Failure Prediction of SME using Logistic Models for Matched Case-Control Studies

The paper illustrates the use of logistic models to measure the default probability of Small and Medium Enterprises (SMEs). As the default event is rare the information on defaulting units is scarce and adjustments are necessary in order to improve the ability of the model to detect firms that are likely to go bankrupted. We here explore the efficacy of logistic model for matched case-control design. Based on balance sheet indicators of SME in an Italian region, several different model selection procedures are presented . All models selected are interpretable from the economic point of view and exhibit a good capacity to discriminate between healthy and unhealthy firms.