Capital structure and debt maturity structure of real estate construction investment and trading enterprises in Vietnam

llows: Firstly, it is noted real estate products have great value, which is an outstanding characteristics of the industry. Therefore, it is a must to identify corporate capital structure and debt maturity structure following business cycles of the real estate industry, which are likely to be longer than commercial and service ones. Secondly, real estate is considered as one of the most competitive industries. It is no doubt that its special products need a sufficiently large capital in order to launch projects. Thus, management experience, reputation in the market, and ability to operate and to penetrate the market are decisive factors for enterprises to compete top positions of the market. Thirdly, the ability to access funding in order to run real estate businesses is limited. The majority of capital is short-term loans from commercial banks. Fourthly, the real estate industry is governed by real estate institutions. The fact that the real estate market is quite young, plus real estate institutions are gradually improving with lots of constant changes and adjustments, is actually an indirect obstacle for businesses to implement capital structure and to decide structure of debt maturity so that corporate value can be increased in the market. 19 1.2.1.2 Characteristics of real estate businesses in Vietnam Enterprises in real estate industry have different characteristics compared to those in other fields: Firstly, it is size of the business. Large and reputable enterprises in the market are easier to access to low-cost capital sources and less subject to liquidity pressure. They are also able to increase business efficiency better than small and medium-sized enterprises. Secondly, it is type of the enterprise in real estate industry and business sector that decides real estate corporate capital structure and structure of debt maturity in Vietnam. Thirdly, investment cost of fixed assets is another feature to be mentioned. It is obvious that fixed assets of real estate enterprises are projects, and it certainly requires very huge costs. Fourthly, it is tax incentives. They show that corporate income tax also has a significant impact on capital structure and debt maturity structure of real estate businesses. 1.2.2 Current situation of capital structure, debt maturity structure and financial objectives of real estate construction investment and trading enterprises in Vietnam 1.2.2.1 Current situation of capital structure and debt maturity structure of real estate construction investment and trading enterprises in Vietnam 20 1.2.2.2 Current situation of financial objectives of real estate construction investment and trading enterprises in Vietnam CHAPTER 2 THEORETICAL FRAMEWORK OF CAPITAL STRUCTURE AND DEBT MATURITY STRUCTURE OF ENTERPRISES 2.1. THEORETICAL FRAMEWORK OF CAPITAL STRUCTURE AND DEBT MATURITY STRUCTURE OF ENTERPRISES 2.1.1 Theoretical basis of capital structure The thesis presents an overview of the applied theories: - Modilligiani and Miller theory; - Agency cost theory; - Trade-off theory (TOT); - Pecking Order theory (POT). 2.1.2 Theoretical basis of debt maturity structure The thesis presents an overview of the theories applied in the thesis. - Modilligiani and Miller theory; - Trade-off theory (TOT); 21 - Signaling theory; - Tax-based theory; - Matching theory. The mentioned theories of capital structure and debt maturity structure have different points of view; however, they do not replace but complement each other, contributing to a better decision of funding sources. 