
Chapter 2
The aim of this section is to analyze and compare researches, described in the previous chapter. Firstly, basic peculiarities of some researches will be rewieved. Next, comparative analysis of methods will be examined.
There is a procedure, which enables to determine the amount of clusters among all banks - the method of self-organizing maps (SOM), which was proposed by Kohonen [2001]. This method was used in the research by Kosheluk [2008]. The rest of researches initially classify banks without determination of the amount of clusters.
There are researchers that determine groups of banks and after that classify banks within groups. Klepach et al. [2005] divided banks into 3 groups: small, medium-sized and large bank and distinguished 5 strategies within these groups: retail banks, controlled by the state, corporate, internally-oriented, currency-oriented, all-sufficient banks. Akindinova [2003] proposed classification not only on 3 groups according to the amount of assets, but also division of each group into 3 levels (echelons). Ivanter [2005] divided banks, oriented to clients and lending banks into several specific groups.
Taking into account the variety of classifications, there is a technique for comparison classifications by the amount of assets. It is a comparison of shares of assets, which are held by each group. According to Klepach et al. [2005], 109 large banks accumulate 83,4% of assets, 168 medium-sized banks – 8,7% and 945 small banks – 7,9%. According to The Bank of Russia [2012] in 2012 the first 5 banks controlled 50,6% of assets, next 15 banks – 20%, next 30 banks – 11%, 51-200 banks of the list concentrate 12,6 % of assets, 201-500 banks – 4,5% and the last group accumulates 1,3%. It means that the last 2 groups, which include about 800 Russian banks, concentrate an insignificant part of all assets. Another classification has similar indicators of the share of assets and the number of banks, therefore it’s reasonable to compare only these two types of classifications. These findings make it possible to conclude that classification, which examines 3 groups of banks: small, medium-sized and large banks, is sufficient for analysis the clusters of banks.
A comparison of the rest methods of classification is more complicated, because it requires matching quantitative and qualitative characteristics, absolutely different methods and instruments. The analysis comes to consideration of advantages and disadvantages of each method of classification.
Advantages of the first type of classification by statistical indicators are simplicity, univocal interpretation, are reckoning with significant indicators of a bank’s performance. However, using only one criterion has some disadvantages. This way does not allow considering peculiarities of banks, the real strategy, therefore developing an appropriate analysis.
The second method - analytical classification not only enables to describe the bank in the context of the published statistical indicators, but also analyzes an actual way to raise capital and allocate resources. The shortcoming of this classification is unilateralism. Groups that are classified in such a way are formed according to one specific aspect (business groups, the focus on the industry), so are not appropriate for complex analysis of a banking system.
The next classification based on factor analysis allows taking into account and reflecting all significant financial and operational factors examiners assess in their evaluation of a bank’s performance, using a comprehensive list of instruments (mathematical, statistical, analytical foundation). The problem of the factor analysis is subjectivity of identification and interpretation of the main factors. Moreover, when the number of indicators are reduced and replaced with the characteristics of generalized parameters, individual properties of objects can be distorted or lost.
Alternative methods of classification contain integral understanding and scope of bank’s operations. The classification by Ivanter [2005] and Matovnikov [2013] selects sources of funds, directions of their spending and types of clients. However, the omission of some models is difficulties in calculation of any factors, for example, the quality of management. Moreover, using these models, researchers can come up with incorrect indicators. It leads to the wrong interpretation of the results. The classification by Bobyshev [2001] makes it difficult to compare the quality of identified clusters; also different initial variables can result in different final clusters.