Francisco Castro
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Francisco Castro has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Siva Theja Maguluri
Industrial and Systems Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332
6 shared publications - Sushil Mahavir Varma
School of Industrial and Systems Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332
6 shared publications - Omar Besbes
Graduate School of Business, Columbia University, New York, New York 10027
3 shared publications - Chiwei Yan
Department of Industrial Engineering and Operations Research, University of California, Berkeley, California 94720
2 shared publications - Hamid Nazerzadeh
Marshall School of Business, University of Southern California, Los Angeles, California 90089
2 shared publications - Ilan Lobel
Stern School of Business, New York University, New York, New York 10012
2 shared publications
Latest Publications
- 1 November 2026
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Operations Research
- Vol. 74 (6) , 3112-3132
Optimal Design of Default Donations
Optimal Design of Default Donations Many nonprofit organizations rely on suggested donation amounts to incentivize giving, yet there is little consensus on how to set those amounts. In this article, Castro and Rodilitz develop a data-driven framework for designing optimal donation menus that maximize fundraising revenue while accounting for differences in donor preferences and responsiveness to suggestions. The authors show that commonly used heuristics can perform poorly and instead introduce a dynamic-programming algorithm that efficiently identifies revenue-maximizing donation menus. Their analysis also quantifies the value of donor information and demonstrates when personalized suggestions, larger menus, or a single targeted default are most effective. Applying the framework to data from a large fundraising experiment, the authors find that an optimally designed menu could substantially outperform the default amounts tested in practice. The results provide actionable guidance for nonprofits seeking to improve online fundraising through more effective choice architecture.
- 1 November 2026
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Operations Research
- Vol. 74 (6) , 3112-3132
Optimal Design of Default Donations
Optimal Design of Default Donations Many nonprofit organizations rely on suggested donation amounts to incentivize giving, yet there is little consensus on how to set those amounts. In this article, Castro and Rodilitz develop a data-driven framework for designing optimal donation menus that maximize fundraising revenue while accounting for differences in donor preferences and responsiveness to suggestions. The authors show that commonly used heuristics can perform poorly and instead introduce a dynamic-programming algorithm that efficiently identifies revenue-maximizing donation menus. Their analysis also quantifies the value of donor information and demonstrates when personalized suggestions, larger menus, or a single targeted default are most effective. Applying the framework to data from a large fundraising experiment, the authors find that an optimally designed menu could substantially outperform the default amounts tested in practice. The results provide actionable guidance for nonprofits seeking to improve online fundraising through more effective choice architecture.
- 8 April 2026
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Manufacturing & Service Operations Management
Matching Queues, Flexibility, and Incentives
Problem definition: In many matching markets, some agents are fully flexible, whereas others only accept a subset of jobs. For example, ride-sharing drivers can specify on the platform the destinations that they are willing to accept. Conventional wisdom suggests reserving flexible agents, but this can backfire; anticipating higher matching chances, agents may misreport as specialized, reducing overall matches. We ask how platforms can design simple matching policies that remain effective when agents act strategically. Methodology/results: We model job allocation as a bipartite matching queueing system and analyze equilibrium throughput performance under different policies when agents choose which queue to join. We show that flexibility reservation is optimal under full information but can perform poorly with private information, sometimes substantially worse than random assignment. To address this, we propose a new policy—flexibility reservation with fallback—that guarantees robust performance across settings without requiring precise knowledge of system parameters or agent utility functions. Managerial implications: Our results underscore the importance of accounting for strategic reporting in the design of matching policies; the proposed fallback policy both preserves flexibility and exploits latent flexibility when explicitly flexible agents are exhausted. Its simplicity and parameter-free nature also make it practical to implement in platforms such as ride-sharing and affordable housing allocation. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2024.0774 .
- 8 April 2026
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Manufacturing & Service Operations Management
Matching Queues, Flexibility, and Incentives
Problem definition: In many matching markets, some agents are fully flexible, whereas others only accept a subset of jobs. For example, ride-sharing drivers can specify on the platform the destinations that they are willing to accept. Conventional wisdom suggests reserving flexible agents, but this can backfire; anticipating higher matching chances, agents may misreport as specialized, reducing overall matches. We ask how platforms can design simple matching policies that remain effective when agents act strategically. Methodology/results: We model job allocation as a bipartite matching queueing system and analyze equilibrium throughput performance under different policies when agents choose which queue to join. We show that flexibility reservation is optimal under full information but can perform poorly with private information, sometimes substantially worse than random assignment. To address this, we propose a new policy—flexibility reservation with fallback—that guarantees robust performance across settings without requiring precise knowledge of system parameters or agent utility functions. Managerial implications: Our results underscore the importance of accounting for strategic reporting in the design of matching policies; the proposed fallback policy both preserves flexibility and exploits latent flexibility when explicitly flexible agents are exhausted. Its simplicity and parameter-free nature also make it practical to implement in platforms such as ride-sharing and affordable housing allocation. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2024.0774 .
