I compare returns to scale in the U.S. and Canadian banking system from 1996 to 2015. I estimate a parametric trans-log cost function and, for robustness, an inputoriented distance function. I do this in a way that is commensurate with the limitations of these models. Among the ten largest commercial banks, I find evidence for small but statistically significant increasing returns to scale (RTS). This reflects the descriptive data that offers little evidence for extremely large scale economies. Comparatively, I find constant RTS for the Canadian banks. They paid fewer costs per asset, particularly lower labour costs and legal penalties. Comparing income statement items, I find that, despite higher firm concentration in Canada, the U.S. banks had higher net interest margin rate, paid a lower rate of interest on funds, and had higher credit losses per financial assets. If the U.S. banking system is more competitive, this questions whether an increase in bank competition will create a net positive outcome for society.
QED Working Paper Number