Different Slopes for Different Folks? Train Accident Learning Curves.

Learning curves imply that the greater the experience of a firm the lower its learning rate. However, a growing body of work is finding heterogeneity in how much firms benefit from experience and even find increasing returns to experience. We explore heterogeneity in learning by separating between two mechanisms: opportunity to learn (recent experiences) and operational ability (quality of operations) using US freight rail accident data. We find support for the traditional learning curve: greater opportunity leads to steeper learning. Accounting for different types of experience, we also find support for increasing returns to experience: the better you are, the faster you keep learning. In other terms, increasing and decreasing returns to experience co-exist, although beyond a certain point there is a trade-off.