MINISTRY OF EDUCATION AND TRAINING STATE BANK OF VIETNAM HO CHI MINH CITY BANKING UNIVERSITY GRADUATION THESIS FACTORS AFFECTING THE CAPITAL STRUCTURE OF CONSTRUCTION ENTERPRISES LISTE
INTRODUCTION
Reasons for choosing the topic
The globalization trend not only requires countries to cooperate in politics, military, and science, but also especially focuses on economic cooperation The opening up and integration of different economies has helped every country to benefit, especially for a developing economy like Vietnam However, integration also leads to a more fierce, globally competitive environment and also contains many potential risks In addition, to confidently integrate economically, countries as well as their businesses must have sufficient competitiveness and potential to cope with the risks of economic crises With more business and investment opportunities appearing, businesses also need to prepare capital to realize those opportunities In the process of national building, innovation and economic development, the construction industry in general and Vietnam's construction industry in particular is not only one of the key industries of each country, but also plays a very important role in the economy Therefore, in recent years, the growth rate of the construction industry, as one of the basic sectors of the national economy, has created the material foundations for most economic, and social activities as well as the welfare of people through the fields of civil, and infrastructure to industrial construction
In terms of economic reality, according to some reports in the period from
2013 to 2021, the construction industry has a relatively high and stable contribution rate to GDP each year, along with the development of the construction industry, along with that are the State's macro policies that help the economic growth rate fluctuate in an upward trend Along with the recovery of the real estate industry in the period
2014 - 2019, the construction, installation and infrastructure development sectors have developed strongly with an average growth of over 9.0%/year compared to the previous period of 2015 According to statistics from the Institute for Research on Infrastructure Development and Urban Planning, the growth rate of the construction and infrastructure segment tends to decrease, with the growth rate peaking at 10,8% in 2016 and gradually decreasing over the following years This system of problems makes Vietnamese construction companies increasingly concerned about corporate financial management and focus on finding the right capital structure to help businesses maximize profits and limit financial risks
Therefore, in order to survive and develop sustainably, businesses must pay attention to financial management, especially the decision to sponsor enterprises; because it affects the ability to implement business strategies and the sustainable development of enterprises The decision to sponsor is the key for businesses to grow, create more jobs and income, and contribute to the economic development of a country Vietnam is one of the region's fast-growing economies, attracting many domestic and foreign investors to invest With socio-political stability, our country has become a destination for large investors According to the General Statistics Office of Vietnam (2021), the construction industry has contributed significantly to Vietnam's growth, an average of 8%/year over the past 10 years A major characteristic of the construction industry is that investment projects often require large amounts of capital At the same time, most enterprises do not have enough capital to invest in projects Moreover, a business's financing decision involves choosing capital sources to finance projects or capital needs of the enterprise and is one of the main issues of corporate finance towards the establishment of an optimal capital structure to maximize income for owners, minimizing risks and capital costs Therefore, using loans and mobilizing capital from investors to implement the project is inevitable However, finding a reasonable capital structure is not an easy task, it is necessary to study when the company borrows, when to use equity and how much debt is reasonable, helping the business have opportunities for sustainable development and ensuring sufficient competitiveness in the market
With the current developing market economy, businesses have many opportunities and channels to mobilize capital However, the concern of corporate finance managers is how to build a capital structure to maximize the value of the enterprise so that the company's stock price also increases, also known as building an optimal capital structure The optimal capital structure is understood as the capital structure in which the average cost of capital is the smallest and the enterprise value is the largest To build an optimal capital structure, managers must first consider how factors affect the capital structure However, depending on the characteristics of the business and each specific time, the capital structure of companies will be different and the factors affecting the capital structure will also be different Managers need to have a plan to build a safe and optimal capital structure to ensure that they do not fall into a situation of capital shortage or worse, financial distress Besides, it is necessary to understand the advantages and disadvantages of each funding source, thereby helping to increase business efficiency
The survey of factors affecting the capital structure, especially the capital structure of the construction industry, is very urgent and can be applied in the era of economic development in the construction industry today With an awareness of the importance of capital structure and setting the highest goal in maximizing the corporate value of the construction industry in Vietnam and helping company managers make accurate decisions with company goals Therefore, the author chooses the topic “Factors affecting the capital structure of construction enterprises listed on the Vietnam stock market”.
Research objectives
The general research objective of the study is to determine and measure the impact of factors on the capital structure of construction companies listed on the Vietnamese stock market, thereby proposing solutions to help listed construction companies build an optimal capital structure to maximize profits and increase business value
Based on the general goal, the author divides it into 3 specific goals, including:
- Determining factors affecting the capital structure of construction enterprises in the Vietnam stock market from 2016 to 2023
- Assessing the influence of capital structure on construction companies in the Vietnam stock market from 2016 to 2023
- Proposing recommendations and providing direction to optimize the capital structure of construction companies listed on the Vietnam stock market.
