Early Warning Model for Financial Distress Using Artificial Neural Networks
Andry Alamsyah (a), Nuning Kristanti (b), Farida Titik Kristanti (c)

School of Economic and Business, Telkom University
Jl. Telekomunikasi 01, Terusan Buah Batu
Bandung 40257, Indonesia
a) andrya[at]telkomuniversity.ac.id
b) nuning.kristanti20[at]gmail.com
c) faridatk[at]telkomuniversity.ac.id

Abstract

Financial investment has become a trend in Indonesia with significant increase of active investors since 2015. Before making an investment, the investors need a comprehensive analysis to reduce the chances of failure that result in financial distress, the same apply towards companies in order to organize its financial strategies. Financial distress indicated by losing its value, ineffective production, cash flow problems or high financial leverage value. These conditions threaten the companies and the investors who face significant financial loss. The purpose of this research is to construct early warning model of financial distress, by examining the phenomenon of 90 companies in Indonesia, from 2015 to 2018 listed on Indonesia Stock Exchange (IDX). We apply Artificial Neural Network (ANN) backpropagation methodology using financial indicators such as profitability, liquidity, and solvability as the inputs. We divide the ANN model into time categories that is t-2, t-3, and t-4. In constructing ANN model, we configure four types of splitting training and testing data. The results show that ANN backpropagation model with 30 neurons, 90% training data and 10% testing data in category t-4 works well an accuracy 95.6% for financial distress prediction in Indonesia.

Keywords: artificial neural network, financial distress, early warning model

Topic: Information Engineering