Bad luck or bad management: how system effects moderate inferences about executive ability

Failure and disaster are often blamed on the people in charge. Organizational researchers have argued that such practice of blaming individuals is often misplaced because the fundamental causes of failures are often tightly coupled systems which make organizations sensitive to rare external shocks. In this project we use a formal model to evaluate this claim and examine how inferences about skill depend on system design. We show that very poor performance in tightly coupled systems is relatively uninformative about individual skill and that moderately low performance can be a more reliable indicator of low skill. Data from Formula One Racing supports this prediction: cars with extremely poor performance do not have the lowest expected quality. Experiments show that participants can learn the non-monotonic association between performance and skill in our model, but they do not act upon this knowledge when making hiring or firing decisions, implying that learning is perhaps not enough to help resist the temptation of firing unlucky executives.


Chengwei Liu
Associate Professor of Strategy and Behavioural Science
Jerker Denrell
Professor of Behavioural Science