Research questions
To achieve the research goal, the thesis will answer the following questions:
- What factors affect the capital structure of construction companies listed on the Vietnam stock market?
- How do these factors impact the capital structure of construction companies listed on the Vietnam stock market?
- What recommendations will help improve the capital efficiency of construction companies listed on the Vietnam stock market?
Object and scope of research
The subject of the thesis is capital structure and factors affecting the capital structure of the construction enterprises listed on the Vietnam stock market
Research space: The study uses data from 82 construction companies listed on the HOSE (Ho Chi Minh Stock Exchange) and the HNX (Hanoi Stock Exchange)
Research time: The research period from 2016 to 2023 spans 8 years, ensuring that the selected sample is representative of the research results.
Research method
The study uses panel data regression with three methods: Pooled Ordinary Least Squares (POOLED OLS), Fixed Effects Model (FEM) and Random Effects Model (REM) Then, the author will compare the regression results of each method through the F-test and Hausman test to choose the most suitable model In addition, the author also checks multicollinearity, autocorrelation, heteroscedasticity and addresses those defects by using the Feasible Generalized Least Squares method (FGLS) Finally, the research uses the Generalized Method of Moments (GMM) model to solve the issue of endogeneity variables in the model.
Contribution of the thesis
This research aims to analyze the factors that influence the capital structure of construction companies in the Vietnam stock market The study aims to provide practical recommendations to companies to help them build an optimal capital structure that maximizes their business value and profits Additionally, the research also contributes to enriching the theoretical and practical basis for the capital structure of construction companies.
Research structure
The research topic includes 5 main chapters, specifically:
In this chapter, the author will briefly introduce the topic The introduction will cover the reason for choosing the topic, the research objectives, research questions, research scope, research methods, and the contributions of the research topic and its structure.
LITERATURE REVIEW
Overview of capital structure
Capital structure is the particular combination of debt and equity used by a company to finance its overall operations and finance its assets A firm’s capital structure is typically expressed as a debt-to-equity or debt-to-capital ratio According to Yapa Abeywardhana (2017) claims that capital structure or financial leverage decisions should be examined concerning how debt and equity mix in the firm’s capital structure influence its market value As defined by Damodaran (2014), a firm's capital structure is a mix of debt and equity used to finance production and business activities Fundamentally, capital structure is a mixture of numerous types of funds that contain a firm’s own retained earnings, external debt and shareholder’s equity Rehan and Abdul Hadi (2019) and Ghasemi and Ab Razak (2016)
In Vietnam, Nguyen Thi Thanh Vinh (2020) believes that “capital structure reflects the proportion of loans and equity that affects profitability as well as business risks that an enterprise may encounter” The thesis of Dang Van Dan và Nguyen Hoang Chung (2017): “Capital structure is the proportion of capital sources (including liabilities and equity) in the total capital an enterprise mobilizes and uses in operations manufacturing business”
In summary, from the perspectives of previous studies, the concepts of capital structure do not have many differences Therefore, the capital structure of a business can be understood as a combination of debt and equity to finance the business's activities Hence, identifying the impact factors and the level of impact of the factors on employees is one of the most necessary tasks that company administrators need to pay attention to
2.1.2 Meaning of capital structure in business
One of the most crucial factors in business management is the balance between debt and equity in a company's capital structure, and it plays a major role in its success Effective use of capital increases a company's mobilization since any immediate capital needs for operations or manufacturing may be met with ease Conversely, an inappropriate capital structure may put the business under financial risk strain Financial leverage is one financial indicator that financial managers frequently utilize when managing capital structure
Capital structure refers to the mix of different sources of funds, including equity and debt, used by a company to finance its operations and investments It represents the way that a company finance its assets and is essential in determining its financial health and risk profile A business can have a flexible source of cash to meet any capital needs for operations or manufacturing because of the larger percentage of debt in the capital structure Additionally, using debt has the tax- shielding effect against interest, allowing businesses to lower tax payments Conversely, having a large debt load can result in hefty interest payments, particularly when the economy's average interest rate is high This can have a substantial negative impact on the company's ability to operate and perhaps cause it to close In situations where the equity proportion is higher, corporations are required to pay dividends to shareholders rather than incur interest payments A large equity ratio also signals to investors that the business is reliable because it has drawn significant funding from the general population Conversely, a large equity ratio lessens the tax shield effect and causes ownership to be dispersed, which can put pressure on the company's managers and change the course of the business's growth.
