“Financial distress prediction of listed companies – empirical evidence on the Vietnamese stock market”

Financial distress is a matter of concern in the recent period as Vietnam gradually enters global markets. This paper aims to examine the factors of Altman Z-score to detect the financial distress of Vietnamese listed companies. The authors use a sample of 30 delisted companies due to financial problems and 30 listed companies on the Vietnamese stock market from 2015 to 2018. They employ Independence Samples T-test to test the research model. It is found that there are significant differences in the factors of Altman Z-score between the group of listed companies and the group of del- isted companies. Further analyses using subsamples of delisted companies show that the factors of Altman Z-score are also statistically different between companies with a low level of financial distress and those with a high level of financial distress. Based on the results, there are some suggestions to assist practitioners and the State Securities Commission in detecting, preventing, and strictly controlling financially distressed businesses. These results also enable users of financial statements to make more ratio- nal economic decisions accordingly. 2000) for private companies, and Z-score (Altman, 2000) for non-manufacturing companies and emerging economies. Various studies have applied the Altman Z-score model (e.g., Malik, Awais, Timsal, & Hayat, 2016; Babatunde, Akeju, & Malomo, 2017; AlAli, 2018; 2018).


INTRODUCTION
Along with the modern trend in international economic integration, studies on corporate financial distress are of great importance for assisting the economies, governments, and investors in detecting financially distressed companies. In the context of developing capital markets, the prediction of listed companies' financial health plays a very important role because of high financial risks in these markets. In this paper, the authors examine the financial distress of listed companies in Vietnam, a developing country.
While there have been many previous studies on prediction of financial distress, the most popular study is the Altman Z-score models with three variants, including Z-score (Altman, 1968) for manufacturing companies, Z-score (Altman, 2000) for private companies, and Z-score (Altman, 2000) for non-manufacturing companies and emerging economies. Various studies have applied the Altman Z-score model (e.g., Malik In addition to Altman's models, several researchers have developed models themselves to predict financial distress of listed companies. According to Aziz and Dar (2006), there are sixteen techniques for developing a new bankruptcy prediction model based on Altman (1968) and Ohlson (1980). However, Ijaz, Hunjra, and Azam (2017) conclude that multiple discriminant analysis (MDA) and binary regression (logistic analysis) are the two most reliable and efficient methods. For example, Shaher, Salem, and Khasawneh (2012) use a multiple discriminative analysis to develop a prediction model with 22 financial ratios. Meanwhile, Ahmed and Govind (2018) improve the corporate bankruptcy model's predictive power by adding another variable to the regression model based on five Z-score variables. In another approach, Januri, Sari, and Diyanti (2017) compared Altman Z-score models with Springate and Zmijewski methods to find the most effective model for detecting companies under financial distress. Also, Hu and Ansell (2006) considered among five credit scoring techniques, including Naïve Bayes, Logistic Regression, Recursive Partitioning, Artificial Neural Network, and Sequential Minimal Optimization (SMO) to find out the best performance models.
Given that there are different models to predict financial distress, in this paper, the authors employ the Altman Z-score for emerging economies, which is one of the most popular models to examine the financial distress of Vietnamese listed companies. Using a sample of 30 delisted companies due to financial problems and 30 listed companies on the Vietnamese stock markets from 2015 to 2018, the authors show significant differences in the factors of Altman Z-score between the group of listed companies and the group of delisted companies. Further analyses using subsamples of delisted companies show that the factors of Altman Z-score are also statistically different between companies with a low level of financial distress and those with a high level of financial distress.
Based on the findings, there are some recommendations for practitioners. First, it is proposed that the authorities should establish a standardized scale based on the Z-score model to detect financially distressed companies. For example, the stock market authorities shall impose delisting for companies that have negative values of working capital, retained earnings, pre-tax, and interest income for three consecutive years. Second, it is suggested that users of financial statements, e.g., investors, not only rely solely on the information in the capital market or the trend of the market but also pay attention to companies with the Z-score value smaller than 4.35 for consecutive years.