- 1 December 2025
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Stochastic Systems
- Vol. 15 (4) , 291-336
Power-of- d Choices Load Balancing in the Sub-Halfin-Whitt Regime
We consider the load-balancing system under Poisson arrivals, exponential services, and homogeneous servers under the power-of-d choices routing algorithm, which chooses the queue with minimum length among d randomly sampled queues. We study this system in the sub-Halfin-Whitt regime. In particular, we consider a sequence of systems with n servers, where the arrival rate of the nth system is [ ] for [ ]. It was shown that under power-of-d choices routing with [ ], the queue length behaves similarly to that of joining the shortest queue and that there are asymptotically zero queueing delays. The focus of this paper is to characterize the behavior when d is below this threshold. We obtain high probability bounds on the queue lengths for various values of d and large enough n. In particular, we show that when d is [ ] for some integer [ ], then the asymptotic queue length is m with high probability. Moreover, if d grows poly-log in n, that is, slower than any polynomial, but is at least [ ], the queue length blows up to infinity asymptotically. We obtain these results by using an iterative state space collapse approach. Funding: This work was partially supported by the National Science Foundation [Grants EPCN-2144316 and CMMI-2140534].
- 1 December 2025
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Stochastic Systems
- Vol. 15 (4) , 291-336
Power-of- d Choices Load Balancing in the Sub-Halfin-Whitt Regime
We consider the load-balancing system under Poisson arrivals, exponential services, and homogeneous servers under the power-of-d choices routing algorithm, which chooses the queue with minimum length among d randomly sampled queues. We study this system in the sub-Halfin-Whitt regime. In particular, we consider a sequence of systems with n servers, where the arrival rate of the nth system is [ ] for [ ]. It was shown that under power-of-d choices routing with [ ], the queue length behaves similarly to that of joining the shortest queue and that there are asymptotically zero queueing delays. The focus of this paper is to characterize the behavior when d is below this threshold. We obtain high probability bounds on the queue lengths for various values of d and large enough n. In particular, we show that when d is [ ] for some integer [ ], then the asymptotic queue length is m with high probability. Moreover, if d grows poly-log in n, that is, slower than any polynomial, but is at least [ ], the queue length blows up to infinity asymptotically. We obtain these results by using an iterative state space collapse approach. Funding: This work was partially supported by the National Science Foundation [Grants EPCN-2144316 and CMMI-2140534].
- 24 November 2025
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Management Science
Electric Vehicle Fleet and Charging Infrastructure Planning
We study electric vehicle (EV) fleet and charging infrastructure planning in a spatial setting. For a centrally managed fleet that serves customer requests arriving continuously at a rate [ ] throughout the day, we determine the minimum number of vehicles and chargers for a target service level along with matching and charging policies. Whereas non-EV systems require extra [ ] vehicles because of pickup times, EV systems differ. Charging increases nominal capacity, enabling pickup time reductions and allowing for an extra fleet requirement of only [ ] for [ ], depending on charging infrastructure and battery pack sizes. We propose the power-of-d dispatching policy, which achieves this performance by selecting the closest vehicle with the highest battery level from d options. We extend our results to accommodate time-varying demand patterns and discuss conditions for transitioning between EV and non-EV capacity planning. Simulations verify our scaling results, insights, and policy effectiveness. Whereas long-range, fast-charging fleets resemble non-EV systems, short-range, low-cost fleets can still perform competitively, underscoring the need for EV-aware management policies. This paper was accepted by Victor Martínez de Albéniz, operations management. Supplemental Material: The online appendix is available at https://doi.org/10.1287/mnsc.2024.08524 .
