On the Ability of Accounting Ratios to Predict Failure of Philippine Business Firms

Previous studies indicated that linear multiple discriminant analysis (LMDA) of financial ratios can be used to predict company failure. A discriminant function is derived from a set of financial ratios which, when transformed into Z-scores, characterize firms as failing or nonfailing based on a predetermined cut-off point. This study sought to develop a prediction model for Philippine firms using LMDA on financial ratios. Twenty-six failing firms in the manufacturing sector were chosen from among the 1,000 top corporations of 1979. For lack of data on bankruptcy, failure was defined as default in loan repayment between 1973 and 1981. A corresponding number of nonfailing firms were chosen from the same industry as the failing counterpart. Fourteen industries were represented. Some 260 financial statements were examined and over 8,000 ratios were computed for the 52 firms over a 5-year period. The ratios were subjected to various statistical tests to determine which of them should be used in deriving the discriminant function. Among the thirteen LMDA models developed, the “best” predictor of company failure consisted of four variables. These variables were Cash Flow/Total Liabilities, Net Income/Total Liabilities, Total Liabilities/Total Assets and Sales/Total Assets. It had the highest percentage of correct prediction at Year 1 (83%) and the narrowest range of overlap in Z-scores over the five-year period tainted (.26 to — .23).