Presenting online retail customers with individualized substitution suggestions (ISS) when an item is forecasted to be out-of-stock (OOS) is a challenging problem. We investigate how to provide more relevant ISS by employing state-of-the-art choice models. In collaboration with a partner online retail platform, we assess the value of our model by conducting a field experiment. Online retailers can obtain benefits both in revenue and in costs from presenting customers with improved ISS.

Problem: Does selling ugly produce in grocery stores reduce food waste?

Methodology/results: Modeling the supply chain, we find that selling ugly produce reduces waste but lowers retailer profit. Dedicated ugly produce retailers achieve the same waste reduction, explaining the rise of startups. A food landfill tax can significantly reduce waste.

Managerial implications: Our findings support the value of ugly produce startups, informing efforts to combat food waste and hunger.

Problem: How should retailers handle imperfect produce to reduce food waste?

Methodology/results: We analyze discarding, bunching, and differentiating strategies. Increasing acceptance of imperfect produce may not reduce waste. Full-shelf ordering may not increase waste, especially with higher prices. Discarding can decrease waste.

Implications: Retailers should choose strategies wisely, consider full-shelf ordering, and educate consumers on imperfect produce.

TBD

TBD

Our work provides new results on the arm-sampling behavior of the celebrated UCB family of multi-armed bandit algorithms, leading to several important insights. Among these, it is shown that arm-sampling rates under UCB are asymptotically deterministic, regardless of the problem complexity. This discovery facilitates new sharp asymptotic characterizations revealing profound distinctions between UCB and Thompson Sampling such as an "incomplete learning" phenomenon characteristic of the latter.

Many pharmaceutical drugs have multiple indications, for which they offer a varying degree of benefit for patients. Yet, in the current US pricing system, the price of the drug is the same regardless of the indication for which it is prescribed. We use a modeling approach to analyze how indication-based pricing compares to uniform pricing for the manufacturer’s profit and investment incentives, the patients’ access to the drug and benefit, and the payer’s coverage incentives and objective.

We propose and analyze a budget-neutral incentive scheme that Organ Procurement Organizations could utilize to increase donor hospital's (DH's) proportion of timely referrals and thereby increase the number of potential donors. We find that, depending on the DH's cost of effort required to increase the proportion of timely referrals, the proposed contract could increase the percentage of family authorizations from a single DH by 1.3%.

Although available evidence suggests that home hemodialysis(HHD) may achieve similar clinical outcomes to in-center hemodialysis and are less resource intensive for patients with end-stage renal disease, they have been used less in the US than in other developed nations. We design two incentive plans in a bi-level game structure consisting of a payer and providers to obtain equilibrium. We show these incentive plans can improve the HHD rate and increase ESRD beneficiaries' quality of life.

Our paper studies how to determine the personalized incentive for each customer to learn her true preference in the assortment optimization and product design problem. We provide algorithms that sequentially decide the personalized incentives based on the evaluated customer's contribution to the revenue increase. We show that compared to the naive supervised learning algorithm that provides fixed incentives, our algorithm can reduce the total cost significantly while achieving high revenue.

We introduce the concept of randomization into the robust assortment optimization liter-ature. We show that the firm can improve its worst-case expected revenues by selecting an assortment randomly according to a prudently designed probability distribution. We demonstrate this potential benefit of randomization both theoretically in an abstract problem formulation and empirically across three choice models: the multinomial logit model, the Markov chain model, and the preference ranking model.

We study dynamic joint assortment and pricing over a finite time period. The goal is to maximize the expected cumulative revenue. We propose a new doubly high-dimensional contextual bandit model to formulate the problem. We developed a computationally tractable algorithm and provide a convergence rate for it. The numerical results of applying our method to a real-world online retail data set demonstrate the efficiency of our method, which is further supported by extensive simulations.

We study a service system where customers with different service demands arrive at a facility with a shared waiting room. We assume two types of customers and two servers. Types differ by their service demand: type 1 seeks service from server 1, and type 2 seeks service from server 2. Customers cannot distinguish between the types, and make their decision whether to join based on the total number of customers in the shared waiting room, without observing the state of the servers.

We study ghost kitchen platforms, which consist of delivery-only restaurants that serve limited number of dishes. We develop a new model of multi-dash queueing system at the platform’s position. In the multi-dash queueing system, an order splits into a random number of sub-orders and is assigned to different home chefs. Our study identifies conditions under which the ghost kitchen platform can be more profitable than traditional food delivery platforms.

TBD

There are more than 1.1 billion people lacking electricity residing in off-grid communities, where extending national electricity grids is infeasible. These communities turn to stand-alone diesel generators for their energy needs, which creates significant economic, operational and environmental challenges. In this study, we identify the optimal capacity investment decisions from the perspective of a remote community and investigate how common policy mechanisms interact with these decisions.

Precision agriculture has been promoted by agricultural technology providers through the use of outcome-based pricing (OBP). In this paper, we examine the effects of OBP on the adoption rate of precision agriculture, and the benefits it offers to both farmers and service providers. We develop a two-period game model that incorporates providers' learning from experience. We also explore the implications of government intervention when the service provider switches to OBP from traditional pricing.

We investigate the effect of upstream supply risk correlation of substitutable items. Suppliers offer menu of price-quantity or wholesale price contracts. Using common agency methodology we characterize the equilibrium contracts. We find that the buyer's profits are increasing in yield correlation under wholesale price contracts but are increasing under price quantity contracts only when product substitutability is low and in fact decrease for high product substitutability.

We consider the problem faced by spare parts inventory that demand intensity for components is unclear at the beginning of a product life cycle. We present a robust optimization (RO) approach in spare parts inventory to against demand uncertainties and design two more time-efficient algorithms capable of finding solutions in case of a large number of items in the model. Our experiments show that the RO model exhibits remarkable efficacy in case of limited information on the demand distribution.

