1
Bachelor Thesis Economics & Business
The effects of oil price shocks on the Chinese stock
returns at sector level
Abstract
This paper explores the effects of oil price shocks on the stock returns of 10 sectors in China. The vector auto-regression model is adopted and the monthly observations over the period from 01-02-2009 to 31-03-2016 are taken for the analysis. The results indicate that among 10 sectors, the effects of oil price shocks on the stock returns are significant in 4 sectors, which are the Energy, Information Technology, Telecom and Industrial sectors. For the Energy, Information Technology and Telecom sectors, the effects of oil price increases are positive and for the Industrials sector the effects are negative. The asymmetric effects of oil price shocks on the stock returns are only found in the IT sector and are not significant in other sectors.
Name: Yichun Zheng Student Number: 10621806 Date: 29-06-2016
Specialization: Economics & Finance Subject: Finance
2
Statement of Originality
This document is written by Student Yichun Zheng who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
1. Introduction
Crude oil is an indispensable resource for many economic activities. As oil price can influence both the supply and demand sides of the economy, its effects on the stock prices can be transmitted in different ways (Cong et al., 2008). Theoretically, the stock price equals the expected cash flows discounted by the interest rate. On the one hand, since oil is an essential productive input, its price changes can influence production costs and profits of individual firms and thus can affect the expected future cash flows and stock prices. On the other hand, oil price shocks may influence stock prices by inducing expected inflation and changing interest rates as oil is also an important consumer good (Zhu et al., 2015). The relationships between oil price shocks and stock markets have received considerable attention from researchers. In this paper, the effects of oil price shocks on the stock returns of different sectors in China is researched.
This relationship is worth exploring because of two reasons. Firstly, the crude oil price has been volatile in recent years. By January 2016, the WTI crude oil price has reached a low point of 31.4 dollars/barrel, which was a plunge of about 71% from its peak at 108 dollars/barrel in June 2014 (Eiagov, 2016). The slump in crude oil price has aroused concerns about the health of global economy due to its negative impacts on investments and asset prices (Economistcom, 2016). In the face of volatile oil price, studies about the oil price sensitivity of different sectors can provide investors with valuable information about risk diversification. As different sectors have different demand levels for the crude oil, the effects of oil price shocks on the stock returns are also diverse. Investors can adjust the stock holdings of different sectors in their portfolios according to the responses of the sectors to oil price shocks so that the oil price risk can be reduced. Secondly, China is the second-largest oil consumer in the world and has become the world's largest net importer of petroleum at the end of 2013 (Eiagov, 2016). With China’s rising oil demand, Chinese stock market is increasingly exposed to oil price changes. Although plenty of previous studies discussed the relationship between oil price shocks and stock markets, most of them focused on developed countries. And among those researches done on the Chinese market, little attention was paid to the different effects of oil price shocks on the stock returns of different sectors.
Therefore, the central research question of this paper is: how do oil price shocks influence the stock returns at sector level in China? This question is analyzed from two aspects: first, on which sectors do oil price shocks have significant impacts and in which
4 directions; second, whether the asymmetric effects of oil price shocks on the stock returns exist for these sectors. The research adopts the Vector Auto-regression (VAR) model for the estimation and uses monthly data over the period from 01-02-2009 to 31-03-2016, covering 10 sectors in the Chinese stock market.
The outcomes of the estimation and statistical tests suggest that among the 10 sectors investigated, the effects of oil price shocks on the stock returns are significant in the Energy, Information Technology, Telecom and Industrial sectors. For the Energy, Information Technology and Telecom sectors, the stock returns increase with the oil price and for the Industrials sector, the effects of oil price shocks are negative. The asymmetric effects of oil price shocks are only supported by statistical evidence in the IT sector and are not significant for the Chinese stock market in general.
The paper proceeds as follows. Section 2 provides a review of the existing studies on this subject. Section 3 introduces the hypotheses and data used in this study. In this section, the properties of the data are examined by statistical tests and the selected empirical model is specified. Section 4 presents the results of the estimation and tests. The concluding part gives a summary of the important findings and discusses the hypotheses based on these findings. It also provides the answer to the central question and some advice for future work.