2.2 THEORETICAL REVIEW AND EMPIRICAL RESEARCHES ON CAPITAL STRUCTURE AND DEBT MATURITY STRUCTURE 2.2.1 Theoretical review of capital structure and debt maturity structure 2.2.2 Review of Empirical researches on capital structure and debt maturity structure 2.3 MODELS AND PROPOSED RESEARCH HYPOTHESIS 2.3.1 Research Hypothesis Based on the theoretical framework of capital structure and debt maturity structure, the thesis builds hypotheses to study the capital structure and debt maturity structure of listed real estate construction investment and trading enterprises listed in Vietnam in model 1 (capital structure) and model 2 (debt maturity structure)as follows: 22 Table 2.2 Review of research hypotheses Variable Theory Hypothesis Emperial research Model 1 Model 2 Interest term structure (TERM) Modilligiani and Miller theory; Tax- based theory (+) (+) Antoniou & et al (2006), Cai & et al (2008), Cook & Tang (2010), Correia & et al (2014), Deesomsak & et al (2009) Economic growth (GDP) Characteristics of the market (+) (+) Lemma and Negash (2012), Wang & et al (2010), Deesomsak & et al (2009), Fan & et al (2012), Alves & Francisco (2015). Inflation (INF) (+) (+) Deesomsak & et al (2009); Wang & et al (2010), Fan & et al (2012). Financial development (FD) (+) (+) Krich & Terra (2012), Fan & et al (2012), Lemma &Negash (2012), Alves 23 and Francisco (2014), Correia & et al (2014). Country Policy and Institution Assessment (CPIA) (+) (+) Prowse (1990), Demirguc - Kunt & Maksimovic (1999), Deesomsak & et al (2004), Öztekin & Flanery (2012), Fan & et al (2012), Lemma & Negash (2012), Krich & Terra (2012), Duan & et al (2012), Alves & Francisco (2015), Bernardo & et al (2018) Return of Total Equity (ROE) Signaling theory (-) (-) Deesomsak & et al (2009), Fan & et al (2012), Kirch & Terra (2012), Mateurs& Terra (2013), Correia & et al (2014) 24 Liquidation (LIQ) Signaling theory (+) (+) Antoniou & et al (2006), Cai & et al (2008), Deesomsak & et al (2009), Mateurs& Terra (2013) Business risk (RISK) TOT theory; Agency cost theory; Signaling theory (-) (+) Antoniou & et al (2006), Cai & et al (2008), Deesomsak & et al (2009), Lemma &Negash (2012) Firm’s size (SIZE) Agency cost theory (+) (+) Fan & et al (2012), Kirch & Terra (2012), Correia & et al (2014) Growth oportunity (GRO) Agency cost theory (-) (-) Cai & et al (2008), Kirch & Terra (2012), Lemma & Negash (2012) Assets Structure (TANG) POT theory; Matching theory (+) (+) Kirch & Terra (2012), Mateurs& Terra (2013) 25 Assets Maturity Rate (AMR) Matching theory (+) (+) Cai & et al (2008), Wang & et al (2010), Lemma & Negash (2012), Correia & et al (2014) Corporate income tax (TAX) Modilligiani and Miller theory; Tax-based theory. (+) (+) Cai & et al (2008), Kirch & Terra (2012), Fan & et al (2012) Mateurs & Terra (2013), Correia & et al (2014) Capital structure (TDR) Signaling theory (+) Dang (2011), Mateurs & Terra (2013), Correia & et al (2014) Debt maturity structure (LDR) Signaling theory (+) Antoniou & et al (2006), Cai & et al (2008), Deesomsak & et al (2009), Kirch & Terra (2012), Lemma &Negash (2012) Note: (+) Positive impact, (-) Negative impact Source: Author 26 Table 2.3 Measurement of variables in the models Variales Abbr. Des. Variable measurements Sources TDR Capital structure Total debts Total assets Hanoi Stock Exchange website ( and Ho Chi Minh city Stock Exchange ( LDR Debt maturity structure Long − term debts Total debts TANG Assets structure Net Fixed Assets Total assets LIQ Liquidity Short − term Liabilities Current Liabilities SIZE Firm size Ln(Total assets on the book) 27 AMR Assets Maturity ( Curent assets Curent assets + Net Fixed Assets × Curent assets Cost of goods sold ) + ( Net Fixed Asset Curent assets + Net Fixed Asset × Net Fixed Assets Depreciation ) GRO Growth opportunity Liability + Market price of capital Total assets RISK Busines risks | EBITt − EBITt−1 EBITt−1 | − average of | EBITt − EBITt−1 EBITt−1 | ROE Return of Equity Profit before taxesit Average equityit 28 TAX Corporate income tax Corporate income taxesit Profit before taxesit TERM Interest term structure Government bond yield (5-year term) - Treasury bill yield (3-month term) Datastream GDP Economic growth GDPt − GDPt−1 GDPt−1 IMF INF