Why German banks should merge

The Authors

Peiyi Yu, Department of International Business, University of Wolverhampton, UK

Werner Neus, Faculty of Economics and Business, Department of Banking, University of Tuebingen, Tuebingen, Germany

Bac Van Luu, Landesbank Baden-Wuerttemberg, Stuttgart, Germany

Sean Dodd, Department of International Business, University of Wolverhampton, UK

Abstract

Purpose – The paper aims to investigate whether the wave of mergers observed in other European countries is suitable for the German banking industry.

Design/methodology/approach – This question is approached by studying the relationship between market structure and profit (the so-called profit-structure relationship) in the German banking industry using the model suggested by Berger. By extending his econometric model to include portfolio and capital risk and using German banks' financial data, the authors are able to test simultaneously for three competing theories of the profit-structure relationship.

Findings – It is found that including portfolio risk significantly improves the fit of the estimated profit-structure relationship. In addition, it is found that scale economies are present in German banking and that the so-called structural conduct – performance hypothesis is accepted. Owing to the acceptance of this hypothesis, conclude it is that mergers are likely to boost German banks' profitability but possibly at the expense of lower consumer welfare.

Research limitations/implications – The research methodology employed is unable to weigh the potential benefits of higher financial sector profitability and stability against the possible detrimental impact on consumer

Originality/value – The value of the paper is to provide a robust estimate of scale economies and a reliable test of the profit-structure relationship in German banking based on recent data. It should be of particular value to bank managers, seeking to increase their institution's profitability by way of merger and to the regulator of the financial industry.

Article Type:

Research paper

Keyword(s):

Germany; Banking; Profit; Acquisitions and mergers.

Journal:

Studies in Economics and Finance

Volume:

24

Number:

2

Year:

2007

pp:

140-155

Copyright ©

Emerald Group Publishing Limited

ISSN:

1086-7376

Introduction

In a recent paper published by the IMF, Decressin et al. (2003) propose that recent weak bank profitability in Germany appears to be related with structural factors rather than the macroeconomic cycle. Some anecdotal evidence and financial ratio analyses are also presented to support this claim. The motivation of this paper is to study the issue of bank profitability in a coherent and rigorous econometric framework with a large panel data set of the German banking industry. Another main motivation is to go beyond the factors explored in the IMF paper in explaining why the profitability of the German banking system has been relatively low and trended downwards over recent years. For example, over 20 per cent of Germany's commercial banks in the Fitch IBCA database did not earn a rate of return for their owners that exceeded the rate of a risk-free treasury bill[1]. This immediately leads to the question of how the structure and the organisation of the German banking system can be changed to safeguard banks' profitability and the sector's stability.

As shown in Table I, the German banking system is composed of the three following pillars: commercial banks, cooperatives, and public sector banks[2].

These three pillars are all different with respect to ownership and objectives. For example, most of the public sector banks are effectively owned by state and local governments, which operate commercially but also have a public mandate and currently benefit from a government guarantee. The second group in the German banking industry is the cooperatives (Volksbanken, Raiff-eisenbanken, Spar- and Darlehenskassen). This group of banks was founded as self-help organizations for craftsmen, workers and farmers. Cooperative banks concentrate on their respective local markets and do not compete with one another[3]. Finally, the major part of private sector banks are commercial banks. Commercial banks comprise the big four banks, which account for roughly two thirds of this sector's business. The private sector banks also include the Postbank[4], foreign banks, and numerous smaller banks. The biggest four commercial banks comprise Deutsche Bank, HypoVereinsbank, Dresdner Bank, and Commerzbank. Like the cooperative banks, they do not benefit from a public sector guarantee, and thus are at a disadvantage relative to the state banks (Landesbanken) in tapping capital markets. Neglecting the different ownership structure, co-operations and public banks exhibit a quite similar behaviour.

Generally speaking, from the Fitch IBCA database, we can observe that savings banks (Sparkassen) and cooperatives currently have higher returns on equity than commercial banks.

After reviewing the German banking system, we come back to the argument of how structural factors affect the German banking profitability. Many propose that the relatively low profitability of the German banking system could possibly reflect that profit maximization is not always the paramount objective of public sector banks and cooperatives. Furthermore, a high number of banks per capita leads to an intense competition. For instance, Decressin et al. (2003) point out that competition in Germany appears to be more intense than in the United Kingdom and France. In the following part, we will try to find out how the market structure affects banks' profitability by examining a model that can distinguish between three competing profit-structure hypotheses.

