Banking instability: causes and remedies

Abstract

The recent U.S. subprime mortgage crisis rapidly spread throughout the world and put the global financial system under extraordinary pressure. The main implication of the recent crisis is that complex banking regulations failed to adequately identify and limit riskiness of banking systems at both domestic and international levels. In spite of a large empirical literature on the causes and remedies of the recent crisis, there remains substantial uncertainty on (i) how risk measuring models performed during crisis, (ii) how systematic factors such as house prices affected the financial system, and (iii) how effectively government policy responses resolved the financial crisis. This thesis seeks to narrow this gap in the literature by offering three empirical essays. The first essay investigates the performance of alternative parametric VaR models in forecasting riskiness of international equity portfolios. Notably, alternative univariate VaR models are compared to multivariate conditional volatility models with special focus given to conditional correlation models. Conditional correlation models include the constant conditional correlation (CCC), dynamic conditional correlation (DCC), and asymmetric DCC (ADCC) models. Various criteria are then applied for backtesting VaR models and to evaluate their one-day-ahead forecasting ability in a wide range of countries and during different global financial conditions. It is found that most VaR models have satisfactory performance with small number of violations during pre-crisis period. However, the number of violations, mean deviation of violations, and maximum deviation of violations dramatically increase during crisis period. Furthermore, portfolio models incur lower number of violations compared to univariate models while DCC and ADCC models perform better than CCC models during crisis period. From risk management perspective, most single index models fail to pass Basel criteria for internal VaR models during crisis period, whereas empirical evidence on the choice between CCC, DCC, and ADCC models is mixed. The recent crisis also raised serious concerns about factors that can systematically destabilise the whole banking system. In particular, the collapse of house prices in the United States triggered the recent subprime mortgage crisis, which was associated with a sharp increase in the number of nonperforming loans and bank failures. This in turn demonstrates the key role that house prices play in systematically undermining the whole banking system. The second essay investigates the determinants of nonperforming loans (NPL) with a special focus on house price fluctuations as a key systematic factor. Using a panel of U.S. banking institutions from 1999 to 2012, the analysis is carried out across different loan categories, different types of banks, and different bank size. It is found that house price fluctuations have a significant impact on the evolution of nonperforming loans, while the magnitude of their impact varies across loan categories, institution types, and between large and small banks. Also, the impact of house price fluctuations on nonperforming loans is more pronounced during crisis period. The last essay of this thesis investigates the effectiveness of the U.S. government strategy to combat the crisis. As a comprehensive response to the recent financial crisis, the US government created the Troubled Asset Relief Program (TARP). The Capital Purchase Program (CPP) was launched as an initial program under the TARP. The CPP was designed to purchase preferred stocks or equity warrants from viable financial institutions. Using a large panel of the U.S. commercial banks over the period 2007Q1 to 2012Q4, survival analysis is used to investigate the impact of TARP funds on the likelihood of survival in the recipient banks. It is found that larger recipient banks are more likely to avoid regulatory closure, while receiving capital assistance does not effectively help banks to avoid technical failure. This implies that governmental capital assistance serves larger banks much better than their smaller counterparts. In addition, TARP recipients are more likely to be acquired, regardless of their size and financial health. In summary, the empirical findings reveal that capital infusions do not enhance the survival likelihood of the recipient banking institutions.