Publication Type
PhD Dissertation
Version
publishedVersion
Publication Date
6-2026
Abstract
Freight forwarders act as intermediaries between shippers and carriers, facilitating the efficient movement of goods across global supply chains. Their core business is to organize logistics on behalf of the shippers. They do so by purchasing transportation capacity from carriers in advance and resell it to shippers. By consolidating shipments from multiple clients, forwarders optimize capacity utilization, generating profits through economies of scale.
However, freight forwarders face a critical challenge due to timing mismatch in their business model. They must commit to purchasing shipping capacity months ahead, while shipper demand remains uncertain. This often leads to over-procurement on some routes, incurring unnecessary costs, or under-procurement on others, resulting in missed opportunities and capacity shortages. Both scenarios erode profitability.
As such, collaboration among forwarders, such as capacity sharing, could offer a potential solution to mitigate these risks. By pooling container resources, forwarders can better align capacity with demand. However, equitably allocating costs among the participants, where different forwarders contribute varying levels of capacity and demand, poses a complex challenge. This dissertation explores how game-theoretic approaches can provide a principled framework to address this challenge, ensuring fair cost distribution while incentivizing cooperation. The key contributions of this dissertation are a mathematical program to model forwarder collaboration, methods for cost sharing in static and dynamic settings, and a framework for integrating forwarders with pre-existing collaboration into broader platform-enabled cooperation.
First, when multiple forwarders agree to collaborate by sharing capacities anddemands, determining the optimal assignment of shipments to available containers becomes computationally challenging. We formulate this as the Freight Forwarder Collaboration Problem (FFCP), an integer linear program that minimizes total shipping costs while respecting capacity constraints. FFCP is a variable-cost bin-packing problem and hence NP-hard. We develop a two-step solution approach that first uses a greedy heuristic to establish feasible bounds, then applies exact optimization within these bounds. This method achieves optimal solutions while reducing computation time by up to 81% compared to simply solving it as an integer program. Experiments show that the approach makes it practical to solve realistic instances.
Second, in static settings where all forwarders participate simultaneously, computing fair cost allocations with Shapley value using traditional methods becomes computationally prohibitive for large networks. We introduce Locally Collaborative Games (LCG), a new class of cooperative games that exploits the network structure of forwarder collaborations, where an agent’s marginal contribution depends only on its immediate collaborators rather than the entire network. Our Fast Shapley algorithm for LCG reduces runtime by 82% compared to general graph-restricted game algorithms. Experimental results demonstrate that forwarders achieve 15% cost savings through collaboration, with the algorithm scaling effectively to networks of 50 participants.
Second, in static settings where all forwarders participate simultaneously, computing fair cost allocations with Shapley value using traditional methods becomes computationally prohibitive for large networks. We introduce Locally Collaborative Games (LCG), a new class of cooperative games that exploits the network structure of forwarder collaborations, where an agent’s marginal contribution depends only on its immediate collaborators rather than the entire network. Our Fast Shapley algorithm for LCG reduces runtime by 82% compared to general graph-restricted game algorithms. Experimental results demonstrate that forwarders achieve 15% cost savings through collaboration, with the algorithm scaling effectively to networks of 50 participants.
Third, in dynamic environments where forwarders join sequentially, the challenge is to compute allocations that are simultaneously fair, stable against coalition defections, and consistent, meaning no existing forwarder’s cost should increase when a new participant arrives. We introduce the Dynamic Stable and Fair Allocation Problem (DSFAP) to formalize these requirements jointly. We develop the Coalition-Based Subsidy (CBS) mechanism, which models the subsidy as a cooperative game, and show it reduces to a tractable Agent-Based Subsidy (ABS) linear program with one decision variable per agent. A second-stage minimum-norm quadratic program then selects the unique minimum-subsidy allocation closest to the Shapley value, resolving LP non-uniqueness. An incremental caching algorithm based on a recasting formulation achieves up to 1.6× speedup and saves 38% of cumulative Shapley computation time. Experiments show subsidies remain modest (under 5.3% of coalition cost), with arrival order most affecting large forwarders.
Fourth, when forwarders with pre-existing co-loading relationships join a platformmediated collaboration, determining fair cost allocation and platform pricing becomes challenging. We introduce the Platform Value Game (PVG), which models the platform explicitly as an agent and distinguishes pre-existing from platform-enabled collaboration through two characteristic functions. Applied to freight forwarding as the Platform-Enabled Forwarder Collaboration Game (PEFCG), dynamic platform charges emerge naturally from the Shapley value. To make computation tractable, we exploit three structural properties of PEFCG (service separability, packing-pricing separation, and component decomposition), organized into a three-stage pipeline of a coalition-value oracle, a permutation Shapley estimator, and a subsidy quadratic program. Numerical experiments show that the platform creates the most value when pre-existing collaboration is sparse, the market is fragmented, or container costs are heterogeneous, while stabilizing subsidies remain small.
Degree Awarded
PhD in Computer Science
Discipline
Operations and Supply Chain Management | Transportation
Supervisor(s)
CHENG, Shih-Fen
First Page
1
Last Page
132
Publisher
Singapore Management University
City or Country
Singapore
Citation
TAN, Pang Jin.
Collaborative freight forwarding: A game-theoretic perspective. (2026). 1-132.
Available at: https://ink.library.smu.edu.sg/etd_coll/923
Copyright Owner and License
Author
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.