Testing for profitability and contestability in banking: evidence from Austria

Abstract

The paper investigates the determinants of banking profitability and banking market conditions in Austria. We conduct a panel econometric analysis which allows for testing the hypotheses which have become the most prominent in the literature on bank profitability: the structure–conduct–performance hypothesis, the efficient‐structure hypothesis and the relative market‐power hypothesis. Further, we test whether Austrian banking markets are, on average, contestable. A newly compiled dataset covering more than 700 Austrian banks ranging over the period from 1995 to 2002 is used to carry out these econometric analyses. The empirical findings support the view that the Austrian banks do exert, on average, some local market power. However, the gains in terms of excess profits are rather minor as a result of low deterrence powers of the incumbent banks.

Keywords:

• banking performance
• banking profitability
• banking market structure
• panel econometrics

JEL classifications:

• G21
• L13
• L25
• L44

Acknowledgement

I am very grateful to Christa Magerl for excellent research assistance on this project. This article has also greatly benefited from the comments of two anonymous referees and valuable suggestions by the managing editor.

Notes

1. Note, according to the analysis in Hannan (1991) the SCP frame when applied to banking calls for the return on assets as a profitability measure. For an outline of the theoretical reasoning proposed in Hannan (1991), see Hahn (2005).

2. Note that scale efficiency is not identical to scale elasticity (or economies of scale). Scale efficiency, if output‐oriented, measures the change in output required to produce at minimum efficient scale, whereas scale elasticity is a measure related to the relative change in costs associated with an incremental change from a particular output level. The latter concept is usually associated with the measurement of economies of scale.

3. Throughout the paper, we use the term contestability for a market condition with no or, in a practical sense, negligible barriers to entry and exit.

4. According to Mayerhofer (2002) the area of an Austrian administrative district is 847 km² on average, and its population is roughly 87,000.

5. All data of this database are deflated by GDP deflator, 1995 = 100.

6. Descriptive statistics of the used balanced panel of Austrian banks are made available on request.

7. Alternative measures of profitability, such as the return on equity, do not alter the basic findings of the econometric analyses to come.

8. The HHI for a home market is defined as is the market share of the ith bank in the jth district, j = deposits and branches, respectively.

9. For the computation of the efficiency measures considered we use the software package DEA–Solver–PRO 4.0.

10. For example, as compared with low‐income customers a high‐income clientele is expected to show both, a higher demand for advanced banking services such as investment banking products and a higher product quality awareness. Further, high‐income districts are more likely to be economically more developed than low‐income regions which again results in higher demand for high‐end banking products in the former and for low‐end banking products in the latter.

11. The WIFO regional classification scheme results in nine economic regions: metropolitan area, city, suburban, medium‐sized town, intensive industrial region, intensive touristic region, extensive industrial region, touristic periphery, industrial periphery. HUMAN encompasses metropolitan districts, city districts, suburban districts, and medium‐sized town districts. PHYSICAL encompasses intensive industrial and intensive touristic districts. RURAL encompasses extensive industrial regions and the industrial and touristic periphery.

12. The difference in coefficient estimates on X‐EFF_DEA and X‐EFF_SFA , as reported in Table 1, is primarily due to a scale effect.