Courier Dispatch in On-Demand Delivery

In recent years, home service platforms have experienced substantial growth. These platforms typically employ commission-based contracts to incentivize suppliers to offer a variety of services to consumers. To maximize profits, platforms must strategically optimize their dispatch strategies, contract design, and product quality decisions. We develop a model to analyze these decisions in an environment in which suppliers differ in service efficiency and consumers vary in their willingness to pay for quality. Consumers with higher (lower) willingness to pay for quality are classified into high (low) segments. When the low segment becomes particularly influential—because of its large population share and relatively high willingness to pay—the platform may find it optimal to induce high-efficiency suppliers to provide low-quality products. This tendency is further amplified when the matching probability for low-quality products is high. Conversely, when the heterogeneity between consumer segments is sufficiently small, the platform prefers to induce high-efficiency suppliers to provide high-quality products, resulting in upward quality distortion for the high segment. In this case, the quality gap between the two offerings increases with the low segment’s willingness to pay for quality. Funding: H. Xiong acknowledges the financial support of the National Natural Science Foundation of China [Grants 72371107 and 71971091]. Y.-J. Chen acknowledges the financial support of the Hong Kong Research Grants Council [GRF 16212821 and GRF 16204521] and Crown Worldwide Professorship. L. Hsiao acknowledges the financial support of the National Science and Technology Council [Grant MOST 110-2410-H-005-016-MY3]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/serv.2024.0151 .

In the last two decades, last-mile delivery (LMD) firms have seen immense growth fueled by the success of e-commerce, leading to faster and cheaper deliveries. Operating on thin margins, LMD firms strive for successful first-time deliveries to avoid the financial and reputational costs of reattempts. Delivery agents (DAs) are integral to LMD efficiency, influencing customer experience, delivery success, and productivity. However, most LMD performance enhancement research focuses on process, technology, and incentives, which presume workers will conform to procedures and monitoring tools will function flawlessly. Nevertheless, in practice, DAs deviate from expected behaviors, that is, indulge in misconduct, negatively affecting delivery efficiency, often resulting in returned parcels. One of the major forms of misconduct is entering fake remarks about deliveries, wherein DAs intentionally do not deliver the parcels and provide fake reasons for it. For instance, even without reaching a delivery address, a DA remarks “customer unavailable” and records a delivery failure. In this study, we collaborated with a leading Indian LMD firm and, using instrumental variable regression, found that such misconduct leads to a spillover productivity loss. This effect reduces the next day’s successful deliveries by 1.60% and first-time-right deliveries by 1.86%. We discuss misconduct’s correlation with factors such as task complexity and offer novel insights into how opportunistic circumstances can influence worker behavior. This paper was accepted by Elena Katok, operations management. Funding: V. Choudhary acknowledges the support received from Ministry of Education, Singapore [Grant RS12/20]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.01829 .

We propose a data-driven portfolio selection model that integrates side information, conditional estimation, and robustness using the framework of distributionally robust optimization. Conditioning on the observed side information, the portfolio manager solves an allocation problem that minimizes the worst-case conditional risk-return tradeoff, subject to all possible perturbations of the covariate-return probability distribution in an optimal transport ambiguity set. Despite the nonlinearity of the objective function in the probability measure, we show that the distributionally robust portfolio allocation with a side information problem can be reformulated as a finite-dimensional optimization problem. If portfolio decisions are made based on either the mean-variance or the mean-conditional value-at-risk criterion, the reformulation can be further simplified to second-order or semidefinite cone programs. Empirical studies in the U.S. equity market demonstrate the advantage of our integrative framework against other benchmarks. Funding: The material in this paper is based on work supported by the Air Force Office of Scientific Research [Award FA9550-20-1-0397]. Additional support is gratefully acknowledged from the National Science Foundation [Grants 1915967, 1820942, and 1838676], the Natural Sciences and Engineering Research Council of Canada [Grant RGPIN-2016-05208], and the China Merchant Bank. V. A. Nguyen gratefully acknowledges the generous support from the Chinese University of Hong Kong [Improvement on Competitiveness in Hiring New Faculties Funding Scheme] and the Chinese University of Hong Kong [Direct Grant 4055191]. S. Wang is partially supported by the National Natural Science Foundation of China [Grant 72371022]. Finally, this research was enabled in part by support provided by Compute Canada. Supplemental Material: The computer code and data that support the findings of this study and the online appendix are available within this article’s supplemental material at https://doi.org/10.1287/opre.2021.0243 .

