Banking activity has strongly changed in the last few years. The liberalization process has turned banks more competitive and has improved the role of intermediate products in the market. However aconcentration movement begins to surge and new strategies are being developed. In this paper cluster analysis and PCA methodology have been used in order to understand different strategies and different typologies in the banking system.
This paper tries to understand the main strategies followed by the banks between 1998 and 1997.
Banks have been distributed by groups, according to their dominant activity: banks with universal characteristics; banks specialized in credit for companies and individuals. Are strategies correlated with the institutional structures of the banks (public banks, Portuguese private banks or foreign banks)? Did banking strategies contribute to the formation of various classes of banks?
In thise research the authors have analysed the balance sheets and the income statement accounts provided by the Association of the Portuguese Banks “A.P.B.”. Data comprise the accounts of 52 associated banks. The “A.P.B.” banks totalise more than 95 % of the banking industry in Portugal. Taking in attention the Portuguese legislation for banks and considering the distinction between universal banks and specialized banks, the following typology have been used:
TABLE 1
[ 1 ] Universal Banks, subdivided in:
- Public banks
- Portuguese private Banks
- Foreign Banks
[ 2 ] Specialized Banks, subdivided in:
- Banks of investment more directed towards the companies
- Private banks more directed towards individuals.
In order to analyse strategies followed by the banks, a cluster analysis in association with a
factorial analysis (P.C.A.) has been done. Cluster analysis enables us to confront data information. By factorial analysis (P.C.A.) we reduce the number of variables included in the study.
1st stage: In this stage a principal component analysis is generated between all the ratios in order to simplify the research on reducing the number of variables.
2nd stage: Here a hierarchical classification is generated where all the possible iterations are included.
The P.C.A. is an exploratory analysis. It crosses individuals (the banks) with quantitative variables (the ratios). The objective is to highlight classes of variables (ratios), which represent the same reality, each one represented by a principal component. Each principal component represents a synthetic variable, which constitutes a summary of the whole initial variables. In order to obtain a succession of synthetic variables and a representation of the correlations between the variables it is used a criterion of extracting a maximum projected variance.
With P.C.A. analysis a way to reduce variables is generated . In cluster analysis a way to summarize all data into clusters is obtained. The interest of each cluster depends, mainly, of the characteristics that make possible to define the cluster. In a process of classification, one seeks a similarity between the data. Each cluster represents an artificial classification, insofar as the number of classes is preset. However,what is to be seen is that if this classification approaches to a natural classification, resulting from the strategies followed by the banks.A mixed classification is made where a maximum number of clusters is chosen and these clusters are validated by a hierarchical cluster analysis with the number of clusters chosen a priori. The general principle of “clusterising” is founded in the construction of a table of similarities between the data (The proximity matrix). As data are quantitative, the method of Ward is used which is based on the Euclidean distance of each data allowing the establishment of a partition between classes of variables.
Main Results:
The P.C.A. reveals five significant principal components, i.e. with a total variance higher than one.
TABLE 2
Each component expresses a linear combination between all the variables of this study.Taking in consideration the results with orthogonality between the axes, the rotation of the axes under the Varimax »method under a Kaiser’s standardization was made. The objective is to maximize the approach of factorial variables to the axes. The limitation of this method, it is that the axes are now correlated between them.
The most correlated variables with each component are the following ones:
1st component - [Resources ]: DEPORDAT; CAPITAT; DEPRZAT; CASHAT
2nd component - [Interbank ]: CREDIFAT; INTERBAT
3rd component - [ Titres with income fixes ]: RENDFAT; SECURAT
4th component - [Risk]: PROVISAT
5th component - [ Inheritance ]: IMOBILAT
The crédit/Assets ratio is no significantly correlated with any component, but it intervenes partially in the first and second components. Afterwards, taking in consideration the main variables chosen with PCA analysis, seven clusters a priori were fixed. The analysis only reveals six. The seventh cluster represents only one case.
TABLE 3
The number of banks in each cluster, except for cluster 4, maintains stable for the period. There is an increase in banks (from 3 to 11) from 1989 to 1990, in cluster 4. This cluster includes the foreign banks where new entries on the market in 1990 were recorded. The post-entry years in the E.U. are characterized by a high economic growth. The good performance of the Portuguese economy is observed through productivity increases in all activities. En consequence, foreign investment increased, which can justify the establishment of foreign banks. 8Cluster 4 also includes private banks which have begun their activity for only three years. After 1989, the share of public banks in the market has been more reduced. This period is also marked by the efforts of the installed banks to maintain their share of the market, and the efforts of some private banks to gain shares very quickly (it is the case of the B.C.P.). The first two clusters include the traditional banks, already installed in the market in a very long time. These banks are essentially public banks, except for one bank [M.G.] which is a medium size bank with a traditional behaviour. The importance of public banks is still quite visible, as much in the number of banks, as in terms of assets and income statement.
TABLE 4
Clusters 1 and 2 show the following characteristics: Intermediation is very important. The average value of the ratio ‘Term Deposits /Assets’ is higher than 50 %, whereas the ratio of other clusters is less than 31 %. the average value of Credit/Assets ratio is equal to 32 % in cluster 2, and than 58 % in cluster 1, whereas in other clusters (except for cluster 6) the ratio is lower than 25 %. The performance of banking is quite good in terms of financial margins, corresponding to 82% of the banking product in 1990, 84 % in 1989 and 82 % in 1988. Clusters 1 and 2, as well as cluster 6 obtain a very significant financial margin, accounting for 3,9% and 3,8% of the total credit. Differentiation between clusters 1 and 2 lies firstly in the importance of the credit compared to total assets (58,1 and 31,9% respectively). As cluster 1 includes the public bank [C.G.D.], which is the largest Portuguese bank, the values of this cluster are basically determined by the results of [C.G.D.]. The banks of cluster 2 are mainly public banks, in which traditional activities of intermediation prevail. Clusters 3,4,5 and 6 represent the banks of smaller size, except for the cluster 3 which integrates [B.C.P.]. The majority of foreign banks is locate in cluster 4. It is in private investment banks and in foreign banks that the solvency ratio is higher. The size of these banks (smaller size than the public banks) must explain the ratio:
TABLE 5
Cluster 6 differs significantly from clusters 3, 4 and 5. Cluster 6 is characterized by the importance given to the credit, which accounts for 58% of total assets. However, this cluster is also marked by the “Provisions” whose variable is associated with risk. Provisions account for 2,3% of total credit in this cluster. Cluster 6 is also differentiated by the variable “Equity” and by the variable “Equipment”. Cluster 6 characterizes the investment’ banks. Clusters 4 and 5 include the majority of the foreign banks. Their activity is not very important in Portugal compared to the other banks, except for [Barclays] and [Lloyds ]. Cluster 4 and cluster 5 present a very large dispersion in the ratios [Term Deposits/Assets ] (23,1 % in cluster 4 and 30,6 % in cluster 5). Cluster 4 and 5 registers also a very large dispersion in the Credit/Assets ratio with 19,2 % and 35,3% respectively.
TABLE 12
Conclusion:
Banks have changed its strategies along the observed period. Decreasing importance of the Public Banks increased competition. As consequence of the globalisation of the financial markets banks have to adapt to a new environment. Movements inside clusters reveal a changing of strategies observed by the banks. The formation of banking groups which has initiated in the third period of our analysis has strongly contributed to the formation of new clusters.
No comments:
Post a Comment