- 24 November 2025
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Management Science
Electric Vehicle Fleet and Charging Infrastructure Planning
We study electric vehicle (EV) fleet and charging infrastructure planning in a spatial setting. For a centrally managed fleet that serves customer requests arriving continuously at a rate [ ] throughout the day, we determine the minimum number of vehicles and chargers for a target service level along with matching and charging policies. Whereas non-EV systems require extra [ ] vehicles because of pickup times, EV systems differ. Charging increases nominal capacity, enabling pickup time reductions and allowing for an extra fleet requirement of only [ ] for [ ], depending on charging infrastructure and battery pack sizes. We propose the power-of-d dispatching policy, which achieves this performance by selecting the closest vehicle with the highest battery level from d options. We extend our results to accommodate time-varying demand patterns and discuss conditions for transitioning between EV and non-EV capacity planning. Simulations verify our scaling results, insights, and policy effectiveness. Whereas long-range, fast-charging fleets resemble non-EV systems, short-range, low-cost fleets can still perform competitively, underscoring the need for EV-aware management policies. This paper was accepted by Victor Martínez de Albéniz, operations management. Supplemental Material: The online appendix is available at https://doi.org/10.1287/mnsc.2024.08524 .
- Open Access
- 6 April 2025
- Published by Elsevier in Transportation Research Part C: Emerging Technologies
- Vol. 174
Idle wage as a tool to regulate the relationship between ride-hailing platforms and drivers
- 6 April 2025
- Published by Elsevier in Transportation Research Part C: Emerging Technologies
- Vol. 174
Open AccessIdle wage as a tool to regulate the relationship between ride-hailing platforms and drivers
- Open Access
- 26 July 2024
- Published by SAGE Publications in Production and Operations Management
- Vol. 33 (10) , 2014-2030
Getting Out of Your Own Way: Introducing Autonomous Vehicles on a Ride-Hailing Platform
Once autonomous vehicles (AVs) are deployed for ride-hailing platforms, human drivers will compete with AVs until AV costs decrease enough to eliminate human drivers entirely. We examine a ride-hailing platform’s strategy to recruit human drivers while operating a private AV fleet. Using a game-theoretic model, we analyze how the platform sets the human-driver wage and the size of its AV fleet. We show that setting a higher wage can surprisingly lead to less human-driver participation. Moreover, we show that having the option to augment its AV fleet after observing human participation levels can, counterintuitively, hurt the platform’s bottom line. Our findings emphasize the need for ride-hailing platforms to carefully navigate the complex interactions between their roles as supply providers (of AV-served rides) and supply seekers (of human drivers), as failure to do so can be costly.
- 26 July 2024
- Published by SAGE Publications in Production and Operations Management
- Vol. 33 (10) , 2014-2030
Open AccessGetting Out of Your Own Way: Introducing Autonomous Vehicles on a Ride-Hailing Platform
Once autonomous vehicles (AVs) are deployed for ride-hailing platforms, human drivers will compete with AVs until AV costs decrease enough to eliminate human drivers entirely. We examine a ride-hailing platform’s strategy to recruit human drivers while operating a private AV fleet. Using a game-theoretic model, we analyze how the platform sets the human-driver wage and the size of its AV fleet. We show that setting a higher wage can surprisingly lead to less human-driver participation. Moreover, we show that having the option to augment its AV fleet after observing human participation levels can, counterintuitively, hurt the platform’s bottom line. Our findings emphasize the need for ride-hailing platforms to carefully navigate the complex interactions between their roles as supply providers (of AV-served rides) and supply seekers (of human drivers), as failure to do so can be costly.
- Open Access
- 8 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 205
Human-AI Interactions and Societal Pitfalls
When working with generative artificial intelligence (AI), users may see productivity gains, but content generated with the help of AI may not match their preferences exactly. The boost in productivity may come at the expense of users' idiosyncrasies, such as personal style and tastes, preferences we would naturally express without AI. To let users express their preferences, many AI systems let users edit their prompt (e.g., Midjourney) or allow more natural interactions (e.g., ChatGPT), and users can always review and edit the AI-generated output themselves. However, aligning a user's intentions with an AI's output can take time and may not always be worth it if the AI's first or default output "does the job." In short, users face a trade-off between AI output fidelity and communication cost. The purpose of this work is to examine the impact of this human-AI interaction on the AI-generated content we produce as a society. We propose a Bayesian model to study the societal consequences of human-AI interactions. For a given task, rational users can exchange information with the AI to align its output with their heterogeneous preferences. The AI has a knowledge of the distribution of preferences in the population and uses a Bayesian update to create the optimal output with maximal expected fidelity given the information shared by the user. Users choose the amount of information they share to maximize their utility, balancing the cost of communication with the fidelity of the output. We show that the interplay between individual users and AI may lead to societal challenges. Outputs may become more homogenized. The AI-generated output distribution has a lower variance than the users' preference distribution. And this phenomenon is exacerbated when AI-generated content is used to train the next generation of AI: we show numerically that the users' rational decisions and the AI's training process can mutually reinforce each other, leading to a homogenization "death spiral." We also study the effects of AI bias, identifying who benefits or loses when using an AI model that does not accurately reflect the population preference distribution. At the population level, the censoring type of bias (e.g., biasing against the more unique preferences) negatively impacts the population utility as a whole, especially users with uncommon preferences who rely on AI interactivity the most. On the other hand, directional biases (e.g., a slightly left-leaning AI) will influence the users' chosen output, leading to a societal bias. Nonetheless, our research also demonstrates that creating models that facilitate human-AI interactions can limit these risks and preserve the population preference diversity. A full version of this paper can be found at https://arxiv.org/abs/2309.10448.
