The economic consequences of including fair value adjustments to shareholders' equity in regulatory capital calculations

摘要

We investigate the economic consequences of the implementation of a particular aspect of Basel III in the U.S. Specifically, the Basel III proposal and the corresponding U.S. rule (hereafter referred to as the removal of the AOCI filter) to make the inclusion of unrealized fair value gains and losses of available-for-sale (AFS) securities in regulatory capital mandatory for all banks was highly controversial. The regulators' view that such an inclusion would result in greater bank regulatory discipline was met with the concern that the regulatory costs of such regulatory tightening would exceed any possible benefits. Specifically, opponents of this rule argue that the inclusion of unrealized gains and losses would result in unrealistic volatility in regulatory capital and would force banks to make costly changes to their investment and risk management behavior. Using a comprehensive sample of U.S. banks we provide three pieces of evidence: First, we find that inclusion of unrealized fair value gains and losses on AFS securities for the period 2009 to 2013 would have resulted in increased volatility of regulatory capital. Second, bank share prices reacted negatively (positively) to pronouncements that increased (decreased) the likelihood that this rule would be implemented and these market reactions are strongly positively related to the relative amount of unrealized gains and losses. Third, we find evidence that banks affected by the AOCI filter removal (i.e., advanced approaches banks) changed their investment portfolio management. Specifically, affected banks reduce the maturity of their investment portfolio and decrease the proportion of AFS securities more significantly than unaffected benchmark banks. Interestingly, our results also suggest that affected banks reduce the size of their illiquid investment securities held in the AFS category more than unaffected banks. Given that we observe these changes before the actual implementation date of the new rule, we believe our results speak to the ex ante effects of fair value accounting on banks' risk taking behavior.