Bankruptcy Prediction — An Investigation of Cash Flow Based Models [1]

Ratios included in bankruptcy prediction models are based on a type of ad hoc pragmatism rather than a sound theoretical work. Altman, Haldeman and Narayanan (1977) used financial ratios and measures found helpful in providing statistical evidence of impending failures in other studies plus new measures thought to be potentially helpful. Dambolina and Khoury (1980) have used similar criteria for ratio selection but added that data availability for the calculation of ratios across firms and across years was an additional consideration. This also applies to such other studies as Beaver (1967), Deakin (1972), Edminster (1972) and Altman (1968). Pinches et al. (1975), and Chen and Shimerda (1981) list and analyse the multitude of ratios used by various researchers. Blum (1974) points out that in the absence of a theory of symptoms one cannot use statistical analysis of financial ratios and expect a sustained correlation between independent variables and the event to be predicted. Ball and Foster (1982) also criticize this brute empiricism approach using stepwise discriminant/regression analysis as a result of which the selected variables tend to be sample-data specific and the empirical findings do not permit generalization. Wilcox (1973), Vinso (1979) and Emanuel, Harrison, and Taylor (1975) have tried to arrive at a single measure — the probability of bankruptcy based on theoretical approaches to avoid ‘brute empiricism’. Casey and Bartzack (1984, 1985) have investigated the use of operating cash flows as a possible predictor of bankruptcy. More recently Gentry, Newbold and Whitford (1985) have used a cash-based fund flow model developed in 1972 by Helfert (1982) to obtain a set of financial ratios and measures for generating a multivariate bankruptcy model. Our study is similar to that of Gentry et al. (1985). However, instead of using Helfert's fund flow model we have used a cash flow identity developed by Lawson (1971).