Hotels that offer flexible cancelation terms often witness a high number of canceled bookings. A common assumption in the hotel industry is that cancelations are driven by random and exogenous factors beyond customers’ control. However, in transaction data obtained from a high-end U.S. hotel partner, we find that the cancelation rate increases substantially with the booking price. This points to the possibility that the booking price can be an important driver for cancelations. To investigate this relationship, we employ a hazard model and find empirical support. Specifically, a $50 increase in the booking price results in a 16% increase in the hazard of a cancelation. This finding is robust to several alternative model specifications, different subsets of the data, and an alternative operationalization of the booking price. Combining the reservation data with price data from competing hotels nearby, we test for customers’ continued price search after booking as a potential mechanism and show that it mediates 27% of the total effect of booking price on the hazard of a cancelation. Driven by this empirical evidence, we conduct a counterfactual analysis and find that the hotel may lose as much as 11% in revenue during high season by ignoring the effect of pricing on cancelation.

Tabular foundation models (TFMs) generate predictions on structured data via in-context learning, without task-specific estimation. We ask whether TFMs can be effectively applied to discrete choice, a central demand estimation framework in marketing and operations, and find that directly applying TFMs yields limited performance. The gap is structural: TFMs assume row-independent observations, whereas discrete choice is inherently set-valued and subject to persistent consumer preference heterogeneity. We propose a reformulation that encodes both choice-set dependence and individual heterogeneity within a row-based learning framework. Evaluated on a yogurt scanner panel, individual-level heterogeneity encoding is the dominant driver of predictive accuracy. The best reformulation outperforms hierarchical Bayesian estimation by 8% in holdout log-likelihood and 3.6% in hit rate, running 16 times faster, a practical advantage for large-scale demand estimation. The advantage is largest in the medium-data regime (10–40 purchase occasions per consumer), where parametric Bayesian shrinkage most distorts estimates for atypical consumers. Fine-tuning on population choice data provides additional gains for consumers with shallow purchase histories, where in-context learning has limited individual-specific signal to condition on. These results establish a principled approach for applying foundation models to consumer choice problems more broadly.

A stream of recent research considers the practice of random price discounts (a.k.a randomized pricing) when selling to forward-looking customers and shows that such pricing strategies can mitigate strategic customer waiting and boost seller profit. In practice, random price discounts are often offered together with price guarantees, in which customers are refunded the price difference if the price is lowered within a given time window after purchase. This paper investigates the efficacy of price guarantees under randomized pricing. To that end, we consider a model in which a firm adopts Markovian pricing and interacts with customers over an infinite time horizon. The following results are obtained. First, while Markovian pricing allows firms to price discriminate customers based on their monitoring costs, price guarantees further allow firms to price discriminate customers based on their willingness to pay. Second, offering price guarantees under Markovian pricing can help retain customers effectively by inducing high-valuation customers to purchase early, regardless of their arrival time. Third, even with price guarantees, Markovian pricing can dominate static pricing only when high-valuation customers are more likely to have a high monitoring cost, which illustrates that customer composition plays a crucial role in the effectiveness of the firm's pricing strategy. Fourth, the optimal duration of price guarantees is closely related to customers' lifetime duration. Finally, perhaps surprisingly, offering price guarantees can decrease the aggregate customer surplus since the firm offers sale prices less often under price guarantees.

Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Copy DOI

Admission decisions for loss systems accessed by multiple customer classes are a classical queueing control problem with a wide variety of applications. When a server is available, the decision is whether to admit an arriving customer and collect a lump-sum revenue. The system can be modeled as a continuous-time infinite-horizon dynamic program, but suffers from the curse of dimensionality when different customer classes have different service rates. We use approximate linear programming to solve the problem under three approximation architectures: affine, separable piecewise linear and finite affine. The finite affine approximation is a recently proposed generalization of the affine approximation, which allows for non-stationary parameters. For both affine and finite affine approximations, we derive equivalent, but more compact, formulations that can be efficiently solved. We propose a column generation algorithm for the separable piecewise linear approximation. Our numerical results show that the finite affine approximation can obtain the tightest bounds for 75% of the instances among the three approximations. Especially, when the number of servers is large and/or the load on the system is high, the finite affine approximation always achieves the tightest bounds. Regarding policy performance, the finite affine approximation has the best performance on average compared to the other two approximations and the achievable performance region method (Bertsimaset al., 1994, Kumar and Kumar, 1994). Furthermore, the finite affine approximation is 4 to 5 orders of magnitude faster than the achievable performance region method and the separable piecewise linear approximation for large-scale instances. Therefore, considering bounds, policy performance, and computational efficiency, the finite affine approximation emerges as a competitive approximation architecture for the class of problems studied here.