We examine the impact of business disruptions in explaining the gender gap in small firm performance in developing countries. We find that business disruptions significantly increase gender gap in firm performance – on average, business disruptions decrease sales and sales growth of a small women-led firm by 11.6% and 15.2 percentage points, respectively, compared to their male counterpart. Importantly, building resilience helps small women-led firms to close this performance gap.

Strategic inventory refers to the inventory held by firms purely out of strategic considerations other than operational reasons (e.g., economies of scale). In this paper, we investigate the roles of strategic inventory in a system with two parallel supply chains under both full bargaining and partial bargaining, which differ in whether inventory is included in the bargaining terms.

This study proposes a markdown strategy to optimize joint replenishment and pricing decisions in a dynamically changing fresh/non-fresh inventory assortment with stochastic lifetimes, lead times, and demands. We model the problem as a generally modified (r, Q) policy and reformulate that as a MIP model to solve the model exactly. We propose an EOQ approximation and provide some bounds on the optimality gap with respect to market demand, maximum WTP, and lifetime that vanishes asymptotically.

Platforms are able to gather large quantities of data, which can result in high precision predictions regarding consumers purchasing patterns. A fundamental question is whether a platform has the incentives and ability to share such non-verifiable information with its sellers. We demonstrate that cheap-talk information can be exchanged when the platform shares with its sellers region-forecast. In this equilibrium, the platform shares only a region that contains its private information.

Firms rely on expert teams for decision making but face privacy concerns. We propose a mechanism to protect experts' privacy and analyze incentives. The firm garbles experts' reports to address privacy concerns, encouraging truthful reporting. Larger teams may perform worse, and more capable experts may be detrimental to team performance under privacy-conscious conditions.

Tele-driving lets drivers operate vehicles remotely, offering a viable alternative to fully automated ones. It reduces spatial mismatch in ride-hailing, enabling any driver for any customer. We quantify gains by comparing traditional systems to tele-driving. Findings: 1) For impatient customers, optimized driver capacity enhances service or reduces drivers; 2) For patient customers, tele-driving stabilizes systems or cuts drivers with similar service quality.

This paper investigates the undesirable customer behavior of opportunistic picking for products with longer expiration dates in grocery retail stores. We find that on average 20% of food waste at the retail stage is caused by consumer picking. Further, we reveal store and product-related determinants of picking waste and derive insights for store operations decisions. Our paper is the first to empirically quantify consumer picking as a food waste driver.

While fast delivery helps retailers to stay competitive, it often leads to worse environmental outcomes. We study how to nudge online retail customers to voluntarily choose a green but slow shipping option. We develop and test a theory to inform which strategies are more effective based on their ability to overcome psychological barriers. The results of our experiments provide guidelines on which information strategy to use in two logistical contexts: no-rush shipping and consolidated shipping.

"Eye level is buy level" is a common belief in the retail industry, but there is a lack of rigorous empirical evidence to support it. While some studies have used observational data and eye-tracking technology, more controlled field experiments are needed. This study uses a two-stage experimental design in collaboration with a convenience store chain to investigate the impact of eye-level placement on product sales.

Supplier copycatting occurs when the supplier (S) to a manufacturer (M) copies M’s product and sells a copycat product to the customers. M can change its suppliers; and S runs the risk of facing repercussions if it decides to copy. We propose a two-period game-theoretic approach to supplier copycatting. We investigate a setting with M, copycatting S, and non-copycatting S, and examine how the equilibrium is influenced by the presence of future opportunities and repercussions facing the firms.

Included in pdf

We study a manufacturer-supplier interaction during the rollover between a legacy part and its successor in a durable good supply chain. In practice, manufacturers try in vain to leverage the future business of the new part to delay a supplier’s ''last time buy'' and retirement of the old part. We propose a two-stage noncooperative game to guide managers and prove that there exist only six subgame perfect Nash equilibria which achieve this delay under a simple necessary and sufficient condition.

We study an off-policy contextual pricing problem where the seller has access to samples of prices that customers were previously offered, whether they purchased at that price, and auxiliary features. This is in contrast to the well-studied setting in which samples of the customer's valuation are observed. We focus on convex loss functions for pricing in this setting, prove expected revenue bounds when the valuation distribution is log-concave, and provide generalization bounds.

We study the design of stochastic optimization methods with a focus on guaranteed out-of-sample performance when data is corrupted. We design a novel robust approach that offers protection against corruption while ensuring strong generalization. Our approach is based on distributionally robust optimization with a combination of Kullback-Leibler and Levy-Prokhorov ambiguity sets. Our method is applied to training neural networks, resulting in robust networks with state-of-the-art performance.

The paper presents dynamic pricing with demand learning in presence of covariates and develops safe UCB pricing algorithms. The pricing algorithm minimizes total regret where the expected regret of the pricing policy should be at least as good as a fraction of that of the baseline policy. We extend it to the case that the cumulative reward of the algorithm is at least as high as a fraction of the cumulative reward of the baseline policy with known and unknown baseline expected reward.

Patient-level data from 72 MRI hospitals in Ontario, Canada from 2013 to 2017 shows that over 60% of patients exceeded their wait time targets. We conduct a data-driven analysis to quantify the reduction in the patient Fraction Exceeding Target (FET) for MRI. Our resource pooling model lowers the FET from 66% to 36% while constraining the average incremental travel time below three hours. In addition, our model and method show that only ten additional scanners are needed to achieve 10% FET.

Problem: We optimize personalized interventions for behavioral health using digital platforms, considering cost and capacity constraints. Existing approaches are data-intensive or overlook long-term dynamics. Our DecompPI algorithm approximates policy iteration, reducing intervention costs while maintaining efficacy. A case study shows potential for 50% cost reduction, enabling scalable implementation.

Many developing countries lack the health-emergency infrastructure of the developed world. In this context, our industry partner Flare (operating in Nairobi, Kenya) coordinates existing ambulance providers by operating a platform. We study the operational challenges for such platforms as they often lack knowledge about all ambulances' future availability and their location at a tactical level and typically do not fully control these ambulances.

