For market discipline to be effective, market factors such as changes in firm equity and debt values and returns, must influence firm decision making. In banking, this can occur directly via bank management or indirectly though supervisory examinations and oversight influencing bank management. In this study, we investigate whether equity market variables can provide timely information and add value to accounting models that predict changes in bank holding company (BOPEC) risk ratings over the 1988–2000 period. Using a variety of equity market indicators, the findings suggest that one-quarter lagged market data adds forecast value to lagged financial statement data and prior supervisory information in the logistic regressions. Furthermore, using extensive out-of-sample testing for the years 2001–2003, we find: (1) that multiple models estimated over different phases of the business and banking cycles are superior to a single model for forecasting BOPEC rating changes; (2) that equity data adds economically significant power in forecasting BOPEC rating upgrades and performs well for identifying no changes; (3) that for downgrades, the accounting model forecasts the best; (4) that modeling the three possible risk ratings categories simultaneously (downgrade, no change and upgrade) minimizes both Type I and Type II classification errors; and (5) that using multiple models to forecast risk ratings enhances the overall percentage of correct classifications.

This paper examines whether bank holding company (BHC) risk ratings are asymmetrically assigned or biased over business cycles from 1986 to 2003. In a model of ratings determination which accounts for bank characteristics, financial market conditions, past supervisory information, and aggregate macro-economic factors, we find that bank exam ratings exhibit inter-temporal characteristics. First, exam ratings exhibit some evidence of examiner bias for several periods analyzed. When the business cycle turns, examiners sometime depart from standards that they set during the previous phases of the cycle. However, this bias is not widespread or systematic. Second, exam ratings exhibit some inertia. Our results suggest that examiners rate on the side of not changing (rather than upgrading or downgrading) an institution’s exam rating. Third, we find robust evidence of a secular trend towards more stringent examination BHC ratings standards over time.