How to Manage Financial Risk Under the Corporate Governance Constraint: Does Leverage Have

CREDIT-SCORING MODELS AND THEIR MULTIDIMENSIONAL PRACTICAL APPROACH.

CASE STUDY ON THE ROMANIAN LISTED COMPANIES ON THE BUCHAREST STOCK EXCHANGE

Cristina Maria TRIANDAFIL, Petre BREZEANU

Academy of Economic Studies, Bucharest,Romania

Abstract:This paper focuses on the two-side practical approach of credit-scoring models: corporate defaultprediction and corporate economical performance hierarchization. Taking as database the most 15 representative listed companies on the Bucharest Stock Exchange within the equipment field, the case-study implies the elaboration using Linear Discriminant Analysis of a specific credit scoring model, especially adapted to the afore-mentioned sample of firms, and the valorization of this credit-scoring model as a corporate default predictor and also as an economical performance hierarchization method.Thus the quantitative approach will be completed with the qualitative one since individual financial analysis, elaborated to the level of the Balance Sheet and Profit and Loss Account will be required in order to give an accurate interpretation of the output.

Keywords:default valuation, default predictors, credit-scoring model, cluster analysis

1. INTRODUCTION

The complexity of the corporate default as a financial phenomenon has been acknowledged by the whole literature. During 80’s, financial analysts considered that firms with low liquidity and solvability were likely to default (Franks and Torous, 1980). Meanwhile there were enough cases when firms with a very good financial overview were not able to avoid default.Credit-scoring models have represented a continuous preoccupation for the financial analysts because of their financial utility. They are considered to be a method of identifying, quantifying and managing the risk of corporate default. Their multidimensional approach subscribes to the idea of reflecting the financial health of a company, but also of ensuring a classification of the companies in accordance with an assembly of financial indicators integrated synthetically in a score function. Their multidimensional approach becomes operational especially in the case of the credit institutions which focus the whole credit activity on their internal credit-scoring models. From the standard credit-scoring models, based on MDA, it has been accomplished the transition to the credit-scoring models based on logit regression where the dependent variable is a binary one.

But most of the applications valorized by the international rating agencies have as fundamental the Alman score function. RiskCalc Model, considered the most commercial application, have as operational set-up an assembley of financila indicators which reflect liquidity, profitability, solvency and also the activity of the firm.Crosbie in his study ,,Modelling default risk’’ has analyzed the opportunity for the credit-scoring models to be differentiated in accordance with the filed of activity of the company in order to deliver a higher accuracy of the corporate default prediction in comparison with the credit scoring models implemented by Altman.

The test which have been performed have rejected the initial hypothesis.The Altman score function, although ithas been contested by numerous analysts because of the simple cut-off approach, keeps its predictive accurate capacity because of the applicability to the level of a sample of enterprises and, especially, to the degree of particularization. This paper is organized as follows: the first section is dedicated to the database description. The second section will depict the methodology and the third section contains the proper case-study.The fourth section is dedicated to the conclusions.

Section 1

The database includes financial data extracted from regarding 9 companies listed on the Bucharest Stock Exchange, activating in the equipment field. The matrix of financial data will be the following:

Tabel no.1 – The analysis of the financial indicators for the 9 enterprises

The financial indicators which are valorized are: I1 – Total Debt/Total Assets, I2– Total Debt/Equity, I3 –General Liquidity, I4-Return on Assets, I5-Return on Equity, I6-Return on Investment, I7- Net profit variation during the last 12 months, I8 – Growth rate of the assets, I9- log(Activ).The companies which are analyzed are: A- Aerostar Bacau, B - Altur Slatina, C - Armatura Cluj Napoca, D- Compa S.A. Sibiu, E- Electroputere Craiova, F- Impact Bucuresti, G - Mecanica SA Ceahlau, H - Mefin Sinaia, I - Rulmentul Brasov.

Section 2

The methodology which will be used in order to build the credit-score function will imply in a first stage the identification of the most representative financial indicators within the database of 9 financial indicators using the method of PrincipalComponents Analysis. Then there will be identified the principal axis which will define the new space of the causality dependence structure between the analyzed variables in accordance with which there will be determined the coefficients corresponding to the weight of every financial indicator within the score function. This will represent the point of reference for the hierarchization.

Section 3

Analyzing the Eigenvalues of the correlation matrix, expressing the initial causality space by a single principal component, we explain only 49,53 of the initial variance. Extending the analysis to 2 principal components, we will ensure the assessment of 70,24% of the total variance while 3 principal components will assess 82,26% of it.

Table no. 2– The Eigenvalues of the correlation matrix

The analyzed variables will be projected on the new axis and this will represent the coefficients of the score function:

Table no.4 –The results obtained by the appliance of the new score function

In order to get a deeper insight concerning the content of the Principal Components, there will be determined the correlation coefficients between the 3 Principal Components and the financial indicators.

The first Principal Component is highly positively correlated (coefficient of correlation of 0,91) with the variation of the Net Profit during the last 12 months, but also with the return on equity and with the return on assets (coefficients of correlation of 0,87 and 0,83). As for the negative correlation, the first Principal Component is negatively correlated with the liquidity indicator (coefficient of correlation of -0,86), which, indeed, has lower correlation correlation coefficients with the other Principal Components. Taking into consideration te high correlation coefficients, we are able to appreciate that the first Principal Component is a synthetical indicator of the return on assets and of the return on equity.

