Human Decision-Making in Dynamic Resource Allocation

We experimentally study individuals' dynamic resource allocation decisions under a budget constraint, using product development as the decision context. During the development process, a series of design opportunities will emerge to improve the product with extra engineering costs. These opportunities arrive sequentially, and a decision of whether to implement the design change or not must be made when the opportunity arrives. All of this is managed under a pre-specified budget. We find that human subjects perform reasonably well in a simple condition where the cost-to-implement is fixed throughout the project, and the only uncertainty comes from the ex-ante random benefits of the opportunities. However, in a more complex and practical situation where the cost is known to increase for the latter half of the project, subjects' performance worsens substantially. One key reason is that many subjects in the more complex condition act as if the cost were constant throughout. Two potential drivers behind such behavior are proposed: “limited attention", where subjects fail to pay attention to the cost-increase information, and “limited cognition", where subjects fail to derive action plans to act on the cost-increase information. To improve performance, two “targeted" treatments are developed to address these drivers, along with a third treatment using extra monetary incentives to induce effort. We find that addressing limited attention alone is insufficient to achieve a performance improvement. Subjects can overcome their cognition barriers via curated social learning, which achieves a significant performance improvement and is more economical than the treatment where incentives are magnified