Inflation Consumer Price Index CPI (%) IMF FD Financial Development Financial development index (from 0 to 1 point) IMF CPIA Country Policy and Institution Assessment Institution Quality Index (from 0 to 6 points) Worldbank Source: Author 29 2.3.2 Proposed research models 2.3.2.1 Models studying on factors impacting capital structure and debt maturity structure; and adjustment speed of capital structure and debt maturity structure of real estate construction investment and trading enterprises in Vietnam Model 1 studies on factors impacting capital structure and debt maturity structure The thesis is based on the research model of Ozkan (2001), Fan & et al (2010), Ramzi &Tarazi (2013), Mateurs & Terra (2013), Nagano (2013), Alves & Francisco (2015) to decide factors impacting capital structure and debt maturity structure; and to estimate capital structure adjustment speed towards the target capital strutureto be compatible with real estate constructure investment and trading enterprises. TDRit = β0 + β1TDRit-1 + β2LDRit + βxXit + βzZt + еit (1a) According toOzkan (2001), Fan & et al (2010), Ramzi &Tarazi (2013), Mateurs & Terra (2013), Alves & Francisco (2015), model of the target capital structure TDR*itis defined as below: TDR*it = β0 + βxXit + βzZt + еit (1b) Identification of Capital structure towards the target capital structure threshold is carried out, then first-order lagged variable of the capital structure variable is included in the model as the adjustment.According to TOT theory, determining adjustment speed of capital structure is based on the adjustment coefficient λ.Therefore, 30 the model determining the adjustment speed of capital structure is rewritten is as follows: TDRit - TDRit-1 = λ (TDR*it - TDRit-1) (1c) The combination of (1b) and (1c) for capital structure adjustment towards the target capital structure threshold is as the belows: TDRit = λβ0 + (1- λ) TDRit-1 + λβxXit + λβzZt + еit (1d) Model 2 studies on factors impacting capital structure and adjustment speed of debt maturity structure Regarding research models of factors impacting debt maturity structure of real estate construction investment and trading enterprises in Vietnam, the thesis is based on research models of Ozkan (2000); Antoniou & et al (2006), Cai & et al (2008), Deesomsak & et al (2009), Terra (2011), Fan & et al (2012), Krich & Terra (2012), Alcock & et al (2014), Alves & Francisco (2015), Hussain & et al (2018). LDRit = β0 + β1LDRit-1 + β2TDRit + βxXit + βzZt + еit (2a) According to Antoniou & et al (2006), Alcock & et al (2014), Alves & Francisco (2015), Hussain & et al (2018), the model of the target debt maturity structure LDR*itis defined asbelow: LDR*it = β0 + βxXit + βzZt + еit (2b) Similarly, following Trade-Off Theory, the identification of adjustment speed ofdebt maturity structure is based on λ as shown: LDRit - LDRit-1 = λ (LDR*it - LDRit-1) (2c) 31 From (2b) and (2c), adjustment speed of debt maturity structure towards the target threshold is written as below: TDRit = λβ0 + (1- λ) TDRit-1 + λβxXit + λβzZt + еit (2d) 2.3.2.2 Models determining the target capital structure and debt maturity structure of real estate construction investment and trading enterprises in Vietnam Model 3: The threshold regression model determining the target capital structure ROEit = β0 + β1TDRitI(TDRit≤ℽ)+ β2TDRitI(TDRit>ℽ) + β3LDRit + βxXit+ βzZt+еit (3) Model 4: The threshold regression model determining the target debt maturity structure ROEit = β0 + β1LDRitI(LDRit≤ℽ)+ β2LDRitI(LDRit>ℽ) + β3TDRit + βxXit + βzZt+еit (4) CHAPTER 3: RESEARCH METHODOLOGY 3.1 RESEARCH DESIGN 3.2 DESCRIPTIONS OF RESEARCH DATA The thesis sample focuses on data of real estate businesses listed on Vietnam's stock market collected from audited consolidated financial statements (balance sheet and income statements) within 10 years from 2008 to 2017, published on the websites of Hanoi Stock 32 Exchange at as well as the transaction