Factors affecting the size of Vietnam's corporate bond market

rge or small something or someone”. Therefore, it can be understood that the size of the corporate bond market is how large or small of market. Because the corporate bond market includes both primary and secondary markets, the size of the primary market is considered through the issuance volume and the size in the secondary market is shown by liquidity (Mizen and Tsoukas, 2014). 1.1.2.2. Measurement of the size of corporate bond market
Primary bond market Issuance volume = Total value of outstanding bonds/GDP (1) Issuance volume = Total value of bonds issued (2)
Secondary bond market (i) Number of transactions (3) (ii) Number of bonds traded (4) (iii) Turnover (5) 1.1.3. Factors affecting to the size of corporate bond market 1.1.3.1. Factors affecting to the size of primary corporate bond market
The size of economy. The small economy often lack useful tools for growing bond market.
The openness of the economy. Economies with large openness usually have more development corporate bond markets.
The development stage of the economy. The higher the 4 national stage of development, the greater the corporate bond market
Interest rate variability. The larger interest rate variablility will increase the risk of long-term assets in the financial market, especially corporate bonds and the size of the corporate bond market will decrease.
The size of banking system. In some cases, the larger the banking system, the less developed the corporate bond market. However, in an open economy, commercial banks now become the necessary subjects to promote the development of the corporate bond market.
Exchange rate variability.
Foreign exchange reserves. Increasing of the level of foreign exchange reserves will lead to the development of corporate bond market.
Creditors’ rights. Countries that have strong creditors’ rights will promote the increasing of the size of corporate bond market. 1.1.3.2. Factors affecting to the size of secondary corporate bond market
Issuance Volume. Higher issuance volume, higher trading transactions.
Bond’s age. Bonds issued in the latest time period will be traded most often.
Credit risk. Investors do not want to invest to bonds that have high credit risk.
Profit volatility.
Stock’s trading volume. 1.2. Literature review 1.2.1. Foreign studies Research by Hawkins (2002) has made a multi-dimensional assessment of the relationship between the banking system and the size of corporate bond market. Eichengreen and Luengnaruemitchai (2004) tested 15 factors affecting the size of the bond market. The results show that the size of the economy, the size of the banking system, the exchange rate, compliance with international accounting standards, corruption and state management are factors that influence the bond market of Asian countries. Research by Bhattacharyay (2013) focused on understanding the 5 correlation between the size of the bond market and the macro factors. The testing results for corporate bonds show that the issuance volume of the corporate bond market is affected by factors: the size of the economy, the openness of the economy, the size of the banking system, the stage of development of the economy and interest rate variability. Kowalewski and Pisany (2017) suggested that creditors' rights and market size are positively related. Mizen and Tsoukas (2014) studied the factors affecting the issuance decision of businesses. In this study, the authors believe that there are both macro and micro factors that influence corporate decision making. The results show that the size and growth of revenue of the business has a positive relationship with the size of the issuance. The regression results of Braun and Briones (2006)’s study showed that the most important factor determining the size of the market. Bond is the level of economic development, or GDP per capita. The size of the banking system and the creditors' rights also have a positive relationship with the size of the bond market. Focusing on emerging economies, a study by Tendulkar (2015) concluded that international corporate bond size is affected by GDP per capita, number of listed enterprises, and domestic credit and interest rate variability. The total size of the corporate bond market is affected by the size of the government bond market, the number of listed companies, interest rate spreads, domestic credit, CPI and hedging premiums. Mu et al. (2013) suggested that domestic credit has a positive effect, interest rate variability have a negative relationship to the size of the African corporate bond market. Fredrick's (2014) showed different results. Accordingly, only the exchange rate, interest rate fluctuations, the size of