A Bayesian Approach to Managing Learning-Curve Uncertainty
提出一个贝叶斯决策模型，用于在存在学习曲线不确定性时优化生产决策。模型扩展了经典学习曲线，允许学习过程随机变化，并分析了最优产量是否超过短视水平以及是否随累积产量增加。
This paper introduces a Bayesian decision theoretic model of optimal production in the presence of learning-curve uncertainty. The well-known learning-curve model is extended to allow for random variation in the learning process with uncertainty regarding some parameter of the variation. A production run generates excess value (above its current revenue) for a Bayesian manager in two ways: it pushes the firm further along the learning curve, increasing the likelihood of lower costs for future runs; and it provides information, through the observed costs, that reduces the uncertainty regarding the rate at which costs are decreasing. We provide conditions under which one of the classical deterministic learning-curve results—namely, that optimal production exceeds the myopic level—carries over to the extended framework. We demonstrate that another classical deterministic learning-curve result—namely, that optimal production increases with cumulative production—does not hold in the Bayesian setting.