We study privacy protection mechanisms inspired by recent regulatory regimes, limited data retention and customer self protection. Privacy protection affects the estimation of demand model and thus the charged prices. We find that the change of the resulting price and which customer groups benefit from the protection depend on the product type. A real dataset of online auto loans validates our theoretical findings. We also extend the framework to nonlinear demand functions and duopoly.

Personalized recommender systems face privacy concerns. We explore optimal design under local differential privacy constraints. Our findings suggest a coarse-grained threshold policy for recommendations. Pursuing privacy comes at an economic loss but may benefit consumers. Our study guides algorithm design and informs regulators on privacy policies.

TBD

To increase equity on their platform, we re-designed the search algorithm on VolunteerMatch (VM), the largest online platform for connecting volunteers and nonprofits. The implementation of our algorithm in Dallas led to a 10.2% increase in the number of different volunteer opportunities that receive a sign-up each week without reducing the total number of sign-ups: a Pareto improvement for VM. A similar effect nationwide would lead to 800 more opportunities with at least one sign-up each week.

We build a game-theoretical model to analyze the impact of a merger between two platforms on consumer surplus and labor welfare. While a merger reduces competition, the merged platform can pool customers and agents together and improve matching between them; moreover, the merger can amplify the cross-side network effect and thus moderate the merged firm's pricing power. Under a sufficiently strong cross-side network effect, a merger can make merging firms, customers, and agents all better off.

TBD

Recycled wastewater plays a pivotal role in water sustainability by closing the urban water cycle. This paper studies the wastewater recycling capacity investment decision of a water utility using a two-stage stochastic cost-minimization model under rainfall and recycling cost uncertainties. We characterize the optimal wastewater recycling capacity investment and conduct sensitivity analysis to investigate the impacts of uncertainties on the optimal expected cost and recycling capacity.

Disasters disproportionately affect low-income communities' access to groceries. We study price changes after hurricanes and find higher increases in low-income areas, along with stockouts and substitutions.

Low-income communities face greater price increases, reduced promotions, more stockouts, and increased substitution to higher-priced products.

Disaster response should address pricing, stock management, and equitable access to groceries based on community needs and substitution patterns.

Existing stormwater management fails to address climate-adaptive needs in cities experiencing intensified rainfall. This paper utilizes robust optimization to identify worst-case rainfall scenarios and their impact on flooding loss. It provides guidelines for risk mitigation, optimal infrastructure designs, and emphasizes the urgency of taking action as climate change exacerbates flooding risks.

In global retail marketplaces, returns may result in long shipping distances and loss of welfare through time-induced price deterioration. We propose to minimize these negative impacts by exchanging returns within the marketplace’s supplier network. We show that an auction-based system where suppliers bid on the returned items reduces returns’ shipping distance by up to 88% and increases resell value on average by 5% mitigating the drawbacks of benevolent return policies of retail marketplaces.

Motivated by our interaction with a labor market, we focus of matching jobs to workers with heterogenous preferences. In each period, jobs arrive sequentially. Each worker has three parameters: her work quality, capacity, and minimum number of jobs she desires. The platform wishes to maximize the matches quality via its online matching policy. We use our model to understand limitations of simple policies and propose an optimal index-based policy. We supplement our theory with extensive numerics.

Motivated by the rapid growth of the cloud computing industry, we study an infinite-horizon, stochastic optimization problem from the viewpoint of a firm that employs cloud resources to process incoming orders (or jobs) over time. Orders and resources are heterogeneous. The firm's goal is to minimize the long-run average expected cost per period, considering reserved-capacity costs, on-demand capacity costs, and order-delay costs. We show that our proposed policy is asymptotically optimal.

This paper uses a newsvendor setting to analyze the effect of payment methods and parking prices on payment amounts, occupancy, and wait times. Empirical results show that cash or mobile app payments are less than credit card payments, and lower prices increase payments for all methods. Simulation shows that lowering prices has a greater effect on occupancy and wait times than increasing them. The study also indicates surroundings of parking spaces significantly affect performance measures.

With the emergence of on-demand platforms, it has become viable for drivers to participate in package delivery and passenger rides operations at the same time. The integration and coordination of the two services, known as co-modality, is considered a promising solution for improving the efficiency of urban logistics and mobility systems. In this work, we propose a simple zoning-based coordination policy and analyze its performance against pure delivery and ride-hailing policies.

TBD

Rent-to-own (RTO) firms rent products in exchange for a fee and offer the already-rented products for purchase at buyout prices to their renters. Prediction of demand requires a decision model that captures the renters' decision-making process and their ownership and rental utilities. We develop a modular framework that separates utility estimation from identifying the renter's decision process. We build renter decision models, reflecting different degrees of sophistication and alertness.

Digital goods firms routinely utilize renting models for product differentiation: one can try a free or cheap version first, then subscribe to unlock additional functionality. Recent technological advances enable physical goods firms to create products with identical hardware that are digitally controlled to allow for similar differentiation. We present a stylized model to explore when physical goods firms should adopt such differentiation instead of the traditional high- and low-end products.

Consider a platform facilitating matches in the presence of supply congestion. A key attribute of supply is how competitive it will be for demand to obtain the supply after the match. Should the platform reveal current competition levels? To answer this, we partnered with a marketplace and propose a structural model in which workers account for future competition. We conduct counterfactual analysis to study the impact of signaling competition on workers' lead purchasing decisions and revenue.

We theoretically and experimentally study the secretary problem under competition, focusing on the effect of two market design choices: i) transparency about agents’ priorities; and ii) the mechanism to collect decisions (simultaneous vs. sequential). Our results show that the latter affects the saliency of competition and induces frustration. Moreover, we theoretically show that transparency may lead to higher welfare, but the benefits do not translate to practice due to information friction.

We propose and analyze a stochastic model of service interactions that captures two (a)symmetries between the customer and agent: co-production vs self-production, synchrony vs asynchrony. This model reveals connection to the behavioral operations literature, such as non-monotonic system performance from monotonic agent-load slowdown, yielding insights for decision making and analysis.