the economy and average income affect to the size of the corporate bond market. Another similar study was conducted in the Indian corporate bond market by Maurya and Mishra (2016) which also gave conclusions that are not very similar to those of previous emerging market studies. According to Maurya and Mishra (2016), the Indian corporate bond market was most significantly affected by foreign exchange reserves. Study of Alexander et al. (2000) focused on testing factors that affect trading volume of high – yield corporate bonds. The results showed that issuance volume and bonds’age are two factors that have strong impact on trading volume of these bonds. Alexander et al. (2000) 6 supposed that bonds within 2 years of issuance are traded the most. Besides, this study also showed that credit risk and profit volatility all have positive impact to trading volume. In contrast, Wahyudi and Robbi (2009) mentioned that issuance volume has a negative impact to trading volume. Hotchkiss and Jostova (2007) expanded the scope of research, confirmed that issuance volume and age have strong impact to trading volume of corporate bonds. 1.2.2. Studies in Vietnam Tran Thi Thanh Tu (2007) used descriptive statistic method to compare and analyze data on Vietnam corporate bond market in the period of 2003 - 2007. According to the author, most of corporate bonds were issued at that time were long-term bonds, and they belong to State enterprises or big enterprises. The secondary bond market has almost no bonds traded. Trinh Mai Van (2010) and Phan Thi Thu Hien (2014) suggested that the lack of legal framework on corporate bonds and information transparency was reason of the underdevelopment of the corporate bond market in Vietnam. The research of Vuong and Tran (2010) analyzed Vietnam corporate bond market in the period of 1992 – 2009 with a large dataset from different aspects. They confirmed that the issuance of corporate bonds focused on large enterprises, mainly state-owned enterprises and the competition of state-owned enterprises limited the ability to raise debt of SMEs. Stuydy of HNX (2016) also described the current situation of the market in recent years. Nguyen Thi Nhung and Tran Thi Thanh Tu (2019) gathered 11 groups of criteria to assess the liquidity of corporate bond market. In particular, size of issuance, outstanding debts of corporate bonds market, growth speed of corporate bonds in circulation are used to measure the liquidity of primary bond market and volume of corporate bonds traded and coefficient rotation are used to measure the liquidity of secondary bond market in Viet Nam. This research compared these criteria’s data of Vietnam and some other countries to assess the liquidity of Vietnam bond market. Research of Nguyen Hoa Nhan et al. (2014) is the first research in Vietnam that uses regression analysis method to analyzie the relationship between the issuance volume of corporate bonds and some macro – factors. This study showed that the openness of economy, the size of banking system and issuance volume in last 7 period have positive impact to the issuance volume in this period. In contrast, exchange rate and the stage of development of the economy have negative impact to the issuance volume of corporate bonds in Vietnam. 1.2.3. Research gaps Tại Việt Nam, các nghiên cứu về TPDN chủ yếu ở dạng thống kê, mô tả số liệu và có rất ít nghiên cứu về các yếu tố tác động tới quy mô thị trường TPDN. Đối với quy mô của thị trường TPDN sơ cấp, nghiên cứu của Nguyễn Hòa Nhân và cộng sự (2014) là nghiên cứu đầu tiên sử dụng mô hình phân tích hồi quy. Tuy nhiên, các yếu tố được phân tích trong nghiên cứu còn ít và tỷ lệ phần trăm giải thích sự biến động của biến phụ thuộc thông qua biến độc lập còn thấp. Có thể xem xét thêm nhiều yếu tố khác để xác định thêm các yếu tố tác động tới quy mô phát hành TPDN Việt Nam. Ngoài ra, nghiên cứu về quy mô thị trường TPDN Việt Nam chủ yếu tập trung vào quy mô phát hành. Quy mô giao dịch của TPDN trên thị trường thứ cấp là vấn đề ít được nghiên cứu. Đặc biệt là chưa có nghiên cứu nào sử dụng mô hình hồi quy để xem xét các yếu tố tác động tới quy mô giao dịch của TPDN Việt Nam. CHAPTER 2: THE REAL SITUATION OF VIETNAM CORPORATE BOND MARKET 2.1. Overview of Vietnam’s economic environment in the period of 2005 – 2017 2.2. Overview of Vietnam corporate bond market 2.3. The structure of Vietnam corporate bond market 2.3.1. The primary cororate bond market
Issuance volume The annual issuance volume greatly affects the size of the market. It can be seen that the annual issuance volume of corporate bond in Vietnam has tended to increase in the period of 2005 - 2018, from over VND 137 billion in 2005 to over VND 40,000 billion in 2018.