Choice models, which specify the consumers’ choice probabilities of options over a given choice set, are widely studied and applied in many fields. We propose a temporal tree representation of choice that covers all rational choice models. Compared with existing structural choice models, the tree representation exhibits two major advantages that address the key challenges of model identification. First, all rational choice models have a tree representation, and a subclass of tree representation (with set-dependent branching) has a one-to-one correspondence with the rational choice models. This bridges the gap in the existing characterizations of structural models, which are unidentifiable, impose uninterpretable conditions, or do not cover the entire space of rational choice models. Second, the tree representation allows for the flexibility of systematically specifying the choice model structure based on available knowledge and data. In particular, the number of parameters needed to specify a tree representation can be primarily determined by the sufficient knowledge level, which corresponds to a specific layer of the tree branching. The sufficient knowledge level can be empirically determined based on the amount of available data, which, in turn, determines the number of parameters needed for model estimation. Therefore, the tree representation allows for a natural way of data integration, avoiding misspecification from restrictive assumptions and overfitting for general models. This paper was accepted by David Simchi-Levi, operations management. Funding: M. Xue’s research is partly support by the National Natural Science Foundation of China [72201257, 72571069, 72371240]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.03967 .

Problem definition: Motivated by the significance of side information in numerous operations management problems, this paper studies conditional stochastic optimization to enable more informed decisions. The side information constitutes observable exogenous covariates that alter the conditional probability distribution of the random problem parameters. Decision makers who adapt their decisions according to the observed side information solve a stochastic optimization problem where the objective function is specified by the conditional expectation of the random cost. If the joint probability distribution is unknown, then the conditional expectation can be approximated in a data-driven manner using kernel regression. Although this approximation scheme has found successful applications in diverse decision problems under uncertainty, it is largely unknown whether the scheme can provide any reasonable out-of-sample performance guarantees, and how such statistical guarantees can guide the decision-making process. Methodology/results: We employ the Nadaraya–Watson kernel regression for data-driven approximation of the conditional expectation and leverage moderate deviations theory to establish its performance guarantees. Our analysis and resultant statistical bounds motivate the use of a conditional standard deviation regularization scheme to enhance out-of-sample performances. As the designed regularization scheme leads to a nonconvex optimization problem, we further adopt ideas from distributionally robust optimization to obtain tractable formulations. We examine our proposed models on portfolio optimization, inventory management, and wind energy commitment problems. The numerical results demonstrate the effectiveness of our proposed regularization scheme. Managerial implications: Our paper illustrates the importance of side information in real-world decision-making problems. Incorporating side information through a regularized Nadaraya–Watson scheme offers managers a robust framework to enhance decision making under uncertainty. The theoretical guarantees provide guidance on the number of samples required to obtain high-quality solutions and how to optimally adjust the regularization parameter. For problem instances with high-dimensional covariates, we further present a simple dimensionality reduction procedure that helps improve the sample complexity of the scheme. All our proposed formulations are concise and straightforward for the operations manager to implement using any popular programming language interfaced with standard off-the-shelf solvers. Funding: Y. Wang was supported by the National Natural Science Foundation of China [Grant 72501204] and the Fundamental Research Funds for the Central Universities. G. A. Hanasusanto was funded by the National Science Foundation [Grants CCF-2343869 and ECCS-2404413]. C. P. Ho was supported by the Research Grants Council [General Research Fund 11508623] and the CityUHK Start-Up Grant [9610481]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2024.0997 .

We study endogenous merger decisions among three competing firms with asymmetric capacity: one large firm has ample capacity, and two small firms have limited capacity. Each firm can either operate independently or merge with another firm. Because of antitrust laws, all three firms merging is not permitted. When two firms merge, the merged entity and the remaining independent firm use the resulting capacity to capture their own “loyal buyers” and compete in price for price-sensitive “disloyal switchers.” We first consider a base model in which the size of loyal buyers for different firms is symmetric. We find that, in equilibrium, either there is no merger or the large firm will merge with a small firm as a “mixed merger.” We explain how this equilibrium structure is driven by the mergers’ impacts on market competition and firms’ pricing aggressiveness. Specifically, the mixed merger that can most effectively reduce price competition is the equilibrium under relatively competitive market conditions (i.e., the size of loyal buyers is small and capacity is large but limited). We also extend to the case with asymmetric size of loyal segments and show that two small firms merging (“small merger” structure) can be the equilibrium structure when the large firm has a much larger loyal segment than the small firms and the small firms’ capacity is sufficiently large. The impacts of mergers on social welfare and consumer surplus provide practical implications from the antitrust perspective. The antitrust agencies can approve a mixed merger to improve social welfare under certain conditions, but if the agencies pay more attention to consumer surplus, then mixed mergers should not be permitted. Small mergers, though sometimes not the equilibrium structures, can improve social welfare and consumer surplus. In such cases, the government can appropriately promote mergers that are beneficial for society and/or consumers by adopting some restrictive measures or offering incentives. This paper was accepted by Jeannette Song, operations management. Funding: The work of D. Wang was supported by the National Natural Science Foundation of China [Grant 72401275]; the work of X. Wu was supported by the National Natural Science Foundation of China [Grants 72025102, 72293560/72293565]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/mnsc.2022.01943 .