office Ho Chi Minh City Stock Exchange at According to Hair (1998), the sample size must follow empirical principles, meaning the maximum experimental research model has 10 variables. From that, the sample must be at least 5 times of the number of present variables in the model, meaning that the minimum sample size is from 50 observations (10 x 5 = 50 observations). Thus, financial data of 70 real estate construction investment and trading enterprises in Vietnam, forming the total number of 70 x 10 = 700 observations, meets the requirements in terms of sample size conformity. Information related to macro factors such as growth rate, inflation index, and financial development index is collected from IMF website; others relating to profitability of government bonds and institutional factors is sourced from the Datastream and Worldbank's website during the period of 2008 and 2017. 3.3 RESEARCH METHODS 3.3.1 Quantitive research method 3.3.1.1 Research methods for models analyzing factors impacting capital structure and debt maturity structure; and adjustment speed ofcapital structure and debt maturity structure of real estate construction investment and trading enterprises in Vietnam 33 The thesis applies the System - Generalized method of moments (Sys-GMM) introduced by Arellano & Bover (1981; 1995). It is obvious that these estimation results will be more efficient than processing individual equations because the variance change is avoided, and autocorrelation or endogenous problems are addresed in the selected model. The Sys-GMM estimation method needs two basic tests: Arellano- Bond - AR serial correlation test and Sargan/Hansen test. (i) Testing of Arellano-Bond of serial autocorrelation with null hypothesis H0 is applied fordifferential balances with no autocorrelation.Testing the AR process (1) in the first order difference usually rejects null hypothesis H0. Therefore, the AR(2) test is more important because it checks autocorrelation at several levels. Notably, the results of AR(2) test shows that there is no correlation found in the model if Prob ≥ 0.05; (ii) Sargan/Hansen testis used for testing over-identifying restrictions in a statistical model. Sargan/Hansen test with hypothesis H0 tool variables are exogenous, meaning that it is not correlated with errors in the model. Sargan/Hansen statistical results with Prob> chi2 ≥0.05 can be interpreted that endogenous variables and defects have been removed, which means estimation model is in conformity. 34 3.3.1.2. Reasearch methodsfor the model identifying the target capital structure and debt maturity structure for real estate construction investment and trading enterprises in Vietnam In order to identify the target capital structure and debt maturity structure, the author ultilizes model of Hansen’s (1999) Panel Threshold Regression (PTR). 3.3.2. Qualitative research method A ten-question interview in person is carried out to collect opinions of ten experts in real estate industry (General Directors/ Directors, Chief Financial Officers, Chief Accountants of Real Estate Enterprises) to analyze real estate relevant issues and to interpret further meanings of the research results, . 3.4 RESEARCH PROCEDURES CHAPTER 4: RESEARCH RESULTS AND DISCUSSIONS ON CAPITAL STRUCTURE AND DEBT MATURITY STRUCTURE OF REAL ESTATE CONSTRUCTION INVESTMENT AND TRADING ENTERPRISES IN VIETNAM 4.1 DESCRIPTIVE STATISTICS AND MULTI- COMMUNICATION TESTING 4.1.1. Descriptive statistics 35 Table 4.1: Statistics results of descriptive Variables Variable Observation Average value Minimum value Maximum value Standard deviation TDR 700 0.5357 0.0091 0.9518 0.1988 LDR 700 0.2960 0.0000 0.9583 0.2485 ROE 700 0.2538 -1.7543 122.2656 4.6220 RISK 700 2.6193 -25.1318 252.4167 16.1914 LIQ 700 2.7903 0.3249 109.0462 6.0520 SIZE 700 14.1041 8.0580 19.1805 1.3757 GRO 700 0.9110 0.2426 2.9319 0.3405 TANG 700 0.0914 0.0001 0.6880 0.1159 AMR 700 19.5434 0.2003 749.0189 57.1706 