Terms of bond Terms of corporate bonds in Vietnam are very diverse, ranging from 1 to 20 years, depending on the target and issuance conditions. In cases of supplementing NWC, or debt restructuring, firms will raise short – term loans (<3 years). Loans for financing projects will be more than 3 years, or more than 5 years. In general, in the period of 2005 - 2018, the average term of corporate bonds of less than 3 years. 8 2.3.2. The secondary corporate bond market
Trading Corporate bonds are traded in 2 ways: (1) at stock exchange and (2) OTC
Size of market In the period of 2012 - 2015, the trading volume was only a few trillion, an average of over 3.8 trillion per year, in 2016 reached over 10 trillion. In the period of of 2016 - 2017, each year there were average of 7 trillion worth of corporate bonds trading on the secondary market. 2.4. Policies to develop corporate bond market in some other countries. CHAPTER 3: METHODOLOGY 3.1. Research framework and models 3.2. Methodology 3.2.1. Descriptive satistics 3.2.2. Regression analysis 3.2.2.1. Regression analysis for time series data
The stationary of time series data Augmented Dickey – Fuller (ADF) test is used to test the stationary of variables. If > at significant levels of 1%, 5% and 10%, variables are stationary at the corresponding significant levels (Levin et al., 2002).
Regression analysis Data of factors affect the issuance volume is quarterly data and it is called time series data. In which, t is used to refer time points in the series. Model of time series data is as follow (Nguyen Quang Dong and Nguyen Thi Minh, 2013): (a) Because (a) is a linear regression model, it will be estimated by the least square method (Ordinary Least Square - OLS). When the assumptions are satisfied, the estimation results are reliable.
Test for specification errors Test for specification errors: multicollinearity, multivariate normality, autocorrelation, homoscedasticity, functional form. 3.2.2.2. Regression analysis for panel dât
Regression methods 9 Data of factors affect to trading volume is panel data. According to Baltagi (2005), panel data model is as follow: (b) Model without : (c) Depending on , there are 3 estimation methods: Pooled OLS – POLS, Random Effects Model – REM and Fixed Effects Model – FEM. 3.3. Dataset 3.3.1. Dataset of factors affect to issuance volume (i) Data of issuance volume is secondary data that is collected quarterly from website Asianbondsonline.adb.org in the period of 2005 – 2018. (ii) The size of economy (Measurement as GDP), the openness of economy (Measurement as Experts) are secondary data that are collected from website finance.vietstock.vn. GDP is a non-linear data so it will be taken as a logarithm. (iii) Interest rate variability, the size of banking system (Measurement as domestic credit), exchange rate variability and foreign exchange reserves are secondary data that are collected quarterly from website data.imf.org in the period of 2005 – 2018. (iv) The stage of development of economy (Measurement as GDP per capita) is secondary data that is collected yearly from website data.worldbank.org in the period of 2005 – 2018 and is interpolated in quarterly data. GDP per capita is a non-linear data so it will be taken as a logarithm. (v) Creditors’ right is analyzed and calculated basing on Djankov et al. (2007) and Laws of bankruptcy 1993, 2004, 2014 of Vietnam. 