Predicting bankruptcies in Swedish manufacturing firms A comparison between traditional statistical and machine learning methods

Abstract

The purposeofthispaperistoexaminewhetherwecanpredictbankruptciesinSwedish industrial andmanufacturingSMEsbeforetheyoccur.Toconductthepredictionsweuse a traditionalstatisticalmodel,multiplediscriminantanalysis(MDA)andamoremodern machinelearningapproachinwhichweutilizesupportvectormachines(SVM).Wethen compare thetwomodelsinordertofindoutwhichofthemperformsthebest.Further,we examine howmanyyearspriortofailureweareabletopredictthebankruptcy,andthus test bothmodelsuptofiveyearspriortofailure.Ourresultssuggestthatweareableto predict bankruptcieswithsignificantlyhigheraccuracythana50/50guessingstrategy. Further,wefoundwewereabletopredictbankruptciesupthefiveyearspriortothem occurring,usingboththetraditionalMDAmodelandtheSVMmodel.Whencomparing the twowefoundtheMDAmodelachievehigherpredictionaccuracyoneyearpriorto failure.