Real options or fallen angels: Examining the complexities of learning from terminated projects

Do organizations learn from innovative failures? Extant research has been mixed. We posit that these disparate findings likely reflect an underlying (and under-researched) spectrum of failure severity, which is comprised not only by the magnitude of the failure in question but by the frequency with which it occurs. Using fine-grained, project-level data, consisting of 33,905 projects in 2496 pharmaceutical firms, we decompose 1251 terminated projects into two types based on their phase of development: early- and late-stage terminations. We use hazard rate modelling and an analysis of abnormal equity returns to assess the likelihood of future failure and the market's reaction to the project termination. Specifically, we are interested in understanding the role that magnitude and frequency of project terminations play in the likelihood and magnitude of future project terminations.