Fundamental banking changes

2. Austrian Banking overview

Austria's fundamental banking changes began in 1994 with the Bankwesengesetz (BWG) act[1]. (Web 1) This helped remove barriers between different banking sectors to try and improve the profitability of the sector and meet legislation to enter the EU. They entered the single European market in 1995 and with this came deregulation which allowed foreign banks to enter, and more competitive pressure.

The Austrian banking system has seen a lot of structural change in terms of mergers and acquisitions over the past decade to try and improve performance. Austria was thought to be over banked and over staffed; making them one the least profitable banking institutes worldwide (Hahn 2004). Today it has some of the top operating banks in Europe, however many of it banks operate regionally or locally and not many operated at a national or even international level. The sector has strived to be more efficient by reducing the number of banks and employees, as this was thought to be one reason why profits levels were so low.

Austria had well over 1200 credit institutions and almost 4500 branches, when the population of the country is only 8 million. It also had one of the lowest banks to population ratios within the OECD area until the early 1990s. (Hahn 2007) Austria tried to reduce its banking inefficiencies by reduce the number institutes within the country through mergers. It reduced the number of banks from 1210 in 1990 to 883 in 2004, and further down to 834 in 2006. Despite this achievement the number of bank braches did not reduce at the same rate, remaining rather high. Austria has too many small banks, many of the mergers happened at a local regional level but there were national mergers. There have been many domestic in house mergers in Austria but few acquisitions. Hahn (2007) identifies a staggering 118 domestic mergers but only eight domestic acquisitions over the period of 1996-2002.

Bank concentration in Austria is quite low compared to other European countries. As measured by the total assets of the top five banks in 2004 they had only 46.9% market share. This was because many of the mergers that had been happening within the country were local, regional banking. (Hahn 2007). One of the biggest mergers that has happened in the country was when Bank Austria merged with Creditanstalt in 1997 to create Austria's biggest bank; Bank Austria Creditanstalt (BA-CA). Also around this time In 1997 Erste Bank acquired Griocredit ensuring its place as the third largest bank. The acquisitions and mergers of Erste bank and Bank Austria Creditanstalt have been the larger transactions In Austria.

Cross boarder activities have risen significantly over the past eight years in Austria as many national banks tried to get a foothold is eastern Europe. Austrian banks, conscious of the limitations of their mature home market, have led the way in expanding operations into their neighbours, central and eastern Europe (CEE). The three biggest institutions - Bank Austria Creditanstalt (BA-CA), Erste Bank and Raiffeisen Zentralbank (RZB) - have larger operations in CEE than in Austria (Web 2).

We can see how entering the eastern European market has benefited Erste Bank who is the Austrian leader in foreign acquisitions, since they went public in 1997. In 2000 they began cross boarder acquisitions in eastern European countries, such as Czech Republic and Hungary. The expansion of the European Union has helped with its operations abroad and remains the third biggest bank in Austria, with around a 20% share of the market. As Erste Bank's General Manager says "Since 1997 we have far surpassed our original goals. Our home market now covers a contiguous area of more than 115 million people, and we have successfully integrated more than ten acquisitions in Central and Eastern Europe. First was Mezoebank in Hungary in 1997." (Eduard Oswald 2007) It is one of the few banks that has made a serious push into central and eastern Europe, with its eastern European subsidiaries contributing to 59.2% of net profit in 2005 (web 3).

In 2000, three years after it was formed via a merger Bank Austria-Creditanstalt AG (BA-CA) was acquired by Germany's second largest bank Hypovereinsbank. This was a major cross boarder acquisition that resulted in the creation of Europe's third largest bank in terms of assets (Web 4). Since December 2006, BA-CA has been a member of Italian group UniCredit. The Austrian bank is responsible for all the eastern European activities of its owner UniCredit apart from Poland, and it is already expanding in Russian commercial banking through International Moscow Bank.

Raiffeisen International is not only one of the top three Austrian banks but is one of the leading performers in Europe. The growth potential of the Czech Republic, Hungary and Slovakia, has been a powerful draw for investors who have bought into their acquisition lead growth. They saw net earnings in 2006 climb to 594m from 227m in 2003. Also their takeover trail has stretched as far Ukraine and Belarus (Simonian 2008).

