Idiosyncratic risk assessment in the mortgage market

This thesis investigates the role of borrower characteristics in revealing risk assessment in the mortgage market. Two innovations make important contributions to the international literature. First, the thesis exploits a unique, detailed mortgage application database which provides a means by which to investigate the role of individual characteristic and personal circumstances in the variety of mortgage contracts found in the market. Second, it takes advantage of the distinctive characteristics of Australian financial markets to analyze contract choice in an environment where, due to the relative underdevelopment of a mortgage securitisation market and absence of government guarantees, the risks of the mortgage contract are spread mainly between the originator (the bank) and the mortgagee (the household). As a result of the impact and spread of the global financial crisis of 2008-09, governments and regulators are focusing their attention on macroprudential policy and systemic risk. Results presented in this thesis demonstrate that underpinning these macroeconomic concerns are important issues for the understanding of idiosyncratic mortgage risk. We focus on these micro level problems by studying loan-level owner-occupier mortgage application data. This work reinforces the relevance of assessing and pricing the risk for borrowers and lenders based on the personal characteristics of the individual household. In particular, the thesis investigates the role of borrower characteristics in the determination of: (i) mortgage product choice; (ii) loan-to-value ratios; and (iii) mortgage interest rate. The thesis brings together three substantive essays which are intended to be published as separate papers, accompanied by two chapters which provide background to the Australian financial sector (and the role of mortgage markets) and details on the micro loan-level application data on which the subsequent empirical work is based. The first essay aims to provide a formal clarification of the association between borrower characteristics and types of mortgages by building typologies of borrowers based on their characteristics and the type of mortgage they take. It introduces the use of Multiple Correspondence Analysis techniques to household finance data. The empirical work finds that young, low-income and low-wealth applicants deviate from the mean borrower, and that households with high income but low wealth are associated with variable- or fixed rate mortgages, while households with low income but high wealth are associated with home equity loans and discounted variable-rate mortgages. However, borrowers remain heterogenous within mortgage types. Fixed-rate mortgages are taken by young, constrained families, settled families and mobile first-time buyers. Discounted variable-rate mortgages are taken by young (female) households, settled families and mobile first-time buyers. The second essay considers the empirical evidence on mortgage product choice with the inclusion of a full range of variables representing mortgage costs, market conditions and borrower characteristics, consistent with models proposed in the theoretical literature. A discrete choice model is used to predict the marginal effect of borrower characteristics on the probability of choosing a particular mortgage product. Moreover, it explores the way mortgage choice effects are affected by loan-to-value ratio incentives given to banks and imposed by the Australian regulatory body - in accordance with the Basel capital requirements. It reveals a relationship between the prudential regulatory structures and the mortgage choice at an individual-level. It finds that, contrary to popular wisdom, the Basel capital adequacy rules play a 'sorting' role in terms of the types of loan contracts that emerge from the interaction between banks and borrowers. Although mortgage choice is mainly determined by the relative cost of the mortgage and by a broad range of borrower characteristics - which reveal life-cycle stage, income and wealth constraints and uncertainty, risk-aversion, financial experience, and mobility expectations - these borrower characteristics play different roles for different loan-to-value ratio levels. The third essay addresses possible endogeneity and bias selection in estimating mortgage choice. Endogeneity is addressed by predicting the loan-to-value ratio and instrumenting the value of the property under the mortgage contract. Selection bias, which arises from observing only the revealed preference of the borrower and not the alternatives in the choice, is corrected by estimating the reduced form of the interest rate associated with the mortgage. This chapter focuses on the effect borrower characteristics have on interest rate and loan-to-value ratio determination. It finds that the value of the property at the time of application is not statistically significant in determining the loan-to-value ratio the bank offers to the borrower; however house price in ation and expectations determine the loan-to-value ratio. Moreover, income and wealth levels are strong factors determining both loan-to-value ratios and individual interest rates. Importantly, borrower characteristics play the role predicted by theory in defining price and terms of the mortgage at the individual level. This finding is necessary, but not suffcient, for market effciency.