Theory of capital structure
Durand's (1952) study was the first work on the capital structure of businesses with assumptions such as: businesses operate in an environment with corporate income tax, financial markets are imperfect and businesses potential risk of financial distress due to the use of debt Typically, debt is cheaper than equity However, when businesses increase the ratio of loan capital to equity capital, the risk level increases; This forces owners to increase profit margins Therefore, reducing business value According to this theory, there exists an optimal capital structure that maximizes corporate value and minimizes the cost of capital (WACC) However, the main problem with the traditional view is that there is no underlying theory that shows how much the cost of equity should increase due to the ratio between debt and equity, or how much the cost of debt should increase due to the risk of default Therefore, the M&M theory was born based on providing evidence as well as supplementing the shortcomings that this view still lacks Bringham, Houston and Eugene (2009)
2.2.2 Modigliani and Miller’s Theory (M&M Theory)
Modigliani and Miller’s theory (1958) is known as the foundation for further studies on capital structure theory M&M theory is stated in two important propositions The first clause talks about company value The second clause talks about the cost of capital These propositions will be considered in two cases respectively, with two assumptions: businesses operate in tax and non-tax environments
M&M theory without corporate income tax follows assumptions that there is no corporate income tax and personal income tax; no transaction costs; no bankruptcy costs and financial difficulty costs; individuals and businesses can borrow money at the same interest rate With the above assumptions, the content of M&M theory is stated in the following two propositions
The first proposition: The value of an enterprise using debt is equal to the value of the enterprise without using debt, meaning capital structure does not affect enterprise value In a perfect capital market, the benefits of using debt (tax benefits) or the costs (transaction costs, bankruptcy costs and agency costs ) do not exist Therefore, the use of debt does not bring any additional benefits and additional costs to the business Enterprises have an unchanged value, regardless of capital structure, which means, enterprise value is not affected by the ratio of debt and equity No matter what method of capital mobilization, it will not change the value of the business; a business using debt has the same value as a business not using debt
The second proposition: The cost of equity capital is proportional to the debt ratio This proposition states that the cost of equity capital increases if the enterprise increases debt Under the assumed conditions, the cost of using debt capital is lower than the cost of using equity capital, so if a business increases the proportion of debt, it will reduce the average cost of capital, and at the same time the cost of capital Shares also increase with the proportion of on-lending debt, causing the average cost of capital to increase Therefore, the average cost of capital does not change
However, the M&M theory does not consider the impact of several other costs, causing the benefit of the tax shield to gradually decrease and become eliminated as the company increases its debt ratio That is the impact of the cost of financial distress When a company increases its use of debt, the company's risk increases This gives rise to financial distress costs that simultaneously impact the benefits of the tax shield
At some point, the costs of financial ruin will outweigh the benefits of the tax shield
Based on the foundation of M&M theory, the Trade-off theory has considered the impact of taxes and financial distress costs (Increase in capital costs ) Initiated by Kraus & Litzenberger (1973) and developed by Myers (1977), Trade-off theory suggests that businesses should only use a certain level of debt to maximize business value, as opposed to M&M theory: The value of a company increases the amount of debt it uses, the company receives benefits equal to the present value of tax savings thanks to interest on debt This benefit causes the value of the leveraged company to increase Trade-off Theory has shown that the target capital structure is the point at which the benefits from tax shields can offset the costs of financial distress However, when the debt ratio increases to a certain level, the cost of financial distress will exceed the benefit of the tax shield from interest From there, the company value will decrease and increase the probability of bankruptcy
Value of leveraged firm = Value of unleveraged firm + Present value of tax shield – Present value of financial distress cost
Factors affecting capital structure from the structural trade-off perspective include: corporate income tax, financial distress costs, tangible fixed assets, company size and profits
The Trade-off Theory has explained the limitations of M&M theory regarding the cost of financial distress for debt-borrowing enterprises However, there are also many things that the Trade-off theory cannot explain, such as why some businesses are still successful and have good business results when they borrow very little debt; or in reality, when a business's stock price is high and the business needs external financing, the company is more likely to issue shares (rather than borrow debt)
Besides the above theories, the pecking order theory developed by Stewart Myers and Nicolas Majluf (1984) goes in a different direction by stating that: "an optimal capital structure does not exist, but indicates an order of priority when using investments" The study divides funding sources into: internal capital sources (contributed capital and retained profits) and external capital sources (loans and new stock issuance)
According to Myers et al (1984), based on information asymmetry between financial managers and outside investors Managers will have more information than investors Therefore, investors will often request higher discounts, making the cost of raising external capital higher, thereby leading to the formation of a funding priority order
Although the pecking order theory explains some aspects that affect businesses' decisions to choose funding sources, this theory still has many limitations when it cannot explain the impact of taxes, bankruptcy costs and securities issuance costs to corporate debt
According to market timing theory, capital structure is the cumulative result of past market timing efforts In other words, there is no optimal capital structure according to Trade-off theory Baker & Wurgler (2002) believe that company value is determined by the following two factors: stock price and time of market entry
Time to enter the market: This time has a huge and persistent impact on the leverage ratio
Thus, many theories on capital structure have been presented and applied This study focuses on applying two theories of capital structure: The Trade-off theory of capital structure and The Pecking order theory.