In brief, these three profit-structure hypotheses have emerged in the banking literature to explain the profit-structure relationship (Molyneux and Thornton, 1992; Lloyd-Williams et al., 1994; Molyneux and Forbes, 1995; Katib, 2005). They are the structural-conduct-performance hypothesis, the relative-market-power hypothesis, and the scale-efficiency version of the efficient-structure hypothesis. The structural-conduct-performance hypothesis states that banks set prices that are less favourable to consumers in more concentrated markets because of an imperfect competition. The relative-market-power hypothesis suggests that only banks with large market shares and well-differentiated products can exercise market power in pricing these products and earn supernormal profits (Shepherd, 1982). Finally, under the scale-efficiency version of the efficient-structure hypothesis, all banks have equally good management and technology (the same X-efficiency), but some banks simply produce at more efficient scales than others. Under the scale efficiency version of the efficient-structure hypothesis, since these banks which locate on more efficient scale are also assumed to gain large market shares that may result in high concentration, the positive profit-structure relationship is spurious (Lambson, 1987).

At the end of this paper, we try to look beyond market structure and factors of macroeconomic cycle to consider risk as an additional explanatory variable in the profit-structure relationship. Our consideration is that the risk-taking behaviour of financial institutions has in recent years come to the forefront of the debate on the stability of the banking system. Recent interest in the risk issue has been resurrected by the seminal article Keeley (1990) who reasons that the surge of bank failures in the US during the 1980s had (at least partly) been caused by various deregulation measures and market factors that reduced monopoly rents. Edwards and Mishkin (1995) argue that the erosion of profits due to competition from financial markets can be held responsible for the excessive risk-taking observed in the 1980s in the USA. Chapelle et al. (2005) suggest that substantial savings can be achieved through active risk management techniques, although the effect of a reduction of the frequency or severity of operational losses depends on the calibration of the aggregate loss distributions.

The remainder of this paper is structured as follows. The next section outlines the functional form and measurement methodologies adopted in this study. Followed by the section that discusses the data sources. The penultimate section shows the estimation and results. In a final section, we summarize our findings and give suggestions for the future industrial organization of the German banking sector.

Specifications of models

Methodology: scale economies and the profit-structure relationship

Before we investigate the profit-structure relationship in the German banking industry, we need to estimate scale efficiency[5]. Most previous contributions to the banking literature have tested the profit-structure hypotheses alone by examining the price-concentration relationship without the benefit of efficiency. However, a potential drawback is present because the excluded efficiency variables may be correlated with both prices and market structure. For example, if an efficient bank has lower marginal cost (since this bank locates on the scale efficiency region of the average cost curve), this bank is usually bigger and has larger market share. In such cases, findings will incorrectly support the structural-conduct-performance hypothesis and the market-power hypothesis. To argue with this point, we modify a model specification from Berger (1995) that nests the three profit-structure hypotheses, including a direct measure of scale efficiency. We also consider risk as an explicit factor in this model.

The profit-structure relationship

The relationship between market structure and the profitability of banks is of concern to bank managers and to banking regulators. Particularly, the banking regulators have to weigh the potentially beneficial effects of mergers on the combined banks' profitability and viability against the possible detrimental impact on consumer welfare. For example, increased competition from financial deregulation in the banking sector may force banks to invest into higher yielding assets by increasing their risk exposure beyond a reasonable level (Brewer et al., 2003). Based on this consideration, we will pay particular attention to the delicate balance between profitability and risk. We incorporate aspects of banks' ex-post risk-taking behaviour into a framework developed by Berger (1995) to evaluate alternative theories of the profit-structure relationship. The modified model is described as follows: Equation 1 Equation 2 Equation 3 All three hypotheses, the structural-conduct-performance hypothesis, the market-power hypothesis and the scale-efficiency version of efficient-structure hypothesis, are represented by different variables. The major equation (1) is shown to be a valid reduced form for all of the hypotheses and any or all of them may be found to be consistent with the data. For instance, if the structural-conduct-performance hypothesis holds, the coefficient of concentration is significant and positive but the coefficient of market share is not in this case. This result indicates that the positive profit-concentration relationship occurs because concentration affects price and price affects profit. On the other hand, if the coefficient of market share is positive and significant, but the other coefficients are not in this case, the relative-market-power hypothesis holds. Under the relative-market-power hypothesis, market share becomes the key exogenous variable since banks with large market shares have well-differentiated products and are able to exercise market power in pricing these products.