3.3.2. Dataset of factors affect to trading volume (i) Trading volume is primary data that is collected monthly from HSX in the period of 2012 – 2017. (ii) Bond’s age is based on time of issuance. (iii) Credit rating of issuers is used to measure default risk of bonds because corporate bonds have not rated in Vietnam. Credit rating of issuers is secondary data that is collected from CIC. 10 (iv) Profit variability is measure as fluctuation of VWAP. (v) Stock’s trading volume is secondary data that is collected from finance.tvsi.com.vn/data 3.4. The process of testing 3.4.1. Factors affect to issuance volume (i) Test the stationary of time series data (ADF test) (ii) Estimate model by using OLS and test the model’s errors. (iii)Test for specification errors: multicollinearity (VIF), multivariate normality (Jarque – Bera test), autocorrelation (Breusch-Godfrey test), homoscedasticity (Breusch – Pagan – Godfrey test and White test), functional form (Ramsey test). (iv) If model has errors, remedies will be performed. 3.4.2. Factors affect to trading volume (i) Test the stationary of panel data (ii) Analyze the correlation between variables. (iii) Using Breusch – Pagan test to find out the most suitable estimation method for panel data. (iv) If prob > 0.1: using POLS. (v) If prob < 0.1: using Hausman test to test the correlation between and independent variables. If Prob 0.1: using REM CHAPTER 4: ANALYZE FACTORS THAT EFFECT TO THE SIZE OF VIETNAM CORPORATE BOND MARKET 4.1. Factors affect to primary corporate bond market 4.1.1. Variables and measurement Expected model: (1) In which: t = 1, 2, ..., 56 represent for 4 quarters/year of the period from 2005 to 2018. Table 4.1: Variables and Measurement of model (1) Model Variables Measurement Symbol Unit (1) Size Issuance volume in each quarter IVOL Billions VND (1) The size of economy Log(GDP) GDP % 11 Model Variables Measurement Symbol Unit (1) The openness of economy Exports EXPRT % (1) The stage of development of economy Log(GDP per capita). GDP per capita is interpolated in quarterly data PGDP (1) Interest rate variablility Leding interest rate – Deposit interest rate DRATE % (1) The size of banking system Domestic credit CREDIT Millions VND (1) Exchange rate variability Average exchange rate in quarter EXR VND/USD (1) Foreign exchange reserves Foreign exchange reserves FER Millions USD (1) Creditors’ right The index ranges from 0 (weak creditor rights) to 4 (strong creditor rights 2012 – 2014: 1 score; 2015 – 2017: 2 scores RIGHTS 0 – 4 scores 4.1.2. Descriptive statistics 4.1.3. Test factors affect to the issuance volume 4.1.3.1. The stationary of time series data Data of variables is tested the stationary on EVIEWS. The results show that IVOL is stationary at 0 level (10%), log(PGDP) is stationary at 0 level (1%), EXPRT is stationary at 2 level and other variables are stationary at 1 level. 4.1.3.2. Regression analysis of model (1.1)
The results of regression analysis of model (1.1) 12 Dependent Variable: IVOL Method: Least Squares Sample (adjusted): 2005Q2 2018Q4 Included observations: 55 after adjustments Variable Coefficient Std. Error t-Statistic Prob. C 1339.511 1675.374 0.79953 0.4281 LOG(GDP) 2.82E+02 1.35E+02 2.085928 0.0426(**) D(EXPRT) 0.017436 0.008629 2.020693 0.0492(**) LOG(PGDP) -483.992 367.7746 -1.316 0.1947 CREDIT 0.037872 0.011563 3.275279 0.002(***) DRATE -5.31E+01 4.83E+01 -1.09984 0.2771 EXR -0.051366 0.042415 -1.21106 0.2321 FER 1.47E-02 4.74E-03 3.109871 0.0032(***) RIGHTS 65.55346 94.01795 0.697244 0.4892 R-squared 0.53508 Adjusted R-squared 0.454225 S.E. of regression 150.381 Sum squared resid 1040264 Log likelihood -348.852 F-statistic 6.617725 Prob(F-statistic) 0.00001 (Source: Author’s calculations) In which: (*): significant at 10%, (**): significant at 5%, (***): significant at 1%. Prob(F-statistic) of model (1.1) = 0.00001 < 0.05