RESEARCH METHODOLOGY
Research data
The study uses a data set collected from audited financial statements and the reference source Vietstock of 82 construction companies listed on the Vietnam stock market from 2016 to 2023, including HOSE and HNX stock exchanges.
Research methodology
The study uses panel data regression with three methods: Pooled Ordinary Least Squares (POOLED OLS), Fixed Effects Model (FEM) and Random Effects Model (REM) Then, the author will compare the regression results of each method through the F-test and Hausman test to choose the most suitable model In addition, the author also checks multicollinearity, autocorrelation, heteroscedasticity and addresses those defects by using the Feasible Generalized Least Squares method (FGLS) Finally, the research uses the Generalized Method of Moments (GMM) model to solve the issue of endogeneity among the independent variables in the model
General data on the number of observations, averages, maximum and minimum values, and the standard deviations of variables Furthermore, the author synthesizes and provides general comments according to the statistical findings
Correlation coefficient matrix analysis determines whether there is a multi- collinear relationship between variables in the model According to the results, if the correlation coefficient of the variables is less than roughly 0.8, there is no pairwise correlation between the variables in the model However, this method does not always produce reliable findings when a low correlation coefficient yet multicollinearity persists Use the variance inflation factor (VIF) for more efficiency
The Pooled OLS model is a simple regression model and is thoughtless about the time and space factors Therefore, the given results lack the realism and correctness of the data
Yit = α + β2X2it + β3X3it + àit i: ith cross-unit t: Time t α: Slope constant μit: Random error
In the FEM model, the author assumes that the original slope varies from commercial banks, but it does not change over time
Yit = β1i + β2X2it + β3X3it + àit i: ith cross-unit t: Time t àit: Random error
In this model, β1i is assumed to be a random variable with mean β1 and is presented as follows:
In which: εi: Error component of the cross-unit àit: The combined error component of the cross-unit and time series
Feasible Generalized Least Square (FGLS)
Feasible Generalized Least Squares (FGLS) method that will estimate the model according to the OLS method even in the presence of autocorrelation and heteroscedasticity The model's errors will be utilized to estimate their variance or covariance matrix The matrix will then be used to estimate the coefficients of the independent variables to address the issues of autocorrelation and heterogeneity The coefficients of the variables will reveal the impact of the proposed factors on capital structure once the model results are available
Generalized Model of Moments (GMM)
The GMM method is used to tackle the endogeneity issue of some explanatory variables by utilizing instrumental variables Hypothesis testing examines whether there is a correlation between the instrumental variable and the residual in the model H0: The instrumental variables are consistent, and no endogenous phenomenon occurs Not rejecting the hypothesis Ho (P-value >10%) means that the instrumental variables are suitable The study employs the second- order correlation test (AR2) to evaluate the second-order correlation of the model's residuals, with the hypothesis Ho: There is no second-order correlation between the residuals When p-value >10%, accept the hypothesis Ho: The model residual does not exist in the corresponding object of order 2, implying the model meets the requirements.
Research model
Based on the empirical studies of Almanaseer, S R (2019), Saif-Alyousfi et al (2020), An Thai (2017) and Nguyen Thi Nhu Quynh et al (2020) the author proposes the following model:
TLEV it = α + β 1 LIQ it + β 2 ROA it + β 3 SIZE it + β 4 TANG it + β 5 GROWF it + β 6 GDP t
TLEVit : Financial leverage of firm i at time t
LIQit : Liquidity of company i at time t
ROAit : Profitability of a company i at time t
SIZEit : Firm size of company i at time t
TANGit : Tangible assets of a company i at time t
GROWFit : Growth opportunities of a company i at time t
GDPt : Gross Domestic Product at time t
INFt : Inflation at time t α: Intercept coefficient βi : The regression coefficients ɛit : The model error term.
