Events & Seminars

Consumers often receive a full or partial refund for product returns or service cancellations. Much of the existing literature studies cash refunds, where consumers get their money back minus a fee upon a product return or service cancellation. However, not all refunds are issued in cash. Sometimes consumers receive credits that can be used for future purchases, often times with an expiration date after which the credits are forfeited. A prominent example is the airline industry, where consumers who purchase non-refundable fares are often issued a credit that is valid within a fixed time window (typically a year) upon ticket cancellation. We study the optimal design of credit refund policies. Different from models that consider cash refunds, we explicitly model repeated interactions between the seller and consumers over time. We assume that consumers’ valuation for the product/service varies over time, and that there is an exogenous probability for product returns. Several interesting results emerge. First, a credit refund policy facilitates intra-consumer price discrimination for a single type of consumers with stochastic valuation. Second, an optimal policy often involves an intermediate credit expiration term, under which a consumer with a high product valuation always makes a purchase, while a consumer with a low product valuation may be induced to make a purchase as the credit approaches expiration, leading to a demand induction effect. Finally, a credit refund policy can be more profitable than a cash refund policy, and can lead to a win-win outcome for both the firm and consumers under certain conditions. We also consider several extensions to check the robustness of our findings.

This paper investigates the nature of social learning (SL) on content platforms and its impact on optimal content design. The content provider can choose low- or high-quality content and, in addition, may also polarize (as opposed to keeping it neutral) the content in favor of some consumers while opposing other consumers’ opinions/preferences. On the content platform, SL manifests as consumers’ inference about the unknown quality of content using the history of past consumption, based on which they make consumption decisions. We specify a behavioral model of SL that accounts for and illustrates the impact of false consensus effect (FCE) --- a cognitive bias wherein consumers project their own preference onto others --- on the SL outcome. In this environment, we find that whether the SL mechanism reveals the true quality of content depends largely on the interaction between content polarization and the degree of the FCE. Depending on the extent of the interaction, SL may be incomplete or even cursed in the sense that beliefs converge to a limit where history offers no information about quality. The optimal content design internalizes the SL dynamics and as a result, we find that quality and polarization can be used as substitutes by the content provider. In particular, content may be polarized to mask its low quality. Interestingly, we also find that SL increases the incentive for the content provider to increase quality compared to a benchmark without SL. Furthermore, we find parametric regimes in which SL may not be beneficial to consumers, but is in fact preferred by the content platform (and vice versa). In this sense, the value of SL may be misaligned between the platform and its consumers.

We consider a joint assortment optimization and customization problem under a mixture of multinomial logit models. In this problem, a firm faces customers of different types, each making a choice within an offered assortment according to the multinomial logit model with different parameters. The problem takes place in two stages. In the first stage, the firm picks an assortment of products to carry subject to a cardinality constraint. In the second stage, a customer of a certain type arrives into the system. Observing the type of the customer, the firm customizes the assortment that it carries by, possibly, dropping products from the assortment. The goal of the firm is to find an assortment to carry and a customized assortment for each customer type that can arrive in the second stage to maximize the expected revenue from a customer visit. The problem arises, for example, in online platforms, where retailers commit to a selection of products before the start of the selling season, but they can potentially customize the displayed assortments for each customer. We give an approximation algorithm that obtains 1/log m fraction of the optimal expected revenue, where m is the number of customer types. Contrasting this problem with the variant where customization is not possible, it is NP-hard to approximate the latter variant within a factor better than 1/m. Thus, from computational complexity perspective, the variant with customization is fundamentally different.

