The dependence between international crude oil price and vietnam stock market nonlinear cointegration test approach

MINISTRY OF EDUCATION AND TRAINING UNIVERSITY OF ECONOMICS OF HO CHI MINH CITY HÀ THỊ NHƯ PHƯƠNG THE DEPENDENCE BETWEEN INTERNATIONAL CRUDE OIL PRICE AND VIETNAM STOCK MARKET NONLINEAR COINTEGRATION TEST APPROACH ECONOMIC MASTER THESIS Ho Chi Minh City -2015 MINISTRY OF EDUCATION AND TRAINING UNIVERSITY OF ECONOMICS OF HO CHI MINH CITY HÀ THỊ NHƯ PHƯƠNG THE DEPENDENCE BETWEEN INTERNATIONAL CRUDE OIL PRICE AND VIETNAM STOCK MARKET NONLINEAR COINTEGRATION TEST APPROACH Major: FINANCE - BANKING Code: 60340201 ECONOMIC MASTER THESIS INTRUCTOR: Assoc Prof NGUYỄN THỊ NGỌC TRANG Ho Chi Minh City -2015 COMMITMENT I commit that the economic master thesis titling “the dependence between international crude oil price and Vietnam stock market: Nonlinear cointegration test approach” was made by myself with the direction of Associate Professor Nguyen Thi Ngoc Trang The study’s results are truthful and data was collected from the credible sources such as: Ho Chi Minh City stock exchange, Energy Information Administration, the State Bank of Vietnam and General Statistics Office of Vietnam th Ho Chi Minh City, October 28 , 2015 Author HA THI NHU PHUONG TABLE OF CONTENT SUB TITLE PAGE COMMITMENT TABLE OF CONTENT LIST OF ABBREVIATIONS LIST OF TABLES LIST OF FIGURES Abstract .1 Introduction 2 Literature Review 2.1 Literature Review 2.1.1 The relationship between crude oil price and stock market .7 2.1.1.1 Negative effect from crude oil price to stock market 2.1.1.2 Positive effect from crude oil price to stock market 2.1.1.3 Insignificant nexus between oil price and stock market 11 2.1.1.4 The imperial evidences about the relationship between oil prices and Vietnam stock market 12 2.1.2 The relationship between stock market and exchange rate 13 2.2 Overview about Vietnam stock market, oil sector and exchange rate regime 17 2.2.1 Vietnam stock market 17 2.2.2 Oil section 19 2.2.3 Exchange regime 22 Data and research methodology 25 3.1 Data 25 3.2 Methodology 30 3.2.1 Gregory and Hansen Test - GH test 30 3.2.2 Toda-Yamamoto (TY) version of Granger non-causality test 32 3.2.3 Error Correction Model 34 Researching result 36 4.1 Descriptive statistics 36 4.2 Unit root test 41 4.3 Gregory and Hansen Test-GH test 45 4.4 TY procedure of Granger non–causality test 46 4.5 Error correction model 49 Conclusion 51 References Appendices LIST OF ABBREVIATIONS Abbreviation bbl/d ECM GCC GDP HNX HOSE IOCs LNG MSCI OPEC P5+1 PVEP PVN SBV SVAR Tcf The US TVTP VAR VECM LIST OF TABLES Table 3.1 Variable descriptions and sources: 30 Table 4.1 Descriptive statistic of three variables, exchange rate, crude oil price and VN index for the entire sample 38 Table 4.2 Descriptive statistic of three variables, exchange rate, crude oil price and VN index for four phases 40 Table 4.3 Unit root test result for entire sample 43 Table 4.4 Unit root test result for four phases: 44 Table 4.5 Threshold cointegration results 45 Table 4.6 Critical values of GH test with significant level at 5% and regressors 45 Table 4.7 TY version of Granger non–causality tests 48 Table 4.8 Error correction model 49 LIST OF FIGURES Figure 1.1 Global crude oil and petroleum liquids consumption, supply and inventory in 2014 and 2015 (Source: Energy Information Administration) .2 Figure 1.2 Crude oil export revenues and productions from 2009 to 08 months of 2015 (source: General Statistics Office of Vietnam) Figure 2.1 Vietnam Stock market capitalization to GDP (%) from 2004-2014 (Source Federal Reserve Economic Data) 18 Figure 2.2 Proportion of sectoral market capitalization in 2015 (source: HOSE website) 19 Figure 2.3 Average interbank exchange rates from 2006 to 2015 23 Figure 3.1 Graphical presentation of the series for first phase 26 Figure 3.2 Graphical presentation of the series for the second phase 27 Figure 3.3 Graphical representation of the third phase 28 Figure 3.4 Graphical representation of the fourth phase 29 Figure 3.5 Graphical representation of the entire sample 29 Abstract This paper investigates the relationship between crude oil prices and Vietnam stock prices by using the daily data in the period from 01/03/2006 to 08/31/2015 The data is divided into four phases, corresponding