Choices

We propose and test a novel decision-making bias called “outlier neglect.” In general, when evaluating people, places, and opportunities with many features, people focus on the average of those features. For example, when hiring a team, people want the average performance of the team to be high, so they try to hire individuals who perform well on average. But in many cases, the relevant metric is not the average, but the best – for example, when a pharmaceutical company tests different types of malaria drugs, their success depends on the best-performing drug (not the average) because they can scale up the best and ignore lesser performers. We propose that evaluators neglect the importance of outliers in such cases and instead focus on portfolio averages. In some cases, this leads to suboptimal outcomes (e.g. in the drug scenario above, or it may be better to hire someone to join a team who is amazing at one specialized task than someone who is higher on average). We test this bias and show that across many contexts – including personnel selection, creative idea generation, consumer decisions, and investment strategy – people neglect outliers in favor of the average, leading to suboptimal decisions.…Read More

Business Model Innovation for Renewables

Consumers who want access to renewable energy have two main options: install renewable energy generation equipment “behind” the electrical meter (e.g., solar panels on the roof) or buy energy from a utility company, which would then source energy from generation companies. The first approach has obvious diseconomies of scale. It is only available to homeowners, while the second approach requires over-reliance on utility companies, which may not contract with renewable suppliers or may offer an expensive mix of renewable and nonrenewable energy. We study an alternative innovative business model, “community solar,” whereby consumers subscribe to own a portion of the energy generated by a large solar plant. We partner with Origo Energy, a Brazilian company that pioneered this model in South America, and we analyze the behavior of consumers who switch from the traditional business model to Origo’s subscription to understand boundary conditions for community solar.…Read More

Grid-Scale Mobile Battery Energy Storage Systems

Grid-scale electricity storage technologies play a vital role in balancing electricity supply and demand, particularly as renewable energy sources like wind and solar introduce greater variability into power systems. Lithium-ion batteries, accounting for 90% of U.S. electricity storage capacity, are widely regarded as essential to the clean energy transition. By storing excess electricity during periods of low demand, and thus low prices, and releasing it when demand, and thus price, is high, storage technologies smooth fluctuations in generation and earn significant revenues from arbitrage. Battery operators strategically locate systems in areas with high nodal price variability, but current practices often fail to adapt to changing market conditions, risking inefficient investments with diminishing price spread at selected locations.

Mobile Energy Storage Systems (MESS) present a transformative innovation, enabling both temporal and geographic flexibility in energy storage. Unlike existing Stationary Energy Storage Systems, MESS can be relocated to provide storage services at different points in the grid as market dynamics evolve with rapid addition of transmission and renewable generation capacity. Although MESS technologies currently find niche applications, such as disaster relief, advancements in material technology and declining battery costs make utility-scale adoption plausible. This study addresses a critical gap by modeling MESS fleet operations, analyzing their feasibility, and comparing their financial performance against stationary systems in renewable-rich grids. Our findings aim to guide developers and grid operators in leveraging MESS for enhanced energy flexibility and resilience in renewable-rich grids.…Read More

Private Equity Ownership and Human Capital Acquisition Strategy

The private equity industry has grown significantly in the past decades, and this explosive growth has generated much interest on the impact of private equity’s footprint on the economy and the labor market. Using career history data from nearly 11 million employees at 16,137 private equity-backed firms from 2000 to 2024, I find that firms controlled by private equity recruit specialized managerial workforce. Post-deal, newly hired senior executives and middle managers are more likely to have previously worked at a private equity-backed firm. Moreover, they are also more likely to have worked at a firm backed by the same private equity owner. I find evidence that some private equity firms facilitate managerial mobility within their portfolio, creating an “internal” network of senior and middle managers that mirrors the way public corporations have traditionally groomed executives internally. Altogether, my results suggest that private equity firms rely on specialized managerial workforce who are familiar with private equity operations and with a specific owner’s playbook.…Read More

The Hidden Tolls of Reputational Risk: Using Media Sentiment to Detect Threats to Corporate Reputation and Its Financial Impact

