RDP 9202: Some Tests of Competition in the Australian Housing Loan Market 2. Data Issues

All studies in applied industrial organisation face a common problem: how to define the industry under consideration. This is not as easy as it seems. Banks are multi-product firms, which not only take deposits and make loans but deliver several other services, e.g. insurance, funds management, corporate advice, trading in foreign exchange markets, management of property trusts, and no doubt many others. Since banks compete, in part, not just against each other but against merchant banks, insurance companies and other providers of financial services, a case can be made for examining competition not just between banks but in the financial services industry as a whole. However, this is impractical because the necessary data do not exist.^[1]

Data on lending by individual banks are available on reporting forms submitted by each bank to the Reserve Bank. These forms contain certain disaggregated data on the bank's assets and liabilities. This is potentially useful information because recent discussions of Australian banking have asserted that “retail” banking is less competitive than “wholesale” banking, the former being done primarily at bank branches, the presence of which constitute an (alleged) barrier to entry. Retail banking is essentially lending for housing plus personal loans. However, reliable data on the latter have been available only since January 1989. We therefore use the housing loan market as a proxy for retail banking and construct a monthly series of housing loans by individual banks. To obtain a measure of the real value of loans, we deflate these data by a housing price series. The Appendix contains details of data sources and construction.

Most analyses of competition start with some measures of concentration. Figure 1 plots the concentration of Australian banking assets as measured by the Herfindahl index. This index is defined as

where s_i is the share of each bank in the market. H=0 is the limiting case of an infinite number of firms with zero share of the market, while H=1 when there is a monopoly. 1/H gives the number for “equal sized firm equivalence”, so that, for example, H=0.25 is equivalent to the existence of four banks of equal size.

Figure 1 shows that the concentration ratio has had its ups and downs since the early 1980s. The big rise in 1982/83 was associated with the merger of the Bank of New South Wales with the Commercial Bank of Australia in October 1982 to form Westpac Banking Corporation, and the merger of the National Bank of Australasia with the Commercial Banking Company of Sydney in January 1983 to form the National Australia Bank. The introduction of new banks – both foreign and converted building societies and merchant banks – in the mid 1980s led to a decline in concentration. The recent takeover of the State Bank of Victoria by the Commonwealth Bank in January 1991 has led to an increase in concentration.

Figure 2 plots the concentration of housing lending over sample period May 1986 to December 1990. This is the period of the deregulated market for housing loans that we use in the estimation of our model. As Figure 2 shows, concentration in loans for housing gradually diminished over the period. (The relatively large fall in March 1987 was due to the United Permanent Building Society becoming National Mutual Royal Bank.) Figure 2 also shows the relative unimportance of foreign banks in the market for housing loans. The average value for H is 0.14.

Footnote

One could also argue that banks compete against building societies in the market for home loans. However, this issue is difficult to deal with in practice because many building societies became banks during our sample period. [1]