In community first responder (CFR) systems, traditional emergency service response is augmented by a network of trained volunteers who are dispatched via an app. A central application of such systems is out-of-hospital cardiac arrest (OHCA), where a very fast response is crucial. For a target performance level, how many volunteers are needed, and from which locations should they be recruited? We model the presence of volunteers throughout a region as a Poisson point process, which permits the computation of the response-time distribution of the first-arriving volunteer. Combining this with known survival-rate functions, we deduce survival probabilities in the cardiac arrest setting. We then use convex optimization to compute a location distribution of volunteers across the region that optimizes either the fraction of incidents with a fast response (a common measure in the industry) or patient survival in the case of OHCA. The optimal location distribution provides a bound on the best possible performance with a given number of volunteers. This can be used to determine whether introducing a CFR system in a new region is worthwhile or can serve as a guide for additional recruitment in existing systems. Effective target areas for recruitment are not always obvious because volunteers recruited from one area may be found in various areas across the city depending on the time of day; we explicitly capture this issue. We demonstrate these methods through an extended case study of Auckland, New Zealand. This paper was accepted by Carri Chan, healthcare management. Funding: This research was financed in part by the Netherlands Organization for Scientific Research (NWO) in the form of a Rubicon grant (019.172EN.016) to C. J. Jagtenberg and a Veni grant (VI.Veni.191E.005) to P. L. van den Berg. S. G. Henderson and H. Li were supported in part by National Science Foundation [Grant CMMI-2035086]. Vrije Universiteit Amsterdam and Erasmus University received funding from TKI Dinalog [Grant 2023-1-307TKI]. Some of this work has been executed under the TKI Dinalog [Grant 2023-1-307TKI]. Part of this work was completed during a research visit made possible by a [Distinguished Visitor Award] from the University of Auckland. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.04024 .

Efficiently managing inventories requires an accurate estimation of stockout costs. This estimation is complicated by challenges in determining how to compensate consumers monetarily to ensure they will maintain the same level of utility they would have obtained had stockouts not occurred. This paper presents an analysis of these compensation costs as applied to the design of optimal stockout rates by an online retailer marketing to consumers aesthetically substandard fruits and vegetables rejected by mainstream grocery chains. Because growers face high uncertainty in their harvesting conditions and in the aesthetic quality of their crops and there are little data on hand to predict the value consumers attach to the availability of subpar produce, it is difficult to optimally match the supply of these products with consumers’ demand. Our analysis draws from a multiple discrete-continuous extreme value (MDCEV) choice model to calculate consumer compensations that the retailer can use to estimate the opportunity costs of stockouts to manage its inventory. We show that not taking into account these compensation costs could unduly inflate the optimal stockout rates for these products. Armed with these compensation cost estimates, we show how these costs can serve as incentives for retailers to source greater inventory amounts of imperfect produce from growers and how this will ultimately translate into less waste in the supply chain. This paper was accepted by Charles Corbett, operations management. Funding: Funding from Agriculture and Food Research Initiative (National Institute for Food and Agriculture, USDA) [Grant 2016-09900] is gratefully acknowledged. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2021.03174 .

The U.S. K–12 public education system has been notoriously hard to improve. Some argue new education technologies (EdTech) can help transform schools for the better. Yet, as large-scale policy reforms have had only limited success, it seems unlikely that small changes from EdTech could have any measurable impact. In this paper, we study DonorsChoose, a nonprofit that operates a teacher crowdfunding platform. We ask whether DonorsChoose improves educational outcomes, specifically at low-income schools. Combining DonorsChoose data with data on student test scores in Pennsylvania from 2012–2013 to 2017–2018, we find that an increase in the number of DonorsChoose projects funded at a school leads to higher student performance, after controlling for selection biases. For a school with zero funded projects, one funded project—of about $400 in value—translates to between two to nine more students scoring basic and above in all subjects in high school and science and language arts in primary and middle school. We find this effect is driven mostly by low-income schools, indicating funded projects help close the gap in educational outcomes between students at low- versus high-income schools. Based on a textual analysis of 20,000 statements from all funded teachers describing how project resources are used, we find two channels of improvement most effective in the lowest income schools. We demonstrate that, although DonorsChoose projects are small, they improve outcomes and reduce inequality because they come directly from frontline workers—teachers—who know most intimately the obstacles their students face and how to help. This paper was accepted by Charles Corbett, operations management. Supplemental Material: The data files and online appendix are available at https://doi.org/10.1287/mnsc.2022.4331 .