Innovations in eastern European markets were key to the recent growth of the Austrian banking sector allowing its top banks to expand rapidly across the region. Despite the distances and unfamiliar locations, the Austrians' results have seen few nasty surprises. Profits have soared as basic banking skills have been imported into undeveloped markets and more sophisticated products gradually introduced. However the top three Austrian banks are just starting to feel the pressures of the global credit squeeze, and operating in uncertain markets. All three have seen falls in share prices, Erste having reached a low of 34 Euros almost half its peak. It is now time for the Banks to make sure they produce efficient operations in the foreign empires they have built.

3. Methodology

The literature on productivity and efficiency analysis is divided into two main areas: the parametric Stochastic Frontier Analysis (SFA) and non-parametric Data Envelopment Analysis (DEA). Both approaches require the specification of a production or a cost function or frontier. However unlike the parametric approach, DEA does not require the specification of a particular functional form for the cost or production function. Therefore the efficiency estimates are not functional form dependent. Another important difference between parametric stochastic frontier approach (SFA) and DEA is that SFA allows for the presence of a random error term. This means that DEA attributes any deviation from the efficient frontier as being associated with inefficiency. Furthermore, DEA may overstate the true values of relative inefficiency (Berger and Mester, 1997; Grosskopf 1996).

Stochastic Frontier Analysis

Stochastic frontier models date back to Aigner, Lovell and Schmidt (1977) and Meeusen and Van den Broek (1977), who independently proposed a stochastic frontier production, function with a two-part 'composed' error term. This error is composed of a standard random error term, representing measurement error and other random factors, and a one-sided random variable representing what Farrell (1957) called 'technical inefficiency', i.e. the distance of the observation from the efficient production frontier. This idea of technical efficiency represents the firm ability to maximise from a given set of inputs. It is measured by out relative to what it could attain if it were 100% efficient (i.e. on the frontier itself). If this is combined with allocative efficiency, which is the ability of the firm to use its inputs in optimal proportions, given there prices, this would provide a measure of economic[2] efficiency.

In this paper will adopt 3 inputs and 3 outputs. Input 1: total deposits. Input 2: personnel expenses and other operating expenses. Input 3: loan loss provisions. Output 1: Total Customer loans Output 2: total other earning assets output 3 commission income and other operating income.

Stochastic input distance frontier

The idea of distance functions was first introduced by Malmquist[3] (1953) and Shepherd (1953). Using a distance input function means that it is possible to describe a multi-input, multi-output production technology function, without having to incorporate the pricing objectives of a firm for instance profit maximisation or cost minimisation. The input distance function is the greatest radical contraction of the input vector, with the output vector held fixed. So that the input vector stays in the input requirement set V(y).

3.1. Data

The data was obtained from the global database; Bankscope, which contains information on 28,000 public and private banks around the world. We collected data for the 'largest' 25 Austrian banks, in terms of total assets, over the three year sample period between 2004 and 2006. Hence, the sample consists of a total of 75 observations.

We employed both the intermediation and production approach to model the production process of the banking firm with the former adopting 3 outputs and 3 inputs and the latter 5 outputs and 3 inputs.

For the production approach five outputs are defined: 'total customer loans, 'Total other earning assets', 'Commission income', 'Other operating income' and importantly 'total deposits. The model assumes these outputs are produced using the inputs 'total non-interest expenses', 'total other operating expenses' and 'total provisions'.

For the Intermediation approach the three outputs produced were ' Total customer loans, 'Total other earning assets' and 'Commission income plus Other Operating income' The inputs for this approach were 'total deposits', 'total operating expenses' and 'total provisions'.

The limitations of our study were almost solely concerning the collection of data. Initially, when collating Austria's largest 25 banks, there was unavailable data for certain variables we required, and therefore that bank was removed and we would work with the next largest, etc. Through adopting this process, we needed to work with many more banks than the 'largest' 25 Austrian banks, making our sample less representative.

The other limitation of our study was in the collection of our inputs and outputs in that not all of the data we desired to specify our inputs and outputs was available. For example, for the 'loans' output we were not able to collect the desired 'total customer loans' plus 'total other lending', and therefore had to adopt the figure solely for 'total customer loans'. The issue of missing data was a continuous limitation throughout the collection process.

