International Finance and Accounting Handbook

• Net worth was calculated in constant terms as the difference between assets and
  liabilities.


• Sales were deflated using the wholesale price index for the industry.

The variables used in the model along with the means and univariate F statistics
are presented in Exhibit 10.20.
The resulting discriminant function using the F value as the criterion to enter con-
tained the following three variables:

1.Sales/debt

2.Net Earnings/total assets

3.Long-term debt/total debt

The classification accuracy of the model in the original sample was 98% for Type
I and 85% for Type II. In the Lachenbruch holdout test, the corresponding values
were 98% and 83% respectively. The Lachenbruch test (sometimes called the “jack-
knife” test) is used to eliminate the sample bias, by estimating the model with one
observation held out and then classifying that observation. This process is repeated
as many times as there are cases which virtually eliminates any potential bias. The
author performed holdout tests by validating the model with random sub-samples.

10.22 URUGUAY 10 • 45

Variable FP Mean NP Mean F

Asset turnover 1.11932 1.64829 16.39
7
Current ratio 1.02636 2.29415 39.59
4
Changes in working capital 0.03091 0.46927 4.514
Sales/nonbank working capital 2.94295 4.78073 10.43
3
Leverage 1.33432 3.03975 54.26
0
Inventory/bank debt 0.98568 4.58146 21.54
8
Bank debt/total debt 1.68295 2.84097 8.735
Long-term debt/total debt 0.07455 0.12659 2.912
(Accounts receivable +
inventories/accounts payable

• spontaneous sources) 3.85841 3.06780 2.070
  Inventory turnover 3.90432 7.68439 16.65
  6
  Rate of return on assets –0.25068 0.23341 6.414
  Sales/debts 1.53454 4.67829 68.24
  3
  Net earnings/total assets –0.08705 0.10756 27.05
  7

F1.60(0.05) = 4.00, F1.120(0.05) = 3.92
F1.60(0.01) = 7.08, F1.120(0.01) = 6.85

Exhibit 10.20. Means of the Variables and Significance Tests.