The author found that eight factors existed that account for 79.3% of the informa-
tion contained in the initial set of ratios. Just two factors provide for 42.1% of the in-
formation. The eight factors are:
1.Capacity to repay debt
2.Liquidity
3.Fixed assets financing
4.Efficiency of the firm
5.Rotation of fixed assets
6.Profitability of permanent funds
7.Structure of working capital
8.Structure of short-term debtFourteen ratios with a higher loading from the principal components were selected
as input for the discriminant analysis procedure. A six variable discriminant function
emerged as the best, with an overall classification accuracy of 84% in the original
sample. The discriminant function is as follows:
where
The results of the model on the development sample and the hold out sample are
given in Exhibit 10.4. As expected, there is a slight drop in performance of the model
in the hold out sample. Of greater concern is where the drop in performance is: nor-
mally the Type I accuracy will be maintained and the Type II accuracy will be lower.
In this case, the Type I accuracy has dropped from 84 to 70%. Some follow-up analy-
V17Cash-flow>salesV12Earnings before taxes>salesV9 Return on investmentV6 Cash-flow>Current liabilitiesV4 Quick ratio>Industry valueV3 1 Permanent funds>Net fixed assets2>Industry value0.514119*Vl20.43665*Vl7Z10.26830V30.54666*V40.55483*V60.62925*V910.9 SPAIN 10 • 21Predicted
Group MembershipActual Group No. of Cases 1 2
Group 1 25 21 4
84.0% 16.0%
Group 2 25 4 21
16.0% 84.0%
Exhibit 10.4. Classification Results.