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 the money back minus a fee upon a product return or service cancellation. However, not all refunds are issued in cash. Sometimes consumers receive credit that can be used for future purchases, oftentimes with an expiration term after which the credit is forfeited. 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 intraconsumer 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, whereas 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 was accepted by Jeannette Song, operations management. Funding: Y. Liu was substantially supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. PolyU15502420). Y. Liu was also supported by a grant from NSFC [72293564/72293560]. Supplemental Material: The data and online appendices are available at https://doi.org/10.1287/mnsc.2020.03042 .

Choosing an appropriate selling model for online retailing is a crucial problem faced by many e -commerce platforms and manufacturers. We examined three popular selling models on e-commerce platforms-the reselling model, the agency selling model, and the advertising service model-and exam-ined the impact of product features, service efficiency, and interfirm power relationships on the choice of selling models. The problem was analyzed as a two-stage bargaining problem between a platform and a manufacturer under service-sensitive demand. We show that the choice of selling models is in-fluenced by the interfirm power relationship, firms' service efficiencies, and demand sensitivities. Specifi-cally, when the manufacturer dominates the bargaining game, it is always beneficial to choose the agency selling model, except when the manufacturer has lower service efficiency than the platform and customer demand is extremely sensitive to service. In contrast, when the platform has the dominant bargaining power, it should always opt for the reselling model if it has better service efficiency than the manufac-turer. As the price and service sensitivities of customer demand increase, their influences in determining the optimal selling model increase. The advertising service model becomes a viable option only when the manufacturer has superior service efficiency but less bargaining power than the platform and when there is low price sensitivity but high service sensitivity. In addition, we considered a two-part tariff sell-ing model that includes a fixed membership fee and a unit transaction rate, and different direct sales costs for the platform under the three selling models. (c) 2022 Elsevier Ltd. All rights reserved.

This paper studies a multi-stage multi-product production and inventory planning problem with random yield derived from the cold rolling process in the steel industry. The cold rolling process has multiple stages, and intermediate inventory buffers are kept between stages to ensure continuous operation. Switching products during the cold rolling process is typically very costly. Backorder costs are incurred for unsatisfied demand while inventory holding costs are incurred for excess inventory. The process also experiences random yield. The objective of the production and inventory planning problem is to minimize the total cost including the switching costs, inventory holding costs, and backorder costs. We propose a stochastic formulation with a nonlinear objective function. Two lower bounds are proposed, which are based on full information relaxation and Jensen’s inequality, respectively. Then, we develop two heuristics from the proposed lower bounds. In addition, we propose a two-stage procedure motivated by newsvendor logic. To verify the performance of the proposed bounds and heuristics, computational tests are conducted on synthetic instances. The results show the efficiency of the proposed bounds and heuristics.

A Novel and Promising Approximation for Network Revenue Management In “Product-Based Approximate Linear Programs for Network Revenue Management,” Zhang, Samiedaluie, and Zhang propose a novel separable piecewise linear (SPL) approximation for the network revenue management problem. The coefficients of the proposed SPL approximation can be interpreted as each product’s revenue contribution to the value of each resource in a given period, which provides more granular information compared with the existing resource-based SPL approximation in the literature. The new approximation provides more flexibility for policy construction. Furthermore, the new approximation opens the opportunity to derive a set of valid inequalities to further improve the computational performance and achieve additional gains in the expected revenue. Computational experiments with instances of various network structures and parameters demonstrate its efficacy: the new approximation leads to bid-price policies generating higher expected revenues and demonstrates better performance in terms of both computational efficiency and numerical stability.