Content-sharing social network platforms rely heavily on user-generated content to attract users and advertisers, but they have limited authority over content provision. We develop an intervention that leverages social interactions between users to stimulate content production. We study social nudges, whereby users connected with a content provider on a platform encourage that provider to supply more content. We conducted a randomized field experiment (N [Formula: see text]) on a video-sharing social network platform where treatment providers could receive messages from other users encouraging them to produce more, but control providers could not. We find that social nudges not only immediately boosted video supply by 13.21% without changing video quality but also, increased the number of nudges providers sent to others by 15.57%. Such production-boosting and diffusion effects, although declining over time, lasted beyond the day of receiving nudges and were amplified when nudge senders and recipients had stronger ties. We replicate these results in a second experiment. To estimate the overall production boost over the entire network and guide platforms to utilize social nudges, we combine the experimental data with a social network model that captures the diffusion and over-time effects of social nudges. We showcase the importance of considering the network effects when estimating the impact of social nudges and optimizing platform operations regarding social nudges. Our research highlights the value of leveraging co-user influence for platforms and provides guidance for future research to incorporate the diffusion of an intervention into the estimation of its impacts within a social network. This paper was accepted by Victor Martínez-de-Albéniz, operations management. Funding: H. Dai thanks the University of California, Los Angeles (UCLA) [Hellman Fellowship and Faculty Development Award] for funding support. R. Zhang is grateful for financial support from the Hong Kong Research Grants Council [Grant 16505418]. Supplemental Material: The data files and online appendix are available at https://doi.org/10.1287/mnsc.2022.4622 .

Shipping experience improvement has been an essential business strategy in e-commerce. Beyond investing directly in improving shipping speed, online retailers have recently expanded their focus on other shipping strategies, such as offering consumers the option to pick up orders at a local station. This paper uses the opening of hundreds of such pickup stations as a natural experiment to study the impact of these stations on consumers. We find that the introduction of pickup stations increased total sales by [Formula: see text]. In contrast with past literature, we show that shipping time reduction is not the driving factor in the impact of pickup stations. Yet, the logistic flexibility introduced by pickup stations explains the sales impact. To explicitly examine how logistic flexibility affects consumers’ decisions on purchases, we develop and estimate a structural model of consumer choice. In our model, consumers value two types of logistics flexibility—the flexibility to pick up their items at their preferred times, referred to as the value of preferred time flexibility, and the flexibility to delay pickup decisions to the last moment, referred to as the value of last-minute choice flexibility. We show that the value of preferred time flexibility accounts for [Formula: see text] of the impact on sales, whereas the value of last-minute choice flexibility accounts for the remaining [Formula: see text]. Using our estimated model, we develop a counterfactual strategy in building pickup stations that could achieve the sales lift with [Formula: see text]–[Formula: see text] fewer stations. Last but not least, using our estimated time flexibility, we also develop a novel shipping strategy without pickup stations that could improve sales by [Formula: see text]. Our estimates suggest that our counterfactual logistic strategies could increase consumer welfare by [Formula: see text]–[Formula: see text]. This paper was accepted by Elena Katok, operations management. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.00163 .