By contrast, if the scale-efficiency version of the efficient-structure hypothesis is accepted, the coefficient of the scale efficiency variable will be positive and significant. An important limitation of the reduced-form profit equation (1) is that it tests only one of the three necessary conditions of the efficient-structure hypotheses. In order to explain the profit-structure relationship spuriously, two more conditions (equations (2) and (3)) should be met, since both profits and the market structure variables (concentration rate and market share) must be positively related to the variable of scale efficiency. For instance, equation (3) means that more efficient firms have greater market shares. This can be explained by the fact that more efficient banks obtain greater market share through price competition or through acquisition of less efficient banks. Finally, the risk factors in this specification model are evaluated by incorporating capital risk and portfolio risk.

Data description and sources

The data resources

The data resources were individual banks' balance sheets and income statements obtained from the Fitch IBCA database from 1998 to 2003. The data on branch numbers for German banks were gathered from Deutsche Bundesbank's (various years) “Verzeichnis der Kreditinstitute”. Our sample includes the 298 biggest German banks (by asset size), which represent at least 90 per cent of total loan market in Germany. Given the chosen intermediation approach, we use four categories of outputs, three kinds of input variables and one control variable in our models. All variables in this study are measured in million dollars. Data from income statements are gathered from 1 January to 31 December for each year. Data from balance sheets and the other official reports are obtained on 31 December for each year. All variables in this paper are defined in the following section.

Definitions of variables

Profitability (ROE)

In this study, we employ the pre-tax return on equity as a profitability indicator. The rate of return on equity is the most appropriate measure of profitability as it is more consistent with the notion that ownership will seek to maximise profits. We observe that average return on equity falls significantly in the German banking market.

Market share (MS) and concentration (CONC) variables

We measure the degree of concentration in the banking sector by using the size of bank loans, and rely upon the Herfindahl index (HERF) for our econometric analysis. The Herfindahl index[6] of market concentration is calculated as follows: Equation 4 where, TD _i bank i's total loans; TD all sample banks' total loans.

We can observe that concentration rate in Germany is much lower than the concentration rate of the US, which is around 20 per cent (Berger, 1995). Moreover, the extent of consolidation in Germany is lower than that in the global banking industry (Balino and Ubide, 2000; Belaish et al., 2001). However, the concentration rate in Germany has still slightly decreased over time.

Scale efficiency: S-EFF^e and S-EFF^d

Based on the previous discussion, we obtain scale efficiencies from the major translog cost function in the previous case of scale economies[7]. For each bank's output mix and input prices, a U-shaped multi-product average cost curve is traced out and the scale-efficient output vector Y ^se at the bottom of the U-curve can also be determined. In this case, we distinguish between scale economy efficiency for banks that are below efficient scale, and scale diseconomy efficiency for banks that are above efficient scale.

Thus, we include the scale economy efficiency (S-EFF^e) variable and the scale diseconomies efficiency (S-EFF^d) variable to replace the scale efficiency (S-EFF) variable, because they may have different implications under the scale version of the efficient-structure hypothesis. Advocates of the scale-efficiency version of the efficient-structure hypothesis argue that banks in the scale economy region grow larger and more profitable and at the same time increase their market share and their market's concentration rate rises, creating the spurious positive profit-structure relationship. In contrast, banks in the scale diseconomy region shrink to increase scale efficiency and profits. If dominant firms shrink, it would reduce concentration rate. These relationships can be explained as follows: Equation 5 Equation 6

Indicator of portfolio risk

Portfolio risk is defined as the standard error of the returns on assets. For example, portfolio risk for the kth period is obtained from the standard error of return of asset for k, k−1, and k−2 period.

