Comparing the reliability of accounting-based and market-based prediction models

Abstract

Recently developed financial distress prediction models adopt a market-based approach. It gained its popularity in the academic world due to its theoretical appeal. However, the comparison of market-based with traditional accountingratio- based models is limited in the literature. Therefore, this paper humbly attempts to add finding to the literature by comparing the accounting-based model with market-based model in order to present a comprehensive computational comparison of methodologies to fulfil the strategic information needs of investors and other stakeholders. Our accountingbased model employed multivariate discriminant analysis (MDA) and logistic regression analysis (LRA) and for marketbased model, we adopted Merton technique. Our sample consists of one hundred and fifty eight public listed companies in Malaysia. Sixteen financial ratios with five-feature groups including activity ratio, cash flow ratio, solvency ratio, liquidity ratio and profitability ratio are selected as variables for our accounting-based model. For the market-based model, we generate the logarithm by adopting the information from the market such as stock price and interest rate. The result of one year prior to financial distress classification indicates that LRA has the highest accuracy compared to other methodologies and both the accounting-based models (LRA and MDA) outperformed market-based (Merton) model.