An Analysis of Financial Distress Accuracy Models in Indonesia Coal Mining Industry: An Altman, Springate, Zmijewski, Ohlson and Grover Approaches

The purpose of this research is to determine companies financial distress base on Altman, Springate, Zmijewski, Ohlson and Grover Models and to assess the accuracy of those five prediction models in coal mining sector firms listed in Indonesia Stock Exchange (IDX) for the period 2015 – 2019. This research has 22 samples of 23 coal mining firms listed in IDX base on the purposive sampling technique. This study is a descriptive design using quantitative and panel data. The research data is analyzed using the Kruskal Wallis test because there are more than two prediction models to compare and the data are not normally distributed. The result indicates that the Modified Altman and Ohlson Models are the most accurate predictive models because these models have the highest accuracy rate of 90.91%, followed by Zmijewski Model, which has an accuracy rate of 86.36%, then Grover Model has 81.82% accuracy rate, and the lowest prediction rate is Springate Model with the value of 63.64%.

The third point is about the gap of the best predictor model. Fatmawati (n.d.) stated that the Zmijewski Model is the best predictor model compared to Springate and Altman. Puspita Sari (2015) said that Altman Model is more accurate than Springate, Grover, and Zmijewski Models. Edi and Tania (2018) and Priambodo & Pustikaningsih (2018) stated that Springate Model has higher accuracy than Grover, Altman, and Zmijewski Models.
The conclusion of this research is expected can be used by management for internal evaluation. Investors and creditors can assess the corporate financial performance prior to commencing investment. The government is expected to review and establish mining regulation to help the company grow or survive during coal price pressure.

Definition of Financial Distress
According to Wruck (1990), the definition of financial distress is a situation when net cash flow is not sufficient to cover current debts. An extreme situation of financial distress is bankruptcy which can be very expensive, involving legal fees and forcing the company to release its assets at a depressed price. Wruck (1990) provides several general indicators of corporate financial distress, they are continuous dividend reduction or even fail to provide dividend at all, unable to pay operations then impact the closure of several branches, layoffs occur to save the company from larger losses, resignation or sacking their executive, and falling of stock price as an indicator of the market value of the company. According to Altman (1968), Whitaker (1999) and Mumford (2003), financial distress occurs when a company cannot pay its debts. According to Kida (1980) and Mutchler (1985), corporate financial distress occurs if the company at least has one signal among some indicators such as negative working capital in the recent year, operating loss during three years before the bankruptcy, the deficit of retained earnings in the third year before bankruptcy, negative profits in three years before the bankruptcy.

Indicators of Financial Distress
Based on the above literature review, the financial distress indicators used in this research happen when the company has negative profits for three or more consecutive years, reduce or fail to pay a dividend at all, and increment of debt to equity ratio (DER).

Modified Altman Z"-Score Model
The most popular model for predicting corporate failure is the Z-score formula was developed in 1968 by Edward I. Altman, an assistant professor of finance at New York University. Altman uses the Multiple Discriminant Analysis (MDA) technique to predict company bankruptcy. In 1983, Altman revised his model to produce Z'-score by adjusting the model for the private company sector model since the previous Z-score was only used for going public manufacture. Furthermore, Altman modified his model to produce Z"-score by removing sales to total assets variable (Altman, 1968). EBIT to total assets ratio has the most contribution at this model version (

Springate S-Score Model
The Springate Model was developed in 1978 use MDA technique to select four ratios from 19 popular financial ratios. They are working capital to total assets, EBIT to total assets, EBIT to current liabilities, and sales to total assets (Ari Rachmad, 2021). S-score has a cut-off value 0.862. If S-score is less than cut-off value, the company is predicted to be in financial distress (Primasari, 2018, p. 28).

Zmijewski X-Score Model
Zmijewski (1984) expanded his study in predicting bankruptcy by analysing return on assets, debt ratio, and current ratio. X-score has a cut-off value 0. If X-score is greater than the cut-off value, the company is predicted in financial distress condition (Edi & Tania, 2018).

