Variables affecting decisions in a multistage inventory task

This study is concerned with the experimental investigation of human dynamic decisions in multistage inventory problems. Its purpose is to detect discrepancies between observed decisions and optimal decisions which could reveal the existence of constraints on human ability to process information and make dynamic decisions. Thirty-four subjects were run individually in six different computer-controlled multistage inventory tasks. They were required to maximize the gain of an operation involving the producing, storing, and selling of a stock of goods. Two-models are tested—a single-stage inventory model and a multistage inventory model. Both are normative models prescribing constant-level optimal ordering policies. Both are rejected as descriptive models. The subjects' level of stock is not constant but slightly decreasing and strongly related to the previously observed demand.