Do rating agencies exhibit herding behaviour? Evidence from sovereign ratings

Abstract

We assess the presence of herding by considering the lead-lag relationship of sovereign ratings assigned by the three main rating agencies at the individual country level. Given that different rating agencies may have different levels of expertise (reputation) for different countries it is not obvious that the lead-lag relationship is homogeneous across countries. We therefore conduct poolability tests within this context to assess this assumption and find evidence of heterogeneity. This leads us to conduct country-by-country time-series tests to assess the lead-lag relationship among agencies. To our knowledge we are the first to do this and thereby extend the literature on herding among rating agencies' sovereign assignments. We also consider changes in the lead-lag relationship through time by splitting the sample into pre-crisis and crisis periods to assess the extent to which any herding is intentional and our results indicate some degree of heterogeneity through time. To the extent that there is herding we find that it is generally towards Standard and Poor's ratings confirming our expectations given that this agency is regarded as possessing the greatest reputational capital. However, our results do not support the expectation that Fitch is a follower for more (a leader for less) countries than Moody's.