5
2. Literature Review
Most of the existing literature has explored the relationship between oil price shocks and the aggregate stock market returns in developed economies and indicated significant links between them. The study of Sadorsky (1999) covered the period from 1947 to 1996 and suggested that oil price changes are an important contributing factor to changes in stock returns in the U.S. market. The research used a vector auto-regression model and its estimation results showed that oil price shocks have a negative impact on real stock returns. Jones and Kaul (1996) found that the postwar oil price shocks depress output and stock returns in the U.S., Canada, the U.K. and Japan. Miller and Ratti (2009) used a vector error correction model to investigate the relationship between the oil price and the stock market in six OECD countries over the period from January 1971 until March 2008. Their research showed that the inverse correlation between the stock market indices and oil price changes has a break in the 1980s and disappears since September 1999. On the contrary, the study of Huang et al. (1996), which also employed a vector auto-regression approach, indicated there is little relationship between oil futures returns and stock market returns except the oil company returns in the U. S. market during the 1980s.
There are much fewer studies on the developing economies than that on the advanced economies. Nevertheless, as the developing countries tend to have higher energy dependency and lower efficiency in energy utilization, they may be more susceptible to oil price changes (Zhu et al., 2015). The developing countries have drawn increasing attention from researchers. Basher and Sadorsky’s (2006) study used an international multi-factor model including both unconditional and conditional risk factors. Their research was based on 21 emerging stock markets and suggested that in emerging economies, oil price risk has strong influence on stock price returns. A similar study was conducted by Aloui et al. (2012), who used a conditional multifactor model and data from 25 emerging markets. They concluded that oil price risk is significantly priced in emerging markets. Bhar and Nikolova (2009) studied the relationship between oil prices and the equity returns in Brazil, Russia, India and China. The analysis results indicated that the impacts of oil price changes depend on whether the country is a net importer or a net exporter of oil. There are also a growing number of studies focusing on the Chinese market. The research of Cong et al. (2008), which adopted the multivariate vector auto-regression method, indicated generally no significant effects of oil price shocks on the stock market returns, except for the returns of manufacturing index and some oil
6 companies, which have positive relationships with oil price shocks. Zhang and Chen (2011) found that Chinese stock market only responds to the expected volatility of world oil price, contrary to the case for the United States. Although Chinese stock returns are positively affected by world oil price changes, the effect is not strong. These conclusions can be ascribed to the inefficiency of the stock market and the irrationality of investors in China.
Furthermore, the asymmetric effects of oil price shocks on the macro economy and the stock market were discovered by many prior studies. The study of Mork (1989), Davis and Haltiwanger (2001) all suggested that positive oil price shocks have larger impacts on the macroeconomic variables than do negative ones. Aloui et al. (2012) found the effects of oil price changes on stock returns are asymmetric in emerging markets. And Sadorsky (1999) found that oil price increases tend to explain more variance in real stock returns compared to oil price decreases. However, the research of Park and Ratti (2008) showed that the asymmetric effects of oil price shocks on real stock returns are only statistically significant for the U.S. and Norway and are not supported by statistical evidence for the oil importing countries in Europe.
As industries differ in the level of energy intensity in production and the demand levels for their products, they can have noticeably different reactions to the oil price shocks (Ratti & Hasan, 2013). Hence, a number of researchers have put emphasis on studying the impacts of oil price changes on the stock returns at industry level. Sadorsky (2001) found that the stock returns of the oil and gas industry are sensitive to and have a positive correlation with oil price changes in Canada. Nandha and Faff (2008) examined the influence of oil price shocks on 35 global industry indices and the results suggested that oil price shocks negatively affect the equity returns of all industries except the mining and oil and gas industries. The research of McSweeney and Worthington (2008) explored the effects of the crude oil price on the Australian stock returns in 9 industries over the period from 1980 to 2006 and found that for different industries, the magnitudes and directions of the effects of oil price movements are diverse. Among all the industries researched, the transportation, banking and retailing industries present remarkably inverse links with the oil price shocks and the energy sector shows an obviously positive relationship with oil price increases. Similarly, Ratti and Hasan (2013) focused on the 10 sectors in the Australian stock market. Their study indicated that the overall market index is inversely related to oil price movements and the stock returns of the energy and materials sectors increase with the oil price, while the returns of all other sectors respond negatively to oil price changes. As for China, there are few studies focusing on
7 impacts of oil price shocks on different sectors. A recent study by Wang and Zhang (2014) explored the effects of oil price shocks on 4 Chinese fundamental industries: grains, metals, petrochemicals and oil fats. They found that the petrochemicals industry is most sensitive to the oil price fluctuations, while the grains market is least affected by the oil shocks. However, their research only covered 4 fundamental industries and did not discuss other sectors in the Chinese stock market.
