Reputation, Information, and Herding in Credit Ratings: Evidence from CMBS

  • PDF / 761,796 Bytes
  • 29 Pages / 439.37 x 666.142 pts Page_size
  • 12 Downloads / 162 Views

DOWNLOAD

REPORT


Reputation, Information, and Herding in Credit Ratings: Evidence from CMBS Xudong An 1 & Larry Cordell 1 & Joseph B. Nichols 2 # Springer Science+Business Media, LLC, part of Springer Nature 2019

Abstract We find strong evidence of herding behavior among credit rating agencies (CRAs) in the CMBS market. CRAs are more likely to change their rating if another CRA had changed its rating on the same bond in the previous period. CRAs tend to adjust their ratings to make them converge with those of other CRAs when there is rating disagreement in the previous period. CRA reduced their herding behavior during the financial crisis and herding after the crisis was still lower than it was pre-crisis. We also find evidence that herding in corporate bonds is largely limited to below-investment grade companies. In contrast, we see evidence of herding in CMBS in both high and low rated tranches. These results are consistent with predictions from a theoretical model where CRAs use both public and private information in determining their credit ratings. Keywords Credit rating . CMBS . Herding

Introduction The investments literature has long studied herding behavior among fund managers and professional financial analysts (see, e.g., Scharfstein and Stein 1990; Banerjee 1992; Shiller 1995; Grinblatt et al. 1995; Welch 2000; Choi and Sias 2009). Very few have studied such behavior among credit rating agencies (CRAs). This is a critical gap in the * Xudong An [email protected] Larry Cordell [email protected] Joseph B. Nichols [email protected]

1

Federal Reserve Bank of Philadelphia, Philadelphia, PA, USA

2

Federal Reserve Board, Washington, DC, USA

X. An et al.

literature given the critical impact of CRA rating actions on many risk management and asset allocation decisions. A key aspect of the financial crisis was a decline in the confidence in the reliability of CRA ratings of mortgage-backed securities (MBS) and collateralized debt obligations (CDOs). In this paper, we present evidence of herding among the CRAs by studying CRAs’ upgrades and downgrades of commercial mortgage-backed securities (CMBS), comparing CMBS rating migration to that of corporate bonds. We develop a simple model, based on Morris and Shin (2002), where the CRAs make their decisions based on their own privately developed credit rating models, but also take into account the difference between their private estimate of bond credit worthiness and the public consensus. This results in an incentive towards herding, and we show through the model that we would see more herding behavior when the value placed on public consensus is higher. Our model encompasses the following two channels of herding. A CRA concern for their reputation can naturally lead to herding as they try to align their ratings with others, fearing that contrarian behavior will damage their reputations as sensible certifiers (Scharfstein and Stein 1990; Devenow and Welch 1996). Further, when this reputational risk is higher, the value assigned to the public consensus is hi