Research Interests

Substantive

• Operations-Finance Interface: Sustainable Operations, People-centric Operations

• Technology Adoption in Service Operations: Robotic Chefs, LLM-assisted Call Centers

Methodological

• Causal Inference, Large Language Model, Natural Language Processing, Machine Learning

Working Papers

1. "Adopting ‘Green’ Shareholder Proposals Can Improve Firm Operations"

with Damian Beil and Andrew Wu,

under review at Management Science

The explosive growth in ESG investing has fueled a heated debate: do ESG activities genuinely enhance firm performance, or are they merely costly signaling ("greenwashing")? While the sustainable OM literature features impactful context-specific studies, there is a recognized need for generalizable empirical research evaluating the effectiveness ("do well") of broader sustainability instruments ("do good"). This study addresses this call by examining the operational consequences of firms' involvement in ESG activities.

The primary challenge in this area is endogeneity, as high-performing firms may self-select into ESG activities. To isolate the causal effect, I leverage shareholder voting outcomes on ESG-related proposals and apply a Regression Discontinuity Design (RDD), utilizing close-call votes to provide quasi-random variation in ESG adoption. I focus on long-term operational outcomes, implementing a dynamic treatment-on-the-treated (TOT) framework to track performance over five years.

I find that only environmental proposals yield significant gains in operating efficiency (e.g., inventory turnover, cash conversion cycle), with effects materializing 3-5 years post-passage. Social and governance proposals show no such impact. Drawing on the Porter Hypothesis, the mechanism I identify is that environmental mandates act as a pressure that spurs investments in modernization. I observe that treated firms increase capital expenditures while simultaneously increasing asset retirements, suggesting a focus on property, plant, and equipment retrofitting rather than expansion. These findings provide causal evidence that ESG mandates–specifically environmental ones–can drive operational improvements, underscoring the importance of the substance of a proposal in affecting operational performance.

2. "Measuring Leaders’ Operational Experience and Its Impact on Firm Operational Performance"

with Damian Beil and Andrew Wu,

in preparation for 3rd submission after Reject and Resubmit at Manufacturing & Service Operations Management

While the operations management literature has extensively studied the impact of frontline workers, the role of senior leadership in shaping operational outcomes remains less understood. Measuring a leader's operational expertise presents significant challenges due to its complex and multifaceted nature; existing research often relies on resource-intensive surveys, context-specific experiments, or simple trait counts, making it difficult to determine whether past operational success is transferable to a new organization.

In this paper, I introduce the Operational Experience Score (OES), the first scalable, quantitative measure of leadership operational experience developed in the literature. Leveraging detailed work histories from BoardEx, the OES encapsulates the operational performance (e.g., inventory turnover outperformance) of leaders' previous workplaces during their tenure. To address the endogeneity concern that firms may selectively hire experienced leaders when aiming to improve performance, I employ an instrumental variable approach leveraging unforeseen leadership changes due to untimely deaths.

The results reveal a causal link: firms led by leaders with higher OES exhibit better operational performance. The sudden death of a high-OES leader negatively impacts inventory turnover, while the departure of a low-OES leader improves it. Furthermore, the effect is concentrated among senior leaders and those in operations-focused roles. Falsification tests confirm that the OES is operations-specific and not merely a proxy for general ability. This study demonstrates that operational skill is both measurable and transferable, establishing a dynamic metric that opens new avenues for research on leadership in operations.

3. "Neutral Point-of-View Operational Language in Earnings Calls and Market Response"

solo-authored,

under review at Management Science (Virtual Special Issue on AI for Business and Finance Decisions)

Operational details–such as supply chain efficiency and capacity utilization–are critical drivers of future cash flows. However, this information is often specialized and complex. Prior research suggests that financial analysts may struggle to accurately interpret and price this information, creating a persistent information asymmetry between managers and investors. While research confirms that the presence of OM topics in earnings calls matters, the linguistic quality of this communication remains underexplored.