Ohlson O-Score Model
Ohlson (1980) published his research use logit or multiple logistic regression techniques to construct a bankruptcy predictor model. He uses a cut-off value greater than 0.38 to predict the company's financial distress condition (Safitri & Hartono, 2014, p. 330).

Grover G-Score Model
The Grover Model is created by redesigning and evaluating the Altman's Model by adding 13 new financial ratios use 35 bankrupt and 35 non-bankrupt companies in 1982 -1996. The cut-off value for a bankrupt company if G-score produces value less than or equal to -0.02. Meanwhile, cut off more or equal to 0.01 indicate the non-bankrupt company (Salim & Sudiono, 2017, p. 381).

Theoretical Framework and Hypothesis
The research framework in this research as follows:

Picture 2. Framework
The hypotheses are formulated for this research as follow: H1: There are differences in financial distress prediction result using Modified Altman, Springate, Zmijewski, Ohlson, and Grover Models. This study is a descriptive design using quantitative and secondary data type. Data are cross-section and time-series (panel data) with 22 samples for 5 years observation, so 110 total audited financial statements are to be explored to analyse financial ratios. The operational variables in Picture 3 are used in this research, as follow:

WCTA (Working Capital / Total Assets)
This variable is used to measure the company's liquidity. Altman, Springate, Ohlson, and Grover Models use this variable.
The higher WCTA ratio indicates the greater company working capital from total assets and expected will increase the company profits. This variable can be measured using the company's financial statement data.

RETA (Retained Earnings / Total Assets)
RETA is used to measure the company's cumulative profitability. This variable is only used in Altman Model.

EBITTA (Earnings Before Interest and Taxes / Total Assets)
EBITTA is used to measure the company's profitability. This variable is used in the Altman, Springate, and Grover Models.

BVETL (Book Value of Equity / Total Liabilities)
BVETL is used to determine the company's value by investors in the capital market. This variable is only used in Altman Model.

SATA (Sales / Total Assets)
SATA is used to determine the company's ability to generate sales by existing assets. This variable is only used in Springate Model.

EBTCL (Earnings Before Taxes / Current Liabilities)
EBTCL is used to measure company's profitability. EBT data is obtained from the profit or loss statement. This variable is only used in Springate Model. 7. SIZE (Log (Total Assets / GNP Price Level Index) SIZE is used to measure the company's size. This variable is only used in Ohlson Model. Gross National Product (GNP) price level index data is obtained in www.bps.go.id. 8. TLTA (Total Liabilities / Total Assets) TLTA is a variable to measure the company's total liquidity. This variable is used in Ohlson and Zmijewski Models. This ratio is determined to measure the company's leverage. The company is in difficult financial position when this ratio continues larger and will increase risk of inability to pay company's liabilities.

CLCA (Current Liabilities / Current Assets)
CLCA is used to measure the company's short-term liquidity. This variable is only used in Ohlson Model. If current liabilities exceed current assets, the company will difficult to pay the short-term debt.

NITA (Net Income / Total Assets)
NITA is used to measure the company's profitability. This variable is used in Zmijewski, Ohlson, and Grover Models. Net income and total assets are obtained in the profit or loss statement and balance sheet.

FUTL (Cash Flow from Operation / Total Liabilities)
FUTL is used to measure the company's liquidity and determine company's ability to generate sufficient cash to pay liabilities. This variable is only used in Ohlson Model. The data is obtained in cash flow statement and balance sheet.

INTWO
INTWO is used to measure the company's profitability. This variable is only used in Ohlson Model. If during the last two years the company getting losses, it may be financial distress condition.

OENEG
OENEG is used to measure the company's liquidity. This model is only present in Ohlson Model. If total debt exceeds total assets, the company is likely in financial distress condition. 14. CHIN CHIN is used to measure the changing of company's profitability. This model is only present in Ohlson Model. The data is obtained in profit or loss statement. This variable is determined by measuring the changing of net income during the last two years.

