The Riskiness of Risk Models: Assessment of Bankruptcy Risk of Non-Financial Sector of Pakistan

Abstract

Bankruptcy prediction has long been an important concern for various stakeholders in an increasingly intricated business environment. Using a sample of 3,806 company-year observations of listed non-financial companies of Pakistan during 2005-2015, the paper compares models and identifies an optimal approach in terms of forecasting accuracy for predicting financial distress and bankruptcy. The purpose is to develop a model with relatively high predictability and figure out determinants of bankruptcy. By employing financial ratios, equity market variables and macroeconomic indicators; the hybrid artificial neural network (ANN) validates superior performance as opposed to dynamic panel probit and Merton-KMV models individually. Among financial ratios; quick ratio, cash ratio, current to total asset, quick to total asset, cash flow to short-term debt, gross profit margin, asset turnover, interest to debt, net working capital to net sales, and cash to net sales are crucial in examining firm’s financial status. Additionally, money supply, forex reserves, exchange rate, balance of trade, and real GDP growth rate are found statistically meaningful in predicting bankruptcy.