Contact Shandra Bremer at [email protected] with any questions.

February 21

Dorothee HonHon, Naveen Jindal School of Management

Title: It’s still good! Expiration dates & food waste

Time: 11 a.m. - noon
Location: R1210

Abstract: We examine a retailer that produces and sells a perishable product, facing three key decisions: the date label (i.e., effective shelf life) to print on the packaging, the frequency of batch production, and the quantity to produce in each batch. In making these choices, the retailer takes into account the product’s biological shelf life, the associated costs and revenues, and consumers’ attitudes and expectations regarding shelf life. We model different types of consumers based on their attitudes towards date labels and analyze how these variations affect the retailer’s optimal decisions and resulting product waste. Our findings show that when consumers rely on date labels to guide their purchases, retailers are incentivized to set shorter date labels than the actual biological shelf life. This, in turn, leads to increased food waste, as consumers discard products that are still safe to consume.

February 28

Ilan Lobel, NYU Stern

Title: Auction Design using Value Prediction with Hallucinations

Time: 11 a.m. - noon
Location: R1210

Abstract: We investigate a Bayesian mechanism design problem where a seller seeks to maximize revenue by selling an indivisible good to one of n buyers, incorporating potentially unreliable predictions (signals) of buyers’ private values derived from a machine learning model. We propose a framework where these signals are sometimes reflective of buyers’ true valuations but other times are hallucinations, which are uncorrelated with the buyers’ true valuations. Our main contribution is a characterization of the optimal auction under this framework. Our characterization establishes a near-decomposition of how to treat types above and below the signal. For the one buyer case, the seller’s optimal strategy is to post one of three fairly intuitive prices depending on the signal, which we call the “ignore”, “follow” and “cap” actions.

MARCH 14

Rouba Ibrahim, University College London

Title: Communication in Service Operations

Time: 11 a.m. - noon
Location: R1210

Abstract: We study the effectiveness of information design as a managerial lever to mitigate the overuse of critical resources in congestion-prone service systems. Leveraging the service provider’s informational advantage about relevant aspects of the system, effective communication requires the sharing of carefully curated information to persuade some customers to forgo service for the benefit of customers with higher service needs. To study whether effective communication can arise in equilibrium, we design controlled laboratory experiments to test the predictions of a queueing-game theoretic model that endogenizes the implementation of information-sharing policies. Our main result is that communication increases social welfare even when the service provider lacks the ability to formally commit to their information policy (as usually is the case in practical settings), i.e., under conditions where standard theory predicts that communication fails because it lacks credibility and thus fails to affect customer behaviour. (Joint work with Arturo Estrada Rodriguez and Mirko Kremer.)

MARCH 21

Julie Simmons Ivy, University of Michigan

Time: 11 a.m. - noon
Location: R1210

Abstract: From Data to Decision Making in Health and Humanitarian Logistics: Insights and Challenges 

Julie Simmons Ivy, Ph.D., University of Michigan 

Decision making to satisfy the basic human needs of health, food, and education is complex. We present an overview of two illustrative studies using data to inform decision making in health care delivery associated with sepsis and hunger relief.  In the first study, we integrate electronic health record (EHR) data with clinical expertise to develop a continuous-time Markov decision process model of the natural history of sepsis. We formulate this as a stopping problem to find the optimal first intervention to minimize expected mortality and morbidity. We explore the effect of the complex trade-offs associated with the intervention costs and patient disposition costs which are subjective and difficult to estimate. This framework provides key insights into sepsis patients’ stochastic trajectories and informs clinical decision making associated with caring for these patients as their health dynamically evolves. In the second study, we develop a single-period, weighted multi-criteria optimization model that provides the decision-maker the flexibility to capture their preferences over the three criteria of equity, effectiveness, and efficiency, and explore the resulting trade-offs. We introduce a novel algorithm to elicit the inherent preference of a food bank by analyzing its actions within a singleperiod. The non-interactive nature of this algorithm is especially significant for humanitarian organizations such as food banks which lack the resources to interact with modelers on a regular basis. We explore the implications of different decision-maker preferences for the criteria on distribution policies.  

MARCH 28

Will Ma, Columbia University

Title: Model-based vs. Model-free Online Decision-making for Optimal Stopping, Pricing, and Inventory

Time: 11 a.m. - noon
Location: R1210

Abstract: Buoyed by advances in computation and AI, businesses have growing interest in
adopting black-box algorithms for decision-making. Motivated particularly by Reinforcement
Learning (RL) at an industry partner, we compare model-based to model-free approaches on
several stochastic optimization problems over time. Here we define "model-based" to construct
independent distributions over time and solve dynamic programming, while "model-free"
optimizes over historical trajectories in a black-box fashion.

We provide a rigorous theoretical comparison, in a finite-horizon time-inhomogeneous RL
setting with offline trajectories that allows perfect hindsight evaluation. We derive surprising
horizon-independence results for PAC-learning, which exploit the problem-specific structure of
the Inventory Replenishment and Optimal Stopping problems. Perhaps more surprisingly, this
requires different approaches for different problems: model-free for Inventory, and model-based
for Stopping. Meanwhile, we show that a horizon-independent learning guarantee is impossible
for the Pricing problem.

These theoretical findings are consistent with observations in simulations, and explanatory of
successes/failures in deploying model-free RL at our industry partner. The takeaway is that
businesses should not adopt a one-size-fits-all verdict on whether to deploy black-box
algorithms, as they are relatively advantaged for some problems but not others.

April 4

L. Beril Toktay, Georgia Tech: Scheller College of Business

Title: Personalized Assortment Optimization for a Subscription Business Model of Experience Goods

Time: 11 a.m. - noon
Location: R1210

April 11

Peng Sun, Duke: Fuqua School of Business

Title: Optimal Push and Pull Funding for Global Health

Time: 11 a.m. - noon
Location: R1210

Abstract: Malaria caused over 600,000 deaths in 2021, yet commercial incentives are weak for drug and vaccine development for malaria and other tropical diseases. Governments and nonprofits address these market failures with push mechanisms (e.g., grants) and pull mechanisms (e.g., prizes). We propose insurance, that is paying for failure, as another tool. Many funders face adverse selection and moral hazard issues, because firms have more information about a drug's potential and the firm's capabilities and effort. Using a principal-agent framework and duality theory of (infinite dimensional) linear optimization, we determine the optimal funding mix based on disease characteristics. For most tropical diseases, including malaria, we recommend primarily pull funding with supplementary push support. For tuberculosis, insurance is optimal. These findings challenge current practices dominated by push funding and extend to funding innovations in other sectors.
Authors: Chenxi Xu, David Ridley, Peng Sun

April 25

Mahesh Nagarajan, UBC Sauder School of Business

Title: Operational data driven interventions to decrease adverse events associated with Opioid overdose

Time: 11 a.m. - noon
Location: R1210

Abstract: Adverse events including deaths from illicit drug overdose (specifically Opioids) is a significant issue in North America, especially in the west coast of Canada. In this talk, we present three systematic data driven approaches to decrease such events. First, prevention of drug use and habit formation. Second management of addiction among drug users and third, a reactive and dynamic response to forecasted overdose incidents that uses a predictive model built using near miss events and other variables as signals along with a novel caregiver scheduling mechanism. Theoretical instances of the scheduling problem presents interesting technical challenges on hardness and approximability which we will partially resolve in this talk. We discuss the implementation of these approaches in urban centres in Canada.