CACL (Current Assets / Current Liabilities)
CACL is used to determine the effectiveness of current assets to pay current liabilities. This variable is only present in Zmijewski Model. The data is obtained in company's balance sheet.

Descriptive Statistic Analysis
The descriptive statistic in Table 2

Financial Condition
As mentioned earlier, the financial distress indicators used in this research when for three or more consecutive years the company has negative profits, reduce or fail to pay dividend at all, and increase debt to equity ratio (DER). PKPK has better DER than ARII but relatively increase from 1.05% to 4.07% for five years. These conditions trigger financial distress and if they can not manage them, bankruptcy cannot be prevented. The decline of coal price must impact these conditions but some companies can survive because they have adequate financial management.

Financial Distress Model Analysis
The financial ratio processed from the financial statement is an effective tool to describe the condition of the company. It is used as operational variables contained in each prediction model. The results of this research using calculation of model's formula and summarized in Table 4.
According to The Modified Altman Model shown in Table 4, two companies have an average Z"-score below 1.10. ARII and BUMI are predicted in difficult financial condition. ARII has Z"-score of -0.9833 and BUMI has Z"-score of -2.4054. Based on Grover Model result, four of 22 companies have financial problems due to their average G-score less than -0.02. The distressed companies are ARII, BUMI, SMMT, and SMRU.
The calculation result mentioned in Table 4 indicates that five models have different results, but all models have the same prediction that ARII and BUMI will have financial distress problems.   The hypothesis used in this test is as follows: H0: There is no difference in the prediction of financial distress using Altman, Springate, Zmijewski, Ohlson, and Grover Models for coal mining issuers in 2015 -2019.
Ha: There are differences in the prediction of financial distress using Altman, Springate, Zmijewski, Ohlson, and Grover Models for coal mining issuers in 2015 -2019.
By a significance level of less than 0.05, H0 is rejected and it is concluded that the prediction result is significantly different, whereas if the significance level more than 0.05, then H0 is not rejected and it is concluded that there is no difference is prediction result. Base on the analysis results in Table 6, the significance value is 0.000 or less than 0.05. It can be concluded that H0 is rejected and it means the five models have a significant difference.

Accuracy Rate of Financial Distress Models
The previous analysis result state that the five models have different prediction results. The following Table 7 compares the accuracy level of the five prediction models compared with a real condition as shown previously in Table 3.
The measurement result in Table 7 indicates that Modified Altman and Ohlson Model have the same accuracy rate of 90.91% with an error measurement of 9.09%. These models are the most accurate rate compared to other models, followed by Zmijewski Model that has an accuracy rate of 86.36% and error measurement of 13.64%. Grover Model has an accuracy rate and error measurement of 81.82% and 18.18%, respectively. Springate Model has the lowest accuracy rate of 63.64% and the highest error measurement of 36.36%.

Hypothesis Test
The hypothesis test in this study is based on the result of statistical tests that have been carried out. Hypothesis test 1 is conducted a different test using Kruskal Wallis test, while the testing of hypotheses 2 to 6 is based on the result of the model's accuracy in predicting financial distress. The following Table 8 is the summary of hypothesis test results to answer this research.

Conslusion
The calculation of accuracy level generates that Modified Altman and Ohlson Models are the most accurate models compared to Springate, Zmijewski, and Grover. Both models have a prediction accuracy rate of 90.91% and an error measurement of 9.09%. The Zmijewski Model has an accuracy rate and error measurement of 86.36% 13.64%, respectively, followed by Grover Model with an accuracy rate of 81.82% and an error measurement of 18.18%. Springate Model is the worst prediction model with an accuracy rate of 63.64% and the highest error measurement of 36.36%. These are in accordance with the Kruskal Wallis test which states that all models have a significant difference.
This study will be very relevant for further researchers who are concerned with how to measure the level of financial difficulty in coal mining sector companies listed on the Indonesia Stock Exchange (IDX) for the 2019-2021 period beyond.
Funding: "This research received no external funding."