In this paper, we investigate how line-sitting and express lines affect customers' satisfaction and fairness perceptions. We show that customers who encounter line-sitting report higher satisfaction with their overall experience than those who encounter an express-line customer, despite the actual wait time being the same. Moreover, we find that the effect of queueing schemes on customer satisfaction is mediated by differences in fairness perception.

Several prominent online platforms operate both reselling and agency channels. We explore when and why the platform and a supplier would choose to interact via the platform's agency channel versus its reselling channel.

Problem: Economic drivers of blockchain adoption for supply chain financing. Methodology/Results: Model of firms in a linear supply chain facing information leakage and entry threats. Adoption depends on short-term gains vs. long-term threats. Blockchain adoption is welfare enhancing. Proposed solutions: smart contracts and self-financing transfers.

Problem: Loyalty programs and consumer payment choices. Methodology/Results: Analyzing loyalty program data, we find factors influencing payment decisions, such as mental accounting and perceived value of points. Different consumer segments exhibit distinct attitudes toward points. Managerial implications: Firms can use this understanding to design effective pricing strategies, expand partnerships, and enhance consumer loyalty.

We study a single-item multi-warehouse multi-location order fulfillment problem faced by an online retailer with limited logistic capacity. Orders shall be fulfilled within the time window before being lost. We proposed a heuristic policy based on Lagrangian relaxations and showed that it is asymptotically optimal when the retailer serves a large number of locations. We also showed that a "two-day fulfillment" strategy, together with the policy, could mitigate the shortage of logistic resources.

We examine how an online grocer's fulfillment options affect customers' weekly and lifetime spending. Based on our results, we develop a structural model that estimates the preferences underlying customers’ choices while learning customer preferences. Based on a counterfactual analysis, we recommend whether the retailer should offer pick-up, delivery, or both services in each geographic market.

In this work, we propose an approach that leverages historical customer purchase conversion rates when designing a middle-mile consolidation network that aims to maximize the profit of large e-commerce retailers. We embed lead-time dependent sales volume predictions into a new mixed-integer program (MIP) that simultaneously determines shipment lead times and consolidation plans to maximize profit.

We consider a problem where customers repeatedly interact with a platform. The probability that a customer engages depends on past purchase history. The platform maximizes the total revenue obtained from each customer over the horizon. We study the dynamic program when consumer preferences are known and prove structural properties. We provide a formulation in a contextual episodic RL setting and prove a regret bound. We evaluate effectiveness on simulations, using real data from Expedia.

In a feedback collection process, a company sequentially and adaptively shows display sets to a population of customers and collects their choices. The objective is to identify the most preferred item at a high confidence level with the least number of samples. We propose an elimination-based algorithm, namely Nested Elimination (NE). NE is intuitive, simple in structure, easy to implement, and has a strong theoretical/numerical performance for sample complexity.

Problem: Inpatients wait for hospital diagnostic services, causing chaos. Methodology: We propose "advance notice" scheduling, providing patients with preparation and service time windows. Markov Decision Process optimizes decisions. Numerical study shows significant operational improvement. Managerial implications: Advance notice policy reduces waiting and improves appointment-based services.

Problem: Evaluating an ED delay announcement system in Hong Kong's healthcare. Methodology: Studying 1.3M patient visits, we estimate patient sensitivity to announced waiting times (WT) and factors influencing it. Results show potential improvements in reducing WT and patients leaving without being seen. Implications: Increase patient awareness, reduce WT update window, focus on older population and Kowloon district for promotion.

We study hospitals' operational characteristics contributing to the re-admission of psychiatry patients. We propose that length of stay (LOS) mediates the effects of hospital characteristics on the risk of readmission. We reveal how patient characteristics moderate these effects. Using a dataset of 15,000 psychiatry patients, we provide evidence on the negative volume-outcome and nonlinear LOS-outcome relationship. Our analysis provides helpful insights for managing the flow of psychiatric patients.

We study an extension of the cardinality-constrained knapsack problem where each item has a concave piecewise-linear utility structure. For the offline problem, we present a fully polynomial-time approximation scheme and show that it can be cast as the maximization of a submodular function with cardinality constraints; the latter result allows us to derive a greedy (1 − 1/e)-approximation algorithm. For the online problem in the random order model, we present a 6.027-competitive algorithm.

TBD

TBD

We present a model for studying a ride-hailing platform that is faced with price and delay sensitive riders and drivers, and is considering offering two different service classes which are differentiated in prices and delays. We explore the “price of two sides”, show that the preferences of drivers impact the delays the riders experience and demonstrate that achieving the “full optimum" of price differentiation may not be feasible or optimal for a platform in all market conditions.

Ride-hailing platforms offering shared rides devote effort to reducing improving shared-ride efficiency: reducing the trip-lengthening detours that accommodate fellow customers' divergent transportation needs. Contrary to naive intuition, we show: greater customer sensitivity to shared-ride delay and greater labor cost can reduce the value of improving shared-ride efficiency; and an increase in shared-ride efficiency can prompt a platform to add individual-ride service.

TBD

We study whether legalizing marijuana affects the operational efficiency of manufacturing facilities. We leverage a state-level quasi-experimental setting that arises with the staggered passage of marijuana laws. We find that medical marijuana legislation (MML) adversely affects operational efficiency of facilities, the average waste increased by 5.22 % after MML. Facilities start fewer managerial and technical modifications after MML, which clarifies the degrading mechanisms.

Retailers are increasingly sponsoring take-back initiatives to facilitate the recycling and reuse of secondhand clothing. To stimulate the return of goods, providers share information about circular economy strategies and offer consumers a small incentive in exchange for used clothing items. Through three experiments involving over 3,5000 subjects, we test how consumers’ willingness to return their used garments is affected as the degree of circular economy transparency and reward level is manipulated.

See attached.