Problem definition: In high-product-variety businesses like fashion, mass production (MP) systems create environmental waste in the form of overproduction on a colossal scale. Mass customization (MC) has been proposed—without solid evidence—as a solution. In this paper, we analyze whether MC can indeed offer a win-win solution that helps both the bottom line and the environment. We also study the impact of three real policy options: promoting MC, charging a disposal fee for overproduction, and recycling. Academic/practical relevance: There is increasing interest in mass-customizing fashion goods, not only because consumers value customization, but also because MC is perceived to be environmentally friendly. Our paper puts this advocacy for MC to the test. We contribute to the literature, which has been largely silent on the issue, by uncovering when MC offers a win-win and relating such market outcomes to policy ideas. Methodology: We develop an analytical model of an MP firm adopting MC (going hybrid). The firm’s profit-maximizing variety, price, and inventory decisions then form the basis of our understanding the environmental impact of adopting MC and assessing various policy options. Results: Adopting MC can be a win-win, but it can also increase overproduction and hurt the environment. Our policy analyses reveal two kinds of insights. The first kind is about whether a policy expands win-win outcomes—encouraging sustainable adoption of MC. Among the policy ideas we explore, only promoting MC so as to increase consumers’ tolerance for waiting for mass-customized products can do that unambiguously. The second kind of insight is about whether a policy reduces the hybrid firm’s environmental impact. Only a disposal fee and costly recycling programs can do that unambiguously. Managerial implications: For MC adoption to be a win-win, policy makers must (1) work on convincing consumers to wait for bespoke fashion; (2) target MP firms with low cost of variety (high product-mix flexibility) with disposal fees or costly recycling programs; and (3) encourage those with relatively higher cost of variety to develop/acquire technology that would make recycling profitable. History: This paper has been accepted for the Manufacturing & Service Operations Management Special Section on Responsible Research in Operations Management. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2022.1088 .

E-commerce retailers often need to make inventory decisions for new products under high uncertainty. We study a new business practice of using crowdvoting, wherein a retailer first seeks input from customers on the desirability of the product and then bases the purchasing decision on their votes. We collaborated with a subscription-based apparel rental platform and obtained a proprietary data set comprising the platform’s products and users. We leverage the staggered introduction of crowdvoting in different product categories to identify whether and, if so, by how much the adoption of crowdvoting improves business performance. We find that, after the adoption, both short- and long-term rental outcomes increase. We also uncover several mechanisms that drive this improvement. We find that the platform uses the wisdom of the crowd to make better inventory depth decisions, and the products are inherently a better match with users’ tastes. Moreover, we find that brands bought through crowdvoting are more likely to be newer brands compared with items that are bought solely through expert input. In other words, the diversity of the brands carried on the platform also improves. We also find that voters become more engaged with the platform: they post more reviews and are less likely to cancel their subscription after participating in crowdvoting. Our results have important implications for retailers who may be considering similar crowd-based strategies to improve their decision making. This paper was accepted by Victor Martínez-de-Albéniz, operations management. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.01928 .

Organizations can learn from prior successes and failures to improve organizational performance. Few learning-curve studies have investigated this phenomenon at the individual level. A notable exception found that surgeons learn from their own success and others’ failure. Success in surgery is common and individually independent from other surgeries. We study learning from success and failure in a context where success is rare and competitive: Formula One (F1) racing. Only one driver will win a race, preventing the other competitors from winning. Even severe failures causing drivers to abandon the race are common. We investigate two types of abandonments: car failures and driver failures. Our data set covers F1 from the start of F1 in 1950 through 2017, yielding 21,487 driver-race observations. We find that win probability follows an inverted U-shaped function of racing experience. We also find that drivers learn from their own success, teammates’ success, as well as own car failures. However, drivers do not learn from their own driver failures. A teammate’s win increases the probability of winning the next race by 1.8%. An own car failure increases the probability of winning the next race by 1.9%. We use two characteristics of success, frequency and competitiveness, to define a spectrum of organizational settings. Placement of our F1 findings and the surgery findings on this spectrum reveals when managers can expect benefits from their own versus others’ success and failure. This paper was accepted by Charles Corbett, operations management. Supplemental Material: The online appendix and data are available at https://doi.org/10.1287/mnsc.2022.4324 .

Real-time decisions are usually irrevocable in many contexts of online decision making. One common practice is delaying real-time decisions so that the decision maker can gather more information to make better decisions. For example, in online retailing, there is typically a time delay between when an online order is received and when it gets picked and assembled for shipping. However, decisions cannot be delayed forever. In this paper, we study this fundamental trade-off and aim to theoretically characterize the benefits of delaying real-time decisions. We provide a theoretical foundation for a broad family of online decision-making problems by proving that the gap between our proposed online algorithm (called “delayed Bayesian prophet”) and the offline optimal hindsight policy decays exponentially fast in the length of delay. We also conduct extensive numerical experiments on the benefits of delay, using both synthetic data and publicly available real data. Both our theoretical and empirical results demonstrate an important managerial insight: a little delay is all we need. Finally, we extend our analysis and results to the setting where the arrival distribution is independent but nonidentical, the setting where the arrival distribution is unknown, and the setting where decisions are made in batches. This paper was accepted by Jeannette Song, operations management. Funding: The second author was partially funded by a grant from Amazon.com Inc., awarded through collaboration with the Columbia Center of AI Technology. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.00549 .