The application of a neural network approach to predicting bankruptcy risks facing the major US air carriers: 1979–1996

Airline bankruptcy, an unheard of event prior to the deregulation of the US airline industry, has become rather commonplace. Over 123 air carriers have filed receivership since 1982, and several large carriers have sought court protection more than once in the past decade. In spite of record airline profits over the past two years, the financial condition of many carriers still remains fragile. The huge financing requirements of the industry over the next decade, driven by the carriers’ need to replace aging fleets of aircraft, will create further stress for many. The ability to assess the level of this financial stress is important to many groups, including stockholders, bondholders, other creditors, financial analysts, governmental regulatory bodies, and the general public. For this reason, models that can forecast financial distress are useful. Building on prior research by several of the authors, who utilized multiple discriminant models driven by financial ratios, a neural network approach is employed to increase the reliability of the forecasts. In this paper, a neural net is trained with the result that it successfully classifies 26 out of 26 carriers in the holdout (test) set.