Can an accounting-based duration model effectively measure interest rate sensitivity

Using a sample of banks and bank holding companies, I test whether an accounting-based duration measure is associated with ex-post measures of rate sensitivity while controlling for the presence of interest rate derivatives. The accounting-based duration measure tested was developed by economists at the Federal Reserve Board of Governors, and Federal Reserve System bank examiners have used it to monitor interest rate risk in U.S. commercial banks since 1998. The tests show that the Fed's accounting-based duration measure does indeed capture interest rate risk. In several different tests, the model robustly and repeatedly separates financial institutions that are vulnerable to rate increases from those that are vulnerable to rate decreases even in the presence of derivatives accounting information. Both stock returns and accounting performance are associated with rate changes in a manner that the model suggests. I conclude that such a model effectively measures the direction and relative magnitude of bank interest rate sensitivity.