As a check of the consistency of the efficiency estimates produced by the DEA and SIDF models it's intuitive to plot the results in a scatter chart.

It's necessary to check the consistency of the DEA results with the SIDF results as DEA itself is inherently a non-theoretical model that merely seeks to find correlations between inputs and outputs. SIDF, on the other hand, is a much more sophisticated model that is specifically based on production theory and as a consequence can detect any irregularities or problems with the data inputs into the model. The SIDF model for example can detect for data problems such as multicollinearity that would easily pass undetected if the results were solely estimated using DEA, leaving you with biased estimated results.

Figure 1 below depicts the fact that there is indeed a weak positive relationship between the non-parametric and parametric results estimated for the model using the intermediation approach. This indicates that there is partial consistency between the estimated DEA and SIDF results which is reassuring as it shows that the estimated DEA results that shall be analysed in more depth later in the analysis are at least partially consistent with the more sophisticated SIDF results. However it's important to recognise that there will still be significant inconsistencies between the efficiency estimates produced by the SIDF and DEA models at the level of the individual bank.

On the other hand, Figure two below depicts virtually no correlation between the efficiency estimates produced by the DEA and SIDF models when the production approach towards defining the banks inputs and outputs was utilised. This result raises significant doubt about the credibility of the estimated DEA efficiency results for the model using the production approach as they don't appear at all consistent with the theoretically sound SIDF results.

4.2. Initial Comparison of the Estimates for the Intermediation and Production Approaches

A brief inspection of the mean yearly technical efficiency scores estimated by DEA indicates that the mean technical efficiency estimates for both approaches to specifying inputs and outputs are roughly the same with results of 86.8% and 88.8% respectively.

However, when looking at the efficiency estimates for individual banks the results in table one highlight that in some cases there is a significant divergence in the efficiency estimates produced when the different approaches are used. For example, Porsche Bank was estimated as 100% technically efficient when the Intermediation approach was employed where as it was only found to be 76.4% efficient under the production approach.

Table one shows that the variations in the SIDF efficiency estimates for the two different approaches are much less marked than shown by the range in the DEA estimates at the level of the individual banks. Using the example of Porsche bank again, the SIDF estimate of the technical efficiency under the intermediation approach was 98.6% where as it was only found to be 93.1% efficient when estimated under the production approach. Importantly, the results although significantly different in scale are consistent between the intermediation and production approach when estimated by both the DEA and SIDF method, as they both indicate that Porsche bank is considerably more technically efficient when it's efficiency is estimated on the basis of the inputs and outputs included in the intermediation approach rather than the production approach.

It's clear that the few significant inconsistencies created when different input/output specifications are used for the efficiency estimates highlights a key limitation when trying to formulate optimal strategies for banks to employ to optimise their efficiency. The fact that there is no one single 'correct' approach to specifying bank inputs and outputs for estimation means that it's important to take into account the results of different approaches to ensure that you can make a well-informed decision on a banks efficiency.

Focussing again on the example of Porsche Bank, both the DEA and SIDF estimates produced under the intermediation approach suggest that the bank is virtually 100% efficient and represents a clear benchmark for an efficient bank on the estimated frontier. From these results it could therefore be confidently concluded that the bank should seek to continue producing at it's exact same size and maintain it's managerial techniques, however, if you instead consider the estimated results when the production approach is employed it could easily be concluded that the bank would benefit significantly from a merger or some sort of take-over activity to target reducing the overall inefficiency of the bank.

However, as the estimated DEA and SIDF efficiency results for both the production and intermediation approach are shown to be very similar by the results in table one, we will primarily favour analysing the results for the intermediation approach in my subsequent analyses as the estimated results from the DEA and SIDF models when the intermediation approach was employed proved to be better correlated than those produced when the production approach was undertaken.

4.2. Evaluating the Inefficient and Benchmark efficient banks from the Sample

A brief analysis of the efficiency estimates in table one highlights a staggering link between the most inefficient banks estimated by the DEA and SIDF approaches and the banks themselves. The most inefficient banks highlighted in our sample are all part of the Raiffeisen co-operative banking group. "Raiffeisen" is a reference to Friedrich Wilhelm Raiffeisen, the founder of the co-operative movement of credit unions. Raiffeisen Zentralbank is the central institution of the cooperative banking group and is owned by eight regional banks which are in turn owned by a vast number, approximately 550, local Raiffensenbanks.