Test the errors of model (1.1) The results show that model (1.1) do not have errors except multicollinearity because VIF index of some variables are more than 10. So that, variable with high VIF index is removed off model (1.1). New model is as follow: 13 (1.2) 4.1.3.3. Regression analysis of model (1.2)
The results of regression analysis of model (1.2) Dependent Variable: IVOL Method: Least Squares Sample (adjusted): 2005Q2 2018Q4 Included observations: 55 after adjustments Variable Coefficient Std. Error t-Statistic Prob. C -470.421 9.64E+02 -4.88E-01 0.6279 LOG(GDP) 196.2759 119.4337 1.643388 0.107 D(EXPRT) 0.019231 0.008586 2.23984 0.0299(**) CREDIT 0.039318 0.0116 3.389591 0.0014(***) DRATE -15.8087 39.40295 -0.40121 0.6901 EXR -0.09919 0.022044 -4.49945 0.000(***) FER 0.010778 0.003688 2.922729 0.0053(***) RIGHTS 131.8318 80.00656 1.647762 0.1061 R-squared 0.517576 Adjusted R-squared 0.445726 S.E. of regression 151.5473 Sum squared resid 1079429 Log likelihood -349.868 F-statistic 7.203536 Prob(F-statistic) 0.000007 (Source: Author’s calculations) In which: (*): significant at 10%, (**): significant at 5%, (***): significant at 1%. Prob(F-statistic) of model (1.2) = 0.000007 < 0.05 Log(GDP), DRATE and RIGHTS have prob. > 0.1 and they are not significant at 10%.
Test errors of model (1.2) The results show that model (1.2) do not have errors. Because Prob. of DRATE is more than 0.1, this variable is removed off. New model is as follow: 14 (1.3) 4.1.3.4. Regression analysis of model (1.3)
The results of regression analysis of model (1.3) Dependent Variable: IVOL Method: Least Squares Date: 12/17/19 Time: 14:00 Sample (adjusted): 2005Q2 2018Q4 Included observations: 55 after adjustments Variable Coefficient Std. Error t-Statistic Prob. C -609.552 891.742 -0.68355 0.4975 LOG(GDP) 206.748 115.5235 1.789661 0.0798(*) D(EXPRT) 0.018589 0.008361 2.223195 0.0309(**) CREDIT 0.041407 0.010275 4.0298 0.0002(***) EXR -0.10087 0.021453 -4.70171 0.0000(***) FER 0.010688 0.003649 2.929367 0.0052(***) RIGHTS 143.6858 73.69839 1.949646 0.0571(*) R-squared 0.515924 Adjusted R-squared 0.455415 S.E. of regression 150.2169 Sum squared resid 1083126 Log likelihood -349.962 F-statistic 8.526337 Prob(F-statistic) 0.000003 (Source: Author’s calculations) In which: (*): significant at 10%, (**): significant at 5%, (***): significant at 1%. Prob(F-statistic) of model (1.3) = 0.000003 < 0.05
Test errors of model (1.3) The results show that model (1.3) do not have errors. The last model is as follow: IVOLt = - 609.552 + 206748Log(GDP)t + 0.018589D(EXPRT)t + 0.041407CREDIT – 0.10087EXRt + 0.010688FERt + 15 143.6858RIGHTSt (1.3) 4.1.3.5. Discuss the research’s results (i) The stage of development of the economy (PGDP) and the interest rate variability (DRATE) have no impact on the issuance volume of corporate bonds. (ii) The size of the economy (GDP), the openness of the economy (EXPRT) and the size of the banking system (CREDIT) have a positive impact on the issuance volume of corporate bonds. (iii) Exchange rate fluctuations (EXR) have a negative impact on the issuance volume. (iv) Foreign exchange reserves (FER) have a positive impact on the volume of corporate bond issuance. (v) Creditors' rights (RIGHTS) have a positive impact on issuance volume. 