to two important events, the financial crisis in 2007 and the significant decline of crude prices from the third quarter of 2014 The research methods employed are the threshold cointegration test of Gregory and Hansen (1996), TY procedure for Granger non-causality proposed Toda- Yamamoto (1995) and Error correction model (Granger, 1987) The results show that there exists a long run relationship between crude oil prices and Vietnam stock market in the entire sample; however, there is no cointegration between these variables in all four phases There is evidence that crude oil prices unidirectionally affect stock prices in the entire sample and in the second and third phase; and the crude oil price variable is an exogenous one In the last phase attached with the decline of crude oil prices, no evidence in statistic shows that the oil prices Granger cause to stock prices It likely proves that volatilities of world crude oil prices can affect negatively or positively to profit outlooks of the listed companies on Vietnam stock market, and there is a balance between benefits and damages in this period ECM model indicates that oil prices and stock prices have a positive relationship in short term, and the speed of adjustment of stock price to return the equilibrium state after a shock is slow around 0.25% These findings also have an important policy implication that helps the government intercept the market to reduce the negative effect from the energy shocks in general and oil price shocks in particular Those are to pay more attention to domestic production and trade revenues to get more stable budget, research the alternative energy and enhance international cooperation in the energy sector Key words: Oil price, stock market, threshold cointergration Introduction The oil crude prices has fallen less than $50 per barrel, about 50% of August 2015 The main reason is the oil supply more than demand (see the figure 1) Growing oil inventories and supply typically put downward pressure on near-term prices The United States discovered and applied the new oil drill technology, called “shale oil revolution” This pushes the oil production is near 10 million barrels per day So that it can be offset the substantial oil supply disruption in the Organization of the Petroleum Exporting Countries (OPEC) However, the resumption of significant Libyan oil production, combined with the weakening outlook for global oil demand, the large economies in the word such as China, Russia, Europe area show not good performances about industrial production and expectation for economic growth Figure 1.1 Global crude oil and petroleum liquids consumption, supply and inventory in 2014 and 2015 (Source: Energy Information Administration) On July 14, the P5+1 (the five permanent members of the United Nations Security Council and Germany) and Iran announced an agreement that could result in relief from United States and European Union nuclear-related sanctions (which include some oil-related sanctions) If the agreement is implemented and sanctions relief occurs, it will put additional Iranian oil supplies on a global market that has already Null Hypothesis: LVNI has a unit root Exogenous: Constant Lag Length: (Automatic - based on SIC, maxlag=21) Augmented Dickey-Fuller test statistic Test critical values: *MacKinnon (1996) one-sided p-values Augmented Dickey-Fuller Test Equation Dependent Variable: D(LVNI) Method: Least Squares Date: 09/10/15 Time: 08:49 Sample (adjusted): 1/07/2010 6/30/2014 Included observations: 1074 after adjustments Variable LVNI(-1) D(LVNI(-1)) D(LVNI(-2)) C R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) Null Hypothesis: D(LVNI) has a unit root Exogenous: Constant Lag Length: (Automatic - based on SIC, maxlag=21) Augmented Dickey-Fuller test statistic Test critical values: *MacKinnon (1996) one-sided p-values Augmented Dickey-Fuller Test Equation Dependent Variable: D(LVNI,2) Method: Least Squares Date: 09/10/15 Time: 08:50 Sample (adjusted): 1/07/2010 6/30/2014 Included observations: 1074 after adjustments Variable D(LVNI(-1)) D(LVNI(-1),2) C R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) 2.4.2 GH test THE GREGORY-HANSEN COINTEGRATION TEST MODEL 4: Regime Shift ADF Procedure t-stat Lag Break -5.597777 1.000000 2/08/2011 Phillips Procedure