Concept of financial distress
Financial distress usually takes place in developing countries where the business environment is not well controlled. The existence of this condition poses detrimental consequences on the social and economic of such nations. According to Beaver (1966) and Altman (1983), financial distress is a state when the afore-planned monetary commitment of an enterprise cannot be met. In most studies, bankruptcy is the most common term used to define financial distress. However, as mentioned by Muller, Steyn-Bruwer, and Hamman (2012), financial distress is regarded as "a major structural change to the company such as mergers, absorptions, delisting or liquidations." Generally speaking, financial distress can be defined as a condition when a company cannot continue to exist in its normal business form.

Altman Z-score models
Altman's model was based on Beaver's prior study (1966), which was the first time the T-test method was applied to predict companies' bankruptcy. Beaver (1966) employed this method to assess each accounting indicator's importance based on univariate analysis. Then, Altman's improvements over the previous studies are applying statistical analysis, numerical analysis, and the implementation of a combination of different items simultaneously.
Altman employed an analysis technique based on data collected from 66 companies in the US, divided into two groups consisting of 33 each. Group 1 includes companies that went bankrupt during 1946-1965. Group 2 includes non-bankrupt companies that were still operating normally until 1966. From the statement of financial position and statement of profit or loss, 22 financial indicators were calculated and divided into five groups: liquidity, profit, leverage, solvency, and performance index. In the list of 22 financial indicators calculated, five indicators were selected for the predictive model of bankruptcy based on four criteria: 1) observe the statistical significance of multiple equations turn; 2) evaluating the correlation between related variables; 3) observe the model's ability to predict accurately; and 4) analyst's prediction.
For emerging economies, Altman added constants +3.25 to the Z-score model for standardization. Accordingly, this paper proposes using this standardization factor for the sample of listed companies in Vietnam. The formula of the Z-score model is as follows: ( ) where X 1 -Working capital per total assets (WCTA), X 2 -Retained earnings per total assets (RETA), X 3 -Earnings before taxes and interests per total assets (EBITTA), X 4 -Market capitalization per total liabilities (MVETL), X 5 -Revenue per total assets (REVTA).
Based on the statistics of the Z-score, Altman classified companies into three groups as follows: • If Z > 5.85: in safe zone, a company is healthy; • If 4.35 < Z < 5.85: in gray zone, a company is at risk of bankruptcy; • If Z < 4.35: in distress zone, a company is in a dangerous position with high risk of bankruptcy.
Altman (1968) concluded that the analysis of every single financial indicator is not an important analytical technique in the academic environment anymore because the implementation is quite simple.
To assess financial indicators rigorously, a set of financial indicators is incorporated into the analysis to predict the bankruptcy potential. Theoretically, if researchers analyze these financial indicators in a multivariate framework, they will have a greater statistical significance than a sequential comparison of each financial indicator.
From 1985, Altman's models have been widely accepted by auditors, management accounting, courts, and data systems used for credit evaluation. The models have been used extensively in many countries, although they were originally designed for manufacturing companies with assets of more than USD 1 million before changes were made to suit both private and non-manufacturing companies.    servations. Selected companies are those with full financial-statements data during the research period, and data are collected from the website www.vietstock.vn. The detailed steps are as follows.

Previous studies
The research data are firstly collected from the audited financial statements of the companies in the samples. Financial factors are taken from the balance sheets and the statements of profit or loss to calculate Z-score. The sample companies are divided into two groups, including a group of financially distressed companies -Group 1 (delisted companies due to financial problems) and a group of normal companies -Group 2 (listed companies). From Group 1, the authors continued to divide it into two groups basing on Z-score value: Group 1a which consists of companies with a low level of financial distress (e.g., companies with Z-score greater than 4.35) and Group 1b which consists of companies with a high level of financial distress (e.g., companies with Z-score lower than 4.35). Figure 1 shows the elements of the research sample.

Calculation of Altman (2000)
Z-score where WCTA is working capital on total assets, which is equal to current assets minus current liabilities, all scaled by total assets; RETA is the ratio of retained earnings to total assets; EBITTA is the ratio of earnings before taxes and interests to total assets; MEVTL is the ratio of market capitalization to total liabilities, where market capitalization is equal to numbers of shares outstanding time share values. H1a: The WCTA index has a significant difference between Group 1 and Group 2.

