Online Appendix Research Articles

The Effect of Capital Gains Taxes on Business Creation and Employment: The Case of Opportunity Zones

The Tax Cuts and Jobs Act of 2017 established a new program called Opportunity Zones (OZs) that reduces or eliminates capital gains taxes on investment in a limited number of low-income Census tracts. We provide a model illustrating how a change in capital taxation affects employment in existing and new establishments. We then use establishment-level data to show that, in its first two years, the OZ designation increased employment growth relative to comparable tracts by between 3.0 and 4.5 percentage points in metropolitan areas. The job growth occurred in multiple industries and persisted into 2021 rather than quickly disappearing. However, most of the jobs created by the program were likely taken by residents who live outside of the designated tracts, consistent with only 5% of U.S. residents working in the same Census tract as the one in which they live. This paper was accepted by Tomasz Piskorski, finance. Funding: The authors are grateful for the financial support of an award from UW-Madison’s Fall Research Competition. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.03223 .

Optimal Salesforce Compensation with General Demand and Operational Considerations

Problem definition: We investigate the optimal salesforce compensation scheme in the context of private information and unobservable actions, considering common operational factors encountered in practice, including inventory costs, contractible versus censored demand information, and controlled versus delegated ordering. Methodology/results: Based on an agency model with general demand and cost functions, we derive optimality conditions for implementable contracts that can achieve the second-best outcome in all scenarios. The contracts are in the forms of a menu with linear compensation for demand or sales, incorporating inventory costs. Moreover, the contracts feature adjustments in compensation corresponding to the ordering level if it is delegated. Managerial implications: Our study reveals that, under reasonably mild conditions, optimal salesforce contracts can still maintain relatively simple forms, even when confronted with common operational factors and generalized demand and cost functions. However, the contracts must be tailored to suit the operational settings. Intriguingly, neither the loss of demand information nor the delegation of inventory decisions would compromise system efficiency at optimum. Funding: H. Song is partially supported by the Key International Cooperation and Exchange Projects of the NSFC [Grant W2411062] and the Foundation for Innovative Research Groups of the NSFC [Grant 71821002]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2022.0400 .

Colonial Legacy and Informal Finance

An influential line of research emphasizes that colonial legacy plays a key role in formal financial development. Can colonial legacy also shape informal finance? We investigate the impact of colonial legacy on informal financial development using a manually georeferenced data set within a credible empirical framework. In the 19th century, Europeans arbitrarily designed colonial borders that partitioned many ethnicities across multiple countries in Africa. Leveraging several spatial regression discontinuity designs across national borders and within British-French–partitioned Cameroon and a unique natural experiment where the same former British colony is compared with two otherwise similar areas with different exposures to French colonization, we discover that former British colonies today have better informal financial development than former French colonies. Exploring the channels, we find that places with a British colonial legacy maintain a style of social control that facilitates information flow, supports private enforcement and market interactions, and promotes strong legal cultures. This paper was accepted by Kay Giesecke, finance. Funding: C. Lin acknowledges the Research Grants Council of Hong Kong for financial support [Grant AoE/B-704/22-R]. J. An acknowledges the general financial support from The University of Hong Kong Start-up Fund for New Professoriate Staff. M. Tai acknowledges the Research Grants Council of Hong Kong for financial support [General Research Fund No. 17500722]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.01856 .

Supplying Cash-Constrained Retailers: Understanding Shopkeeper Behavior at the Bottom of the Pyramid

Problem definition: As the largest retail channel in the world, nanostores are the source of income for millions of shopkeepers in developing countries. Being cash constrained, nanostore shopkeepers face the challenge of deciding how much of their available cash to retain to support their families and how much to invest in acquiring products from their suppliers, who visit the nanostore with different frequencies and offer products with different margins. We study how shopkeepers deal with this challenge while seeking to make their nanostores financially sustainable. Methodology/results: We conduct an empirical study to explain how suppliers’ visit frequency influences shopkeepers’ ordering behavior. The study integrates more than 29 million orders placed with a multinational manufacturer. Results show that visit frequency adjustments by the supplier cause nanostores to experience a significant discontinuity in their orders such that reduced order frequency leads to significantly lower orders per unit of time. Building on this insight, we ran two experimental studies that make use of theoretical benchmarks to explore how shopkeepers’ ordering decisions are influenced by supplier visit frequency, product margin, and shopkeepers’ cash constraints. We analyze how and when shopkeepers deviate from the theoretical benchmarks and show how the available budget moderates the relationship between suppliers’ visit frequencies and shopkeepers’ order volumes, normalized by unit of time. Our results give support to the theory of diversification bias and show how this bias might bring additional benefits to suppliers offering low-margin products. Finally, we propose a behavioral model that explains how shopkeepers’ decisions might be driven by their preference toward inventory (i.e., leftover versus shortage aversion) and toward cash (i.e., liquidity preferences). Managerial implications: Our findings suggest that suppliers should deploy high-frequency visits to cash-constrained nanostores regardless of the margin that their products provide for shopkeepers. For low-margin suppliers, this mitigates shopkeepers’ tendencies to reduce the order volume per unit of time following a visit frequency decrease. For high-margin suppliers, high-frequency visits mitigate shopkeepers’ overdiversification bias and liquidity preference. Financial and digital schemes that reduce the need for high-frequency visits may lead to substantial efficiency gains. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2022.0362 .