Hence, this paper focuses on the effects of oil prices shocks on the Chinese stock returns at sector level. And the asymmetric effects of oil price shocks, as indicated by previous researches, are also be examined. The Vector Auto-regression (VAR) model is applied following many prior studies as it is useful in capturing the dynamic relationships among multiple time series.
8
3. Methodology
This section elaborates on the research methodology of this paper in terms of hypotheses, data used, the definitions and properties of the variables and the selected empirical model.
3.1 Hypotheses
This research focuses on the ten sectors in China, which are the Energy, Materials, Industrials, Consumer Discretionary, Consumer Staples, Health Care, Financials, Information Technology, Telecommunication Services and Utilities sectors. As stated before, the analyses are based on two aspects: on which sectors do oil price shocks have significant effects and in which directions; whether the effects of oil price shocks are asymmetric for these sectors.
For the first aspect, most of the prior studies suggested that the oil price shocks have significant positive effects on the returns of oil and gas, energy and materials sectors and have negative effects on the other sectors. Hence, I expect that for sectors with high oil and energy intensity in production, the profits growth resulting from oil price increases will dominate other consequences of oil price increases and thus the expected cash flows and stock price will increase. And for sectors with lower oil and energy intensity, the cost increase due to the oil price increase will outweigh other effects and thus the profits and stock returns will decrease. For the 10 sectors in this study, the Energy, Materials and Industrials sectors tend to have higher oil intensity in production and the other sectors tend to have lower oil intensity. Therefore, the following hypotheses are made:
Hypothesis 1: Oil price increases have significant positive impacts on the stock returns of the Energy, Materials and Industrials sectors.
Hypothesis 2: Oil price increases have negative impacts on the stock returns of all sectors except the Energy, Materials and Industrials sectors.
For the second aspect, a number of papers showed that oil price shocks have asymmetric effects on the macro economy and stock markets. Therefore, for the sectors significantly affected by oil price shocks, the following hypothesis is made:
Hypothesis 3: The effect of oil price increases on the stock returns is significantly different from that of oil price decreases.
9
3.2 Data and descriptive statistics
To investigate the effects of oil price shocks on the different sectors in China, the monthly data for different variables are used and the timespan is from 01-02-2009 to 31-03-2016. There are 86 months during the sample period. This period is selected because the data for the sector indices are only accessible from the January of 2009. Moreover, there are few relevant researches covering the recent two years, when the world crude oil price has experienced huge fluctuations while the other macroeconomic variables are relatively stable and thus oil price shocks can be a stronger contributing factor to stock return changes.
The basic model of this research consists of four variables including , , and . represents the changes of the crude oil price and is calculated as the log first difference between the crude oil price in the current period and that in the last period: . As for the crude oil price, I use the West Texas Intermediate (WTI) spot prices (dollars per barrel) that can be found on the website of the U.S. Energy Information Administration, because the WTI price is the most widely used indicator for the world oil price. And the Census X-13 approach is used to remove the seasonality of the crude oil price. is the log first difference of the exchange rate (Yuan/USD): and it denotes the changes in the exchange rate, which can affect the crude oil price in the Chinese market. The data are taken from the DataStream of the University of Amsterdam. stands for the excess market return of the Chinese stock market. This variable is considered in the model because the whole stock market return indicates the macroeconomic conditions and has essential influence on the returns of different sectors. As the Shanghai Stock Exchange (SSE) is the largest stock market in China in terms of both market capitalization and the numbers of listed stocks, the returns of the SSE Composite Index are used as a proxy for the stock market return in China. To take into account the interest rate, I use the excess market return instead of the total market return, which is the difference between the total market return and the risk free interest rate. And the risk free rate is represented by the 3-month interbank lending rate in China. The last variable means the excess stock return for sector i. There are 10 sector indices in total in the Shanghai Stock Exchange and they can represent 10 different sectors: Energy, Materials, Industrials, Consumer Discretionary, Consumer Staples, Health Care, Financials, Information Technology, Telecommunication Services and Utilities. I calculate the returns of these SSE sector indices and use them as proxies for the stock returns of different sectors. And the risk free rate is also subtracted to get the excess
10 stock returns. Hence, both and are defined as
in the mathematical
form. And the data of all stock indices and the interest rate are collected from the Resset Financial Research Database.
