government interventions beyond their natural span of existence, the data for such
firms was collected as of two years prior to the time their net worth became negative.
For others, data was gathered for one year prior.
The authors found that the performance scores generated by the three models were
highly correlated and ranked the problematic firms similarly. Because the models ap-
peared to be equivalent, the authors chose just the probit model for presenting the re-
sults. It was found that the probabilities of failure increased for the problematic firms
from 0 in 1973–1974 to more than 0.5 in the mid-seventies, with complete deterio-
ration of performance of about two-thirds of the problematic firms in the sample by
1979.
While there is no doubt that the models anticipated the problematic firms quite
well, the results would be more compelling had the authors published the Type I ac-
curacy of the models. A model may have 100% Type I accuracy, but if it has 0 Type
II accuracy, then it is of no value.
10.13 ARGENTINA
(a) Swanson and Tybout (1988). This paper analyzes the determinants of industrial
bankruptcy on Argentina on three levels. First, the importance of macroeconomic
variables on the business failures is considered. Real interest rate, credit stock, man-
ufacturing output, real wage rate and the peso exchange rate are regressed on busi-
ness failures two variables at a time using a multivariate regression with third order
polynomial distributed lag terms. Second, sectoral failure rates are examined to de-
termine whether reform policies had a differential effect on highly protected indus-
tries. The data is divided into high protection and low protection industries and the
differential impact of economic policies is evaluated by adding the degree of protec-
tion as a dummy variable in a regression of the number of business failures against
the real interest rate and credit stock. The authors then consider the firm-level
anatomy of failure by creating a probit regression model on a sample of 19 to 22 fail-
ures and 190 to 324 survivors. Measures of financial structure consisting of cash flow
indices, firm financial structure variables, firm size, and the degree of protection were
utilized. The firm failure model was estimated for the pre- and post-maxi devaluation
period of the Argentinian peso, that is, 1979–1981 and the period following 1981, re-
spectively.
Following the military coup that ousted Isabel Peron in 1976, Argentina passed
through a reform period. The reform started with selective tariff reductions. Soon,
contractionary monetary policies and temporary wage and price controls were im-
posed to combat hyperinflation. In late 1978, an exchange rate regime was intro-
duced. The end result of all these policies led to a maxi devaluation of the peso that
threw the economy into a recession similar to the Mexican peso crisis precipitated by
the December 1994 devaluation. The authors examine the effects of the reform po-
lices with the hope that policymakers will evaluate future policy options in terms of
the stress they place on the corporate sector. Their results were obviously ignored or,
more than likely, unknown in the more recent Mexican crisis.
Using quarterly data on the macroeconomic variables (24 data points), 10 regres-
sions were estimated using a different combination of two macro variables. Although
the business failure rate, rather than the absolute number of business failures would
have been more appropriate as the dependent variable, the authors did not have the
data on the total number of businesses in each time period, and therefore they were
10.13 ARGENTINA 10 • 31