Temporal Miscoupling: The Challenges and Consequences of Enacting a Practice in Decline

A practice firmly entrenched in one period can experience decline or even deinstitutionalization in another. However, at the organization level, practice abandonment can be a slow process. Organizations may continue to use a practice that is in decline or out of date, coupling asynchronously with a prior institutional environment. In this paper we develop the concept of temporal miscoupling and examine the challenges of performing a practice under this condition. Drawing on a field study of a labor strike in Canada, we examine the difficulties of inhabiting the practice of striking—picketing a workplace—with eroding political, legal, and cultural support; declining familiarity with the practice; and fading narratives to motivate and justify the practice. We show how strikers developed extemporaneous resources that gave local meaning and form to the practice. These improvised resources supported the practice but distorted historical meanings and performances. Our study expands the analytical repertoire of inhabited institutionalism by problematizing the temporal lag between institutional conditions and organizational practices for the on-the-ground enactment of practices. The extemporaneous resources generated in the context of temporal miscoupling threaten future enactments, indicating an important role for practice enactment in processes of decline and revitalization. Supplemental Material: The online appendix is available at https://doi.org/10.1287/orsc.2021.16026 .

Social Learning and Content Quality Under Polarization

Problem definition: This paper studies how polarization influences content consumption and production on digital platforms that monetize consumer engagement. Specifically, we consider a content that advocates a particular position on a divisive issue. Consumers with polarized preferences toward the content’s position are sequentially exposed to the content. Initially, consumers are uncertain about the content quality, but they have the opportunity to learn about it using aggregate consumption metrics and other informative signals provided by the platform. Methodology/results: Using a stylized model, we find that under polarization, social learning based on consumption metrics can mislead consumers to perceive low-quality content as higher quality, even in the long run. Consequently, content providers may decrease their effort to improve content quality. These effects are amplified for more polarizing issues, especially when the content’s position is “mainstream” (i.e., aligned with the majority of the population). Our results thus provide a potential explanation for the proliferation of low-quality, polarizing content on platforms. Managerial implications: We offer normative guidance for content platforms seeking to enhance content quality, including offering consumers information to aid the social learning mechanism and appropriately selecting audience to mitigate the echo-chamber effect. Additionally, we propose payment schemes based on content popularity as an effective means to encourage content providers to improve quality. Funding: This work was supported by the Hong Kong Research Grants Council [General Research Fund 16502920]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2022.0533 .

Measuring and Mitigating the Risk of Advanced Cyberattackers

Sophisticated cyberattackers (commonly known as advanced persistent threats (APTs)) pose enormous risks to organizations such as financial institutions, industrial and commercial firms, government institutions, and power grids. This study presents a method and an index to measure the vulnerability of organizations to APT risk and shows why a one-size-fits-all solution to mitigate APT risk does not exist. Our vulnerability index is based on a model that describes the optimal behavior of a cyberattacker (APT) with research and development capabilities aspiring to attack a network that manages the organization and a network operator that deploys blocking and detection measures to protect its organization from the attack. We demonstrate how our vulnerability index, which accounts for the network’s structure and the APTs’ resources and strategy, can be used in realistic risk assessments and optimal resource allocation procedures and serve as a benchmark for organizations’ preparedness against APTs’ cyberattacks. We also propose that regulatory agencies of financial (and other) institutions provide the parameters that define an APT’s profile and request, as part of their periodic assessments of the organizations that they regulate, that our (or similar) vulnerability index will be reported to them by the regulated institutions. Finally, the viability of our index in modeling modern cybersecurity defense procedures shows that not only there is no silver bullet defense against all types of APTs, it is also imperative to account for APTs’ heterogeneity because detection and blocking measures can be complements, substitutes, or even degrade each other. For example, when the attacker’s (defender’s) budget is extremely large (small), the defender should deploy only detection measures, strongly advocating Zero Trust practices. Supplemental Material: The online appendix is available at https://doi.org/10.1287/deca.2023.0072 .