Indicator of capital risk

Capital risk can be estimated as the ratio of net value (equity) to total assets. For example, a low ratio means high capital risk and high financial leverage. It should be borne in mind that when banks are profitable, financial leverage has the positive effect of increasing return on equity; when banks report losses, financial leverage increases the magnitude of loss in terms of a negative return on equity.

Finally, we summarize all the definitions and statistics of all variables in Table II and Appendix 2.

Estimation and results

We use a panel data set instead of single year data to investigate scale economies and the profit-structure relationship of the German banking industry. Although the positive serial correlation and heteroscedasticity will still exist, using panel data enables us to investigate the relationships between temporal changes and cross-sectional differences. We employ the seemingly unrelated regression estimation (SURE) technique, which is particularly useful with large panel data sets (Avery, 1977) to estimate several equations simultaneously. In this specific error components model, the regression errors in each equation are assumed to be composed of three independent components − one component associated with time, another with cross-sectional units, and a third with each observation: Equation 7 The model developed above makes the assumptions that both within and between equation error covariances are composed of independent individual, time period, and observation components, and the covariances of all three components are non-zero. Mahajan et al. (1996) and Hunter and Timme (1986) also use SURE to analyse the panel data for the translog cost function system of banks.

Results of the scale economies

Since, we will include the direct measure of scale economies in the specification model of the profit-structure relationship model, we summarise all empirical results from our translog cost function system here.

Translog cost function system

From Table III, we find that the coefficient of labour cost (ln P ₂) and branch number (B) are positively significant. Labor cost (ln P ₂) plays a more influential role in determining total cost. According to the coefficients of all outputs, we may infer that for German banks, producing one more unit of interbank assets (Q ₂) will cost much more than producing the other three outputs: total loans (Q ₁), equity investment (Q ₃) and other investments (Q ₄). Since, the coefficient of time (t) is significantly negative, this may imply that technology (e.g. computer, software of exchange system, information system and so on) has helped German banks to reduce their total costs over time.

Overall economies of scale

We obtain an average value of overall economies of scale for the German banking industry of 0.2545. This empirical result means that from a cost standpoint, all German banks can obtain the benefit from overall economies of scale by increasing their bank asset size. This conclusion is the same as the results from studies cited in the literature review of Molyneux et al. (1997), although the value is smaller[8]. However, this difference can be explained by the choice of a completely different data set, sample period, number of outputs and definitions of outputs and inputs[9] (Guala, 2002). For example, based on the choice of our sample set, there is a wide range of asset size within the 298 biggest German banks and until the 298th banks, already quite a lot of small asset size banks are included. This will also affect the shape of the translog cost function.

Furthermore, we make use of separate samples to provide us with a comprehensive treatment of the banking industry and determine whether the results are stable across environments. From Table IV, we can see the average value of overall economies of scale of public sector banks is 0.3735 and the value of private sector banks is 0.3276. The values of overall economies of scale are all significantly different from one and no big differences exist between these two groups[10].

Results of the profit-structure relationship

In this section, we investigate the three profit-structure hypotheses as competing explanations of the observed variation in bank profitability. Our specification model includes the equations (1), (2) and (3). Firstly, we show the empirical results without considering capital risk and portfolio risk.

In Table V, the coefficient of market share in the major equation (1) is negative and statistically significant at the 1 per cent critical level. This result indicates that the relative-market-power hypothesis is rejected as an explanation of the profit-structure relationship in the German banking market. The structural-conduct-performance hypothesis and the scale-efficiency version of efficient-structure hypothesis do not contribute to illuminating the profit-structure relationship either, since the coefficients of the other variables are insignificant and the adjusted R ² of the equation is very low. In contrast to previous banking studies, we add the factor of capital risk into this specification model. The detailed empirical results are not reported here[11], as they are very similar to the case without consideration of risk. The coefficient of market share is the only statistically significant coefficient and we obtain the same conclusion, namely that the relative-market-power hypothesis is rejected.

Finally, we add the factor of portfolio risk into our specification model. From Table VI, we note that the fit of the model improves significantly.