Besides, an alternative model is built for exploring the asymmetric effects of positive oil price shocks and the negative ones. This model includes five variables: ∆ , and . The only difference between the alternative model and the basic one is that the oil price change variable is divided into two variables: oil price increase and oil price decrease ∆ . equals max( ) and ∆ equals min( .
A summary of descriptive statistics, including the means, medians, standard deviations, minimums and maximums for all variables, are presented in Table 1.The sector names in Table 1 denote the excess stock returns of the sectors. And Consumerd, Consumers, IT, Telecom are the abbreviations of Consumer Discretionary, Consumer Staples, Information Technology, Telecommunication Services respectively.
Variables Mean Median Std. Dev. Minimum Maximum
∆EX -0.000578 -0.000565 0.0050684 -0.01088 0.0304154 ∆OP -0.001222 0.007512 0.0867262 -0.245526 0.2138658 ∆OP+ 0.0314763 0.007512 0.0491737 0 0.2138658 ∆OP- -0.032698 0 0.0549613 -0.245526 0 ERM 0.0047327 0.003299 0.0787422 -0.229031 0.2018085 Energy -0.001137 0.000441 0.0981876 -0.260614 0.2858221 Consumerd 0.0098563 0.010027 0.0857312 -0.265082 0.1923948 Consumers 0.0089246 0.010255 0.0736849 -0.216071 0.1543207 financials 0.0083175 -0.00382 0.1002327 -0.298616 0.4210327 Health care 0.0130232 0.016388 0.0810996 -0.255461 0.1698203 industrials 0.0049219 0.00079 0.1006532 -0.254005 0.4154428 IT 0.0155811 0.017794 0.1069745 -0.28578 0.2503773 materials 0.0041054 0.009364 0.101426 -0.26417 0.2846558 Telecom 0.0085252 0.0216 0.1057288 -0.329683 0.3640897 Utilities 0.0047065 0.004894 0.0810856 -0.242161 0.2879134
11
3.3 Unit root test and cointegration test
All of the components in the VAR model should be stationary if one intends to perform tests on the statistical significance of the coefficients (Brooks, 2008). Hence, I conduct a PP unit root test (Phillips & Perron, 1988) to examine the stationarity of the variables. The null hypothesis of PP test is that the series has a unit root, which means the series is an I(1) non-stationary series and one differencing operation is needed to make the series non-stationary. The test statistics and p-values are shown in Table 2.
The left part of Table 2 presents the PP test results for all variables in levels including the natural logarithm of oil price and the natural logarithm of exchange rate , the excess return of the SSE Composite index and the excess return for each sector index . The outcome shows that for and , the null hypothesis that the series has a unit root is rejected at the 1% level of significance, which indicates that these series for excess returns are all stationary in level. However, for and , the null hypothesis is not rejected at the 5% level of significance and thus both of them are non-stationary in levels. One approach to removing the non-stationarity is differencing the series. The right part of Table 2 shows the PP test results for and in their first-differenced forms. It can be seen that the two variables in first differences are stationary at the 1% level of significance.
Although the differencing operation can make the series stationary, this method may ignore important long-run relationships between the series (Brooks, 2008). The Vector Error Correction (VEC) model can solve this problem as it combines levels and first-differenced terms. Since the VEC model is used for non-stationary cointegrated series, I perform a
Table 2 Unit root test (Philips and Perron)
variables
In levels Adj. t-Stat Prob.* In first differences Adj. t-Stat Prob.