Dynamic pricing is widely adopted in many industries, such as travel and insurance. These industries are also gaining extensive capabilities in identifying and segmenting customers, partly fueled by the increasing availability of data. It is natural to ask whether firms should take advantage of such developments by charging different prices to different customer segments. If so, under what conditions? We seek answers to these highly managerially relevant questions. We consider a market with two customer segments served by a monopolist. The monopolist can choose among a set of pricing strategies to exploit consumers' inter-temporal preferences and/or inter-segment variations. At one end of the spectrum, the firm can charge a constant price to all customers, which is called static pricing. At the other end of the spectrum, the firm can charge different prices to different customer segments and vary these prices over time, which is referred to as dynamic targeted pricing. We systematically compare these alternative pricing strategies. We show that dynamic pricing without targeting can be more effective than static targeted pricing when customers are not very forward looking, which corroborates the findings in the empirical literature. Interestingly, we find that the monopolist can be worse off when she adopts targeting in addition to dynamic pricing. We conduct laboratory experiments to test several key model predictions. The studies show that individuals behave in a manner consistent with the predictions of our model. (c) 2022 Elsevier B.V. All rights reserved.

We study the effects of unscheduled service in an outpatient clinic, where physicians have discretion to admit walk-in patients who are not scheduled in advance when all appointment slots are fully booked. Admitted patients are appended to the end of treatment queues. Unscheduled service is a form of capacity expansion and helps the hospital mitigate the imbalance between supply and demand. However, it increases physicians' workload and may have a negative impact on the quality of care. Given the prevalence of unscheduled service in hospitals, it is important to study its effects. We utilize a unique dataset from a children's hospital in China to empirically study the effects of unscheduled service. An instrumental variable approach is employed to quantify the causal impacts of unscheduled service on service efficiency and the quality of care. Unscheduled patients have significantly shorter average service time, characterized by fewer diagnostic tests and shorter face-to-face diagnostic time. Surprisingly, there is no evidence that the care quality measured by 30-day readmission rate and 7-day revisit rate is lower for unscheduled service. This result can be attributed to physicians' behavior in patient consultation and prescription. Our analyses also show that the unscheduled service only has limited impact on scheduled patients.Understanding the effects of unscheduled service on service efficiency and quality of care allows physicians and hospital managers to make informed decisions on workload planning, capacity allocation, and long-term capacity planning.

Opaque selling, in which a seller offers opaque goods (OGs), in addition to physical goods, has been shown to be an effective strategy to segment a market and improve the seller's profit. This article studies opaque selling with stochastic demand and fixed initial inventories of multiple products, where the seller dynamically controls the product offers and determines the product assignment to fulfill the demand for OGs over time. The problem is formulated as a stochastic dynamic program. Due to the curse of dimensionality, we study the fluid control problem that gives a time‐based fluid policy and a stationary probabilistic fulfillment strategy. We show that the fluid policy is asymptotically optimal when the arrival rates and initial inventory level are scaled up linearly. Furthermore, we propose a decomposition heuristic based on the corresponding fluid solution. The decomposition heuristic is shown to provide a tighter upper bound than the fluid control problem. Numerical study on a set of test instances illustrates the performance and efficacy of opaque selling.

Approximate linear programs (ALPs) have been used extensively to approximately solve stochastic dynamic programs that suffer from the well‐known curse of dimensionality. Due to canonical results establishing the optimality of stationary value functions and policies for infinite‐horizon dynamic programs, the literature has largely focused on approximation architectures that are stationary over time. In a departure from this literature, we apply a nonstationary approximation architecture to an infinite‐dimensional linear programming formulation of the stochastic dynamic programs. We solve the resulting problems using a finite‐horizon approximation. Such finite‐horizon approximations are common in the theoretical analysis of infinite‐horizon linear programs, but have not been considered in the approximate linear programming literature. We illustrate the approach on a rolling‐horizon capacity allocation problem using an affine approximation architecture. We obtain three main results. First, nonstationary approximations can substantially improve upper bounds on the optimal revenue. Second, the upper bounds from the finite‐horizon approximation monotonically decrease as the horizon length increases, and converge to the upper bound from the infinite‐horizon approximation. Finally, the improvement does not come at the expense of tractability, as the resulting ALPs admit compact representations and can be solved efficiently. The resulting approximations also produce strong heuristic policies and significantly reduce optimality gaps in numerical experiments.