From the empirical results in Table VI, the market share coefficient in the major equation (1) is still negative and significant at the 1 per cent critical level which, again, suggests that the relative-market-power hypothesis is rejected. However, we have several new findings. First, the coefficient of concentration rate is positively related to return on equity and significant at the 1 per cent critical level in the major equation (1). This result means that there is a positive profit-structure relationship in the German banking industry. The implication is that German banks could achieve a higher profitability (return on equity) if the German banking market was more concentrated. We may draw the policy conclusion that German banking regulators should encourage banks to merge thus reducing the number of banks in the German banking market and raising profitability. The drawbacks of such an approach should not be neglected, however. Regulators also need to pay attention to protect consumers' benefit, because the acceptance of the structural-conduct-performance hypothesis indicates that adverse effects of higher concentration on consumer welfare are likely. The structural-conduct-performance hypothesis states that banks can set prices that are less favourable to consumers in more concentrated markets because of competitive imperfections.

The next important result is that the coefficient of portfolio risk is significant at 1 per cent critical level and has the “right” (positive) sign. Thus, we can say that German banks are able to achieve higher bank profits by increasing their portfolio risk. On the other hand, if a bank is very conservative (low portfolio risk) that bank's profit may be negatively affected. However, portfolio risk should be maintained within in a certain level to assure banks' safety. Nonetheless, achieving a higher yield on their assets will likely be key factor for German banks in increasing their profitability. It has often been argued, for example in a recent paper by Goldman Sachs economists Broadbent et al. (2004), that German financial institutions have not adequately priced loans to the corporate sector. The intention of the Basel II recommendations on risk-adjusted capital adequacy ratios will presumably force German banks to differentiate loan pricing on the basis of internal or external ratings and finally improve the return on loan portfolios. At the same time, it may become increasingly more difficult for German medium-sized companies without access to the capital markets to obtain bank financing.

Furthermore, we observe that the coefficient of market share is still negative and significant at the 1 per cent critical level in the major equation (1). This means that the relative-power hypothesis is still rejected. Moreover, the coefficient of capital risk is statistically significant at the 5 per cent critical level in the major equation (1), but the factor of capital risk only brings a slightly negative effect on the German banking profitability. Finally, the coefficient of scale economy efficiency is always statistically insignificant in the three empirical cases under consideration. It maybe inferred that the scale-efficiency version of efficient-structure hypothesis does not contribute to explain the profit-structure relationship in the German banking market.

It should be pointed out that after the factor of portfolio risk is added into this specification model, adjusted R ² is considerably improved from 4 to 32 per cent. We can conclude that German banking profitability is not only determined by the market structure but also the factor of portfolio risk. We summarize these empirical results on market profit-structure relationships in Table VII.

Conclusions

To answer the questions posed in our introduction, the empirical evidence gathered in this paper shows that market structure plays a significant role in determining German banks' profitability. Analysis on a panel of 298 German banks from 1998 to 2003 supports the structural-conduct-performance hypothesis. As a new finding compared to previous studies of the German banking sector, portfolio risk is shown to contribute positively to banks' profitability. Furthermore, we try to look beyond the cost function approach taken by most contributions in the literature by not only examining overall economies of scale, but also investigating the German banking profit-structure relationship. The model specification used here enables us to distinguish among the three profit-structure hypotheses discussed above. Combining our results, we can conclude that the recent wave of bank mergers observed in France, Sweden and Austria may also suitable for Germany. Owing to the existence of overall economies of scale and the acceptance of the structural-conduct-performance hypothesis, bank mergers should be encouraged by German banking regulators. Particularly, a higher concentration rate is likely to bring about a positive effect on German banking profitability. If banks can increase their asset size by merging, they can also obtain the benefit from overall economies of scale. However, because of the empirical support for the structural-conduct-performance hypothesis, German banking regulators also need to pay attention to protecting consumers' benefit as further concentration may give banks the ability to set less favourable price for customers.

Another important finding in this paper is that incorporating portfolio risk can significantly increase the adjusted R ² of our specification model of the profit-structure relationship. This empirical result indicates that the management of the risks in asset portfolios is a key factor in determining German bank profits. It can be inferred that the German banking sector should be able achieve higher profitability by increasing portfolio risk. Certainly, appropriate portfolio risk management systems still need to be in place. If the latter is not the case and competition becomes too intense, increased risk-taking by banks may even threaten the stability of a country's financial system.