Table 2 below details the extent to which the Raiffeisen Co-operative banking institutions are less efficient than the mean of the other Austrian banking Institutions in the sample. Focusing on the results calculated under the Intermediation approach in particular, the banks under the Raiffeisen heading were estimated to be 64.16% efficient by the DEA method and 94.22% efficient by SIDF relative to 90.25% and 97.29% for the rest of the Austrian banks in the sample.

Table 2: Comparison of mean yearly technical efficiency estimates of the Raiffeisen co-operative banking group relative to the other Austrian banks in the sample.

This key result highlights a highly inefficient sector within the Austrian banking market relative to the rest of the institutions. A likely explanation as to why members of the Raiffeisen bank are so inefficient relative to other banks in the sample is the fact that individual members of the group such as Raiffeisenlandesbank N-W completely saturate a particular region with their own bank branches. For example, Raiffeisenlandesbank LB-NW that represents 82 co-operative banks in the region for lower Austria and Vienna,

exploits its shear presence in the region to provide incomparable proximity to its customer base as a strategy to attract custom and remain highly profitable despite the inefficiencies caused by being 'over-branched' relative to the efficient banks in the market.

The benchmark efficient banks are clearly identifiable from table one. The DEA results for the Intermediation and production approaches find 5 and 6 banks respectively from the sample technically efficient (DEA Score=100), and these banks will define the respective frontiers for bank efficiency in each model. The SIDF results are broadly consistent of this estimate as the SIDF scores for the same 'efficient banks range from 99.5% to 95.2% for the intermediation approach and 94.1% to 92.1% for the production approach.

Clear 'benchmark' efficient banks in the sample are Volksbank Vorarlberg and Sparkasse Schwaz who are estimated to be 100% technically efficient by DEA for both approaches and found to be 97.2% and 98.8% efficient according to SIDF under the intermediation approach and 93.4% and 94.1% efficient according to SIDF under the production approach.

4.2 Evaluating the DEA Mean efficiency Indexes for both approaches

In this setion of the paper we seek to analyse the mean efficiency statistics for the three efficiency indexes obtained from the DEA estimation. The results reported in table 2 show that the 25 largest Austrian Banks can be shown to have been 79% overall efficient over our three year sample period according to both the intermediation and production approach. In other words, the results of the study indicate that Austrian banks could have reduced cost by 21% from what they actually incurred had they all been operating at overall efficiency.

The figures in Table 3 also depict the components that the overall efficiency score is composed of, i.e. the mean technical and scale efficiency of the Austrian banks included in the sample. It is evident from these figures that overall inefficiency is a consequence of both scale and technical inefficiencies in the Austrian banking market, with technical/managerial inefficiencies appearing to be slightly more culpable for the 21% mean overall inefficiency score calculated in the case of the Intermediation approach.

Focussing in particular on the scale efficiency estimates, table 4A and 4B report the total number of banks that fall into each category of scale economies over our sample period for each of the approaches. The estimated DEA results for the intermediation approach found that in 15 of the 75 observations, banks were found to be scale efficient. Of the remaining 49 observations, only 2 were found to be experiencing increasing returns to scale and would therefore reduce input costs by expanding their scale of production closer towards the MES.

The striking result was that in 58 of the total 75 observations estimated, Austrian banks were found to be producing on the upward sloping section of the average cost curve, and would therefore effectively reduce their average costs by downsizing their scale of production. This summary result of the intermediation approach highlights an overall phenomenon that the Austrian banking market is heavily endowed with 'over-branched' and over-sized banking institutions that are considerably larger than they need to be to minimise bank cost.

This empirical finding is broadly supported by the literature reviews of the Austrian banking sector, for example, Hahn highlighted that 'the reduction in the number of Austrian banks has gone down from 883 in 2004 to 834 by 2006, but despite this achievement the number of bank braches did not reduce at the same rate and remained rather high' (2007,Hahn).