Corporate reputation is a vital strategic asset for organizations. Yet, its socially constructed nature has made it challenging for scholars to agree on a precise definition or develop a reliable measurement strategy for it. Historically, scholars have relied on measures that are useful for assessing reputation earned but fail to capture its dynamic nature or identify emerging threats in real time, exposing a critical blind spot in both theory and practice. To address these limitations, I propose Cumulative Abnormal Media Sentiment (CAMS), a novel approach for identifying and analyzing reputational risks and opportunities by tracking abnormal volatility in stakeholder sentiment. To validate this construct, I conduct a quasi-replication of Caroline Flammer’s 2013 event study, extending her analysis of corporate news coverage of environmental events for U.S. publicly traded organizations through 2024. Using this expanded dataset, I measure reputational signals surrounding coverage of eco-friendly and eco-harmful corporate behavior. My analysis reveals a direct relationship between reputational risk from eco-harmful events and stock price volatility. This research offers new insights into the established relationship between reputation and financial performance, while introducing a replicable and adaptable measurement tool for event study analyses, equipping future researchers with a robust framework for examining the dynamic interplay between reputation and financial outcomes.…Read More

Generative AI for Efficient and Equitable Healthcare on a Global Scale

This proposal presents two innovative research projects designed to harness the transformative power of AI to enhance healthcare outcomes. In close collaboration with the Somaliland Ministry of Health and Development (MoHD), the Taiwan International Cooperation Development Fund (ICDF), and Penn researchers, we aim to tackle critical healthcare challenges in Somaliland, one of East Africa’s most impoverished regions. Our main goal is to develop effective and safe AI methodologies to improve healthcare accessibility, quality, and efficiency. These projects will deepen our understanding of how AI can be applied in safety-critical scenarios and resource-constrained environments, facilitating healthcare advancement on a global scale.…Read More

Do We Write What AI Tells Us To? LLMs as Persuasive Agents

Consumers are increasingly turning to large language models (LLMs) as an aid to everyday writing (i.e., email, text). While it is clear that LLMs can enhance the grammatical and syntactical structure of written communication, might they also lead people to communicate things that depart from their original intentions? We explored this question through an experimental paradigm in which participants were first asked to create an opinionated message, then viewed a suggested revision generated by the Chat-GPT 4 LLM that was either more positive or negative than the original. Results of our experiment reveal that LLMs do exert a substantial influence on written communication, but this effect has important moderators. Notably, participants who initially conveyed negative (vs. positive) opinions were less resistant to persuasion from LLMs, and text revisions that made a message more positive (vs. negative) were embraced more readily.…Read More

Effects of Prior Authorization on Medicaid Prescription Drug Access

Prescription drug spending has increased rapidly over the last two decades. Prior authorization represents an innovative strategy in Medicaid’s prescription drug benefit management, using administrative tools to influence prescribing behaviors and control expenses. State Medicaid programs have widely adopted prior authorization policies to curb spending and enhance the targeting of treatments. Despite the importance of these policies for Medicaid, there is limited evidence of their impact on prescription drug access and patient outcomes. In this project, we use a novel regression-discontinuity design to study the consequences of prior authorization. We study the impact of these policies on the prescribing of drugs covered by prior authorization and substitution to other drugs as well as heterogeneous impacts across geographic areas and socio-demographic characteristics. We also assess the importance of different design features of these policies and their impact on inappropriate and appropriate prescribing.…Read More

Exploring the Demand Side for Commercializing Academic Science

Most of the prior research on the topic of commercializing academic science approaches the topic from the supply side (innovations from academic institutions and scientists). The needs and behavior of firms are rarely considered in this literature. We aim to do so by using a variety of data sources, both proprietary and public, to characterize technologies and situations in which firms are likely to license academic science. Doing so will also affect startup formation to commercialize such technologies, an increasingly important commercialization avenue.…Read More

Off on a Limb: Balancing the Decision to Amputate

Rising healthcare costs remain a significant challenge in the U.S., with one major contributor being the increasing incidence of amputations due to vascular diseases. The number of vascular disease-related amputations in the U.S. is projected to double by 2050, reaching an estimated 3.6 million cases. In 2023 alone, Medicare’s annual expenditure exceeded $900 billion, with $24.5 billion (~2.72%) allocated to amputation procedures and post-amputation care. Unfortunately, nearly 50% of individuals who undergo vascular disease-related amputations die within five years, a mortality rate higher than that of breast, colon, or prostate cancer.…Read More

Scientific Malpractice in Alzheimer’s Research: Systematic Evidence and Impacts on Pharmaceutical Firms