Redemption hurdles, such as finite expiration terms and redemption thresholds, are common for customer reward programs. In “An Analysis of ‘Buy X, Get One Free’ Reward Programs,” Yan Liu, Yacheng Sun, and Dan Zhang study the economic rationale behind redemption hurdles and how they should be optimally set. They show analytically that redemption hurdles can be used as a price-discriminating vehicle that increases firm profitability. Redemption hurdles can facilitate the firm’s price discrimination on consumers whose valuations may vary over time. Redemption threshold alone cannot ensure profitability, unless it is coupled with a finite expiration term or a positive transaction utility from the rewarded free product. Optimal design of redemption hurdles is not straightforward, and the interdependence between the two types of redemption hurdles and the price is nontrivial. Optimally set redemption hurdles may not only increase firm profitability but also, increase the welfare of consumers who purchase frequently.

Firms often vary product prices over time to price discriminate customers. In response, customers may delay purchases to obtain the product at a more favorable price. We consider a model in which a firm interacts with short-lived customers over an infinite time horizon. Customers differ in their valuations and lifetime. A proportion (type I customers) leave immediately with or without a purchase, whereas the remainder (type II customers) are willing to wait for a time that is exponentially distributed before making a purchase or leaving. The firm adopts a Markovian pricing strategy. We show that the optimal pricing policy is either static pricing or high/low pricing with flash sales, where the firm charges a high price all the time, except for occasional price drops. Moreover, customer heterogeneity affects the profitability of Markovian pricing. Specifically, when type II customers are more likely to have low valuations, the firm is more likely to offer high/low pricing. To mitigate customers' waiting behavior, some firms offer price guarantees that refund customers the price difference in the event of a markdown within a given time after product purchase. We show that offering price guarantees is not optimal when all customers take advantage of them. When only type II customers take advantage, offering price guarantees can improve the firm's profit because it enables price discrimination between high- and low-valuation type II customers. Moreover, the firm offers sale prices less often under price guarantees. Perhaps surprisingly, offering price guarantees can decrease the aggregate customer surplus.

This paper studies joint quality and refund policy design for service products in the presence of customer valuation uncertainty.

Much of the network revenue management (NRM) literature considers capacity control problems where product prices are fixed and the product availability is controlled over time. However, for industries with imperfect competition, firms typically retain some pricing power and dynamic pricing models are more realistic than capacity control models. Dynamic pricing problems are more challenging to solve; even the deterministic version is typically nonlinear. In this study, we consider a dynamic programming model and use approximate linear programs (ALPs) to solve the problem. Unlike capacity control problems, the ALPs are semi‐infinite linear programs, for which we propose a column generation algorithm. Furthermore, for the affine approximation under a linear independent demand model, we show that the ALPs can be reformulated as compact second order cone programs (SOCPs). The size of the SOCP formulation is linear in model primitives, including the number of resources, the number of products, and the number of periods. In addition, we consider a version of the model with discrete price sets and show that the resulting ALPs admit compact reformulations. We report numerical results on computational and policy performance on a set of hub‐and‐spoke problem instances.

A little-understood phenomenon of customer reward programs is the prevalent use of finite reward expiration terms. We develop a theoretical framework to investigate the economic rationale behind this phenomenon and the trade-off between short and long expiration terms. In our model, a monopolistic firm sets the expiration term, along with the price and reward size, and interacts with consumers over an infinite horizon. Consumers are heterogeneous in shopping probabilities and product valuations and forward-looking in making purchase decisions. We find that a customer reward program with a finite expiration term can increase firm profits when (i) the valuation difference within the consumer population is intermediate and (ii) the shopping probabilities and valuations are negatively correlated among consumers. Several model extensions confirm the robustness of these results. Finally, we conduct an empirical investigation on the reward program practice of the top 100 U.S. retailers, which provides directional support for several key theoretical predictions. This paper was accepted by Gad Allon, operations management.

Network loan platform has become the main mode of Internet finance, it provides its capital suppliers and demanders with convenient and efficient financing channels, but also advances the vigorous development of the financial industry.In this paper, there is analysis of risk of network loan platform and the current situation, which puts forward the corresponding countermeasures for network credit risk and these will promote the healthy development of Internet finance.