18.12.2012 Views

Proceedings

Proceedings

Proceedings

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

The main advantage of DA is obtaining classification functions, which will allow the<br />

subsequent inclusion of the companies that are not part of the analyzed sample, in<br />

predictive purposes. The coefficients associated with the indexes of the evaluation<br />

methods are presented in Table 4, and the classification functions obtained have the<br />

form: ScoreInsolvent = -0.899 -0.075IEV + 0.180IMC +0.093IVFCF and ScoreSolvent = -<br />

0.005IEV + 0.266IMC -0.017IVFCF. By replacing the two variable functions with the<br />

data presented by a company that will have to be classified into one of the two<br />

categories, two score values will be obtained (a value for ScoreInsolvent and a value for<br />

ScoreSolvent). These values will be compared, and the highest value will also determine<br />

the company’s belonging to one of the two categories.<br />

In what concerns the LRA method, the main results obtained in SPSS concern the<br />

functions for determining the probability of occurrence of the insolvency risk for a<br />

comoany, after the application of each evaluation method. For an analyzed company,<br />

the three probabilities of occurrence of the insolvency risk will be compared,<br />

according to the used method, with the purpose of their classification. Moreover,<br />

based on the probability computing functions, LRA allows a series of comparisons in<br />

what concerns the correct classification of the analyzed companies into the insolvent/<br />

solvent groups. This way, the method used to obtain a classification function that has<br />

the fewest wrongly classified cases will be the method that best indicates the absence<br />

of the insolvency risk and implicitly a faithful image. The obtained results are<br />

synthesized in Table 5, and for the interpretation of the function coefficients, their<br />

exponential value (exp) will be used.<br />

Variables<br />

Table 5. Results obtained through LRA<br />

Exponential coefficients of the<br />

probability<br />

EV MC VFCF<br />

~ 196 ~<br />

Correctly forecast cases<br />

Solvent Insolvent<br />

IEV 0.183 - - 75% 55%<br />

I MC - 0.862 - 40% 80%<br />

I VFCF - - 1.243 60% 50%<br />

Constant 0.557 1.067 0.975 - -<br />

(*Sig < 0.5)<br />

Starting from the results obtained in SPSS, based on the coefficients associated to the<br />

variation indexes of the evaluation methods, it is possible to obtain the functions for<br />

determining the probabilities of occurrence of the insolvency risk. Therefore, for each<br />

method, a probabilistic model of the insolvency risk will be obtained, having the<br />

form:<br />

EV: [pinsolvent/(1-pinsolvent)] = 0.557·(0.183)^IEV;<br />

MC: [pinsolvent/(1-pinsolvent)] = 1.067·(0.862)^IMC;<br />

VFCF: [pinsolvent/(1-pinsolvent)] = 0.975·(1.243)^IVFCF;<br />

Therefore, for IEV = 0 (the company has not recorded any modifications of EV), the<br />

probability for a company to be bankrupt and not solvent is 0.557 (the ratio between<br />

the probabilities for the two states), and an increase of the index by one unit (IEV = 1,<br />

double EV) will amplify this risk by 0.183, eventually generating an insolvency risk<br />

of 0.102 (0.557·0.183). The EV method successfully indicates 75% of the solvent<br />

companies (error 25%) and 55% of the insolvent companies (error 45%), resulting in

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!