Financial Distress Prediction for Small and Medium Enterprises Using Machine Learning Techniques

Abstract

Financial distress prediction is a key challenge every financing provider faces when determining borrower creditworthiness. Inherent opaqueness of Small and Medium Enterprise business complicates credit decision making process, therefore increasing cost to finance and lowering probability of receiving funds. This paper used data on 12.000 SMEs to estimate binomial classifiers for financial distress prediction using Logistic Regression, Artificial Neural Networks and Random Forest techniques. Classical financial ratios were used to estimate initial single-period predictors, which were later enhanced with time, credit history and age factors to retrieve multi-period models. Contrary to other studies, financial distress is understood as a significant challenge to company’s ability to cover liabilities rather than probability to go bankrupt. Highest prediction accuracy was reached using Random Forest algorithm with additional factors. It was concluded that period-at-risk adjustment is necessary to ensure highest financial distress prediction accuracy.

An Optimal Model of Financial Distress Prediction: A Comparative Study between Neural Networks and Logistic Regression

“…In general, our results show the superior performances of logistic regression over neural networks. These findings are in line with the works of Du Jardin and Séverin (2012), Islek and Oguducu ( 2017), Kim et al (2018), Lukason and Andresson (2019), and Malakauskas and Lakštutien ė ( 2021). For example, logistic regression reached for Du Jardin and Séverin (2012) an accuracy of 81.6% against 81.3% for neural networks with data collected over one year.…”

“…For neural networks, our best results outperform those of Kim et al (2018), Lukason and Andresson (2019), Papana and Spyridou (2020), and Malakauskas and Lakštutien ė (2021). For instance, using neural networks with 42 nodes in the hidden layer, Kim et al (2018) found an accuracy of 71.9% through 41 financial ratios selected from 1548 Korean heavy industry companies.…”

Predictions of failure and financial distress: A study on Portuguese high and medium-high technology small and mid-sized enterprises

“…In terms of prediction methods, as summarized by Yazdanfar and Nilsson (2008) as well as Jackson & Wood (2013), the development of bankruptcy research can be classified into three stages: in the first stage, multiple discriminant analysis worked as the main statistical tool in late 1960s and 1970s after the research of Beaver (1966) and Altman (1968), which attempts to "classify the statistical units (objects) into two or more pre-defined groups" (Omelka et al, 2013(Omelka et al, , p. 2588; in the second stage, logit and probit models as the representative of conditional probability models have been developed by Ohlson (1980) and Zmijewski (1984) since 1980s ; in the third stage, with the development of computer technology, artificial neural networks as the representative method of data mining and machine learning techniques have started to be used in bankruptcy research since 1990s, which assume non-linearity of financial distress function (Malakauskas & Lakstutiene, 2021;Mselmi et al, 2017).…”

Artificial Intelligence Based Optimal Functional Link Neural Network for Financial Data Science

“…The main involvement of this research is a significant development of the predictive accuracy in the short term (twelve months) by ML methods, related with the advanced when creating precise mid and long term predictions. Malakauskas et al [18] utilized data on 12.000 SMEs for estimating binomial classifications for financial distress prediction by LR, ANN, and RF methods. Traditional financial ratios have been utilized for estimating the early single period predictor that was improved using age factor, time, and credit history for retrieving multi period modules.…”