Mavericks and Diplomats: Bridging Commercial and Institutional Entrepreneurship for Society’s Grand Challenges

Many of society’s grand challenges, like access to education, improved public health, and building smart infrastructure, occur in established fields, often ones where noncommercial logics matter. Grounded in a theory-building study of two well-matched EdTech ventures in the nascent MOOC market in the U.S. higher education field, we unpack their strategy formation processes. These ventures faced the dual problem of forming a successful strategy in a nascent market while changing an established field. Our primary contribution is a theoretical framework that identifies two distinct yet effective processes. One (Maverick) is a competitive, learning centric process, whereas the other (Diplomat) is a cooperative, diplomacy-centric one. Although distinct, these processes unexpectedly share the same decision weaving (i.e., hybrid) problem-solving structure that fits complex, novel problems like strategy formation. Broadly, we contribute to institutional theory and institutional entrepreneurship by introducing the diplomacy lens, its tactics, and firm performance. We contribute to learning theory and entrepreneurship by identifying the unique features of learning in established markets, including the limits to rapid experimentation and the value of collaborative learning for changing established fields. We add to practice by emphasizing the difficulty of achieving both commercial success and societal impact. Overall, we bridge the gap between institutional and commercial entrepreneurship. Funding: The authors gratefully acknowledge the generous research support of the Stanford Technology Ventures Program. Supplemental Material: The online appendix is available at https://doi.org/10.1287/orsc.2020.13810 .

The Crowdless Future? Generative AI and Creative Problem-Solving

The rapid advances in generative artificial intelligence (AI) open up attractive opportunities for creative problem-solving through human-guided AI partnerships. To explore this potential, we initiated a crowdsourcing challenge focused on sustainable, circular economy business ideas generated by the human crowd (HC) and collaborative human-AI efforts using two alternative forms of solution search. The challenge attracted 125 global solvers from various industries, and we used strategic prompt engineering to generate the human-AI solutions. We recruited 300 external human evaluators to judge a randomized selection of 13 out of 234 solutions, totaling 3,900 evaluator-solution pairs. Our results indicate that while human crowd solutions exhibited higher novelty—both on average and for highly novel outcomes—human-AI solutions demonstrated superior strategic viability, financial and environmental value, and overall quality. Notably, human-AI solutions cocreated through differentiated search, where human-guided prompts instructed the large language model to sequentially generate outputs distinct from previous iterations, outperformed solutions generated through independent search. By incorporating “AI in the loop” into human-centered creative problem-solving, our study demonstrates a scalable, cost-effective approach to augment the early innovation phases and lays the groundwork for investigating how integrating human-AI solution search processes can drive more impactful innovations. Funding: This work was supported by Harvard Business School (Division of Research and Faculty Development) and the Laboratory for Innovation Science at Harvard (LISH) at the Digital Data and Design (D3) Institute at Harvard. Supplemental Material: The online appendix is available at https://doi.org/10.1287/orsc.2023.18430 .

Working Around the Clock: Temporal Distance, Intrafirm Communication, and Time Shifting of the Employee Workday

This paper examines the effects of temporal distance generated by time zone separation on communication in geographically distributed organizations. We build on prior research, which highlights time zone separation as a significant challenge, but argue that employees may time shift—move work-related communication to outside of regular business hours—to counteract temporal distance. We propose a theory in which employees’ tendency to time shift depends on the demands of their tasks and collaborative relationships and individuals’ ability to supply work outside of regular business hours. Analyzing communication-level data from 12,038 employees of a large multinational firm and using cities’ shifts to/from daylight saving time for identification, we find that temporal distance leads to sizable but smaller than expected reductions in volumes of rich, synchronous communication between employees. Consistent with our arguments, increased temporal distance significantly increases time shifting of work-related communication, especially among workers whose jobs are nonroutine and interactions in strong collaborative relationships. We further document that female employees and employees based in countries with stricter legal work hour limits engage in significantly less time-shifted communication. Our study improves understanding of a ubiquitous source of collaboration friction. It also sheds light on a potential source of inequities in workplace outcomes stemming from differences in individuals’ ability to work outside of regular business hours. Supplemental Material: The online appendix is available at https://doi.org/10.1287/orsc.2023.17558 .