Research hypothesis
Capital structure is measured by many different indicators Therefore, according to the previous studies, the author chose to measure it using the debt ratio, which indicates what percentage of a business's total assets are financed by debt In this study, TLEV is considered financial leverage, this is one of the indicators showing capital structure Financial leverage is calculated as total liabilities divided by total assets, which is the best measure to reflect capital structure
The author uses 7 independent variables in the study, including: liquidity, profitability, firm size, tangibility, growth opportunity, GDP, and inflation rate The author selects the ratios used as a scale for independent variables based on empirical studies and is often used in studies on the impact of factors on capital structure
Liquidity is a crucial factor in an enterprise's capital structure The liquidity of enterprise assets is measured by the ratio between the enterprise's short-term assets and short-term debt
According to the Pecking Order Theory, liquidity and financial leverage will negatively correlate The hypothesis states that businesses with high levels of liquidity will have greater access to liquid assets that they may borrow to fund their projects Studies by Azeem Muhammad et al (2019), Khaki, A R and Akin, A (2020), Lin (2021), Le Thi Minh Nguyen (2016), Nguyen Thi Nhu Quynh, Le Dinh Luan and Le Hoang Vinh (2020) have also demonstrated a negative correlation between liquidity and short-term leverage of businesses The author expects liquidity to have a negative relationship with capital structure
H 1 : Liquidity has a negative impact on an enterprise’s capital structure
Return on the business's total assets (ROA) is an indicator of profitability This index is used to evaluate a business's efficiency in using assets to create profits, with the formula calculated by the ratio of profit after tax divided by total assets
Profitability has a positive impact when analyzed according to the trade-off theory but has a negative impact when analyzed according to the pecking order theory There are many different results from experiments but most of them indicate that there is a negative relationship between ROA and the debt ratio of companies such as Almuaither and Marzouk (2019), Khaki, A R and Akin, A (2020), Dang Quynh Anh and Quach Thi Hai Yen (2014), Dung and Thanh (2021) The author expects the profitability factor to have a negative relationship with capital structure
H 2 : Profitability (ROA) has a negative impact on an enterprise’s capital structure
Firm size is a decisive factor in the capital structure Large-scale enterprises often diversify their industries compared to small-scale enterprises, so their cash flow is less volatile and can better overcome the risk of financial crisis Besides, large businesses can easily access loans from reputable credit institutions with preferential interest rates and larger credit limits Therefore, large-scale enterprises are more likely to use debt in their capital structure than others This study uses the logarithm of total assets as a measure of firm size
The debt ratio and the business size variable are negatively correlated according to the Pecking Order Theory, but they may also be positively correlated according to the Trade-Off Theory Nevertheless, the majority of recent empirical studies carried out in Vietnam and elsewhere such as Dung and Thanh (2021), Hung and Cuong (2020), Nguyen Thi Thuy Hanh (2019), Ab Wahab and Ramli (2014), Anafo et al (2015) also clearly showed a positive correlation between company size and financial leverage The author expects the firm size to have a positive relationship with the company's capital structure
H 3 : Firm size has a positive impact on an enterprise’s capital structure
Tangible fixed assets ratio is a variable that reflects the asset structure of a construction enterprise, measured by the ratio between tangible fixed assets and total assets of the enterprise
Theories of agency costs and trade-off costs show that owning a large number of fixed assets can help businesses borrow capital more easily because they have collateral assets Furthermore, since secured loans are considered safer, the cost of using loan capital is also lower, which will be a driving force for businesses to borrow more According to the trade-off theory, the ratio of tangible fixed assets has a positive relationship with the financial leverage of businesses This positive relationship was experimentally researched by Almuaither and Marzouk (2019), Jaworski and Czerwonka (2021), Le Thi Nhung (2020), Nguyen Thi Thuy Hanh (2019), Le Thi Minh Nguyen (2016) both support The author expects a positive relationship between tangible assets and a company's capital structure
H 4 : Tangible assets has a positive impact on an enterprise’s capital structure
The growth rate variable is used to measure the impact of the annual revenue growth rate on company value This indicator is calculated by taking the difference in net revenue between the following year and the previous year and dividing it by the net revenue of the previous year
GROWF = Net sales t − Net sales t−1