4.2. Factors affect to trading volume 4.2.1. Variables and measurement Expected models: (2.1) (2.2) (2.3) In which: i = 1, 2, ..., 28 (represent for 28 bonds), t = 1, 2, , 72 corresponds with 12 months/year in the period of 2012 – 2017. Table 4.4: Variales and measurement of model (2) Model Variables Measurement Symbol Unit (2) Trading volume (1) Number of transactions in month TIMES Transaction (2) Number of bonds are traded in month NBOND Bond (3) Turnover TOVER Billions VND (2) Issuance Issuance SIZE Thousands VND 16 volume volume in par value (2) Age Number of months since issuance AGE Qualitative variable AGE_1: more than 2 years AGE_2: in years AGE_3: Unreleased/Expired (2) Default risk Firm’s credit rating RATING Qualitative variable RATING_1: In rank A RATING_2: In rank B RATING_3: not rated (2) Profit variability Volume Weighted Average Price (VWAP) variability DVWAP % (2) Stock’s trading volume Value of stock traded in month SVOL Billions VND 4.2.2. Descriptive statistics 4.2.3. Test factors affect to trading volume 4.2.3.1. Test the staytionary of panel data Levin – Lin – Chu (LLC) test is used to test the stationary of panel data. The rusults show that SVOL is stationary at 0 level (1%) and other variables’ data are stationary at 1 level (1%). 4.2.3.2. Correlation analysis
The correlation between dependent variables The correlation between TIMES, NBOND and TOVER is quite tight because the correlation coefficient is approximately 0.7. So that, TIMES, NBOND and TOVER are suitable to measure the trading volume.
The correlation between independent variables The absolut of correlation coefficients between independent are less 17 than 0.7. This shows that model (2) does not have multicollinearity.
The correlation between independent variables and dependent variables The correlation between the independent variables and the dependent variables is very low due to the correlation coefficient is less than 0.7. 4.2.3.2. Results Table 4.11: Methods are used to estimate model (2) Test TIMES NBOND TOVER Breusch – Pagan Prob = 0.000 0.000 0.000 Hausman Prob = 0.000 0.0797 0.0412 Method FEM FEM FEM (Soure: Athour’s calculations) Table 4.12: Results of testing the R2 Dependent variables TIMES NBOND TOVER Methos FEM FEM FEM R2 - overall 0.3250 0.1274 0.1292 P - value Prob. = 0.0000 0.000 0.000 (Soure: Athour’s calculations) Table 4.13: Results of testing 3 models Variable TIMES NBOND TOVER Coefficient Prob Coefficient Prob Coefficient Prob SIZE 0 X 0 X 0 X AGE_1 3.073938 0.000(***) 578346.1 0.000(***) 58.74211 0.000(***) AGE_2 7.341274 0.000(***) 1133685.0 0.000(***) 117.7666 0.000(***) AGE_3 0 0 0 RATING_1 2.740408 0.000(***) 917450.6 0.000(***) 92.58299 0.000(***) RATING_2 3.860852 0.000(***) 1148857.0 0.000(***) 116.9805 0.000(***) RATING_3 0 0 0 DVWAP 0.0116864 0.265 6805.795 0.031(**) 0.7757046 0.016(**) SVOL – 0.0001556 0.183 – 136.1458 0.000(***) – 0.0132649 0.000(***) CONS – 4.063231 0.000 – 1071589.0 0.000 – 109.8196 0.000 (Soure: Athour’s calculations) In which: (*): significant at 10%, (**): significant at 5%, (***): 18 significant at 1%. Because all three models use the fixed effects model (FEM), the SIZE variable is removed from the models. Therefore, it is not possible to assess the impact of SIZE on the trading volume In 3 models, AGE_1, AGE_2, RATING_1 and RATING_2 are significant at 1% and have positive impact to dependent variables. The coefficient of these variables show that bonds are issued in 2 years and are rated in B category have bigger trading volume than others. In model (2.2) and model (2.3), SVOL variable has negative impact to NBOND and TOVER but DVWAP has positive impact to them. 4.2.3.3. Discuss the research’s results (i) Because SIZE did not change d