Quanmeng Wang

Problem definition: In the aftermath of disasters, governments must make urgent decisions about how to deploy limited resources for recovery, such as restoring roads or siting emergency facilities. However, these actions often need to be taken before the amount and timing of external funding (e.g., federal disaster relief) are known. This mismatch between the need for immediacy and the delay in budget realization poses a fundamental challenge: How can agencies prioritize recovery actions when budgets are uncertain and decisions, once made, are irreversible? Methodology/results: We develop a practical planning framework that produces a fixed priority list of recovery actions, allowing agencies to act immediately and continue execution as funding arrives over time. The framework identifies early actions that perform well across a range of possible funding paths and preserve the value of later investments. The model is cast as a multiscenario mixed-integer linear program with monotonicity constraints, enforcing consistency in prioritization across all scenarios. To compute such a list efficiently, whereas the natural linear program relaxation of this formulation is weak, we introduce a pegging-based heuristic: For each scenario, we solve the optimal 0-1 allocation, fix it, and relax the remaining scenarios into a linear program. Aggregating across all scenarios yields a robust and interpretable prioritization list. Our analysis provides performance guarantees for committing to a single priority list instead of waiting for full budget information. We derive explicit bounds on the expected performance loss of any prioritization strategy relative to a full-information hindsight benchmark. These results show that, under modest assumptions, the loss from committing to a single priority list is provably small. Furthermore, our pegging-based heuristic yields approximation guarantees under mild conditions and performs remarkably well in empirical evaluations. Managerial implications: This framework offers disaster response planners a rigorous and practical tool for making irreversible decisions under budget uncertainty. The main insight is that the best early action is not always the one that gives the largest immediate gain, but the one that positions the system best when additional funding becomes available. Through experiments on synthetic data and a real-world road network in Manhattan, we demonstrate that the proposed prioritization strategy consistently outperforms conventional heuristics and closely approximates the performance of an ideal benchmark with full budget information. The results highlight the potential of our approach to support timely, resilient, and high-quality disaster recovery planning under uncertain funding conditions. Funding: This work was supported by the National Natural Science Foundation of China [Grant 72422006] and the Hong Kong Research Grants Council Theme-based Research Scheme [Grant T32-615/24-R]. Supplemental Material: The online appendices are available at https://doi.org/10.1287/msom.2024.1574 .

The Locker Alliance Network (LAN), a cutting-edge initiative under Singapore’s Smart Nation vision, aims to revolutionize parcel pickup processes. This government-led project seeks to implement two major changes in Singapore’s urban logistics: (i) establishing an open-access facility for all logistic service providers (LSPs) and (ii) enhancing the last mile delivery efficiency of each LSP, thereby reducing its operational footprint. This research investigates the impact of locker network design on the delivery efficiency of individual LSPs and explores strategies for the government to construct an inclusive network for all LSPs. Our study explores two operational modes: the two-trip mode, in which home and locker deliveries are managed separately, and the mixed-trip mode, which consolidates both into one route. We extend the Beardwood–Halton–Hammersley theorem to estimate the delivery trip length to remaining home locations, assuming consumers can choose between locker pickup or home delivery. This enables us to explicitly formulate the LAN design problem as a nonlinear optimization model. Furthermore, our models for both operational modes reveal that optimal locker network design tends to expand in a nested manner relative to the market share of an LSP. This finding suggests that governments could design locker networks based on the profiles of the largest LSP, ensuring that the expansion of these networks aligns with the existing logistics infrastructure. Leveraging data from LAN’s pilot program in Singapore, our research underscores the pivotal role of network design in last mile delivery efficiency. A poorly designed locker network can inadvertently reduce productivity by increasing the length of delivery trips, thus negating the potential benefits of the system. In contrast, an optimized network can mitigate this negative impact on delivery efficiency. Beyond network design, low customer adoption emerges as another critical bottleneck. Our findings show that, as more customers shift to locker pickup, especially in mixed-trip modes, delivery efficiency improves. Therefore, network design and a higher adoption rate of locker pickup are key to LAN’s success. The government’s role in driving deeper customer engagement, alongside efforts to optimize network design and delivery operations, is vital to unlocking the full potential of this smart nation initiative. This paper was accepted by David Simchi-Levi, operations management. Funding: This work was supported by the National Natural Science Foundation of China [Grant 72422006], the Hong Kong Research Grants Council [Grants 16210720, 16500423], the 2019 Academic Research Fund Tier 3, the Ministry of Education, Singapore [Award MOE-2019-T3-1-010], and the Natural Science Foundation of Chongqing, China [Grant CSTB2022NSCQ-MSX1667]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2024.04567 .