Research hypotheses
H1b: The WCTA index has a significant difference between Group 1a and Group 1b.
H2a: The RETA index has a significant difference between Group 1 and Group 2.
H2b: The RETA index has a significant difference between Group 1a and Group 1b.
H3a: The EBITTA index has a significant difference between Group 1 and Group 2.
H3b: The EBITTA index has a significant difference between Group 1a and Group 1b.
H4a: The MVETL index has a significant difference between Group 1 and Group 2.
H4b: The MVETL index has a significant difference between Group 1a and Group 1b.

Independence Samples T-test
The research uses quantitative methods and Independence Samples T-test to test the research model. The Independent Samples T-test compares the means of two independent groups to determine whether there is statistical evidence that the associated population means are significantly different. The Independent Samples T-test is a parametric test. This research uses this method to test the differences between Group 1 and Group 2, as well as between Group 1a and Group 1b. In Group 2, the statistics show that WCTA index has an average value of 0.282 and a standard deviation of 0.1820; the RETA index has an average value of 0.137 and a standard deviation of 0.0869; the EBITTA index has an average value of 0.142 and a standard deviation of 0.0826; MVETL index has an average value of 4.343 and a standard deviation of 6.0329. Especially, the Z-score of this group has a high average value of 11.008, which is about 2.5 times greater than the benchmark value of 4.35 for the distress zone, suggesting that the companies in this group are not financially distressed. A small standard deviation of Z-score indicates that the difference in Z-score within this group is insignificant. .033, and 0.000, respectively, and they are less than 0.05, which indicates that the variances of the two groups of companies are statistically different as the authors use the T-test result in a row "Equal variances not assumed." As a result, the p-values in the T-test of the above factors are 0.001, 0.039, 0.040, 0.006, and 0.003 respectively, which are all less than 0.05, indicating that there is a statistically significant difference in the above indexes between Group 1 (delisted companies) and Group 2 (listed companies). Therefore, with 95% reliability, it can be concluded that the prediction of financial distress is different between Group 1 and Group 2. Table 3 shows descriptive statistics of characteristics between the Group 1a and Group 1b.    Note: The table shows descriptive statistics of characteristics between the Group 1a and Group 1b. WCTA is working capital on total assets, equal to current assets minus current liabilities, all scaled by total assets. RETA is the ratio of retained earnings to total assets. EBITTA is the ratio of earnings before taxes and interests to total assets. MEVTL is the ratio of market capitalization to total liabilities, where market capitalization is equal to numbers of shares outstanding time share values. The Z-score of this group has a low average value of -38.652 (which is far lower than the benchmark of 4.35 for distress zone) and a large standard deviation of 116.3857. The findings indicate that the companies in this group are at high risk of bankruptcy, and the difference in Z-score within the group is also significant. Table 4 shows the Independence Samples T-test results of Group 1a and Group 1b. The results indicate that the p-values of the WCTA, RETA, EBITTA, MVETL and Z-score in the Levene's test (F-test) have a value of 0.000, 0.001, 0.004, 0.000, and 0.000, respectively, and they are less than 0.05, implying that the variances of the two groups of companies are different as we use the T-test result when "Equal variances not assumed." As a result, the p-values in the T-test of the above factors are 0.001, 0.040, 0.045, 0.000, and 0.002, respectively (which are less than 0.05), indicating a significant difference in the average of the above indexes between the group of companies with a low level of financial distress (Group 1a) and the group of companies with a high level of financial distress (Group 1b). Therefore, with 95% reliability, it can be concluded that the prediction of financial distress is different between Group 1a and Group 1b.

Discussion
According to the research results, the differences between Z-score following four independent var-iables: WCTA, RETA, EBITTA, and MVETL are discussed as follows:

Working capital on total assets (WCTA)
There is a significant difference in this index between the group of financially distressed companies and the group of normal companies. This result is consistent with the studies of Malik, Awais, Timsal, and Hayat (2016), Bhavani and Tabi (2017), and Januri, Sari, and Diyanti (2017). This can be explainable that the higher the WCTA index, the more short-term assets are available to meet the business's short-term debt obligations and meet the operational business requirement. As a result, the possibility of bankruptcy is very low.

