Transparency and crash risk: an empirical investigation into the US banks

Majority of the literature on crash risk exclude financial firms due to the beliefs that they are unique. In addition, one of the worst economic events in history started amongst banks in 2007 which stunned the public as well-respected and sound banks announced huge losses within a short period of time. As a result, it is important to investigate the transparency of banks and their riskiness that prevented the public from picking up the symptoms of the GFC earlier. In this thesis, I examine how poor bank transparency affects crash risk using a sample of 319 US bank holding companies (BHC) over 2001 to 2014. I employ three alternative measures of transparency including discretionary LLP, fees paid to auditors and CEO option backdating. For discretionary LLP, I find that banks that have high discretionary LLP are exposed to higher one year a-head crash risk. Similarly, I find that banks with high fees, both audit and non-audit fees are at a higher risk of crashes. In addition, I find that the positive relationship between fees and crash risk is more pronounced if it is a Big 4 auditor. This suggests that the Big 4 auditors compromise audit quality for higher fees. Finally, I also find that banks whose CEO’s are offered stock options at the lowest price of the grant month are at a higher risk of crashes. Next, I examine the transparency-crash risk relation during the GFC. I find no significant difference between discretionary LLP and crash risk during the GFC. However, I find that auditors are not influenced by high fees during the GFC. This suggests that they are more vigilant and concerned about protecting their reputation rather than earning high fee revenues, as a result there is less crash risk during the GFC. Finally, I find that the positive effect of CEO option backdating on crash risk is stronger during the GFC. The main results remain unchanged after controlling for potential endogeneity particularly, the possibility for reverse causality using an instrumental variable estimation through a two-stage least squares iv (2SLS). In addition, I deal with a potential omitted variable problem using additional control variables. In additional considerations, I find that a low R2 is not always indicative of price informativeness as banks with high discretionary LLP may have low R2 due to noise trading by investors as there is less available firm-specific information to them to trade on. I also find that the negative relationship between transparency and crash risk is stronger for banks with good corporate governance and/or are located in counties with high proportion of the population with religious beliefs. Furthermore, I confirm my main results by controlling for potential survivorship bias and dealing with outliers as well as testing whether the transparency-crash risk effect is driven by large banks. These findings have major policy implications for regulators particularly future regulation on further auditor fees restrictions and tighter policies on executive option-backdating in banks