Table no.5 –The correlation coefficients between the Principal Components and the financial indicators

The second Principal Component is correlated in a positive manner with the leverage (coefficient of correlation of 0,79) and in a negative manner with the indebtness degree relative to the Total Assets (coefficient of correlation of -0,75).

A slight positive correlation is also established between the level of total assets and the second Principal Component (coefficient of correlation of 0,59).

The third Principal Component is negatively correlated with the growing rate of the Total Assets (coefficient of correlation of -0,68). In order to get a more clear insight of the correlations established between the Principal Components and the financial indicators, we will proceed with a rotation of the axis which focus on getting lower correlation coefficients for certain Principal Components.

Tabel no.6 –Coefficients of correlation between the Principal Components and the Finacial Indicators

After processing the factor rotation, we discover that the first Principal Component is positively correlated with the 6 variable (coefficient of correlation 0,9), so it can be interpreted as a synthetic indicator of the economical efficiency. The second Principal Component is negatively correlated with the leverage, so it can be interpreted as a synthetic indicator of the solvency (coefficient of correlation 0,85).

The third Principal Component is positively correlated with the variation of the Net Profit during the last 12 months and the growth rate of the assets (coefficient of correlation 0,80 and 0,82), which subcribes to the idea that this Principal Component is a synthetic indicator of the activity ratios.

Valorizing the interpretations of these Principal Components as synthetic indicators of various financial ratios, we can accomplish a classification of the analyzed enterprises.As the first Principal Component is a synthetical indicator of the economical return and the third Principal Component of the assets dynamics, it is obvious that the enterprises which get a a high score in accordance with these 2 components, will be appreciated as being very performant.As for the third component, the interpretation is totally opposite: as this Principal Component is a synthetic indicator of the indebtness degree, we can appreciate that the enterprises which get a low score according to this component will be assessed as performant.

Thus, the following classifications are obtained:

First Class: It includes the enterprises Compa Sibiu, Mecanica SA Ceahlau which have recorded high scores on all the three components;

Second Class: It includes the enterpriseRulmentul Brasov which has good scores on the first two components and a high degree of leverage, which is in accordance with the economical reality since the enterprise has high debts;

Third Class: It includes the enterprisesAerostar Bacau, Altur Slatina, Armatura Cluj-Napoca which had good scores on the first and on the third component;

Fourth Class: It includes the enterprise Electroputere Craiova, Impact Bucuresti, Mefin Sinaia, which had good scores only on the first component

This classification is confirmed by the below chart:

Graph no 1- Classification of the firms

Section 4

Credit-scoringul proved to be an useful tool for getting a financial diagnosis for the firms listed on Bucharest Stock Exchange, equipment section and also for getting a classification. The financial diagnosis for Rulmentul Brasov is a very interesting one since the enterprise has gotten a good score from the perspective of the economical return and of the assets dynamics, but a bad diagnosis from the perspective of the indebtness degree, which subcribes to the idea that the enterprise does not beneficiated from a viable strategy from the point of vue of economical growth. Its decreasing occurred in the context of the incapacity of financial management to valorize the economical resources growth in order to increase the own financial resources.The excessive leverage and the hazard moral as a consequence of the write-offs have acted as a support for the bad financial management.

For the enterprises which are classified in the 3 and 4 groups, it is possible to express some recommendations regarding the financial policies. Thus, it is necessary to diminish the indebtness degree and to direct the economical growth towards the accumulation of some own financial resources since leverage has negative effects on long time.

REFERENCES

Bouroche J. M., Saporta G., L’Analyse des données, P.U.F., 1980.

Chandon J. L., Pinson S., Analyse typologique, Masson, Paris, 1981.

Crosbie, P., (1999), Modelling Default Risk, KMV

Diday E., Pouget J., Lemaire J., Testu F., Éléments d’analyse des données, Dunod, Paris, 1982.

Falkenstein, E., A. Boral and L. Carty, 2000, RiskCalc Private Model: Moody’s Default Model for Private Firms, Moody’s KMV, San Francisc

Stein, R.M., Jordao, F., 2003. What is a more powerful model worth? Moody’s KMV, New

Fernandez G., Data Mining Using SAS Applications, Charman & Hall, 2003.

Härdle W., Simar L., Applied Multivariate Statistical Analysis, TECH, 2002.

Nakache Jean-Pierre, Analyse discriminante sur variables qualitatives, Polytechnica, Paris, 1994.

Quintart Aimable, Les fondements de la théorie financière classique et les limites de l’hypothèse générale de linéarité. În: Eric de Bodt, Henrion E.F. (edit.),

Saporta G., Probabilités, analyse des données et statistique, Editions Technip, Paris, 1990.

Saporta G., Ştefănescu V., Analiza datelor & Informatică, Editura Economică, Bucureşti, 1996.

Spircu, L Analiza Datelor.Aplicatii Economice, Editura Economica, 2003

Stein, R.M., Jordao, F., 2003. What is a more powerful model worth? Moody’s KMV, New

Triandafil, Cristina, Brezeanu Petre, Multidimensionalitatea operationala a modelelor de credit-scoring cu aplicabilitate practica pe firmele cotate la BVB, sectiunea echipamente, Revista ,,Management si Marketing’’, editia 3, 2007