Are Indonesian construction companies financially distressed? A prediction using artificial neural networks

  • Received January 25, 2023;
    Accepted March 30, 2023;
    Published April 6, 2023
  • Author(s)
  • DOI
    http://dx.doi.org/10.21511/imfi.20(2).2023.04
  • Article Info
    Volume 20 2023, Issue #2, pp. 41-52
  • TO CITE АНОТАЦІЯ
  • Cited by
    3 articles
  • 844 Views
  • 226 Downloads

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License

Construction companies are very dependent on the projects carried out by a company. Therefore, measuring whether a company is distressed or non-distressed can be done by looking at the ratios derived from the components of the financial statements from both the balance sheet and the company’s profit and loss. This study offers a new method for measuring financial distress in companies with Artificial Neural Networks (ANN). The model provided comes from several financial ratios in 17 construction companies listed on the Indonesia Stock Exchange. The model is expected to produce the best model by showing the lowest prediction error rate. The results showed that the best ANN model has 25 inputs, 20 hidden layer neurons, and 1 best model output. The model obtained will be tested directly on the sample used; the results are that 6 construction companies in Indonesia have financial distress and 11 non-distress problems. This result proves that the best model obtained can predict the level of financial distress of companies with a small error rate to produce 6 companies identified as financially distressed. This result can be a warning for companies to increase revenue by adding new projects to get out of financial distress status. Traditional financial distress models such as Altman, Zmijewski, Springate, and Fulmer, which have become researchers’ guidelines for measuring financial distress, can be added to the ANN 25-20-1 model as a comparison to strengthen the research results.

view full abstract hide full abstract
    • Figure 1. ANN architecture
    • Table 1. Operation of variables
    • Table 2. Confusion matrix
    • Table 3. Descriptive statistics of training data sample
    • Table 4. Comparison of combined MSE
    • Table 5. Training data output
    • Table 6. Confusion matrix output training data
    • Table 7. Results of financial distress
    • Conceptualization
      Farida Titik Kristanti, Zahra Safriza
    • Data curation
      Farida Titik Kristanti, Zahra Safriza
    • Formal Analysis
      Farida Titik Kristanti, Zahra Safriza
    • Methodology
      Farida Titik Kristanti, Zahra Safriza, Dwi Fitrizal Salim
    • Project administration
      Farida Titik Kristanti, Dwi Fitrizal Salim
    • Supervision
      Farida Titik Kristanti
    • Writing – review & editing
      Farida Titik Kristanti, Dwi Fitrizal Salim
    • Investigation
      Zahra Safriza
    • Validation
      Zahra Safriza
    • Visualization
      Zahra Safriza, Dwi Fitrizal Salim
    • Writing – original draft
      Zahra Safriza