Heatmap Design for Probabilistic Driver Repositioning in Crowdsourced Delivery

This paper proposes the use of heatmaps as a control lever to manage the probabilistic repositioning of independent drivers in crowdsourced delivery platforms. The platform aims to maximize order fulfillment by dynamically matching drivers and orders and selecting heatmaps that trigger the probabilistic flow of unmatched drivers to balance driver supply and delivery requests across the service region. We develop a Markov decision process (MDP) model to sequentially select matching and heatmap decisions in which the repositioning behavior of drivers is captured by a multinomial logit discrete choice model. Because of the curse of dimensionality and the endogenous uncertainty of driver repositioning, the MDP model is solved using a rolling-horizon stochastic lookahead policy. This policy decomposes matching and heatmap decisions into two optimization problems: a two-stage stochastic programming upper bounding problem for matching decisions and a mixed-integer programming problem for heatmap decisions. We also propose a simple policy for efficiently solving large-scale problems. An extensive computational study on instances derived from the Chicago ride-hailing data set is conducted. Computational experiments demonstrate the value of heatmaps in improving order fulfillment beyond the level achieved by matching alone (up to 25%) and identify conditions that affect the benefit of using heatmaps to guide driver repositioning. Funding: The authors gratefully acknowledge the support of the Natural Sciences and Engineering Research Council of Canada through Discovery Grants [Grants RGPIN-2024-04881, RGPIN-2020-04498, and RGPIN-2019-06207] awarded to the first, second, and third authors, respectively. Supplemental Material: The online appendix is available at https://doi.org/10.1287/trsc.2022.0418 .

How Does Artificial Intelligence Shape Audit Firms?

Does artificial intelligence (AI) displace auditors? We exploit the staggered hiring of AI employees at audit office locations across the United States as a proxy for the use of AI at local audit offices. The main findings are as follows. First, relative to audit offices that do not yet hire AI employees, those that do hire AI employees have a 4.3% increase in the number of auditor jobs, particularly among junior and midlevel auditors. Second, using AI is associated with an increased demand for soft skills (e.g., cognitive skills) in auditor jobs. Third, audit offices that use AI have more accurate going concern and internal control opinions. Semistructured interviews of 11 seasoned audit partners confirm that investment in AI is centralized at the national level, but the decision to deploy it often resides at the local audit office level. Notably, none of the partners believe that AI has replaced or will replace human auditors. Overall, our study—comprising both empirical and qualitative data—suggests that using AI does not replace auditors, but rather changes the skills required for these jobs and improves audit quality. This paper was accepted by Suraj Srinivasan, accounting. Funding: K. K. F. Law’s research is supported by the Start-Up Grant from Nanyang Technological University and the Ministry of Education (MOE), Singapore, under its Academic Research Fund Tier 1 [Grant RG128/20]. M. Shen’s research is supported by the Start-Up Grant from the National University of Singapore [Grant A-0003914-00-00] and the MOE under its Academic Research Fund Tier 1 [Grant A-8000098-00-00]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.04040 .

Comment on “Salesforce Compensation with Inventory Considerations”

This note critically analyzes the conditions for equilibrium existence in a model involving incentive schemes and inventory management. It identifies a technical flaw where equilibrium is asserted to exist under certain parameter values when, in fact, it does not. However, by introducing a reasonable assumption that inventory can only be ordered in sufficiently large increments, the qualitative outcomes and key insights remain intact. This paper was accepted by Jeannette Song, operations management. Supplemental Material: The online appendix is available at https://doi.org/10.1287/mnsc.2023.03914 .

