In global supply chains, high uncertainty coming from liner shipping forces manufacturers who commit to on-time delivery to face great losses from tardiness or earliness. To find a reliable operational solution for shipment assignment, realizing the trade-off between the operations cost and reliability is highly essential. In this paper, a risk-hedging policy for shipment assignment is proposed. We incorporate risk aversion into the stochastic optimization model where the objective is to minimize the total deterministic operations cost and the weighted value-at-risk. The closed-form expressions of value-at-risk are derived and some structural properties of the optimization problem are revealed. Linearization techniques are utilized to make the proposed model tractable and solvable by an exact algorithm. Our computational studies further demonstrate the cost efficiency on systems reliability improvement through the risk hedging approach with job combination. It is interesting to uncover that being moderately risk-averse is wise, but possessing a very high risk-averse attitude is doing more harm than good as the increase of deterministic operations cost is much more significant than the decrease of value-at-risk. Being risk neutral may also be unwise as the chance of achieving the optimal expected total cost may be very low. In the extended models, we relax the assumptions and further consider scenarios with correlated shipping lead-times and stochastic exchange rate.

在全球供应链中，班轮运输带来的高度不确定性迫使制造商承诺准时交货，以使其因拖延或过早而蒙受巨大损失。为了找到可靠的运货分配解决方案，在运价和可靠性之间进行权衡非常重要。在本文中，提出了一种针对货运分配的风险对冲策略。我们将风险规避纳入随机优化模型，该模型的目的是使总确定性运营成本和加权风险价值最小化。推导了风险价值的闭式表达式，并揭示了优化问题的一些结构性质。利用线性化技术通过精确算法使提出的模型易于处理和求解。我们的计算研究进一步证明了通过结合工作组合的风险对冲方法来提高系统可靠性的成本效率。有趣的是发现适度地规避风险是明智的，但具有很高的规避风险态度弊大于利，因为确定性运营成本的增加远比风险价值的降低更为重要。中立于风险也可能是不明智的，因为达到最佳预期总成本的机会可能非常低。在扩展模型中，我们放宽了假设，并进一步考虑了具有相关装运提前期和随机汇率的情况。但是具有很高的规避风险态度弊大于利，因为确定性运营成本的增长远比风险价值的下降更为重要。中立于风险也可能是不明智的，因为达到最佳预期总成本的机会可能非常低。在扩展模型中，我们放宽了假设，并进一步考虑了具有相关装运提前期和随机汇率的情况。但是具有很高的规避风险态度弊大于利，因为确定性运营成本的增长远比风险价值的下降更为重要。中立于风险也可能是不明智的，因为达到最佳预期总成本的机会可能非常低。在扩展模型中，我们放宽了假设，并进一步考虑了具有相关装运提前期和随机汇率的情况。