Essays on Causality-Driven Decision Making in Operations Management Restricted; Files Only

Abstract

This dissertation demonstrates how causal inference, machine learning, and optimization can be integrated to support optimal decision-making from a causal perspective. Through three essays, it addresses distinct operational challenges, each with significant business implications.

Essay 1 examines the strategic disclosure of delivery speed information in online retail. Using causal inference methods, we find that faster delivery promises can boost sales and customer spending but at the cost of higher returns and lower retention rates. Based on these causal parameter estimates, we develop a quantitative model that balances short- and long-term benefits to determine the optimal policy. 

Essay 2 delves into a fully personalized causality-driven optimization problem. After finding that last-mile home delivery significantly increases sales, we employ causal machine learning to estimate the treatment effects at the customer level. Since customers are highly heterogeneous, we introduce a novel capacity- and fairness-aware uplift model for optimally targeting the most profitable customers for treatment, while ensuring equitable access to services across different customer cohorts. 

Essay 3 explores a more complex, yet important scenario where treatment assignments are time-varying. We develop a dynamic causal machine learning model to evaluate the impacts of sequential treatments, considering high-dimensional, time-varying confounders. Our model not only personalizes but also provides a dynamic and state-dependent optimal treatment strategy. We apply our model to investigate whether and when giving customers more choice freedom is beneficial. 

Table of Contents

1 Sooner or Later? Promising Delivery Speed in Online Retail

1.1 Introduction 1

1.2 Literature Review 5

1.3 Hypothesis Development 6

1.4 Research Background and Data 9

1.5 Identification Strategy 13

1.6 Estimation Results 15

1.7 Optimal Delivery Speed Promise 18

1.8 Robustness Check 20

1.9 Conclusion 29

1.Appendices 34

2 The Value of Last-Mile Delivery in Online Retail

2.1 Introduction 41

2.2 Literature Review 45

2.3 Hypothesis Development 47

2.4 Background and Data 48

2.5 Impact of Last-Mile Home Delivery 51

2.6 Targeting Policy: Uplift Model and Causal Machine Learning 55

2.7 Robustness Check 66

2.8 Conclusion 69

2.Appendices 75

3. Incentives in Online Gaming: Optimal Policy Design with Dynamic Causal Machine Learning

3.1 Introduction 85

3.2 Literature Review 89

3.3 Background 91

3.4 Empirical Analysis 93

3.5 Dynamic Policy Design 95

3.6 Conclusion 101

3.Appendices 104

About this Dissertation

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Laney Graduate School
Department	Business
Subfield / Discipline	Information Systems & Operations Management
Degree	Ph.D.
Submission	Dissertation
Language	English
Research Field	Operations Research
Keyword	Digital Platform Causality-Driven Data-Driven
Committee Chair / Thesis Advisor	Ruomeng Cui, Emory University
Committee Members	Nikolay Osadchiy, Daniel Minh McCarthy, University of Maryland Tianshu Sun, Cheung Kong Graduate School of Business

Last modified

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	File download under embargo until 22 May 2031	2025-04-04 22:55:20 -0400	File download under embargo until 22 May 2031

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions