Ignoring the Obvious
One of the conundrums of the credit crisis was why banks kept some exposure to the securitizations they issued even as the quality of the underlying loans was declining? Unfortunately, a new paper published by the NBER that addresses the question ends up muddying the water by casting doubt on some of the accepted explanations of the credit bubble. The paper reads like a case study on some of the dysfunctions of academic economics research.
After running through the data, the authors produce a number of eyebrow-raising conclusions, including:
Banks at which executives’ and traders’ compensation was more closely tied to return on equity (unadjusted for risk) did not display a greater appetite for risk through a willingness to keep part of its securitization tranches on their books. That result is inconsistent with the finding of a 2009 paper, Executive Compensation and Business Policy Choices at US Commercial Banks
, according to which “banks in which CEOs have high pay-risk sensitivity [to the volatility of the bank’s stock returns] invest a larger percentage of their assets in private mortgage securitizations.” Or as former Countrywide Financial CEO Angelo Mozilo justified a $10 million stock award to a Congressional committee: “[The award] had performance-based aspects to it... I had to provide a return on equity to the shareholders.” “The large dispersion of holdings across the largest banks explains why there is no support for the arguments that banks that are viewed as too-big-to-fail had incentives to have large holdings of [private MBS] assets.”
Here, the authors appear to assume that all agents respond to incentives uniformly. Instead, isn’t it possible that some ‘too-big-to-fail’ banks preferred not to avail themselves of the option to fail, while others were willing to test it through risky behavior? Richard Fuld, the former CEO of Lehman Brothers, wouldn’t entertain offers of less than $10 per share for his bank less than a week before it filed for bankruptcy) on the view that the Federal Reserve was empowered to inject capital into Lehman. That would have been $10 more than he and his shareholders were ultimately left with.
In sum, this paper strikes me as an example of the attitude that “the model is right, reality is wrong.” When there are intuitive behavioral explanations for economic relationships (e.g, Bank CEOs with compensation tied to ROE are willing to take more risk on) and enormous anecdotal evidence behind them, it’s probably worth reviewing a statistical model that fails to support them.