- 8 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 205
Open AccessHuman-AI Interactions and Societal Pitfalls
When working with generative artificial intelligence (AI), users may see productivity gains, but content generated with the help of AI may not match their preferences exactly. The boost in productivity may come at the expense of users' idiosyncrasies, such as personal style and tastes, preferences we would naturally express without AI. To let users express their preferences, many AI systems let users edit their prompt (e.g., Midjourney) or allow more natural interactions (e.g., ChatGPT), and users can always review and edit the AI-generated output themselves. However, aligning a user's intentions with an AI's output can take time and may not always be worth it if the AI's first or default output "does the job." In short, users face a trade-off between AI output fidelity and communication cost. The purpose of this work is to examine the impact of this human-AI interaction on the AI-generated content we produce as a society. We propose a Bayesian model to study the societal consequences of human-AI interactions. For a given task, rational users can exchange information with the AI to align its output with their heterogeneous preferences. The AI has a knowledge of the distribution of preferences in the population and uses a Bayesian update to create the optimal output with maximal expected fidelity given the information shared by the user. Users choose the amount of information they share to maximize their utility, balancing the cost of communication with the fidelity of the output. We show that the interplay between individual users and AI may lead to societal challenges. Outputs may become more homogenized. The AI-generated output distribution has a lower variance than the users' preference distribution. And this phenomenon is exacerbated when AI-generated content is used to train the next generation of AI: we show numerically that the users' rational decisions and the AI's training process can mutually reinforce each other, leading to a homogenization "death spiral." We also study the effects of AI bias, identifying who benefits or loses when using an AI model that does not accurately reflect the population preference distribution. At the population level, the censoring type of bias (e.g., biasing against the more unique preferences) negatively impacts the population utility as a whole, especially users with uncommon preferences who rely on AI interactivity the most. On the other hand, directional biases (e.g., a slightly left-leaning AI) will influence the users' chosen output, leading to a societal bias. Nonetheless, our research also demonstrates that creating models that facilitate human-AI interactions can limit these risks and preserve the population preference diversity. A full version of this paper can be found at https://arxiv.org/abs/2309.10448.
- Open Access
- 8 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 1187
Optimal Design of Default Donations
Nonprofit fundraising websites often display a set of donation amounts, allowing prospective donors to effortlessly select an amount from this menu of suggestions instead of manually inputting their ideal donation. Although this strategy is effective at shaping behavior, it can also backfire: suggested amounts ("defaults") attract donors with both lower and higher ideal donations, potentially leading to a net decrease in revenue. To address this challenge, we present a comprehensive framework for designing a menu of defaults to maximize fundraising revenue in the presence of heterogeneous donors. Given the combinatorial number of different menus, it is generally infeasible to enumerate over all possibilities; in addition, our analysis reveals significant limitations of a greedy approach. Instead, we design an algorithm based on dynamic programming principles that efficiently finds an optimal menu. Our approach leverages the observation that --- under our model of a utility-maximizing donor --- the donor will only consider three donation amounts: (i) their ideal amount, (ii) the closest default below their ideal amount, and (iii) the closest default above their ideal amount. This spatial structure allows us to establish a connection to facility location problems on the real line, which we leverage in the design of our algorithm. Our approach continues to be effective even when relaxing our behavioral assumptions in model extensions. Additionally, we shed light on the value of information by comparing against the full information benchmark that knows the largest amount that each donor would select. If the nonprofit has information about each donor's ideal donation, it can obtain a constant-factor guarantee with respect to the full-information benchmark. If the nonprofit only has distributional information, we characterize how the guarantee depends on donor heterogeneity and the size of the menu. Our results highlight the value of market segmentation as well as the potential advantages of larger menus. As a case study, we apply our end-to-end methodology to experimental data from Altmann et al. [2019]. Our counterfactual analysis suggests that the optimal menu could increase revenue by more than 3%. The full paper can be found at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4733483.
