Files
Abstract
This paper proposes and applies a set of distribution-free methods for testing for the presence of financial innovation in balance sheet data of large commercial banks. The goal is to detect innovations shortly after they occur using high frequency data. It is postulated that a financial innovation leads to irreversible changes in portfolio ratios, both at the aggregative and individual bank levels. The changes are expected to show up as runs. Using a null hypothesis that changes are the output from a pure white noise process, it is possible to test for an innovation. The method is vulnerable to negative autoregressive patterns that result from monthly payment cycles and reserve requirements. The methods are applied to data for Federal Reserve weekly reporting banks between 1965 and 1976. The results suggest that innovations modified the seasonality of several components of monetary aggregates and that innovations "trickle down" from large to small banks. Certificates of deposit declined in importance permanently at large but not at small banks after 1970. There is some evidence that repurchase agreements and off-shore banking changed the structure of domestic banking liabilities during the early 1970s.