Equation 1

Equation 2

Equation 3

Equation 4

Equation 5

Equation 6

Equation 7

Equation 8

Equation 9

Equation 10

Equation 11

Equation 12

Table I

Table IIDefinitions for all variables in the model of profit-structure relationship

Table IIIEmpirical results of the hybrid translog cost function system

Table IVEmpirical results of overall economies of scale for German banks from 1998 to 2003

Table VEmpirical results of the profit-structure relationship without considering the capital risk and the portfolio risk

Table VIEmpirical results of the profit-structure relationship with considering the capital risk and the portfolio risk

Table VIISummary of results of three profit-structure relationship hypotheses

Table AI

Table AII

Table AIII

Table AIV

References

Avery, R.B. (1977), "Error components and seemingly unrelated regressions", Econometrica, Vol. 45 pp.199-209.

[Manual request] [Infotrieve]

Balino, T., Ubide, A. (2000), "The new world of banking", Finance & Development, Vol. 37 No.2, .

[Manual request] [Infotrieve]

Belaish, A., Kodres, L., Levy, J., Ubide, A. (2001), "Euro-area banking at the crossroads", IMF Working Paper, WP/01/28, .

[Manual request] [Infotrieve]

Berger, A.N. (1995), "The profit-structure relationship in banking – tests of market-power and efficient-structure hypotheses", Journal of Money, Credit and Banking, Vol. 27 pp.404-31.

[Manual request] [Infotrieve]

Berndt, E.R., Hall, B.E., Hausman, J.A. (1974), "Estimation and inference in nonlinear structural models", Annals of Economic and Social Measurement, Vol. 3 pp.653-65.

[Manual request] [Infotrieve]

Brewer, E., Hunter, W.C., Jackson, W.E. (2003), "Deregulation and the relationship between bank CEO compensation and risk taking", FRB of Chicago Working Paper No. 2003-32, .

[Manual request] [Infotrieve]

Broadbent, B., Schumacher, D., Schels, S. (2004), "No gain without pain − Germany's adjustment to a higher cost of capital", Goldman Sachs, London, Goldman Sachs Global Economics Paper No. 103, .

[Manual request] [Infotrieve]

Chapelle, A., Crama, Y., Hubner, G., Peters, J-P. (2005), "Measuring and managing operational risk in the financial sector: an integrated framework", available at: http://ssrn.com/abstract = 675186 (accessed 27 February 2005), .

[Manual request] [Infotrieve]

Christensen, L.R., Jorgenson, D.E., Lau, L.J. (1973), "Transcendental logarithmic production frontiers", Review of Economics and Statistics, Vol. 55 pp.28-45.

[Manual request] [Infotrieve]

Decressin, J., Bruner, A., Klingen, C., Kudela, B., Hardy, D., Mühleisen, M. (2003), "Germany: selected issues", IMF Country Report No. 03/342, .

[Manual request] [Infotrieve]

Deutsche Bundesbank (various years), Bankstellenbericht, Deutsche Bundesbank, Frankfurt, .

[Manual request] [Infotrieve]

Edwards, F.R., Mishkin, F.S. (1995), "The decline of traditional banking: implications for financial stability and regulatory policy", Economic Policy Review, Federal Reserve Bank of New York, Vol. 1 pp.27-45.

[Manual request] [Infotrieve]

Guala, F.A. (2002), "Measuring economies of scale and scope for the Argentine bank industry: a comparative analysis, 1996-1999", UIUC Department of Economics Working Paper, .

[Manual request] [Infotrieve]

Hunter, W.C., Timme, S.G. (1986), "Technical change, organizational form and the structure of bank production", Journal of Money, Credit and Banking, Vol. 18 pp.152-66.

[Manual request] [Infotrieve]

Katib, M.N. (2005), "Market structure and performance in the Malaysian banking industry: a robust estimation", paper presented at 8th Capital Markets Conference, Indian Institute of Capital Markets Paper, .

[Manual request] [Infotrieve]

Keeley, M.C. (1990), "Deposit insurance, risk, and market power in banking", American Economic Review, Vol. 80 pp.1183-200.

[Manual request] [Infotrieve]

Lambson, V.E. (1987), "Is the concentration-profit correlation partly an artifact of lumpy technology?", American Economic Review, Vol. 77 pp.731-3.