- 8 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 1187
Open AccessOptimal Design of Default Donations
Nonprofit fundraising websites often display a set of donation amounts, allowing prospective donors to effortlessly select an amount from this menu of suggestions instead of manually inputting their ideal donation. Although this strategy is effective at shaping behavior, it can also backfire: suggested amounts ("defaults") attract donors with both lower and higher ideal donations, potentially leading to a net decrease in revenue. To address this challenge, we present a comprehensive framework for designing a menu of defaults to maximize fundraising revenue in the presence of heterogeneous donors. Given the combinatorial number of different menus, it is generally infeasible to enumerate over all possibilities; in addition, our analysis reveals significant limitations of a greedy approach. Instead, we design an algorithm based on dynamic programming principles that efficiently finds an optimal menu. Our approach leverages the observation that --- under our model of a utility-maximizing donor --- the donor will only consider three donation amounts: (i) their ideal amount, (ii) the closest default below their ideal amount, and (iii) the closest default above their ideal amount. This spatial structure allows us to establish a connection to facility location problems on the real line, which we leverage in the design of our algorithm. Our approach continues to be effective even when relaxing our behavioral assumptions in model extensions. Additionally, we shed light on the value of information by comparing against the full information benchmark that knows the largest amount that each donor would select. If the nonprofit has information about each donor's ideal donation, it can obtain a constant-factor guarantee with respect to the full-information benchmark. If the nonprofit only has distributional information, we characterize how the guarantee depends on donor heterogeneity and the size of the menu. Our results highlight the value of market segmentation as well as the potential advantages of larger menus. As a case study, we apply our end-to-end methodology to experimental data from Altmann et al. [2019]. Our counterfactual analysis suggests that the optimal menu could increase revenue by more than 3%. The full paper can be found at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4733483.
- 1 January 2024
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Operations Research
- Vol. 72 (1) , 355-372
Mechanism Design Under Approximate Incentive Compatibility
An assumption that is pervasive in revenue management and economics is that buyers are perfect optimizers. However, in practice, buyers may be limited by their computational capabilities or lack of information and may not be able to perfectly optimize their response to a selling mechanism. This has motivated the introduction of approximate incentive compatibility as a solution concept for practical mechanisms. In “Mechanism Design under Approximate Incentive Compatibility,” Balseiro, Besbes, and Castro study, for the first time, the problem of designing optimal selling mechanisms when buyers are imperfect optimizers. Their work characterizes structural properties of approximate incentive compatible mechanisms and establishes fundamental bounds on how much revenue can be garnered by moving from exact to approximate incentive constraints. Their work brings a new perspective to the theory of mechanism design by shedding light on a novel class of optimization problems, techniques, and challenges that emerge when relaxing incentive constraints.
- 1 January 2024
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Operations Research
- Vol. 72 (1) , 355-372
Mechanism Design Under Approximate Incentive Compatibility
An assumption that is pervasive in revenue management and economics is that buyers are perfect optimizers. However, in practice, buyers may be limited by their computational capabilities or lack of information and may not be able to perfectly optimize their response to a selling mechanism. This has motivated the introduction of approximate incentive compatibility as a solution concept for practical mechanisms. In “Mechanism Design under Approximate Incentive Compatibility,” Balseiro, Besbes, and Castro study, for the first time, the problem of designing optimal selling mechanisms when buyers are imperfect optimizers. Their work characterizes structural properties of approximate incentive compatible mechanisms and establishes fundamental bounds on how much revenue can be garnered by moving from exact to approximate incentive constraints. Their work brings a new perspective to the theory of mechanism design by shedding light on a novel class of optimization problems, techniques, and challenges that emerge when relaxing incentive constraints.
- Open Access
- 26 June 2023
- Published by Association for Computing Machinery (ACM) in ACM SIGMETRICS Performance Evaluation Review
- Vol. 51 (1) , 95-96
Power-of-d Choices Load Balancing in the Sub-Halfin Whitt Regime
We characterize the steady-state queue length distribution for the Power-of-d choices routing algorithm for almost all values of d in the sub-Halfin Whitt asymptotic regime.
- 26 June 2023
- Published by Association for Computing Machinery (ACM) in ACM SIGMETRICS Performance Evaluation Review
- Vol. 51 (1) , 95-96
Open AccessPower-of-d Choices Load Balancing in the Sub-Halfin Whitt Regime
We characterize the steady-state queue length distribution for the Power-of-d choices routing algorithm for almost all values of d in the sub-Halfin Whitt asymptotic regime.