Cancel the Deal? An Experimental Study on the Exploitation of Irrational Consumers

Consumers can sometimes be exploited because they make mistakes in their valuation of products. We present the results from a large-scale experimental study that examines whether third-party spectators from the general population in the United States cancel a voluntary deal where a buyer has made a mistake in the valuation of a product and agreed to pay more for the product than the seller knows it is worth. We find that the majority of the spectators cancel such deals, even when the seller’s involvement is limited to accepting a proposal made by the buyer. A substantial share of these spectators is also willing to fine the seller. However, a large minority of the spectators are willing to uphold the deal even when the seller has proposed the deal and obfuscated the information provided to the buyer. Our results shed new light on when people view market transactions as acceptable and their attitudes to government regulation of businesses. This paper was accepted by Axel Ockenfels, behavioral economics and decision analysis. Funding: This project was financed with support from the Norges Forskningsråd [Centres of Excellence Scheme Centre for Experimental Research on Fairness, Inequality and Rationality Project 262675 Grants 236995, 250415, 262636, and 302145]. The project also received funding from the European Research Council [the European Union’s Horizon 2020 Research and Innovation Programme Grant 788433]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.01050 .

Coopetition in Platform-Based Retailing: On the Platform’s Entry

We study the dynamic incentive interactions between a platform and a third-party seller over two stages, where the seller exerts product-value-enhancement effort in the first stage in anticipation of the platform’s potential entry in the second stage. Upon entry, the platform can exert effort to boost the value of the product sold by the platform, with positive spillovers to the value of the product sold by the platform. We show the existence of both the protective and the receptive regimes, characterizing the necessary and sufficient conditions for each regime. These conditions are of thresholds type with respect to the parameters including the degree of spillover, the referral rate, and the competition intensity. In the protective (receptive) regime, the seller is worse (better) off with the platform’s entry than without, thereby distorting effort downward (upward) from the first best to deter (induce) the platform’s entry to the products of intermediate value. Notably, if the spillover effect is negative, only the protective regime will arise. We also provide the necessary and sufficient conditions under which the platform’s nonentry commitment strictly improves the platform’s profits by restoring the seller’s effort to the first best, achieving the win-win outcome for both the platform and the seller. This paper was accepted by Jeannette Song, operations management. Funding: H. Song is partially supported by the Key International Cooperation and Exchange Projects of the NSFC [Grant W2411062] and the Foundation for Innovative Research Groups of the NSFC [Grant 71821002]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/mnsc.2023.00260 .

Adaptive Preference Measurement with Unstructured Data

Many products are most meaningfully described using unstructured data such as text or images. Unstructured data are also common in e-commerce, in which products are often described by photos and text but not with standardized sets of attributes. Whereas much is known about how to efficiently measure consumer preferences when products can be meaningfully described by structured attributes, there is scant research on doing the same for unstructured data. This paper introduces a real-time, adaptive survey design framework for measuring preferences over unstructured data, leveraging Bayesian optimization. By adaptively choosing items to display based on uncertainty around a nonparametric utility model, the proposed method maximizes information gain per question, enabling quick estimation of individual-level preferences. The approach operates on embeddings of the unstructured data, thereby eliminating the requirement for manual coding of product attributes. We apply the method to measuring preferences over clothing and highlight its potential for both the general task of marketing research and the specific task of designing customer onboarding surveys to mitigate the cold-start recommendation problem. We also develop methods for interpreting the nonparametric utility functions, which allow us to reconstruct consumer valuations of discrete attributes, even for attributes that were not considered or available a priori. This paper was accepted by Duncan Simester, marketing. Fundings: Funding for this project was provided by Analytics at Wharton, the Wharton Behavioral Lab, and the Wharton Dean’s Fund. The author also thanks the Govil Family for financial support. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.03775 .

Identity Disclosure and Anthropomorphism in Voice Chatbot Design: A Field Experiment

Fueled by the widespread adoption of algorithms and artificial intelligence, the use of chatbots has become increasingly popular in various business contexts. In this paper, we study how to effectively and appropriately use voice chatbots, particularly by leveraging the two design features identity disclosure and anthropomorphism, and evaluate their impact on the firm operational performance. In collaboration with a large truck-sharing platform, we conducted a field experiment that randomly assigned 11,000 truck drivers to receive outbound calls from the voice chatbot dispatcher of our focal platform. Our empirical results suggest that disclosing the identity of the chatbot at the beginning of the conversation negatively affects operational performance, leading to around 11% reduction in the response probability. However, humanizing the voice chatbot by adding our proposed anthropomorphism features (i.e., interjections and filler words) significantly improves response probability, conversation length, and the probability of order acceptance intention by over 5.6%, 24.9%, and 10.1%, respectively. Moreover, even when the chatbot’s identity is disclosed along with humanizing features, the operational outcomes still improve. This finding suggests that enhancing anthropomorphism may potentially counteract the negative effects of chatbot identity disclosure. Finally, we propose one plausible explanation for the performance improvement—the enhanced trust between humans and algorithms—and provide empirical evidence that drivers are more likely to disclose information to chatbot dispatchers with anthropomorphism features. Our proposed anthropomorphism improvement solutions are currently being implemented and utilized by our collaborator platform. This paper has been This paper was accepted by Felipe Caro for the Special Issue on the Human-Algorithm Connection. Funding: This study is supported by the National Natural Science Foundation of China [Grants 72172169 and 91646125], Program for Innovation Research at the Central University of Finance and Economic, and Shanghai Pujiang Program. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.03833 .