[Manual request] [Infotrieve]

Lloyd-Williams, D.M., Molyneux, P., Thornton, J. (1994), "Market structure and performance in Spanish banking", Journal of Banking & Finance, Vol. 18 pp.433-43.

[Manual request] [Infotrieve]

Mahajan, A., Rangan, N., Zardkoohi, A. (1996), "Cost structures in multinational and domestic banking", Journal of Banking & Finance, Vol. 20 pp.283-306.

[Manual request] [Infotrieve]

Molyneux, P., Forbes, W. (1995), "Market structure and performance in European banking", Applied Economics, Vol. 27 pp.155-9.

[Manual request] [Infotrieve]

Molyneux, P., Thornton, J. (1992), "Determinants of European bank profitability: a note", Journal of Banking & Finance, Vol. 16 pp.1173-8.

[Manual request] [Infotrieve]

Molyneux, P., Altunbas, Y., Gardener, E. (1997), Efficiency in European Banking, Wiley, New York, NY, .

[Manual request] [Infotrieve]

Noulas, A.G., Miller, S.M., Ray, S.C. (1990), "Returns to scale and input substitution for large US banks", Journal of Money, Credit and Banking, Vol. 22 pp.94-108.

[Manual request] [Infotrieve]

OECD (2002), Insurance Statistics Yearbook, OECD, Paris, .

[Manual request] [Infotrieve]

Shepherd, W.G. (1982), "Economies of scale and monopoly profits", in Craven, J.V. (Eds),Industrial Organization, Antitrust, and Public Policy, Kluwer Nijhoff, Boston, MA, .

[Manual request] [Infotrieve]

Appendix 1. Scale economies

Banks' multi-outputs in this paper are measured by the intermediation approach since the nature of banks is more accurately described as intermediaries of financial services rather than producers of loan and deposit account services, a view taken by the production approach

[12]

. We assume that domestic banks in Germany operate in a competitive environment and all banks aim to minimise costs with profit-maximising behaviour. The translog cost function has the form below: Equation 8 where, ln TC,the natural logarithm of the total costs for interest costs, labour cost and capital cost; Q _i , a vector of outputs; Q ₁, total loans which include all class of loans; Q ₂, interbank assets; Q ₃, equity investments' Q ₄, other investments including liquidity investments and other investments; P _i , ith input prices; P ₁, interest rate=(interest paid/interest-baring total deposits); P ₂, labour expense=(overheads expense/total output); P ₃, capital price=(operating cost/fixed assets); ln B, the natural logarithm of the number of branches; α β γ δ λ ρ τ h, k, coefficients to be estimated.

According to Shephard's Lemma (Christensen et al., 1973), the derived demand for an input can be inferred by partially differentiating the cost function with respect to the input price, P _i . Thus, cost share equations can be generated from the translog cost function (8) as follows: Equation 9 Since, the duality theorem requires the cost function to be linearly homogeneous in input prices, the following restrictions have to be imposed on the parameters of the translog cost function (8): Equation 10 Further, the second order parameters of the translog cost function (8) must satisfy the symmetry condition: Equation 11 The translog cost function (8) is estimated jointly with the cost share equation (9) using the SURE technique. Since, the input cost share equations will sum to unity, one cost share equation should be omitted from the estimated system of equations to avoid the problem of a singular contemporary covariance matrix of disturbances (Berndt et al., 1974).

The concept of scale economies is based on the shape of the average cost curve. For instance, economies of scale are present up to the level where the long-run marginal cost curve lies below the long-run average cost curve. By following Molyneux et al. (1997) and Noulas et al. (1990), we estimate overall economies of scale for each bank by evaluating equation (12) to examine how changes in scale affect total cost: Equation 12 If OES<1, there are increasing returns to scale, i.e. economies of scale exist. If OES=1, constant returns to scale exist. If OES > 1, there are decreasing returns to scale. The existence of scale economies means that the average cost of producing a product, in the long run, decreases as more of the output is produced.

Appendix 2. Descriptive statistics of all variables from 1998 to 2003

[13]

Table AI, Table AII and Table AIII.

Appendix 3. Empirical results of the profit-structure relationship with considering the capital risk

Table AIV

Corresponding author

Peiyi Yu can be contacted at: P.Yu@wlv.ac.uk; peiyi_yu@yahoo.com