This solo-authored study investigates how firms can communicate operational realities more effectively. Traditional textual analysis relies on coarse metrics like sentiment or readability. I bridge the OM and finance literature by introducing a novel approach utilizing LLMs to measure a specific linguistic property: the Neutral Point-of-View (NPOV) of operational narratives. Using 80,651 U.S. earnings conference calls from 2006-2023, I apply a BERT-based classifier trained on Wikipedia's NPOV edits to each OM sentence to construct firm-quarter measures of OM NPOV in prepared remarks and in Q&A sections. Higher NPOV scores correspond to language that is more factual and quantifiable and that uses fewer explicitly subjective or evaluative expressions, holding sentiment and other textual features constant. I document that higher OM NPOV in prepared remarks is positively associated with short-window cumulative abnormal returns.

To the OM literature, this research provides the first quantitative method to assess the linguistic quality of OM disclosures and demonstrates their impact on capital markets. Methodologically, it demonstrates the application of LLMs to measure a nuanced textual attribute like objectivity. Practically, my findings imply that, beyond deciding whether to talk about operations, managers' choice to describe OM information in a neutral, fact-focused way in their scripted remarks is associated with how investors and analysts update beliefs. For managers, this highlights the value of clear, NPOV operational disclosure in helping external stakeholders interpret the firm's operations.

Work in Progress

4. "Food Without Feelings? Unpacking Consumer Bias Against Robotic Kitchens"

with Hyungwoo Leo Ryu (CEO, Hanwha Foodtech Global Inc.),

Target submission: Nature Human Behaviour

While the operational gains of automation in service industries are recognized, customers may resist services provided by non-human agents, creating a tension between operational efficiency and customer satisfaction. This research investigates consumer bias toward robot-made products in a real-world setting, motivated by discussions with Hanwha Foodtech–a Korean company that combines robotics and AI with food service. This study aims to contribute to the service operations literature by quantifying the trade-offs of automation beyond efficiency and exploring the behavioral factors influencing technology acceptance.

I utilize large-scale consumer review data and exploit the deployment of automated kitchens by a salad restaurant chain for a Difference-in-Differences (DiD) analysis. The standardized salad preparation process ensures that product quality remains consistent, isolating the effect of the automation. DiD shows lower consumer satisfaction ratings post-deployment. To analyze the nuances of consumer reactions, I use a BERT model fine-tuned on a large corpus of conventional restaurant reviews from Yelp to adapt the model to restaurant-specific language and avoid spurious "novelty" effects from robot-related terms. The textual analysis reveals heterogeneity along themes like speed, accuracy, and perceived sterility, informing how to design automated services that balance efficiency with human perception.

5. "Spare the rod for the CEO, but not for the COO"

with Hyun-Soo Ahn and Jen Choi,

Target submission: Management Science

Research in finance and accounting consistently documents an asymmetry in pay-performance sensitivity for CEOs: they receive substantial rewards for strong performance but are often insulated from compensation reductions for poor performance. It is unclear whether the same applies to Chief Operating Officers (COOs). COOs face a unique tension: operational failures (e.g., supply chain disruptions) are highly visible and salient. In contrast, operational successes (e.g., efficiency improvements, averted crises) are often less apparent if they represent the maintenance of a status quo rather than a discrete event. This project investigates whether COO compensation is asymmetrically linked to firm performance–namely, with steeper penalties when firms underperform and muted rewards when they outperform. Furthermore, I examine the consequences of this incentive structure, specifically how it affects the firm's operational resilience and ability to withstand operational shocks. 

Preliminary findings suggest a significant disparity: COOs face larger compensation reductions when the company underperforms, yet do not receive commensurate monetary rewards when the firm performs well. Furthermore, I explore the operational impacts of this asymmetry by examining how firms respond to exogenous operational shocks (e.g., natural disasters and sudden tariffs). The results suggest that firms with less COO-favorable pay-performance sensitivity exhibit weaker post-shock resilience. This study contributes to the OM literature by linking COO incentive design to cross-firm differences in operational resilience, highlighting COO incentive design as a salient factor.

6. "The Role of Analyst Focus on Operational Information in Forecast Accuracy"

with Damian Beil and Andrew Wu,

data assembled & cleaned; preliminary analysis in progress.