Assessing bankruptcy prediction models via information content of technical inefficiency

We use a stochastic frontier model with firm‐specific technical inefficiency effects in a panel framework (Battese and Coelli, 1995) to assess two popular probability of bankruptcy (PB) measures based on Merton model (1974) and discrete‐time hazard model (Shumway, 2001). Three important results based on our empirical studies are obtained. First, a firm with a higher PB generally has less technical efficiency. Second, for an ex‐post bankrupt firm, its PB tends to increase and its technical efficiency of production tends to decrease, as the time to its bankruptcy draws near. Finally, the information content about firm’s technical inefficiency provided by PB based on DHM is significantly more than that based on Merton model. By the last result and the fact that economic‐based efficiency measures are reasonable indicators of the long‐term health and prospects of firms (Baek and Pagán, 2002), we conclude that PB based on DHM is a better credit risk proxy of firms.