Optimal Ordering Policy and Value of Information under Delayed Lost Sales Observations

摘要: Under many circumstances, demand observations are often censored due to the lack of tracking lost sales caused by stockouts. To understand impact information on ordering decisions, a periodic-review inventory model is formulated in which only obtained immediately upon realization demand. This equivalent observing when less than available stock and inferring that higher there stockout. Subsequently, after delay. In resulting model, an optimal policy, if exists, reveals very complex structure. By decomposing derivative value function, we demonstrate two different roles our model: satisfying extracting information. We show levels under delayed observation always those for demands fully observed. Moreover, as illustrated numerical examples, policy possesses counterintuitive behavior with respect problem parameters. key drivers further compare costs observations. Two important made. First, lower cost realized observed be level, and, furthermore, difference represents Second, while level induces more accurate forecast, exact not monotone procurement cost. Consequently, quantity decreasing