Motivated by an intriguing observation on the time-varying physician productivity (measured by the number of new patients seen per hour, or PPH), we study a continuous-time optimal control problem to understand the transient behavior of individual physicians within their shifts in emergency departments (EDs). By applying Pontryagin’s maximum principle, we characterize the optimal policy and provide insights into physician capacity, productivity, and throughput. We conclude that the transient behavior is intrinsic, mainly induced by shift-based scheduling. We leverage the insights from the time-varying PPH to model the complex ED system as a time-varying multi-server queue with shift-hour-dependent service rates. Validated using data from two Canadian EDs, the simulation results show that our model can accurately capture the time-of-day-dependent patient waiting times with a simple parameter estimation procedure, which is among the first in the literature. In contrast, the simulated waiting times under constant PPH rates deviate significantly from the data. Hence, it is important to explicitly consider time- varying service rates to obtain accurate and valuable models for ED operations. The essence of our model is dimension reduction by state aggregation. As a result, it also allows transient analysis through the uniformization of a continuous- time Markov chain, which can be integrated with off-the-shelf algorithms for physician staffing. Our case study using data from a Canadian ED shows that the new staffing schedule generated from our method can significantly improve the current schedule in practice.

We study how Emergency Department (ED) patients incorporate announced ED waiting time in their decision of choosing which ED to attend. Using a discrete choice framework, we structurally estimate the patients’ sensitivity to announced ED waiting time and potential travel distance to the ED. We find that approximately 30% of ED patients in Hong Kong are sensitive to the announced waiting time while the remaining 70% are not and their decisions are mainly driven by the distance to the ED only. Patients that are sensitive to the announced waiting time would travel an additional 1 km to save approximately 4 hours of waiting at the ED. We also study patient characteristics that differentiate their sensitivity to ED waiting time.

Original equipment manufacturers (OEMs) often source their components from two different suppliers to mitigate supply risks and enhance their bargaining power. However, a notable consequence of dual sourcing is heterogeneity in ex post quality among the final products from different suppliers. Such quality differences can cause peer-induced fairness concerns among consumers who receive products of lower quality: They feel unfair that they have paid the same price as others but receive inferior products. We characterize the effect of the ex post product quality heterogeneity induced by sourcing from different suppliers and the resulting consumer fairness concerns. We examine how the quality differences and fairness concerns affect an OEM’s sourcing strategy selection, the supplier’s wholesale pricing, and the OEM’s optimal ordering decision.

This paper investigates how the presence of the spillover and carryover effects in the multi-channel ad campaign affects the budget allocation decisions of a marketing agency, which strives to maximize the total expected number of clicks or conversions over the campaign. A salient feature of the problem is that the market agency only has access to aggregate data such that the effectiveness of different online advertising channels cannot be estimated using standard methods that typically require individual-level data. The authors propose a data augmentation method for estimating the microlevel consumer advertising response models using aggregate data. The essence of this approach is to simulate latent state dynamics such that the generated data is consistent with the observed aggregate data. The authors then demonstrate the validity of the method using actual channel-level advertising campaign data from an online fashion retailer in Korea. Lastly, the authors study a fluid mean-field formulation and derive key structural insights on the optimal budget allocation policies, which are leveraged to design an implementable budget allocation policy.

Cancer remains one of the leading causes of human death, and early detection is the key to reducing mortality. To detect cancer in the early stages, two-stage screening programs are widely adopted in practice. Individuals receiving positive outcomes in the first-stage (initial) test are recommended to undergo a second-stage test for further diagnosis. The initial test design—i.e., selecting cutoffs to report test outcomes—is crucial for screening effectiveness (i.e., cancer detection) and efficiency (i.e., second-stage capacity costs). However, not all individuals who receive positive outcomes follow up with the second-stage test; evidence shows that adherence behavior is closely associated with the cutoff used in the initial test. This paper studies the initial test design in the context of colorectal cancer (CRC) screening to balance the trade-off between screening effectiveness and efficiency and takes into account individuals’ guideline adherence behavior.

