Putting Behavioral Operations to Work: Improving Firm Performance by Understanding Human Behavior
Flicker, Blair Allyn
MetadataShow full item record
Despite increasing computerization and automation in recent decades, human labor and decision making is still crucial to many business activities. Unemployment is at record lows despite the staggering pace of technological advance. However, the economic value that humans provide is often overlooked. Computers can outperform humans in an increasing number of tasks and games (e.g., chess, Jeopardy!, go), but humans are better able to respond to unanticipated phenomena. Academic research tends to extol the virtues of automation and minimize the role of humans. For example, stylized analytical models omit many real-world phenomena, and such model misspecification degrades the performance of prescribed “optimal” policies in practice. By accounting for unmodeled dynamics, human managers can improve decision making (although heuristics and biases may undermine potential improvement). Human laboratory participants endowed with potentially valuable private demand information fail to utilize this information when asked to place newsvendor orders, but a novel human-machine hybrid algorithm effectively extracts participants’ private information, converts it to orders, and substantially improves profitability compared to unaided human orders and fully automated orders. Even a firm seeking to fully automate operations needs to consider the influences of human behavior because other players (e.g., supply chain partners, competitors, customers) likely involve humans in their decision making. An understanding of human behavior is required to best respond to humans’ decisions. These concepts are demonstrated in two studies combining theory and laboratory experiments. Additionally, I advance a novel method for providing noisy signals to laboratory participants using ambiguous visual cues as opposed to random variables with well-defined stochastic behavior. These “perceptual” signals are a more appropriate model for certain types of information that humans gain from observing the world (e.g., a manager’s sense of future demand for a product, an real estate agent’s estimate of the selling price of a home).