Retained earnings on total assets (RETA)
There is a significant difference in this index between the group of financially distressed companies and the group of normal companies. This result is also consistent with the studies of Malik, Awais, Timsal, and Hayat (2016), Bhavani and Tabi (2017) and Januri, Sari, and Diyanti (2017). The results show that the higher the retained earnings, the more profitable a company can make from its assets, and the stronger the business is, the greater the index will be. Therefore, the risk of bankruptcy is considerably low.

Earnings before taxes and interests on total assets (EBITTA)
There is a significant difference in this index between the group of financially distressed companies and the group of normal companies. This result is also consistent with the studies of Malik, Awais, Timsal, and Hayat (2016), Bhavani and Tabi (2017), and Januri, Sari, and Diyanti (2017). Not mentioning the enterprise's capital structure and tax rate, this result helps investors easily compare the level of revenue generation of the companies and the efficiency in the production and business activities. Therefore, the risk of bankruptcy, as well as information manipulation, will be lower.

Market capitalization on total liabilities (MVETL)
There is a significant difference in this index between the group of financially distressed companies and the group of normal companies. This result is also consistent with the studies of Malik, Awais, Timsal, and Hayat (2016), Bhavani and Tabi (2017), and Januri, Sari, and Diyanti (2017). This shows that the higher the market capitalization value, the higher the companies' market value, and the total debt of such companies could be covered. Besides, this index also shows high expectations of the market towards businesses; therefore, the possibility of bankruptcy and information manipulation is very low.

For State Securities Commission (SSC)
According to this study, it was found that most companies under a high level of financial distress have the negative values of Z-score, WCTA, RETA, and EBITTA for many consecutive years before delisting. Thereby, the SSC is proposed to establish a standardized scale basing on the Z-score model to detect financially distressed companies. Along with this, a strong regulatory system to strictly control and regular supervision by competent authorities should be maintained to ensure the effectiveness and efficiency of the companies' operating activities. Specifically, the SSC shall impose delisting for companies that have negative values of working capital, retained earnings, pre-tax, and interest income for three consecutive years.

For the users of financial statements
Users of financial statements such as investors shall 'not only' rely on the information in the capital market or the market trend 'but also' pay attention to companies that have the Z-score value smaller than 4.35 for consecutive years. Thereby, investors can avoid investing in such companies as well as be able to make more rational decisions to avoid economic losses. Also, the quality of financial information disclosed is strongly associated with auditors' performance. Therefore, auditors should pay close attention to accounting fraud's red flag in such companies' financial statements.
To do so, the auditors shall calculate Z-score according to the model mentioned in this research. After determining the Z-score, the auditor could base on this result to make initial judgments about the enterprise's financial situation so that a suitable audit plan could be put in place. Furthermore, creditors such as commercial banks could also use the Z-score as a benchmark for making prudent credit granting procedures.

CONCLUSION
The reduction of companies' financial distress will make a great contribution to improve the firm value and create a positive brand reputation. Accordingly, the quality of the capital market is also improved, and thereby, it can attract more investment resources, which are essential for economic development.
In this paper, the Altman Z-score model is employed to study the financial distress of companies in Vietnamese stock markets. The authors used the data collected from 30 listed companies, and 30 delisted companies due to financial problems on VSM in four years from 2015 to 2018, with a total of 240 observations. The research results show that there are statistically significant differences in the factors of Altman Z-score between the group of normal companies and the group of financially distressed companies. Using subsamples, it is also found that the groups of companies with a low level of financial distress and those with a high level of financial distress have statistically different characteristics and Z-score. In particular, it is found that WCTA and EBITTA have the biggest influence on the model with a coefficient of 6.56 and 6.72, respectively. The findings suggest that the more working capital and the higher profitability a company has, the greater the stability in the financial health of businesses. Such businesses will be less likely to face bankruptcy risk. Based on these results, several suggestions have been made to assist practitioners in detecting, preventing, and strictly controlling businesses at high risk of financial distress.