TAX 700 0.2110 0.0000 4.3333 0.2470 INF 700 8.5747 0.8786 23.1163 6.6996 GDP 700 6.0076 5.2474 6.8100 0.5273 FD 700 0.2670 0.1871 0.3807 0.0638 CPIA 700 3.7756 3.7123 3.8214 0.0328 TERM 700 1.2346 -0.9120 2.4620 1.1024 Source: The Author’s calculation and combination The statistical results in Table 4.1 reveal that the average value of capital structure (TDR) is 53.57%, meaning that the average total debts to total assets of 70 real estate construction investment and trading enterprises listed on Vietnam's stock market during the research period from 2008-2017 is 53.57%. The average value of the debt maturity structure (LDR) is 29.60%, which shows that real estate construction investment enterprises in Vietnam use long-term debts to total debts of 29.6%. Although real estate businesses need to use long-term capital to finance long-term assets, it is indicated that 29.6% ofreal estate enterprises in Vietnam uses long-term debts. This proportion is quite low compared to Japan - 57.5% (Nagano, 2013), 36 America - 38.3% (Etudaiya-Muhtar & ctg, 2017). These results prove that real estate construction investment and trading enterprises still prefer short-term debts to long-term ones. 4.1.2 Multicollinear test Correlation matrix analysis resultsamong independent variables in the model does not show a great possibility of multi-collinear among independent variables of the model. In addition, the absolute value of the correlation coefficient among variables in the model results less than 10, following low degree of collinearity between variables, which eventually does not affect the estimation results of the regression model. 4.2 ESTIMATION RESULTS OF FACTORS IMPACTING CAPITAL STRUCTURE AND DEBT MATURITY STRUCTURE OF CONSTRUCTION INVESTMENT AND TRADING ENTERPRISES IN VIETNAM 4.2.1 Estimation results of factors impacting capital structure of real estate construction investment and trading enterprises in Vietnam 37 Table 4.2 Regression results of factors impacting capital structure Variables Pooled OLS FEM REM FGLS Sys-GMM LDR 0.103*** 0.0944*** 0.1000*** 0.0403 0.158*** [0.001] [0.000] [0.000] [0.054] [0.000] ROE 0.0013 0.00458*** 0.00386*** 0.00298*** -0.099*** [0.388] [0.000] [0.000] [0.000] [0.000] RISK -0.00030 0.00028 0.00026 -0.00027* 0.00163 [0,483] [0.272] [0.311] [0.048] [0.456] LIQ -0.0093*** -0.0040*** -0.0049*** -0.0074*** -0.00382* [0.000] [0.000] [0.000] [0.000] [0.015] SIZE 0.0205*** 0.0681*** 0.0512*** 0.0428*** 0.00245** [0.000] [0.000] [0.000] [0.000] [0.006] GRO 0.121*** 0.153*** 0.140*** 0.218*** 0.0391 [0.000] [0.000] [0.000] [0.000] [0.058] TANG 0.0193 -0.128* -0.112* -0.0598 0.0848 [0.761] [0.014] [0.029] [0.146] [0.100] AMR -0.00015 -0.00009 -0.00011 0.00002 0.00022 [0.222] [0.245] [0.180] [0.665] [0.468] TAX 0.0433 0.0111 0.00973 -0.00411 0.174*** [0.102] [0.499] [0.556] [0.731] [0.000] INF 0.00271 0.00502* 0.00432* 0.00189 0,00073 [0.452] [0.015] [0.037] [0.168] [0.475] GDP -0.0248 -0.0327** -0.0294** -0.0217*** -0.0151 [0.168] [0.002] [0.005] [0.000] [0.050] FD -0.178 -0.246*** -0.220** -0.163** -0.108 38 [0.142] [0.001] [0.002] [0.002] [0.062] CPIA -0.0801 0.179 0.0802 0.058 -0.626*** [0.800] [0.332] [0.664] [0.674] [0.000] TERM 0.00701 0.0176 0.0138 0.0105 0.0245 [0.750] [0.160] [0.273] [0.240] [0.510] L.TDR 0.7513*** [0.000] Intercept 0.592 -1.049 -0.442 -0.309 2.583*** [0.631] [0.169] [0.555] [0.571] [0.000] Observation number 700 700 700 700 630 Adj R-squared 0.1816*** 0.1040*** 0.1201*** Chow test [0.000] Breusch - Pagan test [0.000] Hausman test [0.948] Heteroskedasticity Test [0.000] Autocorrelation test [0.000] Wald test [0.000] AR (2) [0.527] Sargan test [0.617] Hansen test [0.465] *,**,***include 10%, 5% and 1% statistical significance levels respectively; [] is p-value Source: Author’s from STATA 39 The test result showing that AR (2) has value Prob> z = 0.527 > 0.05, accepting hypothesis H0, implies that no serial correlation is found in the model. Sargan test result with Prob value > chi2 = 0.617 > 0.05 and Hansen test with Prob value > chi2 = 0.465 > 0.05, accepting hypothesis H0, means that the endogenous variable effect has been eliminated, so no correlation with errors is found in the model. Based on the outcomes of AR (2) and Sargan/ Hansen tests, it is concluded that the model applying Sys-GMM method estimating factors affecting capital structure of real estate construction investment and trading enterprises Vietnam is appropriate. 