This paper studies how service-age information influences customers’ strategic joining decisions in an observable M/G/1 queue. We ask how such information is reflected in customers' strategies at equilibrium, and what would be the corresponding system throughput and social welfare. The managerial insights indicate that a revenue-maximizing provider should disclose the service-age information when congestion is high, while a social planner should disclose the information when congestion is low.

We study a problem where a fixed number of servers must be split between a shared and a dedicated track. High-priority customers can be served on the dedicated track or the shared track with non-preemptive priority over low-priority customers. Low-priority customers are strategic and choose to either join the shared track or balk from the system based on wait time information. We use a queueing game to study the capacity allocation decision under different information revealing policies.

Strategic servers may discreetly disclose or hide queue length information to manage customers' joining behavior. For customer-intensive services, the choice of service rate further complicates this queue observability decision by affecting the service quality and consequently customers’ service rewards. In this work, we provide an integral analysis of managing speed-quality trade-off with the joint use of pricing and queue disclosure strategies.

Online labor markets are growing rapidly, where the numerical rating to workers provides a critical metric of service quality. In this study, we propose a two-stage machine learning framework to predict such quality. We apply our method on a unique from a leading online labor platform. We provide an understanding of how various features impact service quality. Then, we discuss the value of prediction in decision-making and provide actionable insights to improve service quality in OLMs.

Service providers often offer multiple variants of a service to their customers simultaneously through different servers. These servers could differ in their service value and expected service times, making one server preferable to customers over the other. However, even when the preferred server is busy and the less preferred server is free, customers may choose to wait for the preferred server. This behavior is called "stalling". Our goal is to price the servers in order to maximize revenue.

Ride-hailing platforms are highly popular so their decisions can affect social satisfaction levels significantly. The price determined by the platform is so critical decision factor in managing their demand and pool of available drivers. However, consumers are affected by pricing in different ways due to their heterogeneity of price or waiting time sensitivity.

We analyze a waiting time differentiation pricing strategy using a game-theoretic model in both monopoly and duopoly environments.

To address income inequality in "Superstar Markets", we propose income pools - a contract where individuals agree to share a portion of future earnings if they become successful. We develop the first math model and prove that no finite-sized stable pool exists. In response, we consider bounded stable pools and epsilon-stable pools, proving their existence and Pareto properties. Our case study on professional baseball shows a 20%-30% welfare increase, most acutely benefiting the weakest agents.

The gig economy relies on task outsourcing, but gig workers face challenges in finding locations and products. We conduct field experiments using AI-enabled guidance to help shoppers. Technology reduces refunds and benefits less experienced shoppers. However, complex routing algorithms increase consultation time and pickup times. AI improves efficiency and increases revenues, but technology adoption has limits and overuse can reduce productivity.

Motivated by collaboration with a clinic, we study open-shop service networks with two service level measures: macro-level average overall wait time and micro-level probabilities of excessive waits for individual services. In a stylized two-station open-shop network, we analytically show that an advanced customer priority and the buffer strategy can improve macro- and micro-level performance. We provide means to improve customers' experience in open shop service networks not applicable before.

We study the behavioral effects of process automation on human workers interacting with automated tasks. A stylized model with two workers completing their tasks sequentially predicts that workers do not delay their tasks if the early completion bonus is high enough. Our behavioral experiment shows that workers actually tend to delay their tasks. Process automation improves the project completion rate and time but reduces the productivity of the worker who collaborates with the robot.

TBD

How can retailers promote healthier food choices? Price, convenience, and taste are key factors. Healthy nudges can incentivize customers. In a field experiment, add-on bundles were tested: unhealthy, healthy, and choice bundles. Healthy snacks increased healthy purchases, even when unhealthy snacks were promoted. However, the effect was not long-term. Retailers can benefit by offering add-on choice bundles, increasing revenue and profit.

A project involves several participants – clients, contractors, and subcontractors – who each manage multiple projects concurrently. This creates a network of otherwise unrelated projects. Accordingly, a disruption in one project forces all parties to reallocate resources from other concurrent projects, causing externalities across the wider project network. We use data from 2.6 million U.S. public projects – and their networks – to quantify the importance of these network externalities.

We consider the capacity planning problem to decide the initial capacity to maximize the expected total profit when general upgrading is allowed. Given the marginal mean and variance information of the demand distribution for each product, we formulate it as a two-stage distributionally robust optimization (DRO) model. We show how to reformulate the DRO model into a tractable SOCP formulation and conduct extensive numerical experiments to validate the out-of-sample performance of the DRO solution.

Problem: The effectiveness of flexibility in mitigating demand and supply mismatch with unequal profits is unclear. Methodology/Results: We evaluate the effectiveness of a long-chain structure compared to full flexibility. Performance ratio lower bound shows effectiveness of some flexibility. Performance ratio increases with profit differentials. Managerial implications: Cluster high-profit products for optimal flexible structures.

Price promotions are an important and widely established retailer's tool to uplift sales, foster cross-selling through increased store traffic and strengthen customer relationships. Retailers hereby face a self-induced dilemma of balancing high product availability and waste. By empirically showing that price promotions are a food waste driver in grocery retail for highly perishable goods, our study helps to address one pressing global social, ecological and economic sustainability issue.

Inefficient food system operations create huge waste for the earth, and cut profit of firms. In this research, we address the simultaneous determination of pricing and inventory control for perishable products to maximize profit. The optimal policy is computationally intractable due to the curse of dimensionality. Instead, we construct a stationary base-stock list-price policy. We show our approximate policy is asymptotically optimal under several parameter regimes.

We study joint replenishment and issuing policy for a perishable product with general lifetime using a periodic review model over a finite horizon. Customers' sensitivity to the remaining lifetime of the unit received is captured by a goodwill cost, which increases as the remaining lifetime decreases. We find that contrary to the case when the issuing policy is fixed (e.g., last-in, first-out policy), the value function in our model is always jointly concave in the on-hand inventory vector.

