(0.304) | (0.000)*** | (0.424) | (0.300) | (0.005)*** | |
N | 576 | 576 | 576 | 576 | 576 |
R-Square | 0.2451 | 0.1948 | 0.1360 | ||
Prob>F | 0.0000 | 0.0000 | 0.0000 | ||
MODEL SELECTION TESTING | |||||
F-test | 0.0000 | ||||
Hausman test | 0.0000 | ||||
MODEL DEFECT TESTING | |||||
Variance change | Prob>chi2=0.0000: There is a phenomenon of variance change | ||||
Autocorrelation | Prob>F=0.0006: There is autocorrelation phenomenon | ||||
MODEL DEFECT TESTING | |||||
GMM AUDIT | |||||
AR2 | 0.222 | ||||
Hansen test | 0.750 | ||||
Maybe you are interested!
-
Research Results For Dependent Variable Scale Efficiency -
Regression Results of Factors Affecting Rrtk at Agribank (Dependent Variable: Fgapr) -
Cronbach'S Alpha Test For Dependent Variable -
EFA Analysis Results for Dependent Variable Group (Preliminary Quantitative Research Phase) -
Completing the organization of accounting for revenue, sales costs and determining business results at Hai Phong Paint Joint Stock Company - 1
(*, **, *** : statistically significant at 10%, 5%, 1%)
Source: author's calculation results using STATA 12.0 program
Table 2.13. Results of endogenous variable test with dependent variable ROA
Variable
P_Value | Endogenous phenomenon | |
SKIN | 0.0000 | There is an endogenous phenomenon. |
UNI | 0.5972 | No endogenous phenomenon |
TSDH | 0.8908 | No endogenous phenomenon |
GROW | 0.0556 | No endogenous phenomenon |
SIZE | 0.0482 | There is an endogenous phenomenon. |
GDP | 0.5463 | No endogenous phenomenon |
Note: Durbin - Wu - Hausman test (P_value), significance level to reject or accept the hypothesis Ho: the instrumental variable is exogenous is 5%
From tables 2.10, 2.11, 2.12 and 2.13 the results of OLS, FEM, REM, GLS, GMM tests and model selection tests. The results are as follows:
- Based on the F test results, there is: For both models, the value Prob>F = 0.0000 < α (α = 5%): Hypothesis H0 is rejected: FEM will be more suitable than Pooled OLS.
- Based on the Hausman test results, there is: Model with dependent variable ROE: Prob>F = 0.0079 < α (α = 5%): Hypothesis H0 is rejected: FEM will be more suitable than REM.
Model with dependent variable ROA: Prob>chi2 = 0.0000<α (α =5%): Hypothesis H0 is rejected: FEM model is more suitable than REM.
- Based on the results of testing the model's defects: The model has heteroscedasticity and autocorrelation and the thesis will perform regression using the GLS method to overcome these defects. However, according to the author, the models in this thesis show signs of endogeneity with some independent variables having a two-way relationship with the dependent variable.
According to the results of empirical research on testing the endogeneity phenomenon for variables in both models of ROE and ROA variables: With a significance level of 5%: variables DA and SIZE: have endogeneity phenomenon (Table 2.11 and Table 2.13). Therefore, the final results of the thesis depend on the results using the GMM method. The summary table of factors affecting financial efficiency is synthesized by the GMM method for both models with variables ROE and ROA as follows:
Table 2.14. Summary of GMM regression results
Variable
Assume | Dependent variable | ||
ROA | ROE | ||
SKIN | - | -0.1937*** | -0.2890*** |
TSDH | - | -0.0540*** | -0.0587*** |
SIZE | + | 0.0169*** | |
GROW | + | 0.0334*** | 0.0465*** |
UNI | - | -0.0102* | |
GDP
+ | 1.4046*** | 1.4842*** | |
AR2 | 0.222 | 0.712 | |
Hansen test | 0.750 | 0.072 | |
Source: Author's synthesis
The test results using GMM method in table 2.14 show:
- Hansen test gives model results with ROA variable with P-value of 75%, for ROE variable is 7.2%, therefore the hypothesis H0 about the instrumental variable in the model is appropriate and cannot be rejected with a significance level of 5%.
- AR2 test results with ROA variable with P-value of 22.2%, ROE variable is 71.2% showing that the hypothesis H 0 : there is no second-order correlation series cannot be rejected at the 5% level.
