Bankruptcy prediction of small and medium enterprises using a flexible binary generalized extreme value model

Abstract

We introduce a binary regression accounting-based model for bankruptcy prediction of Small and Medium Enterprises (SMEs). The main advantage of the model lies in its predictive performance in identifying defaulted SMEs. Another advantage, which is especially relevant for banks, is that the relationship between the accounting character-istics of SMEs and response is not assumed a priori (e.g., linear, quadratic or cubic) and can be determined from the data. The proposed approach uses the quantile function of the generalized extreme value distribution as link function as well as smooth functions of accounting characteristics to flexibly model covariate effects. Therefore, the usual assumptions in scoring models of symmetric link function and linear or pre-specified covariate-response relationships are relaxed. Out-of-sample and out-of-time validation on Italian data shows that our proposal outperforms the commonly used (logistic) scoring model for different default horizons.

“Birds of a Feather” Fail Together: Exploring the Nature of Dependency in SME Defaults

“…In Refs. and extreme values models (generalized extreme value (GEV) and generalized extreme value additive (BGEVA)) to predict default of a sample of Italian SMEs was extended, and in Ref. the performance of U.K. and Italian small firms was compared.…”

The effectiveness of TARP-CPP on the US banking industry: A new copula-based approach

“…Calabrese and Osmetti (2015) propose the BGEVA model by including an additive component to the GEV model. The BGEVA model has been then improved by Calabrese, Marra, and Osmetti (2016) . Several studies ( Andreeva et al, 2016;Calabrese et al, 2016;Calabrese & Osmetti, 2013;2015 ) have shown that the GEV model outperforms the logistic and the probit models even if the percentage of ones in the sample is one per cent.…”

Topological applications of multilayer perceptrons and support vector machines in financial decision support systems

“…As a result, an accurate FDSS risk assessment that distinguishes nondefault (healthy) entities from the default (bankrupt) ones can fruitfully assist to avoid financial scandal and crisis (Calabrese, Marra, & Silvia, 2016). Financial crisis mostly stemmed from the credit default of individual client together with the corporate bankruptcy.…”

“…If vigilant attention is paid to the relevant studies (see,e.g., Section 2), however, one can observe the lack of topological applications of MLPs and SVMs, that is, in most cases when using MLP algorithms, nothing is stated about activation functions (AFs) on one side, being used only one HL on the other side; lots of previous evidence have consistently demonstrated that escalating the number of HLs and experimentations with AFs make the MLP network more commanding and generate the more feasible outcome (Jeong, Min, & Kim, 2012;Luo et al, 2017). Furthermore, their databases for experiential investigations were to some extent imperfect, that is, research findings regarding FDSSs with "purely bankruptcy databases" are much more scarce, even though the size of corporate lending makes sound lending decisions, which is a matter of national interest (Calabrese et al, 2016). Xu, Tang, & Li, 2016).…”

“…Earlier empirical investigations and appraisals of diverse learning approaches have been followed out by Baesens et al (2003), Charitou, Dionysiou, Lambertides, andTrigeorgis (2013), Bauer and Agarwal (2014), Lessmann et al (2015), Rajan, Seru, and Vig (2015), Calabrese et al (2016), Oliveira et al (2017), Bequé and Lessmann (2017), and Volkov, Benoit, and Poel (2017). Earlier empirical investigations and appraisals of diverse learning approaches have been followed out by Baesens et al (2003), Charitou, Dionysiou, Lambertides, andTrigeorgis (2013), Bauer and Agarwal (2014), Lessmann et al (2015), Rajan, Seru, and Vig (2015), Calabrese et al (2016), Oliveira et al (2017), Bequé and Lessmann (2017), and Volkov, Benoit, and Poel (2017).…”