Our study focuses on a preference learning setting where customers select k most preferred items from a personalized set. Our new Mallows-type ranking model offers simple closed-form (ranked) choice probabilities that can be learned through MLE with theoretical guarantees. We demonstrate the model's excellent performance with real data sets. We use our model to study how feedback structure is related to the efficiency of the feedback collection and find that a little favor goes a long way.

This paper applies a machine learning lens to the problem of choice estimation. We (1) establish the first truly realistic-scale benchmark for practical choice estimation, (2) use this benchmark to run the largest evaluation of choice models to date, and (3) propose and prescribe the use of a simple, irrational choice model, which we dub the Sparse Halo-MNL.

When allocating products to brick-and-mortar stores, retailers face product availability constraints that force them to balance product offerings across stores. We model this problem under multinomial logit demand and show that the problem is NP-complete. We develop a tractable continuous relaxation of the problem which has a unique local maximum and allows us to build near-optimal solution algorithms. We use data from a large retailer and identify improvements of better product-store matching.

We investigate the crucial and complex task of selecting patients by physicians in emergency departments when multiple patient types are present. Conventional approaches in guiding physician decision-making are based on patient triage levels or waiting times. However, an important factor that is often overlooked is the time remaining in a physician's shift. We utilize a transient optimal control on the corresponding queueing system to derive an optimal time-dependent patient selection strategy.

Given heterogeneous treatment time distributions, no-shows, and patients that can arrive early or late, we show how a discrete time Markov chain model can be used to equitably space patients so that no patient has an expected wait longer than some maximum value. The calculation of an equitable schedule can be done extremely fast, in polynomial time, involving only simple linear algebra (matrix multiplication). The equitable schedule adds at most 2% to the clinic’s operation time.

Emergency Department (ED) boarding refers to the delay in transfer experienced by admitted patients from the ED to inpatient units. Using an instrumental variable design, we found that, on average, longer boarding time leads to a longer hospital stay and a higher chance of care escalation. Our findings also reveal that the impact of boarding differs across patients, suggesting that considering such heterogeneity when assigning inpatient beds could improve downstream efficiency and quality of care

We study an adversarial online assortment problem with reusable resources and customer-dependent usage durations. We propose a novel online algorithm which features rejection durations filtering out unprofitable products. We show that it achieves a competitive ratio within a constant factor from the best possible one with large capacities. This is the first work that derives a non-trivial performance guarantee for such problem. We further extend our algorithm framework to other reward functions.

We consider an assortment optimization problem for a class of online video games. Our paper is the first to study assortment optimization for the gaming industry under discrete choice models; it is also the first to devise solution approaches for the constrained mixture-of-nested-logit model with performance guarantees. Numerical experiments show that our approaches perform well across a variety of settings. Our work provides guidance to online video game stores for effective revenue maximization.

We study a new class of stock-dependent pricing policies for reusable resources that set prices based on the stock at hand. To find the optimal policy in this class, we introduce a reformulation that is convex. We provide a sharp characterization of the regret of this policy class via matching upper and lower bounds and show they can significantly improve upon static pricing. A simple two-price policy that changes prices when the stock is below a threshold can achieve the optimal rate of regret.

TBD

TBD

Many multinational organizations have made dual commitments to halt deforestation and improve farmers’ livelihoods in agricultural supply chains. We propose group incentives conditional on forest protection requirements as a feasible mechanism for achieving this. We develop an analytical model, prove structural results concerning the equilibrium outcomes, and calibrate the model using data we collected from field research in Indonesia, demonstrating the effectiveness of group requirements.

We analyze competition between a sustainable firm and an unsustainable knockoff. We study open-source sharing and advertising strategies.

Sharing green tech can be profitable. Anti-knockoff ads may not be profitable, while self-promotion ads improve profits. Simultaneous use of open-source and ads can be beneficial but may harm the environment.

Firms can use open-source and ads to combat knockoffs. Governments can facilitate sustainable tech sharing through regulations.

In malaria-endemic countries, we explore optimal donor budget allocation for subsidizing antimalarial drugs in the presence of counterfeits. We develop a game-theoretic model and evaluate strategies to combat counterfeits. Our findings suggest the importance of understanding market characteristics to design effective subsidies and policies for improving access to legitimate drugs.

Prior discussions of predictive model development highlight advances in methods, but the value of tools that implement those methods has been understudied. In a field experiment, we study the importance of tools by restricting access to machine learning libraries in a prediction competition. We find that teams with unrestricted access perform 30% better, and teams with high general data-science skills are less affected by the intervention than teams with high tool-specific skills.

We investigate the remanufacturing strategy for the original equipment manufacturer (OEM) and independent remanufacturer (IR) in an innovative industry where the consumer valuation of the products increases with the innovation level and we consider the investments of an OEM to enhance innovation, in the face of a potential entry onto the market by an IR, together with two remanufacturing strategies: whether to remanufacture and whether to include innovative features in remanufactured products.

Optimization in sustainability is inherently multi-criteria and the underlying value function is usually unknown and difficult to elicit. Firms seeking to be more sustainable face difficult choices during material and supplier selection with no clear guidelines on how to make trade-offs between conflicting environmental impact categories. We propose an interactive optimization method to help decision makers, using an illustrative example of sustainable sourcing in the apparel industry.

This paper considers an industrial symbiosis system composed of a supplier and a manufacturer. The supplier produces product A with the output of by-product, which can be reused by the manufacturer as input to produce product B. Competition and cooperation are often juxtaposed in the same system. This paper examines the mode choice of the supplier, i.e., continue to cooperate to supply by-products to the manufacturer or generate direct competition by encroaching on the manufacturer's market.

Evidence shows national institutions driving innovation through openness. We focus on 'social globalization' and design innovation. Using the KOF Globalization Index, we find social globalization predicts design innovation. Robust analyses and U.S. design patents support our findings. Personal contact plays a key role. Social globalization fosters design innovation.