Za-stat Za-break Zt-stat Zt-break -52.31178 2/08/2011 -5.165793 2/08/2011 2.4.3 TY procedure VAR Lag Order Selection Criteria Endogenous variables: LVNI LEX LOIL Exogenous variables: C Date: 09/10/15 Time: 08:54 Sample: 1/04/2010 6/30/2014 Included observations: 1069 Lag * indicates lag order selected by the criterion LR: sequential modified LR test statistic (each test at 5% level) FPE: Final prediction error AIC: Akaike information criterion SC: Schwarz information criterion HQ: Hannan-Quinn information criterion VAR Residual Serial Correlation LM Tests Null Hypothesis: no serial correlation at lag order h Date: 09/10/15 Time: 08:57 Sample: 1/04/2010 6/30/2014 Included observations: 1075 Lags 10 11 12 Probs from chi-square with df Date: 09/10/15 Time: 08:58 Sample (adjusted): 1/07/2010 6/30/2014 Included observations: 1074 after adjustments Trend assumption: Linear deterministic trend Series: LVNI LEX LOIL Lags interval (in first differences): to Unrestricted Cointegration Rank Test (Trace) Hypothesized No of CE(s) None * At most At most Trace test indicates cointegrating eqn(s) at the 0.05 level *denotes rejection of the hypothesis at the 0.05 level **MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegration Rank Test (Maximum Eigenvalue) Hypothesized No of CE(s) None * At most At most Max-eigenvalue test indicates cointegrating eqn(s) at the 0.05 level *denotes rejection of the hypothesis at the 0.05 level **MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegrating Coefficients (normalized by b'*S11*b=I): LVNI 8.817486 7.475714 16.87325 Unrestricted Adjustment Coefficients (alpha): D(LVNI) D(LEX) D(LOIL) Normalized cointegrating coefficients (standard error in parentheses) Adjustment coefficients (standard error in parentheses) D(LVNI) -0.000156 (0.00147) D(LEX) 0.000983 (0.00036) D(LOIL) -0.007536 (0.00176) Cointegrating Equation(s): Log likelihood 13542.42 Normalized cointegrating coefficients (standard error in parentheses) Adjustment coefficients (standard error in parentheses) D(LVNI) D(LEX) D(LOIL) VAR Granger Causality/Block Exogeneity Wald Tests Date: 09/10/15 Time: 08:59 Sample: 1/04/2010 6/30/2014 Included observations: 1074 Dependent variable: LVNI Excluded LEX LOIL All Dependent variable: LEX Excluded LVNI LOIL All Dependent variable: LOIL Excluded LVNI LEX All 2.5 Phase IV (07/02/2014-08/31/2015) 2.5.1 Unit root test Null Hypothesis: LEX has a unit root Exogenous: Constant Lag Length: (Automatic - based on SIC, maxlag=15) Augmented Dickey-Fuller test statistic Test critical values: *MacKinnon (1996) one-sided p-values Augmented Dickey-Fuller Test Equation Dependent Variable: D(LEX) Method: Least Squares Date: 09/10/15 Time: 09:10 Sample (adjusted): 7/02/2014 8/31/2015 Included observations: 284 after adjustments Variable LEX(-1) C R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) Null Hypothesis: D(LEX) has a unit root Exogenous: Constant Lag Length: (Automatic - based on SIC, maxlag=15) Augmented Dickey-Fuller test statistic Test critical values: *MacKinnon (1996) one-sided p-values Augmented Dickey-Fuller Test Equation Dependent Variable: D(LEX,2) Method: Least Squares Date: 09/10/15 Time: 09:11 Sample (adjusted): 7/03/2014 8/31/2015 Included observations: 283 after adjustments Variable D(LEX(-1)) C R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) Null Hypothesis: LOIL has a unit root Exogenous: Constant Lag Length: (Automatic - based on SIC, maxlag=15) Augmented Dickey-Fuller test statistic Test critical values: *MacKinnon (1996) one-sided p-values Augmented Dickey-Fuller Test Equation Dependent Variable: D(LOIL) Method: Least Squares Date: 09/10/15 Time: 09:12 Sample (adjusted): 7/03/2014 8/31/2015 Included observations: 283 after adjustments Variable LOIL(-1) D(LOIL(-1)) C R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) Null Hypothesis: D(LOIL) has a unit root Exogenous: Constant Lag Length: (Automatic - based on SIC, maxlag=15) Augmented Dickey-Fuller test statistic Test critical values: *MacKinnon (1996) one-sided p-values Augmented Dickey-Fuller Test Equation Dependent Variable: D(LOIL,2) Method: Least Squares Date: 10/31/15 Time: 12:15 Sample (adjusted): 7/03/2014 8/31/2015 Included observations: 283 after adjustments Variable D(LOIL(-1 C R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) Null