ln OP -1.372687 0.5920 d(ln OP) -7.023783 0.0000
ln EX -1.518312 0.5197 d(ln EX) -5.936124 0.0000
Excess returns of:
SSE Composite Index -7,603,691 0.0000
consumer discretionary index -7,229,978 0.0000
consumer staples index -7,947,389 0.0000
energy index -8,760,487 0.0000
financials index -8,431,201 0.0000
health care index -8,813,132 0.0000
industrials index -7,214,706 0.0000
materials index -8,234,789 0.0000
Telecomunication services index -8,293,475 0.0000
Utilities index -8,319,359 0.0000
12 Johansen-Juselius test (Johansen & Juselius, 1990) on and , which are non-stationary in levels, to decide whether there is any cointegration among them and whether the VEC model can be estimated. In table 3, r is the number of cointegration vectors. The null hypothesis that there is no cointegration vectors among the variables is not rejected at the 5% level of significance either in the Trace test or in the Maximum Eigenvalue test. That means the two series are not cointegrated at the 5% level and therefore the VEC model is not appropriate.
As a result, the VAR model is selected for the study. For the variables and , as they are non-stationary in levels, I take their first-differenced forms so that all variables can be stationary.
3.4 Empirical model
The Vector Auto-regression model is an econometric model for exploring the interactions of various time series. In an unrestricted VAR model, all variables are endogenous and each of them is the dependent variable of an equation in which the value of the variable is explained by its own lags and the lags of the other variables in the model. A basic VAR model with p lags is formulated as:
, ,
Suppose there are k variables in the model, then is a vector, which is denoted as , is a vector of constant terms, are coefficient matrices and is a vector of error terms, which satisfies that and there is no serial correlation in . T denotes the sample size of each variable and p is the lag order. In this study, the basic VAR model has 4 variables, which are identified by Cholesky decomposition and arranged in the ordering: , , and . This is the assumed decreasing order of exogeneity of the variables. is the most exogenous because it is assumed that oil price shocks can affect other variables but are independent of the shocks to other variables. is in the last place as the excess stock returns of sectors can be
13 influenced by the disturbances to the oil price, exchange rate and excess stock market return but the reverse is not true. As a result, in this VAR model, variables are expressed as , where as there are 10 sectors in all. The sample size T is 86, which is the number of months in the sample period and the lag order P is selected based on the Akaike and the Schwarz information criteria.
14
4. Results
This section presents the results of the VAR estimations, which can provide the answer for the research question. The estimated VAR model is interpreted by using the Granger causality test, impulse response functions, and variance decompositions. I also perform a test on the asymmetric effects of the oil price shocks.
4.1 Granger causality test
The Granger causality test, proposed by Granger (1969), is a statistical test for examining whether the lagged values of a variable can improve the predictions of the future values of another variable. If the answer is yes, then the former is said to Granger-cause the latter. To study whether the oil price shocks have significant effects on the stock returns of different sectors, I firstly conduct a Granger test on the oil price shocks and the excess stock return of each sector . The null hypothesis is that does not Granger-cause .
Table 4 presents the test statistics and P-values for different sectors. It can be seen that for the Energy sector, the null hypothesis is rejected at the 5% level of significance, indicating that the oil price shocks Granger-cause the excess stock returns of the Energy sector. For the Information Technology and Telecommunication Services sectors, the correlations are even more statistically significant, as the null hypothesis is rejected at 1% level. And for the Health Care and Industrials sectors, the oil price shocks Granger-cause the excess stock returns at the 10% level. While for all other sectors, the null hypothesis is not rejected at 10% level and does not Granger-cause .
Table 4 Granger Causality Independent variable ∆OP
Dependent variables Chi-sq Prob. 5% significance level 10% significance level
CONSUMERD 1.356 0.2442 Not reject Not reject
CONSUMERS 2.507 0.1133 Not reject Not reject
ENERGY 6.083 0.0478 reject reject
FINANCIALS 2.126 0.1448 Not reject Not reject
HEALTH CARE 2.966 0.0850 Not reject reject
INDUSTRIALS 8.067 0.0891 Not reject reject
MATERIALS 2.682 0.1015 Not reject Not reject
IT 9.460 0.0088 reject reject
TELECOM 13.955 0.0009 reject reject
15
4.2 Impulse response function
Impulse responses depict the responses of the endogenous variables and their future values when a one-unit or one-standard deviation positive shock is given to one of the error terms in the system, with the assumptions that the other error terms equal 0 and this error term also returns to 0 in the subsequent periods. Sadorsky (1999) points out that the coefficients estimated by VAR may lack statistical significance because the technique is inaccurate in standard errors estimation. Therefore, impulse response functions are regarded as a better tool for explaining the impacts of a given shock on the endogenous variables.