Problem: Despite advances, chatbot adoption faces hurdles. This paper explores customer choice between chatbots and live agents. Methodology: Experiments vary chatbot performance and features. Users respond positively to improvements but underutilize chatbots due to algorithm and gatekeeper aversion. Remedies: Highlight time savings for algorithm aversion. Managerial implications: Nudges and queue priority rules reduce costs by up to 22% in congested systems

Facing emergent business challenges, entrepreneurs often seek guidance from experienced advisors. When there are multiple alternatives that could potentially solve the entrepreneur’s problem, advisors can lead the entrepreneur’s exploration by choosing which alternative(s) to suggest and in what sequence. We develop a dynamic game-theoretic model that captures the sequential interaction between an advisor and an entrepreneur.

Even if algorithms make better predictions than humans on average, humans may sometimes have private information which an algorithm does not have access to, which can improve performance. When deciding how to use and adjust an algorithm’s recommendations we hypothesize people are biased towards a predictable heuristic leading to over-adhering to the algorithm’s predictions when their private information is valuable and under-adhering when it is not. We test these results in two lab experiments.

We develop a sequential decision-making task in the form of a virtual electric vehicle driving game in which the participant needs to make sequential charging decisions when facing uncertain traffic and receiving machine-generated recommendations on their strategy. Our experimental results offer key insights into how humans make decisions and respond to those recommendations, allowing us to design a better human-centric recommendation system.

We examine the effect of team familiarity on surgery duration. We develop a new approach, which we call the “MT forest” approach, to estimate heterogeneous effects of multiple treatments. We find (1) an increase in team familiarity score significantly reduces surgery duration, and (2) the effect of team familiarity is heterogeneous across patients with different features. Finally, we develop an optimization model to better match surgical teams with patients.

We study an inventory control problem called the One-Warehouse Multi-Store (OWMS) problem when the demand distribution is unknown. The OWMS system is ubiquitous in supply chain management, yet its optimal policy is notoriously difficult to calculate even under the complete demand distribution case. In this work, we consider the OWMS problem when the demand is censored, and its distribution is unknown a priori. Results show that our approach has great theoretical and empirical performances.

Using customer data for inventory management can improve profits and service but also increase privacy risk. We developed privacy-preserved adaptations for two data-driven newsvendor pipelines and analyzed the tradeoff between privacy loss, profits, and consumer surplus. We show that the joint approach outperforms the two-step approach. By accounting for downstream optimization problems, we can obfuscate customer data with more targeted noise injection, making it less costly in terms of profits.

Problem: Stochastic lost-sales inventory control with uncertain supply and demand is computationally challenging. We propose an efficient online learning algorithm for unknown distributions. Our algorithm achieves a regret of O(L+\sqrt{T}) when L≥log(T), outperforming existing literature. We address censored data using a coupling argument. Our method eliminates suboptimal solutions.

Staffing planning for last-mile delivery drivers is the process of planning the number of drivers that are required each week to deliver all the expected volume for a pre-determined time horizon, with the ability to adjust the decisions over time under guardrails restrictions. We formulate the problem as a multi-dimensional stochastic dynamic program with a newsvendor-based cost function and propose an approximation algorithm that can solve the problem to near optimality in tractable time.

The food delivery market migrates to platforms that allow optimizing courier routing by sharing couriers among many restaurants. We address the question: how does courier sharing contribute to the reduction of delivery costs? We consider a spatial queuing model in which couriers are servers. We show that in several scenarios dedicated courier policy achieves higher profit than a sharing policy. This result can be attributed to the imbalance in the courtier allocation that sharing creates.

The pandemic era experienced a significant amount of port congestion. Such congestion at one port spreads to another, leading to shipping delays and driving up costs for shippers. In this paper, we build an analytical fluid model to study how a disruption at a port would impact a disrupted port in one country and its counterpart port in another country.

We consider a decision-maker (DM) who can sample from multiple information sources to learn a state before making an earning decision. The DM optimizes his sampling and earning decisions to maximize his payoffs. The problem is motivated by financial and healthcare applications with multiple information sources. We develop a Bayesian control framework for this problem and solve it in the estimation and testing settings. We also develop an efficient algorithm for the general control setting.

TBD

The US heart transplant system prioritizes candidates based on the severity of their pre-transplant therapy; the premise is that this severity reflects medical urgency. However, it is widely suggested that this rule opens room for gaming the system, by assigning high-severity therapies to low-urgency patients. We study a novel model of the gaming decisions of heart transplant centers. We identify the underlying trade-offs and shed light on the conditions that give rise to such gaming.

In this paper, our goal is to (i) characterize the equilibrium behaviour of the agents under targeted priority mechanisms, (ii) identify the impact of these mechanisms on several performance metrics, including the discard rates of organs, and overall social welfare, and (iii) establish the impact of the selection of design parameters on different outcomes in equilibrium to investigate the optimal design of such mechanisms.

We optimize blood donor incentivization to reduce shortages and wastage. Our model considers random demand, perishability, and donor variability. Using a coupled queueing network approach and the Pipeline Queue paradigm, we derive a tractable convex reformulation. Results show improved performance compared to the threshold policy. It provides decision support for dynamic donor incentivization in blood supply chain management.

We consider the problem of optimal assortment design for a platform by imposing fairness constraints that guarantees exposures of all products to consumers. We show that optimal solution can be found in polynomial time and the algorithm is easy to implement in practice. We investigate the welfare change of imposing the fairness constraints and find that it is possible to achieve a win-win-win solution where the platform, sellers, and consumers are all better off.

Problem: Allocating perishable resources online with non-linear utilities and complementarities. Methodology/Results: An algorithm achieves optimal envy-efficiency Pareto frontier by adapting to perishing order predictions and desired envy bounds. Numerical performance is demonstrated using data from a food bank. Managerial Implications: Accurate perishing predictions are crucial for fairness in practical non-profit settings.