We adopt a Bayesian persuasion framework with information avoidance to model the initial test design and individuals’ response to screening guidelines. We analytically prove that under certain conditions, an initial test using a single cutoff (i.e., a dichotomous test) is optimal for screening follow-up maximization, and a continuous test (i.e., showing exact readings of the biomarker) is optimal for screening effectiveness maximization. We apply the framework to Singapore’s CRC screening guideline design and calibrate the model using various sources of data, including a nationwide survey in Singapore. Our results suggest that compared with the current practice, increasing the cutoff to the level that maximizes expected follow-ups by cancer patients can detect 969 more CRC incidences and prevent 37,820 colonoscopies, which are the second-stage test for CRC screening. Aiming only for high-sensitivity initial tests using lower cutoffs (as in the current practice) can backfire and lead to large numbers of unnecessary colonoscopies and low follow-up rates from cancer patients. We further explore the benefits of using different cutoffs for different subpopulations and use an interpretable clustering technique to construct implementable rules for partitioning the population. We demonstrate that using a lower cutoff for males older than 60 and females older than 70 (high-risk and high-adherence groups) and a higher cutoff for the rest of the screening population (low-risk and low-adherence groups) can further improve screening effectiveness and efficiency.

Cancer remains one of the leading causes of human death, and early detection is the key to reducing mortality. To detect cancer in the early stages, two-stage screening programs are widely adopted in practice. Individuals receiving positive outcomes in the first-stage (initial) test are recommended to undergo a second-stage test for further diagnosis. The initial test design—i.e., selecting cutoffs to report test outcomes—is crucial for screening effectiveness (i.e., cancer detection) and efficiency (i.e., second-stage capacity costs). However, not all individuals who receive positive outcomes follow up with the second-stage test; evidence shows that adherence behavior is closely associated with the cutoff used in the initial test. This paper studies the initial test design in the context of colorectal cancer (CRC) screening to balance the trade-off between screening effectiveness and efficiency and takes into account individuals’ guideline adherence behavior.
We adopt a Bayesian persuasion framework with information avoidance to model the initial test design and individuals’ response to screening guidelines. We analytically prove that under certain conditions, an initial test using a single cutoff (i.e., a dichotomous test) is optimal for screening follow-up maximization, and a continuous test (i.e., showing exact readings of the biomarker) is optimal for screening effectiveness maximization. We apply the framework to Singapore’s CRC screening guideline design and calibrate the model using various sources of data, including a nationwide survey in Singapore. Our results suggest that compared with the current practice, increasing the cutoff to the level that maximizes expected follow-ups by cancer patients can detect 969 more CRC incidences and prevent 37,820 colonoscopies, which are the second-stage test for CRC screening. Aiming only for high-sensitivity initial tests using lower cutoffs (as in the current practice) can backfire and lead to large numbers of unnecessary colonoscopies and low follow-up rates from cancer patients. We further explore the benefits of using different cutoffs for different subpopulations and use an interpretable clustering technique to construct implementable rules for partitioning the population. We demonstrate that using a lower cutoff for males older than 60 and females older than 70 (high-risk and high-adherence groups) and a higher cutoff for the rest of the screening population (low-risk and low-adherence groups) can further improve screening effectiveness and efficiency.

There is an increased interest in bundle selling mechanisms especially with the rise of subscription services. This rise was mainly fueled by the success of subscription services in the digital markets where inventory is unlimited. However, recently there is a slew of subscriptions services that emerged in the retail industry where inventory is limited. In this paper, we take a first step towards understanding the impact of key operational metrics such as inventory levels and limited selling horizons on the optimal bundle selling strategy. We study a dynamic bundle pricing problem when the firm is selling multiple items but with limited inventory. We propose a new scaling regime to study this problem, called high-demand regime, where we scale the arrival rate in order to capture markets where demand is high but inventory is limited. Our results highlight a fundamental limitation of bundling in such markets. Firms should avoid bundling fast moving items together and should rather sell them separately (or bundle fast moving items with slow moving items). Moreover, depending on the tail of the valuation distribution, the firm should either consider static pricing of the items or dynamic pricing. We provide closed form solutions for the static and dynamic pricing policies.