Bank risk disclosures, such as Value-at-Risk (VaR), are affected by both changes in market volatility and bank's risk exposures. While the latter is typically unknown to the public, we show how to estimate it from public data on VaR and volatility. We propose a methodology, which we call Factor Implied Risk Exposure (FIRE), that breakdowns a change in risk disclosure into an exogenous volatility component and an endogenous risk exposure component. In a study of large US and international banks, we show that (1) the main driving force of bank risk disclosures is the shifts in risk exposures, (2) changes in risk exposures are negatively correlated with volatility changes, which suggests that banks reduce risk taking when volatility increases, and that (3) changes in risk exposures are positively correlated among banks, which is consistent with banks exhibiting herding behavior in trading.