Pecking order theory states that growth rate and financial leverage are positively correlated According to Myers (1984), managers would rather employ internal funding than external funding The debt ratio is often smaller in the capital structure of stagnant businesses and larger in the capital structure of developing businesses, according to theory A business with a fast growth rate and great potential for revenue growth will have a high credit reputation to borrow more debt Many studies show that the growth rate and debt ratio of enterprises have a relationship in the same direction such as Nguyen Thi Nhu Quynh et al (2020), Luan and Vinh (2020), Jaworski and Czerwonka (2021) The author expects the growth rate to have a positive relationship with the company's capital structure
H 5 : Growth opportunity has a positive impact on an enterprise’s capital structure
The GDP indicator is the most typical indicator to show the differences in wealth of countries A country with a developed economy is understood to have created many products and services, improved people's lives, and increased consumer demand In addition, the increase in GDP is assessed as a favorable domestic business environment, improved production capacity and an innovative economy From there, it attracts domestic and foreign businesses to invest Seizing this opportunity, construction industry enterprises have launched public investment plans, increasing profits for businesses, so the need for funding for businesses is essential With insufficient capital resources inside the enterprise, construction enterprises will mobilize capital on the market, so accessing external loans will be easier Thus, Memom et al (2015), Jaworski and Czerwonka (2021), and Nguyen Quoc Huy (2022) found that GDP per capita is positively related to the debt usage ratio of businesses The author expects GDP to have a positive impact on capital structure
H 6 : Gross Domestic Product has a positive impact on an enterprise’s capital structure
The inflation rate is measured as the percentage of the annual inflation rate for each country The inflation rate demonstrates the stability of the national currency as well as the government's ability to control the economy A nation's high rate of inflation indicates significant economic instability in that nation Inflation raises the risk to lenders' credit decisions, makes them less likely to lend money overseas, and makes it harder for companies to raise loans As a result, a rise in the inflation rate lowers enterprises' debt-to-capital ratio Thus, as Gatsi (2012), Chadha and Sharma (2015), Nguyen Huynh Tra (2022) also found a rise in the rate of inflation results in a fall in the debt-to-capital ratio of companies The author expects inflation to have a positive impact on capital structure
H 7 : Inflation has a positive impact on an enterprise’s capital structure
Table 3.1 Describe dependent variable, independent variable, expected sign
Description Variables Measurement Previous studies Expectation
(Tran Viet Dung and Bui Dan Thanh, 2021; Dang Quynh Anh & Quach Thi Hai Yen, 2014);
Rao Purnima et al, 2019; Bibi and Akhtar, 2023)
(Nguyen Thi Nhu Quynh et al, 2020;
Profitability ROA EAT / Total assets
Dung and Thanh, 2021; Dang Quynh Anh and Quach Thi Hai Yen, 2014)
Firm size SIZE Log(Total assets)
(Hung and Cuong, 2020; Dung and Thanh, 2021; Ab Wahab and Ramli, 2014; Anafo et al, 2015)
Tangible fixed assets / Total assets
(Le Thi Nhung, 2020; Nguyen Thi Thuy Hanh, 2019;
(Net salest – Net salest-1) / Net salest-1
(Nguyen Thi Nhu Quynh et al, 2020;
Luan and Vinh, 2020; Almuaither and Marzouk, 2019;
Chadha and Sharma, 2015; Nguyen Huynh Tra, 2022)
(Source: Author compiled from research results)
Research process
Based on synthesizing the theoretical framework, determining the model and research hypotheses, the research process will be established including 8 steps
Step 1: Synthesize the theoretical basis and experimental studies closely related to the research topic
Step 2: From the theoretical basis and empirical research, identify models, and research variables and collect data appropriate to the research topic
Step 3: Through STATA software, the author of descriptive statistics aims to synthesize data of variables in the model including maximum value, minimum value, average value and standard deviation
Step 4: Analyze the correlation of the model's independent variables, helping to detect the statistical link between them Besides, the author uses the variance magnification factor VIF to check the level of multicollinearity
Step 5: Conduct regression of Pooled OLS, FEM and REM models to estimate values and regression coefficients through F-Test and Hausman tests to find a suitable model for the study
Step 6: Check for defects in the model to avoid causing erroneous conclusions because the regression coefficients and F-test become unreliable
Step 7: After performing the above tests, if defects are detected, FGLS estimates will be used to overcome the defects and introduce statistically significant variables in the model
Step 8: Based on the research results, the author draws conclusions, evaluates the level of impact of each factor on capital structure and gives suggestions on policy implications related to the capital structure of construction companies on the Vietnamese stock market
In chapter 3, the author presented data, research methods, research model and research hypotheses In addition, the author also provides expectations for the model based on relevant theories and previous research and provides a research process Continuing in Chapter 4, the author will carry out the research model through Stata 15.1 software.
