A Comparative Study of Australian and Korean Accounting Data in Business Failure Prediction Models

This paper studies the comparability of Australian and Korean accounting data in their predictive power in business failure prediction models. To this end, the predictive ability of 52 individual financial ratios is assessed for the sample from each country, along with three different significance tests. In addition, four alternative multivariate models are constructed. The classification results from each multivariate model are then compared to assess the robustness of our findings. The main findings are: (i) in most cases, Australian financial ratios have more information content than their Korean counterparts; (ii) ROE (return on equity), NI/TTA (Net Income/Total Tangible Assets), CF/INT (Cash Flow/Interests) are the most efficient and stable univariate discriminators for the Australian case, and EBIT/INT (Earnings before Interest and Taxes/Interests), RE/TTA (Retained Earnings/Total Tangible Assets) for the Korean case; (iii) compared with the results from the univariate approach, the multivariate approach made only a marginal improvement of prediction accuracy in the Australian case, while the prediction accuracy improved significantly in the Korean multivariate models.