This concept of a highly 'over-branched' Austrian banking industry is clearly well-documented in the contemporary banking literature and is a phenomenon that Austrian banks appear to employ to maximise their own profits. Higher numbers of branches provides greater accessibility for customers and therefore increases the amount of business they will receive from the highly saturated Austrian banking market. This, as touched on before in my analysis, has particularly been characteristic of the Raiffesien banking group

However, when looking into the issue of changes in the scale economies of the banks over our sample period, table 4A reports the results for the intermediation approach and highlights that the scale economies of the banks in our sample remained virtually constant between 2004 and 2005, until structural changes resulted in the number of banks experiencing decreasing returns to scale increased significantly between 2005 and 2006. Average scale efficiency declined from 91.6% to 88.8% over this period as a consequence of all but one of the top 25 largest Austrian banks being estimated to be experiencing decreasing returns to scale by 2006.

The estimated scale efficiency results when the production approach was used largely correspond to those estimated with the intermediation approach for 2004 and 2005 with the vast majority of banks included in our sample being estimated to be experiencing decreasing returns to scale and therefore would experience cost savings by effectively downsizing their scale of production.

The controversial aspect of these estimated results, as shown by Table 4B, is that from 2005 toin which the results showed a ssignificant improvemebts in the scale efficiency of ncistent with the sophisticated SIDF results 2006 the DEA results indicate that there is a significant improvement in scale efficiency with the majority of the banks downsizing to their optimal scale, which is the exact converse of the estimated results for the intermediation approach. We have struggled to find much support for this finding in the contemporary literature and therefore favour the findings when the intermediation approach was employed. However, it's important to respect the significant differences in the findings of the two approaches and caution must be attached to conclusions drawn about the optimal strategy for Austrian banks to undertake.

4.3. Grouping the banks into different asset Categories

When referring to table 7 that lists the asset-sorted banks included in our sample it shows the substantial difference in the size of the banks despite only including the 25 'largest' Austrian banks in our sample. As a consequence of this size imbalance within the sample we felt it would be intuitive to subdivide the sample into different asset categories to enable us to identify more precisely the efficiency needs of different sized Austrian banks.

Table 5 below reports the overall, technical and scale efficiency of banks included in the various sized asset categories below. Focussing first of all on the intriguing differences in the overall efficiencies for both the intermediation and production approaches experienced by banks included in the different asset categories, the results show that 'small banks' have had an average, overall efficiency of 86.1% and 85.1% respectively over the sample period whereas the largest banks have had a mean overall efficiency of only 64.8% and 76.7% over the same period.

This key finding showing that the largest Austrian banks are substantially less efficient overall than the smaller banking institutions can almost entirely be attributed to the greater scale inefficiencies experienced by the largest Austrian banks. Table 5 shows that scale efficiency declines for each successively larger asset group in both the case of the Intermediation and the Production approach. This key result highlights a fundamental policy initiative for the largest Austrian banks as they should seek to down-size their total capacity in order to improve their scale efficiency and make substantial input cost savings.

Further analysis of the results of this investigation, highlights that the significant decline in the overall efficiency of the 'small-to-mid-sized' banks appears to be mainly a result of a decline in technical efficiency from that experienced by the smaller banks in the sample. The implication of this result is that for the 'small-to-mid-sized' banks to improve their efficiency they should primarily aim to improve their managerial/x-efficiencies, perhaps through seeking to emulate the more efficient managerial techniques used by the slightly larger 'mid-to-large-sized' banks who experience much greater managerial efficiency.

4.4. Evaluating changes in Austrian Bank Efficiency over our sample period

To further our analysis we estimated the changes in mean efficiency scores for the banks over the period of study. Table 6 contains the results of this investigation for both the intermediation and production approaches and shows that there is a significant divergence in the results of the two approaches.

The results for the intermediation approach show that overall efficiency has decreased persistently over the period from 81.5% in 2004 through to 73.6% in 2006. The results indicate that the majority of the decline in overall efficiency occurred between 2005 and 2006 when overall efficiency decreased from 80.2% to 73.6%, with this decrease in efficiency being roughly equally attributable to both declining technical and scale efficiency amongst Austrian Banks. Interestingly, the results of this investigation also reveal that the initial, more marginal deterioration in overall efficiency experienced between 2004 and 2005 was solely down to an increase in X-inefficiencies as the scale efficiency estimate remained constant over this period

Interestingly the results of the production approach reflect the converse of the above analysis with the results showing overall efficiency increasing overtime from 73% to 88% over the sample period. The controversial aspect of the results of the production approach is that it argues that the increases in overall efficiency have been mainly attributable to significant improvements in the scale efficiency of Austrian banks in the sample. This finding links with the analysis surrounding table 4B in which the results showed a significant improvement in the scale efficiency between 2005 and 2006 with 16 of the 25 banks in the sample being estimated to be operating at their minimum efficient scale by 2006.