RESEARCH RESULT AND DISCUSSION
Descriptive statistics
Using STATA 15.1 software, the statistical author describes research data in terms of: observations, average value, standard deviation, maximum and minimum value of the dependent variable TLEV and independent variables
Table 4.1 Descriptive statistics of variables
Variable Obs Mean Std Dev Min Max
(Source of results from STATA 15.1 software)
Statistical results from Table 4.1 show that there are a total of 656 observations from 82 construction companies in the period from 2016 to 2023, including:
The total debt to total asset ratio (TLEV) – representing the capital structure of construction companies from 2016 to 2023, with a total of 656 observations has an average of 0.6255 and a standard deviation is 0.2043 The highest is 0.9939 of Hoa Binh Construction Group Joint Stock Company (stock code: HBC) in 2023, the lowest is 0.0245 of Everland Group Joint Stock Company (stock code: EVG) in
Liquidity (LIQ) of construction companies from 2016 to 2023 has an average of 1.7524 and the standard deviation is 1.8835 The highest is 21.2298 of VNECO
1 Electric Construction Joint Stock Company (stock code: VE1) in 2022 and the lowest is 0.2307 of IDICO Infrastructure Development Investment Joint Stock Company (stock code: HTI) in 2021
Profitability (ROA) of construction companies from 2016 to 2023 has an average of 0.0305 and the standard deviation is 0.0530 The highest is 0.3199 of Licogi 14 Joint Stock Company (stock code: L14) in 2021, the lowest is -0.4709 of VNECO 1 Electric Construction Joint Stock Company (stock code: VE1) in 2018
Firm size (SIZE) of construction companies from 2016 to 2023 has an average of 6.0236 and the standard deviation is 0.6503 The highest is 7.6417 of Bamboo Capital Group Joint Stock Company (stock code: BCG) in 2022, the lowest is 4.4181 of VNECO 4 Electric Construction Joint Stock Company (stock code: VE4) in 2019
Tangible assets (TANG) of construction companies from 2016 to 2023 have an average of 0.1310 and the standard deviation is 0.1676 The highest is 0.8769 of IDICO Infrastructure Development Investment Joint Stock Company (stock code: HTI) in 2021, the lowest is 0.0002 of Song Da Urban Development and Construction Investment Joint Stock Company (stock code: SDU) in 2023
Growth opportunity (GROWF) of construction companies from 2016 to 2023 has an average of 0.0306 and the standard deviation is 4.7307 The highest is
25.8798 of Song Hong Construction Joint Stock Company (stock code: ICG) in 2018, the lowest is -116.415 of Icapital Investment Joint Stock Company (stock code: PTC) in 2022
The average Gross Domestic Product (GDP) from 2016 to 2023 is 0.0571 and the standard deviation is 0.0188 The highest is 0.0802 – Vietnam's GDP rate in 2022, the lowest is 0.026 – Vietnam's GDP rate in 2021, 2021 is a year heavily affected by the COVID-19 epidemic, explaining the low GDP
The average Inflation from 2016 to 2023 is 0.0300 and the standard deviation is 0.0053 The highest is 0.0354 – Vietnam's inflation rate in 2018, the lowest was 0.0184 – Vietnam's inflation rate in 2021.
Correlation matrix
Correlation matrix analysis is a tool to examine the correlation between independent variables and the dependent variable, as well as the correlation between independent variables If the independent variables are highly correlated (in the range of less than -0.8 or greater than 0.8), then the model has a high possibility of multicollinearity In addition, if the correlation coefficient of each variable is lower than 0.8, there is no pair-wise correlation between the variables in the model
Table 4.2 Correlation Matrix between variables
TLEV LIQ ROA SIZE TANG GROWF GDP INF
(Source of results from STATA 15.1 software)
From the results of table 4.2, it can be seen that the correlation between variables is at an acceptable level when the absolute value of the correlation coefficient between variables is less than 0.8 The absolute value coefficient of the largest correlation is between the independent variable INF and the dependent variable TLEV at 0.5077.
Multicollinearity test
To ensure that the regression model does not have multicollinearity, the study applies the variance inflation factor (VIF) method If the research results show VIF index < 10 then the research model does not have multicollinearity and vice versa (Field, 2000)
(Source of results from STATA 15.1 software)
The results of the VIF coefficient from Stata 15.1 software in Table 4.3 show a maximum VIF value of 1.35 (GDP and INF) and an average value of 1.15 that VIF values are all less than 10 Therefore, there are no signs of multicollinearity and all variables can be included in the regression model.
Regression model
The study conducts regression analysis of panel data with three estimation methods Pooled OLS, FEM and REM to identify the impact level of independent variables: LIQ, ROA, SIZE, TANG, GROWF, GDP and INF to the dependent variable TLEV The regression results are in table 4.4 below:
Table 4.4 Regression results of Pooled OLS, FEM and REM regression models
Variable Pooled OLS FEM REM
(Source of results from STATA 15.1 software)
According to the findings, at the 1% significance level, the variables LIQ, ROA, and SIZE have a high statistically significant to the dependent variable for all three OLS, FEM, and REM methods Meanwhile, the variables GROWF, GDP and INF have no impact on the dependent variable
On the other hand, the TANG factor has statistical significance at a 5% level in two methods of FEM and REM However, in the OLS method, this variable affects the dependent variable with statistical significance at the 1% level
Even though there are some similarities in the regression findings according to the methodologies, the author tests the models to choose the best one in order to guarantee the model's unbiasedness, correctness, and efficacy.
