Securitization and Italian banks’ risk during the crisis

Abstract

Purpose – The purpose of this study was to examine the 2007-2009 financial crisis to analyze how securitization relates to the Italian bank risk profile, both in terms of credit and liquidity risks. Design/methodology/approach – To test our research hypotheses, we adopt ordered probit models, in which we regress the changes in credit risk and liquidity on a set of regressors, including two securitization dummy variables plus a vector of control variables. Findings – Our results show that the impact of securitization on the originators risk-taking is not uniform. When credit risk is considered, the securitization effects seem to be statistically significant only during the crisis period. However, when we turn to analyze the bank’s liquidity position, our results show that securitization improves it both during the pre-crisis and the crisis years. Our results support the Basel III initiatives aimed to realize a better integration between the different types of risks (i.e. credit and liquidity risks). Research limitations/implications – The major limitation of our study is related to the analyzed geographic area. Practical implications – First, our results support the Basel III initiatives aimed to realize a better integration between the different types of risks (i.e. credit and liquidity risks). In general, the broad policy implication of the paper is that in some contexts, such as the Italian market, securitization does not necessarily produce negative effects in terms of bank’s risk. Originality/value – This study contributes to the empirical literature on the effects of securitization for banks in several ways. First, we consider the complexity of the bank’s risk profile; second, despite the importance of the Italian securitization market, there is a research void on it. Furthermore, unlike previous studies, our analysis covers the period 2000-2009, including the financial crisis years. Finally, to our knowledge, our methodology (ordered probit models) has not been used in the past in this context.