Once again this analysis has highlighted apparent inconsistencies in the results when the two different approaches are undertaken, which emphasizes the need to respect the results of both models and caution to be attached to conclusions drawn regarding the optimal strategy for Austrian banks to undertake.

5. Summary and Conclusion

In this paper, overall efficiency, technical efficiencies and scale efficiencies are calculated for a sample of the twenty five largest Austrian Banks between 2004 and 2006. The linear programming results for both the Intermediation and the Production approach are highly consistent in agreeing that the average Austrian bank could reduce cost by 21% had they all been operating at full efficiency. The investigation of the source of this overall inefficiency revealed that this cost inefficiency is caused almost equally my technical and scale inefficiencies in the Austrian banking market according to the results of both approaches.

Further analysis revealed that a particular source of inefficiency in the banking market are the banks belonging to the Raiffeisen co-operative banking group which was consistent with much of the contemporary literature. Significant cost savings are achievable for these banks in particular if they reformed their strategic approach from attempting to completely saturate an area with their bank branches in an attempt to dominate the regional market through proximity and high customer accessibility, which results in the problem of 'over branching' and significant scale and technical inefficiencies.

This paper also provided an interesting insight into the strong correlation between bank size and overall inefficiency in the Austrian banking market. Both the results of the Intermediation and production approaches showed that overall efficiency declined with successive incremental increases in the total asset size of banks, with declining scale efficiency being the obvious cause in both cases.

In attempting to determine a general trend in the changes of the efficiency estimates over the sample period we have struggled to reach a definite conclusion as the results of the intermediation and the production approach were completely contradictory. The former argues that overall efficiency has declined significantly over the period as a consequence of both decreasing scale and technical efficiencies whilst the production approach suggests the converse as a result of substantial improvements in scale efficiency over the period.

The results of the Intermediation approach suggest that banks have become increasingly over-sized over the period and have been a significant cause of scale inefficiency whereas the production approach again argues the converse that Austrian banks have reformed and become increasingly scale efficient over the period.

As a consequence of this conflicting analysis it is hard to conclude whether significant reform is needed in the Austrian banking market to decrease the average size of Austrian banks back to their minimum efficient scale. For this reason it's essential that caution is taken when drawing conclusions about the optimal strategy for banks to undertake, but intuitively we favour the results of the Intermediation approach as they were shown to be more consistent with the relatively sophisticated SIDF model that is grounded in production and cost theory.

  1. On January 1, 1994, the Austrian Banking Act (Bankwesengesetz, BWG) went into effect and replaced the previous banking act (Kreditwesengesetz, KWG 1979) in the version last amended in 1986. The Austrian Banking Act formed the core of the financial market reform laws of 1993, which introduced a far-reaching overhaul of regulationsto theAustrian financial market. This act contains the most significant legal provisions for banking in Austria. In drafting the Austrian Banking Act, one of the legislature's goals was to ensure that Austrian banking legislation complied with EU legislation. The other main objective of the act was to ensure the smooth functioning of credit institutions as well as creditor protection and consumer protection. In particular, the Austrian Banking Act includes a number of provisions intended to guarantee the security of deposits entrusted to Austrian institutions. (Austrian national bank-
  2. Farrell used the term price efficiency instead of allocative efficiency and the term overall efficiency instead of economic efficiency. The terminology we use in this paper is used most often in recent literature.
  3. A Malmquist index is a bilateral index that can be used to compare the production technology of two economies. It is named after a Swedish statistician, Sten Malmquist, who used the distance function in 1953 to define input quantity indexes

Please be aware that the free essay that you were just reading was not written by us. This essay, and all of the others available to view on the website, were provided to us by students in exchange for services that we offer. This relationship helps our students to get an even better deal while also contributing to the biggest free essay resource in the UK!