Model selection test
4.5.1 Testing between Pooled OLS and FEM models
The author applies the F-test to select an appropriate model between OLS and FEM with the hypothesis:
H0: The Pooled-OLS model is more suitable for the research variables
H1: The FEM model is more suitable for the research variables
Table 4.5 The result of F-test
Statistical value (F test that all u_i=0) F(81, 567) = 22.93
(Source of results from STATA 15.1 software)
The results show that P-value = 0,0000 < 0.05, so the hypothesis H0 is rejected Thus, the FEM model is more suitable for research than the Pooled OLS model
4.5.2 Testing between FEM and REM models
The author employs the Hausman test to choose an appropriate model between FEM and REM with hypotheses:
H0: The REM model is more appropriate
H1: The FEM model is more appropriate
Table 4.6 The results of the Hausman test chi2 (7) 337.15
(Source of results from STATA 15.1 software)
The results show that Prob > chibar2 = 0.0000 < 0.05, so the hypothesis H0 is rejected Hence, the FEM model is more suitable for estimating research variables than the REM model
As can be observed from the above three test results, the FEM model is the best model for estimating Next, the author continues to test possible model defects (autocorrelation and heteroskedasticity) in the fixed effects model The results are shown in the following table.
Test the heteroscedasticity phenomenon in the model
After selecting the optimal model, the author conducts a Modified Wald test to check the phenomenon of heteroskedasticity of the model and the hypothesis as follows:
Table 4.7 The results of Heteroskedasticity tests
(Source of results from STATA 15.1 software)
According to the Modified Wald test results, the study rejects hypothesis H0 since Prob > chi2 = 0,0000 < 0,05 As a result, the model suffers from the heteroscedasticity phenomenon
Test the autocorrelation phenomenon in the model
The author will assess the model's autocorrelation in this part using the Wooldridge test and the study hypothesis is as follows
Table 4.8 The results of the autocorrelation test
(Source of results from STATA 15.1 software)
The results in table 4.8 indicated that Prob > F = 0.0000 < 0.05 so the hypothesis H0 is rejected Thus, the model has autocorrelation
The author concludes that the FEM model is the most appropriate based on the results of two tests: the Hausman test and the F test After that, the author used the Modified Wald and Wooldridge test to look for defects in the model and found that it has heteroscedasticity and autocorrelation Hence, the author resolves two phenomena using the FGLS method.
Feasible Generalized Least Squares method (FGLS)
The author uses the Feasible Generalized Least Squares (FGLS) regression model method to overcome the autocorrelation and heteroscedasticity phenomena in the selected model The results of FGLS are as follows:
TLEV Coef Std Err z P > |z| [95% Conf Interval] LIQ -0.0415*** 0.0039 -10.55 0.000 -0.0492 -0.0338
(Source of results from STATA 15.1 software)
The results of the FGLS method in table 4.9 show that at the 1% significance level, LIQ, ROA, SIZE, and TANG are statistically significant In particular, firm size (SIZE) has a positive relationship with capital structure In addition, the study also found liquidity (LIQ), profitability (ROA), tangible assets (TANG), growth opportunity (GROWF) and gross domestic product (GDP) negatively affect capital structure The remaining variable inflation (INF) is statistically insignificant in the model.
Endogenous variables and GMM regression model method
Even though, the Feasible Generalized Least Squares method (FGLS) is used to overcome the defects Nevertheless, the model may still have endogenous variables that this method cannot overcome Thus, the author tests the endogenous variables of the model using the Wu-Hausman test for the independent variables of the research model, based on the following hypotheses:
Table 4.10 Test the phenomenon of endogenous variables
Independent variables Wu–Hausman test Conclusion
LIQ is an endogenous variable
ROA is an endogenous variable
SIZE p-value = 0.6587 > 5% Not reject hypothesis H0
SIZE is an exogenous variable
TANG p-value = 0.7334 > 5% Not reject hypothesis H0
TANG is an exogenous variable
GROWF p-value =0.6137 > 5% Not reject hypothesis H0
GROWF is an exogenous variable
GDP p-value = 0.5544 > 5% Not reject hypothesis H0
GDP is an exogenous variable
INF p-value = 0.6263 > 5% Not reject hypothesis H0
INF is an exogenous variable
(Source of results from STATA 15.1 software)
After checking the endogenous variables, the test results are presented in table 4.11 above, the author found that there are two endogenous variables in the model: ROA and LIQ (both with p-value = 0.0000 < 5%) In order to overcome this defect, the author applies the GMM estimate method
The study applies the Generalized Method of Moments (GMM) to eradicate the phenomena of endogenous variables to guarantee the model's accuracy and effectiveness The results of testing the research hypotheses are presented in table 4.12 below through STATA 15.1 software
Table 4.11 Results of regression analysis by using the GMM method
TLEV Coef Std Err t P > |t| [95% Conf Interval] TLEV
Prob > F = 0.000 Arellano-Bond test for AR(2) in first differences: z = -1.28
Pr > z = 0.200 Hansen test of overid Restrictions: chi2(15) = 20.41 Prob > chi2 = 0.157 Sargan test of overid Restrictions: chi2(15) = 17.89 Prob > chi2 = 0.268 Number of instruments = 24