Through AR2 and Hansen test, it can be concluded that the GMM model regression results are valid. Specifically as follows:
Hypothesis: “Capital structure has an impact on financial performance”.
The regression coefficients of the DA variable on ROE and ROA are -0.2890 and -0.1937 respectively. It shows that: Debt ratio has a negative impact on financial performance. This means: If other factors remain constant and when the debt ratio increases by 1%, financial performance will decrease by -0.2890% with ROE and -0.1937% with ROA and vice versa. In addition, the coefficients of the DA variable for all regression methods give the same result: DA has a negative impact on financial performance and are all highly reliable.
The result that capital structure has an inverse effect on financial performance is consistent with previous studies in the tourism industry such as: Youn and Gu (2010); Woo Gon Kim (1997); Luis Pacheco (2015); Ajanthan (2013). This is consistent with small and medium-sized tourism enterprises when enterprises with higher financial performance will prioritize the use of retained earnings as per the pecking order theory. On the contrary, Woo Gon Kim believes that enterprises with low financial performance will not have enough funding resources so they are forced to borrow more, this is true for large-scale tourism enterprises that have to invest a lot in fixed assets.
Hypothesis: “Asset structure has a negative impact on financial performance”.
The regression coefficient of the TSDH variable gives the result of -0.0587 for ROE and -0.0540 for ROA. It shows that: The proportion of assets has a negative impact on financial efficiency and has a significance level of 1%. This means: If other factors remain unchanged and when the proportion of tangible fixed assets increases by 1%, financial efficiency will decrease by -0.0587% for ROE and -0.0540% for ROA and vice versa. Besides, the regression coefficient of the TSDH variable for all methods gives the same result: TSDH has a negative impact on financial efficiency and all have high reliability.
This result is consistent with the study of Motanya (2016) when it was stated that tourism enterprises investing in large fixed assets have reduced financial efficiency. The reality is that Hue's traditional customers are tourists with low service spending, and there is a lack of high-income customers with the need for relaxation and entertainment. Therefore, large-scale hotel and restaurant enterprises have difficulty in business operations due to lack of suitable customers.
Hypothesis: “Asset growth rate has a positive impact on financial performance”.
The regression coefficient of the GROW variable with ROE and ROA is 0.0465 and 0.0334. It shows that: Asset growth rate has a positive impact on financial performance and has a significance level of 1%. This means: If other factors remain unchanged and total assets increase by 1%, financial performance will increase by 0.0465% (ROE) and 0.0334% (ROA) and vice versa. In addition, the regression coefficient of the GROW variable for all methods gives the same result: GROW has a positive impact on financial performance and is highly reliable.
The reality of tourism business mentioned in chapter 2 is consistent with the results of business growth rate increasing financial efficiency. Businesses expanding their products and business services will increase profit margins and financial efficiency.
Hypothesis: “GDP growth rate has a positive impact on financial performance”.
The regression coefficient of GDP variable is 1.4842 (ROE) and 1.4046 (ROA). It shows that: GDP has a positive impact on financial efficiency and has a significance level of 1%. This means: If other factors remain unchanged and when GDP increases by 1%, financial efficiency will increase.
The main will increase by 1.4842% (ROE) and 1.4046% (ROA) and vice versa. In addition, the regression coefficient of GDP variable for all methods gives the same result: GDP has a positive impact on financial efficiency and is highly reliable. The conclusion is consistent with Diyya Aggarwal (2016), in the condition of stable economic development, people's income improves, so the demand for entertainment and relaxation increases. Therefore, the revenue and financial efficiency of tourism enterprises increase.
Hypothesis: “Business characteristics have a negative impact on financial performance”.
The regression coefficient of the UNI variable is not statistically significant with ROE but has a result with the ROA variable of -0.0102. It shows that the ratio of cost of goods sold to revenue has a negative impact on financial performance and is significant at 10%. This means: If other factors remain unchanged and when UNI increases by 1%, financial performance will decrease by -0.0102% and vice versa.
Youn and Gu (2010) suggested that tourism enterprises should reduce operating costs, sales and marketing costs to increase profit margins. Hue tourism enterprises are mostly small in scale, so their management and cost control skills are still limited, causing losses.
Hypothesis: “Business size has a positive impact on financial performance”.