4.2.2 Estimation results of factors impacting debt maturity structure of construction investment and trading bussinesses in Vietnam Similarly, the testing result of AR(2) with the value Prob> z = 0.364 > 0.05, implies that there is no serial correlation in the model; Sargan test with Prob value > chi2 = 0.775 > 0.05 and Hansan test with Prob value > chi2 = 0.511 > 0.05 can be interpreted that the endogenous variable effect has been removed, and the model does not correlate with errors. Therefore, the author comes to conclusion that by Sys-GMM method, the model estimating factors affecting the debt maturity structure is in conformity. 40 Table 4.3: Estimation outcomes of factors impacting debt maturity structure Variables Pooled OLS FEM REM FGLS Sys-GMM TDR 0.162*** 0.226*** 0.220*** 0.0157 0.175*** [0.001] [0.000] [0.000] [0.716] [0.000] ROE 0.00148 0.000807 0.00047 0.0025 0.0896 [0.432] [0.599] [0.748] [0.179] [0.131] RISK 0.000647 -0.00004 0.00004 -0.000013 0.00023** [0.229] [0.925] [0.913] [0.964] [0.001] LIQ 0.0103*** 0.0083*** 0.0081*** 0.0097*** 0.00941*** [0.000] [0.000] [0.000] [0.000] [0.000] SIZE 0.0668*** 0.0949*** 0.0811*** 0.0826*** 0.0282*** [0.000] [0.000] [0.000] [0.000] [0.000] GRO 0.0226 -0.0251 -0.0117 0.00049 -0.1418 [0.383] [0.613] [0.767] [0.987] [0.113] TANG 0.313*** 0.139 0.166* 0.172** 0.280* [0.000] [0.087] [0.031] [0.009] [0.049] AMR -0.00016 -0.00021 -0.00022 -0.00018* -0.00039 [0.320] [0.082] [0.072] [0.034] [0.128] TAX -0.00959 -0.0188 -0.020 -0.0242 -0.4373 [0.772] [0.460] [0.425] [0.219] [0.110] INF 0.0012 0.00323 0.00252 -0.00044 0.0084** [0.790] [0.312] [0.426] [0.833] [0.002] GDP 0.0253 0.00897 0.0131 -0.00365 0.02594 [0.261] [0.576] [0.409] [0.686] [0.205] FD 0.188 0.0054 0.0455 -0.0647 0.2001 41 [0.214] [0.961] [0.677] [0.412] [0.070] CPIA 0.608 0.825** 0.745** 0.392 -0.457*** [0.123] [0.004] [0.008] [0.059] [0.001] TERM 0.032 0.0435* 0.0397* 0.00973 0.0516*** [0.245] [0.025] [0.039] [0.476] [0.000] L.LDR 0.7408*** [0.000] Intercept -3.356* -4.422*** -3.960*** -2.413** -2.259*** [0.029] [0.000] [0.000] [0.003] [0.000] Observation number 700 700 700 700 630 Adj R-squared 0.1813*** 0.1946*** 0.2034*** Chow test [0.000] Breusch - Pagan test [0.000] Hausman test [0.245] Heteroskedasticity Test [0.000] Autocorrelation test [0.000] Wald test [0.000] AR(2) [0.364] Sargan test [0.756] Hansen test [0,511] *,**,***include 10%, 5% and 1% statistical significance levels respectively; [] is p-value Source: Author’s from STATA 42 4.3. Research outcomes of adjustment speed of capital structure and debt maturity structure towards the target capital structure and debt maturity structure threshold of real estate construction investment and trading enterprises in Vietnam 4.3.1 Adjustment speed of capital structure and debt maturity structure of real estate construction investment and trading enterprises in Vietnam Capital structure of real estate construction investment and trading enterprises in Vietnam is dynamic and its adjusment speed is of 24.87%. 4.3.2 Adjustment speed of debt maturity structure of real estate construction investment and trading enterprises in Vietnam Debt maturity structure of real estate construction investment and trading enterprises in Vietnam is dynamic and its adjustment speed of is 25.9%. 4.4 Correlation between capital structure and debt maturity structure Total debt rate (TDR) has the same direction of impact with debt maturity structure, which is in line with