Hypothesis: LVNI has a unit root Exogenous: Constant Lag Length: (Automatic - based on SIC, maxlag=15) Augmented Dickey-Fuller test statistic Test critical values: 10% level *MacKinnon (1996) one-sided p-values Augmented Dickey-Fuller Test Equation Dependent Variable: D(LVNI) Method: Least Squares Date: 09/10/15 Time: 09:12 Sample (adjusted): 7/03/2014 8/31/2015 Included observations: 283 after adjustments Variable LVNI(-1) D(LVNI(-1)) C R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) Null Hypothesis: D(LVNI) has a unit root Exogenous: Constant Lag Length: (Automatic - based on SIC, maxlag=15) Augmented Dickey-Fuller test statistic Test critical values: *MacKinnon (1996) one-sided p-values Augmented Dickey-Fuller Test Equation Dependent Variable: D(LVNI,2) Method: Least Squares Date: 09/10/15 Time: 09:13 Sample (adjusted): 7/03/2014 8/31/2015 Included observations: 283 after adjustments Variable D(LVNI(-1)) C R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) Null Hypothesis: D(LOIL) has a unit root Exogenous: Constant Lag Length: (Automatic - based on SIC, maxlag=15) Augmented Dickey-Fuller test statistic Test critical values: *MacKinnon (1996) one-sided p-values Augmented Dickey-Fuller Test Equation Dependent Variable: D(LOIL,2) Method: Least Squares Date: 09/10/15 Time: 09:12 Sample (adjusted): 7/03/2014 8/31/2015 Included observations: 283 after adjustments Variable D(LOIL(-1)) C R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) 2.5.2 GH test THE GREGORY-HANSEN COINTEGRATION TEST MODEL 4: Regime Shift ADF Procedure t-stat Lag Break -4.229776 1.000000 6/24/2015 Phillips Procedure Za-stat Za-break Zt-stat Zt-break 2.5.3 -31.04772 6/24/2015 -4.120625 6/24/2015 TY procedure VAR Lag Order Selection Criteria Endogenous variables: LVNI LEX LOIL Exogenous variables: C Date: 11/02/15 Time: 15:29 Sample: 7/01/2014 8/31/2015 Included observations: 277 Lag * indicates lag order selected by the criterion LR: sequential modified LR test statistic (each test at 5% level) FPE: Final prediction error AIC: Akaike information criterion SC: Schwarz information criterion HQ: Hannan-Quinn information criterion VAR Residual Serial Correlation LM Tests Null Hypothesis: no serial correlation at lag order h Date: 11/02/15 Time: 15:30 Sample: 7/01/2014 8/31/2015 Included observations: 283 Lags 10 11 12 Probs from chi-square with df VAR Granger Causality/Block Exogeneity Wald Tests Date: 11/02/15 Time: 15:31 Sample: 7/01/2014 8/31/2015 Included observations: 282 Dependent variable: LVNI Excluded LEX LOIL All Dependent variable: LEX Excluded LVNI LOIL All Dependent variable: LOIL Excluded LVNI LEX All Error Correct Model 3.1 OLS Model for variables at level Dependent Variable: LVNI Method: Least Squares Date: 10/25/15 Time: 22:53 Sample: 1/03/2006 8/31/2015 Included observations: 2328 Variable C LEX LOIL R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) 3.2 Unit root test for residual of model 3.1 Null Hypothesis: U has a unit root Exogenous: Constant Lag Length: (Automatic - based on SIC, maxlag=26) Augmented Dickey-Fuller test statistic Test critical values: *MacKinnon (1996) one-sided p-values Augmented Dickey-Fuller Test Equation Dependent Variable: D(U) Method: Least Squares Date: 10/25/15 Time: 22:51 Sample (adjusted): 1/05/2006 8/31/2015 Included observations: 2326 after adjustments Variable U(-1) D(U(-1)) C R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) 3.3 Error Correct Model Dependent Variable: D(LVNI) Method: Least Squares Date: 10/25/15 Time: 23:36 Sample (adjusted): 1/04/2006 8/31/2015 Included observations: 2327 after adjustments Variable C D(LEX) D(LOIL) U(-1) R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic) ... 2.1.1 The relationship between crude oil price and stock market .7 2.1.1.1 Negative effect from crude oil price to stock market 2.1.1.2 Positive effect from crude oil price to stock market. .. nexus between oil price and stock market 11 2.1.1.4 The imperial evidences about the relationship between oil prices and Vietnam stock market 12 2.1.2 The relationship between stock. .. between crude oil price and stock market 2.1.1.1 Negative effect from crude oil price to stock market Chen 2010 investigates whether the high oil price can lead the stock market into the bear territory