Sharp Waiting-Time Bounds for Multiserver Jobs

Multiserver jobs, which are jobs that occupy multiple servers simultaneously during service, are prevalent in today’s computing clusters. But, little is known about the delay performance of systems with multiserver jobs. We consider queueing models for multiserver jobs in scaling regimes where the system load becomes heavy and meanwhile, the total number of servers in the system and the number of servers that a job needs become large. Prior work has derived upper bounds on the queueing probability in this scaling regime. However, without proper lower bounds, the existing results cannot be used to differentiate between policies. In this paper, we study the delay performance by establishing sharp bounds on the steady-state mean waiting time of multiserver jobs, where the waiting time of a job is the time spent in queueing rather than in service. We first characterize the exact order of the mean waiting time under the first come, first serve (FCFS) policy. Then, we prove a lower bound on the mean waiting time of all policies, which has an order gap with the mean waiting time under FCFS. We show that the lower bound is achievable by a priority policy that we call smallest need first (SNF). Funding: This research was supported in part by the National Science Foundation [Grant ECCS-2145713]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/stsy.2023.0006 .

Advising Entrepreneurs: Optimal Recommendation of Alternatives

Problem definition: Facing emergent business challenges, entrepreneurs often seek guidance from experienced advisors. When multiple alternatives 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. Methodology/results: We develop a dynamic game-theoretic model that captures the sequential interaction between an advisor and an entrepreneur. The advisor chooses how to recommend alternative solutions, and the entrepreneur chooses which solution to try. The trial’s success depends on the viability of a solution and the entrepreneur’s execution capability. When a trial of a recommended solution fails, the belief about the viability of the solution is updated. Managerial implications: Our analysis reveals that the advisor should strategically recommend alternatives based on the entrepreneur’s execution capability, trial costs, and correlation between alternatives (among other factors). When the trial of the first alternative fails, the advisor should readily offer a new alternative if the entrepreneur’s capability is either very high or very low. Otherwise, the advisor should encourage the entrepreneur to try the same solution multiple times. In order to motivate and sustain the entrepreneur’s exploration over time and across solutions, the advisor may find it optimal to recommend inferior solutions before superior ones (e.g., when trial costs are different or the entrepreneur can improve her capability with experience) or recommend multiple solutions simultaneously (e.g., when there is correlation between alternatives). Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2022.0361 .

A Bilevel Approach for Compensation and Routing Decisions in Last-Mile Delivery

In last-mile delivery logistics, peer-to-peer logistic platforms play an important role in connecting senders, customers, and independent carriers to fulfill delivery requests. As the carriers are not under the platform’s control, the platform has to anticipate their reactions while deciding how to allocate the delivery operations. Indeed, carriers’ decisions largely affect the platform’s revenue. In this paper, we model this problem using bilevel programming. At the upper level, the platform decides how to assign the orders to the carriers; at the lower level, each carrier solves a profitable tour problem to determine which offered requests to accept, based on her own profit maximization. Possibly, the platform can influence carriers’ decisions by determining also the compensation paid for each accepted request. The two considered settings result in two different formulations: the bilevel profitable tour problem with fixed compensation margins and with margin decisions, respectively. For each of them, we propose single-level reformulations and alternative formulations where the lower-level routing variables are projected out. A branch-and-cut algorithm is proposed to solve the bilevel models, with a tailored warm-start heuristic used to speed up the solution process. Extensive computational tests are performed to compare the proposed formulations and analyze solution characteristics. Funding: The research of E. Fernandez was partially funded through the Spanish Ministerio de Ciencia y Tecnología and European Regional Development Funds (ERDF) [Grant MTM2019-105824GB-I00]. The research of C. Archetti, M. Cerulli, and I. Ljubić was partially funded by CY Initiative of Excellence, France [Grant “Investissements d’Avenir ANR-16-IDEX-0008”]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/trsc.2023.0129 .