The following post comes to us from Sugato Bhattacharyya and Amiyatosh Purnanandam, both of the Department of Finance at the University of Michigan.
Excessive risk-taking by banks is widely blamed as a primary factor behind the financial meltdown of 2007-2008. Yet, not much work has been done on whether banks fundamentally changed their risk-taking behavior prior to the crisis, nor has much formal work been done on whether banks’ risk-taking was “excessive” in any way. In our paper, Risk-taking by Banks: What Did Banks Know and When Did We Know It?, which was recently made publicly available on SSRN, we tackle these questions head on and also examine possible motives for bank managers to have changed their risk-taking behavior in the years leading up to the crisis.
In the years 2000 to 2006, a preliminary examination of stock price volatility does not seem to support the idea that the financial markets deemed the level of risks assumed by banks to be excessive. But we document a remarkable compositional shift in the measures of risk-taking: bank’s systematic risk, measured by their equity betas, almost doubled while their idiosyncratic volatility came down significantly. This reduction in idiosyncratic risk is consistent with the increasing reliance on securitization to shed firm-specific risks. But the remarkable increase in betas clearly shows that bank assets were becoming increasingly similar in terms of their risk characteristics and that future bank performance was viewed as much more dependent on the performance of the macro-economy. Our results indeed indicate major changes in the nature of risk-taking by banks in the years preceding the crisis.
We also find that the degree of compositional shift in risk was correlated with heavier involvement in mortgage lending and securitization activity. Such involvement allowed banks to generate relatively higher earnings per share during this period. However, the increase in accounting performance was not accompanied by relatively higher stock returns. Since taking on more systematic risk in booming markets naturally produces higher earnings, we take the relatively lower stock price performance as a sign of excessive risk-taking. Consistent with this interpretation, we find that the market reaction to unanticipated earnings increases was significantly muted for banks heavily engaged in mortgage lending. We also find a reflection of market’s doubts about the sustainability of such reported earnings in the heavier mortgage default rates suffered by these same banks in the post-crisis period.
To investigate why banks may have engaged in excessively risky behavior in the pre-crisis period, we look at the compensation patterns of their CEOs. We show that their CEOs’ compensation depended more heavily on measures of short-term earnings such as EPS relative to stock returns. Such compensation patterns, thus, seems to have given them greater incentives to focus on generating short-term performance even at the cost of taking on excessive amounts of risk. We also find that the sensitivity of compensation to EPS actually increased post-2000 just as the banking sector’s ability to generate short terms earnings through securitization increased during this period. Our study, thus, provides support to the view that governance structure – particularly compensation structure – failed to adequately combat incentives to take on excessive risks during the pre-crisis years.
The full paper is available for download here.