The purpose of this thesis is to analyze the drivers and dynamics of credit losses in Australasian banking over an extended period of time in order to improve the means by which financial institutions manage their credit risks and regulatory bodies safeguard the stability and integrity of the financial system. The analysis is based on a specially constructed data base of credit loss and provisioning data retrieved from original financial reports published by Australian and New Zealand banks. The observation period covers 1980 to 2005, starting at the time when such information was published for the first time in bank financial statements. It moreover covers the time of major crises which occurred in both Australia and New Zealand in the late 1980s and early 1990s. The heterogeneity of reporting the data both amongst banks and through time requires the development of a reporting typology which allows data extraction with equivalent informational content. As a thorough study of credit risks requires long data series often not available from third party data providers, the method developed here will provide value to a range of researchers. Based on an evaluation of many alternative proxies which track a bank's credit loss experience (CLE), the thesis proposes a preferred model for impaired assets expense (as % of loans) as dependent variable, mainly because of its timely nature and good data availability. Explanatory variables include aggregate macro variables of which changes in unemployment and the return in the share markets are found to have the most significant influence on a bank's credit losses. Bank-specific control variables include a pre-provision earnings proxy whose significance points to the use of provisions for the purpose of income smoothing by Australasian banks. The model also controls for size and nature of lending as smaller, retail-oriented housing lenders, on average, exhibit lower loan losses. Clear results are found with regard to the effect of rapid expansion which appears to be followed by a surge of bad debt provisions 2 to 3 years later. Moreover, inefficient banks tend to suffer greater credit losses. An important part of the thesis looks at the characteristics of alternative CLE proxies such as stock of provisions, impaired assets and write-offs which have been used by earlier literature. Estimating the preferred model with such alternative CLE parameters confirms their peculiarities such as the memory character of stock of provisions and the delayed nature of write-offs. These measures correlate rather poorly amongst themselves which calls for caution in the comparative interpretation of earlier studies that use differing CLE proxies.
| Identifer | oai:union.ndltd.org:ADTP/238070 |
| Date | January 2008 |
| Creators | Hess, Kurt |
| Publisher | The University of Waikato |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Rights | http://www.waikato.ac.nz/library/research_commons/rc_about.shtml#copyright |
Page generated in 0.0016 seconds