We study the marketplace platform's problem that decides the assortment of sellers and how much fee to apply to each seller. In our model, each seller has its own outside option, which depends on customer traffic and product preference. The seller's compensation is determined based on the competition with the other joining sellers. We characterize the platform's optimal fee and assortment decisions and further study how the different fee contracts affect the optimal assortment.

This paper investigates the roles of selling format in a two-sided marketplace with many sellers selling substitutable products on a common retailing platform. We show that a contribution-based scheme (CBS), whereby the payment for each seller is based on her contribution, leads to a stable, efficient, and `win-win' outcome for all firms in the entire marketplace. Our findings could provide useful guidance on the design of strategic partnership between firms in a two-sided marketplace.

TBD

We study how to optimally allocate a limited resource (community solar (CS) capacity) among consumers with heterogeneous income levels and private resource valuations. We identify the shortcomings of existing CS program designs, and study various alternatives. Our recommendation offers consumers income-dependent menus of subscription capacity and rate options; this significantly improves social welfare. We further illustrate its usefulness using numerical studies calibrated by real data.

US nonprofits disclose three expenses annually: program to meet their beneficiaries' needs; fundraising to raise donations; administration to build and maintain capacity. We characterize the optimal budget allocations to program, fundraising, and administration using a two-period model, which also includes the nonprofit’s capacity, return on program (the net value of program to beneficiaries) and uncertain future needs of beneficiaries. The optimal allocation depends on the nonprofit's capacity.

As consumers, policy makers and NGOs associate low environmental impact with “uniformly green” (UG) products or higher recycled/reused content, we explore two issues: (i) Should distinct green preferences be “targeted” with distinct product variants? (ii) Is profitable UG adoption environmentally superior? We show that optimal and eco-friendly UG adoption requires design/quality control and reduced production/material costs, and blindly demanding UG adoption may worsen environmental results.

.

We propose a multi-stage choice model in which customers can choose multiple products and multiple units of each product in a single shopping trip. We characterize the optimal assortment under the cardinality, space, and basket size constraints, respectively. We prove the NP-hardness of the problem under the latter two constraints and develop approximation algorithms to find near optimal assortments. We solve the price optimization problem efficiently and provide a calibration method.

This paper studies assortment optimization and product line design problems under the mixed multinomial logit model and discrete pricing. Both literature streams are connected by improving exact, extending approximate, and novel heuristic methods. We show that an FPTAS algorithm exists even if prices are taken into account. To improve the state-of-the-art conic formulation, valid inequalities are introduced to a branch and cut method. Our results show an average time reduction of 35 % - 66 %.

Many product recalls are caused by quality-related product failures. This paper analyzes quality and pricing strategies for competing firms facing the risk of a severe quality-related recall making the product hazardous and leading to its removal from the market. We develop a two-stage Nash game where the probability of recall depends on the firms’ chosen quality investments, and either firm can experience a recall.

Customers are concerned about personalized pricing (PP) tactics. Using a Bayesian persuasion framework, we study whether and under what conditions price can signal such PP implementation to customers. We also investigate whether disclosing inventory availability information can alleviate customer concerns and benefit the firm and customers. We show that price alone may not signal PP, and firms can create transparency over the pricing strategies by disclosing inventory availability information.

Patients self-triage for acute care but lack medical knowledge, leading to inaccurate decisions. Virtual triage tools aim to improve self-triage. We develop a queueing game model to assess virtual triage's operational efficacy. Excessive emergency primary care recommendations reduce ED GP visits. Informative virtual triage can worsen system performance. Policy actions should consider decentralized behavior, incentive alignment, and accuracy decisions based on ROC curve.

Problem: Crowdfunding with insiders and outsiders faces reciprocity and information asymmetry. Methodology/Results: High reciprocity hampers information transmission, benefiting outsiders. Information asymmetry lets outsiders gain a larger share. Increased reciprocity reduces investor payoff. Managerial Implications: Limiting investors or investment improves information transmission. Leverage social networks for better crowdfunding.

We study the implications of assortment variety on customer choice. Specifically, we are interested in characterizing and quantifying the effects of assortment variety on customer choice by proposing and operationalizing a novel representation of an assortment that measures variety along multiple dimensions. Moreover, we investigate potential moderating effects of assortment variety on customer choice, namely customer segments and seasonality. We test our model on a large clickstream data set.

Retailers often provide store inventory information on their websites, and it can influence customers' decisions to visit the store. For instance, when a customer wants to buy three units of a product and the website indicates three units are available, concerns about inventory insufficiency or inaccuracy may still arise. This paper investigates the impact of website inventory information on customer behavior and how retailers can optimize the displayed inventory information to maximize profits.

Prior research on loyalty programs typically finds rewards increase loyalty, without considering the impact of redemption size and consumer habits. We demonstrate these factors are significant predictors in fostering long-term loyalty, with smaller redemptions often outperforming larger ones. Our results demonstrate redemption is a key lever in loyalty program optimization.

We address the reliable facility location problem in a data-driven setting by presenting a model aiming to balance solution conservatism with efficiency. In particular, our model approximates the total cost by a tractable data-driven estimator, which equals to a probabilistic upper bound on the intractable Kolmogorov DRO estimator. Our approach is proved to be asymptotically optimal, and offers a theoretical guarantee for its out-of-sample performance in situations with limited data.

Supply chain models often receive little testing due to a lack of data. We propose a random model of supply chain networks to overcome this problem and establish that it generates accurate representations of real networks. Generated networks' treewidth is logarithmic in the number of firms, which has important implications for tractability. We illustrate this with the NP-hard guaranteed service model, showing a pseudo-polynomial time algorithm for networks with logarithmic treewidth.

This paper proposes an empirical framework for analyzing the evolution patterns of supply chain networks over time. Using a semiparametric matrix variate factor model, we investigate the latent lower-dimensional structure of the network dynamics and the loading matrices that connect the underlying latent factors with the surface supply chain networks and characterize the latent nodes. Our findings shed light on the latent structure, centrality, trends, and patterns of supply chain networks.