The regression coefficient of the SIZE variable is not statistically significant with the ROE variable, but has a result with the ROA variable of 0.0169. It shows that: SIZE has a positive impact on financial performance and has a significance level of 1%. This means: If other factors remain unchanged and when total fixed assets increase by 1%, financial performance will increase by 0.0169% and vice versa.
The regression results with two dependent variable models ROA and ROE can conclude that the variables that have a positive relationship with financial performance are: GROW and GDP. The variables that have a negative relationship are: DA and TSDH. The two variables SIZE and UNI do not find statistical significance with ROE but are significant with the variable ROA.
Synthesize research results on factors affecting business performance and compare with initial assumptions.
Table 2.15. Comparison table of hypotheses and research results
Explanatory variables
Assume | ROA | ROE | |
SKIN | - | - | - |
TSDH | - | - | - |
SIZE | + | + | |
GROW | + | + | + |
UNI | - | - | |
GDP | + | + | + |
Table 2.15 shows that the relationship between factors affecting financial performance has results similar to the initial hypothesis. In which, capital structure has an opposite impact to capital structure. To more clearly assess the impact of capital structure on financial performance according to debt threshold as well as find the optimal capital structure for Hue DDL, the topic continues to perform regression models 3 and 4.
Research results on the non-linear relationship between capital structure and financial performance.
Table 2.16. Regression results of model 3 and model 4
Dependent variable
Independent variable
ROA | ROE | |
SKIN | -0.0245 | 1.3429 |
(0.868) | (0.0000)*** | |
DA2 | -0.0311 | -1.8692 |
(0.856) | (0.0000)*** | |
UNI | -0.0181 | -0.0486 |
(0.005)*** | (0.0000)*** | |
TANG | -0.0624 | -0.1325 |
(0.0000)*** | (0.263) | |
SIZE | -0.0063 | -0.0261 |
(0.054)** | (0.0000)*** | |
GROW | 0.0220 | 0.0281 |
(0.0000)*** | (0.0003)*** | |
GDP | 1,154 | 0.3548 |
(0.0000)*** | (0.557) | |
CONS | 0.048 | 0.2018 |
(0.083)* | (0.0000)*** | |
N | 576 | 576 |
AR (2) test (Pr > z) | 0.140 | 0.463 |
Sargan test | 0.000 | 0.247 |
(*, **, *** : statistically significant at 10%, 5%, 1%)
Source: author's calculation results using STATA 12.0 program
The regression results using the GMM method give the following results: The regression coefficients of variables related to capital structure and square capital structure such as: DA, DA2 are all statistically significant for the ROE variable, but are not significant for the ROA variable. In addition, the Sargan test and AR (2) test results both show that the model with the dependent variable ROE is suitable. This is the basis for determining the optimal capital structure ratio for maximum financial efficiency. The author also conducted a test of 2 thresholds, but the results did not show that there are 2 debt thresholds (Appendix). Therefore, the research results conclude that there is only one debt threshold for Hue's enterprises.
Determine the optimal capital structure:
The threshold of DA is found by considering the first derivative of both sides with DA. After solving the equation, the optimal threshold of DA is 35.92%, which is the highest financial efficiency.
Based on the threshold value of 35.92%, the sample set can be divided into 2 groups with the debt ratio ranging from 0% - 35.92% and greater than 35.92%. To determine the relationship between financial performance and debt thresholds, the study performed OLS regression on the groups and gave the following results:
Table 2.17. Regression results by threshold
Threshold
Variable
0%-<DA<35.92% | DA> 35.92%. | |
SKIN | 0.0962** | -0.6148*** |
UNI | -0.0100 | -0.0048 |
TSDH | -0.0706*** | -0.2097*** |
SIZE | -0.0169** | 0.0076 |
GROW | 0.0633*** | 0.0497* |
GDP | 1.7912** | 0.8407 |
_CONs | 0.0642 | 0.3427* |
(*, **, *** : statistically significant at 10%, 5%, 1%)
Source: author's calculation
When the enterprise uses debt below 35.92%, the DA variable has a value of 0.0962 with a significance level of 5%, showing that the enterprise's financial efficiency will increase by 0.0962% when the debt ratio increases by 1%.
When the enterprise uses debt exceeding 35.92%, the DA variable has a value of -0.6148 with a significance level of 1%, showing that increasing the debt ratio by 1% will reduce the enterprise's financial efficiency by -0.6148%.