This paper studies the high and persistent U.S. cost of financial intermediation (CFI) documented by Philippon (2015) and its inverted U-shape behavior since the mid-1960s. We build a novel model of endogenous growth and bank intermediation and introduce imperfect bank competition, bank IT adoption and bank entry, and an occupational choice that determines the relative size of the labor force and the economy's average level of managerial ability. The interplay between verification costs, market structure, and occupational choice delivers implications for the CFI which are qualitatively consistent with the stylized facts of the U.S. economy. We find that the banking sector structure is the main determinant of the long-run level of the CFI. We also show that the U.S. productivity growth slowdown from the mid-1960s to the mid-1980s is a major driver of the simultaneous increase in the CFI and the number of banks during this period and their subsequent decline.