Science forms the bedrock of industrial innovation, and yet the integrity of the scientific enterprise has recently been questioned across multiple fields by high-profile scandals and replication crises. How common is scientific malpractice, and what are its costs for firms that rely on scientific research to guide their innovation investments? Using advanced AI and manual validation, we estimate the prevalence and different types of data malpractice in Alzheimer’s research, both within and across papers, from 1980 to 2020. We focus on inappropriate image duplications, an objectively observed form of data malpractice widespread in a field where experimental data are often presented as figures. Preliminary findings suggest that, on average, 1.7% of all peer-reviewed papers present inappropriately duplicated images, a number that has steadily increased over time. The incidence is greater for research done in universities (relative to corporate and clinical settings), employing animal and molecular methods (relative to human subject methods), and originating from China. Our current work investigates the costs of scientific malpractice for pharmaceutical firms that base their investments on publicly available science. By tracking citations from firm patents to papers with duplicated images, we plan to estimate how much attention and resources are distorted away from findings with clinical applications and how much this may contribute to explaining the slow progress in finding a cure for Alzheimer’s disease.…Read More

Strategic Openness of the Innovation Portfolio

We investigate the strategic openness of firms’ innovation portfolios, focusing on the determinants and implications of disclosure strategies for diverse innovation assets, particularly in the context of artificial intelligence (AI). While firms traditionally protect innovation through patents and secrecy, open innovation frameworks have gained prominence as firms increasingly leverage external sources of innovation. This research seeks to bridge the gap between the innovation and open-source literatures by exploring how firms disclose and utilize various innovation assets—such as patents, academic publications, and open-source code—in response to their R&D strategies, market environments, and policy pressures.…Read More

TwoMinds: Understanding How Humans and AI Systems Achieve Mutual Understanding

As artificial intelligence becomes increasingly integrated into daily life, understanding how humans and machines achieve mutual understanding has become a crucial scientific challenge. Our research addresses this challenge by developing a digital web platform that studies how humans and AI systems communicate and build shared understanding in real-time. While machines can now model our preferences and predict our behavior, they struggle to genuinely understand human meaning and intention. Our platform will generate novel data about human-AI interaction, with direct implications for improving AI systems and enhancing human-machine collaboration. By integrating insights from psychology, computer science, computational linguistics, and organizational behavior, our work builds an essential bridge between human and artificial minds—a critical need for organizations seeking to effectively deploy AI technology.…Read More

Blessing or Curse? The Impact of Spousal Teams on Startup Hiring: Evidence from Observational Data and a Field Experiment

Gender disparities persist in entrepreneurship, with women facing significant challenges in founding and growing ventures (Gompers et al., 2021; Rocha and van Praag, 2020). Women represent just 20% of founders in the U.S. (Guzman and Kacperczyk, 2019) and tend to underperform relative to men (Ruef et al., 2003; Kim et al., 2006). Prior research attributes this gap to systemic barriers, such as educational and institutional constraints (“leaky pipeline”), and evaluative biases from investors and gatekeepers…Read More

Inorganic Scaling Strategies of Adolescent Technology Ventures

This study will explore technology-oriented startups (such as deeptech) scale through inorganic modes such as ecosystem partnerships and alliances. We attempt to understand how different antecedents such as founding and scaling characteristics affect the timing of when startups engage in such modes.…Read More

Electric Vehicle (EV) Fleet and Charging Infrastructure: Decision-Making by Drivers in the Gig Economy

The rapid integration of electric vehicles (EVs) into gig economy platforms like Uber and Lyft presents unique challenges, particularly in driver decision-making, earnings, and operational efficiency. This study explores how EV-specific constraints, such as charging infrastructure and fleet size, influence the behavior of gig economy drivers. We analyze the role of algorithms in shaping driver earnings, pricing, and trip allocations, addressing concerns about transparency, bias, and geographic disparities. Using a proprietary dataset combined with public data on charging station locations, our research employs descriptive analysis, regression models, and simulation to examine the impact of charging accessibility on driver efficiency and service levels. The findings aim to inform algorithmic design improvements and policy interventions, fostering more equitable and efficient EV integration in gig platforms.…Read More

Product Innovation, Market Sentiment, and Resource Allocation

Our research aims to deepen the understanding of (i) how financial markets value product innovation across various firm types, industries, and economic conditions, and (ii) how market sentiment influences the private economic value of product innovation, its impact on firms’ and competitors’ profitability, and the resulting resource reallocation—such as capital, labor, and R&D—both within and across firms.…Read More

Exploring the Role of Artificial Intelligence in Turbocharging Innovation in the Generative AI Era

rtificial intelligence (AI) has become a transformative force in fostering innovation and productivity. In our prior research (Wu et al. 2020; Wu et al. 2019), we demonstrated that AI-driven analytics can significantly enhance innovation by combining existing technologies in novel ways and refining existing technologies. With the advent of generative AI and other advanced algorithms, firms are discovering unprecedented opportunities to innovate and create new products. Yet some firms are vastly successful at using AI to innovate while the majority fails.…Read More