For my analysis, the focus is on how the stock returns of different sectors react to the oil price shocks. For each sector, I obtain an impulse response function curve based on the VAR( , , , ) estimation which is identified by Cholesky decomposition. I choose Cholesky-dof adjusted method given in Eviews because the ordering is important in this relationship. The study of Nikbakht (2010) suggests that real oil prices have been the primary source of real exchange rate changes in 7 OPEC countries. Therefore, is more likely to affect and it is placed before . The is placed last as the other three variables can have contemporaneous effects on it.
Figure 1 shows the responses of the excess stock returns of 10 sectors over a 10-month period when a one-standard deviation positive shock is given to . Table 5 is a summary of the directions and statistical significance of the impulse responses of 10 sectors within 2 months. It can be seen that among the 10 sectors, only the Energy, Telecom and IT sectors show statistically significant responses at 5% level to the oil price shock, which corresponds with the Granger test results. All of the three respond positively to the oil price shock within 2 months. For the other 7 sectors, the impacts of oil price shocks are not statistically significant at the 5% level. Moreover, all of them have positive responses to the oil price shock within 2 months except the Industrials sector. The response of the Industrials sector is negative for most of the time within 6 months. In addition, the responses of all sectors return to around 0 within 6 months, indicating that the effects of oil price shocks are short-term.
While the response of the Energy sector is expectable, the significant positive responses of the excess returns of the Telecom and IT sectors to oil price shocks are beyond my expectations. It is also surprising to find that for almost all sectors, the impacts of a one-standard deviation oil price shock on the excess returns are positive within 2 months.
16
Figure 1. Response to Cholesky One S.D. Innovations ± 2 S.E.
-.04 .00 .04 .08 .12 1 2 3 4 5 6 7 8 9 10
Response of CONSUMERD to D_LN_OP
-.04 -.02 .00 .02 .04 .06 .08 1 2 3 4 5 6 7 8 9 10
Response of CONSUMERS to D_LN_OP
-.05 .00 .05 .10
1 2 3 4 5 6 7 8 9 10
Response of ENERGY to D_LN_OP
-.08 -.04 .00 .04 .08 .12 1 2 3 4 5 6 7 8 9 10
Response of FINANCIALS to D_LN_OP
-.04 -.02 .00 .02 .04 .06 .08 1 2 3 4 5 6 7 8 9 10
Response of HEALTH_CARE to D_LN_OP
-.08 -.04 .00 .04 .08 .12 1 2 3 4 5 6 7 8 9 10
17 -.08 -.04 .00 .04 .08 .12 1 2 3 4 5 6 7 8 9 10
Response of MATERIALS to D_LN_OP
-.05 .00 .05 .10 1 2 3 4 5 6 7 8 9 10 Response of IT to D_LN_OP -.05 .00 .05 .10 1 2 3 4 5 6 7 8 9 10
Response of TELECOM to D_LN_OP
-.04 .00 .04 .08 .12 1 2 3 4 5 6 7 8 9 10
Response of UTILITIES to D_LN_OP
Note: D_LN_OP denotes the variable oil price changes . Impulse-response function
2 S.E.
(Note: P(N) means positive(negative) response, I(S) indicates insignificant(significant) response at 5% level.)
4.3 Variance decomposition of forecast errors
Variance decomposition provides the percentages of the forecast errors variance of an endogenous variable contributed by the shocks to other variables. It is useful in interpreting the VAR model as it indicates the relative importance of each exogenous shock in affecting the endogenous variables in the model.
Table 5 Impulse responses of excess stock returns of 10 sectors to oil price shocks in 2 months
Sectors Consumerd Consumers Energy Financials Health care Industrials Materials Telecom Utilities IT
Direction P P P P P N to P P P P P
18 Table 6 gives the variance decompositions of forecast errors in the excess stock returns of the 10 sectors over a 10-month period. It is noticeable that for the Telecom sector, the oil prices shocks account for approximately 21% of the fluctuations in the excess stock return over 10 months. And in the IT and Industrial sectors, oil price shocks can also explain more than 10% of the variance of the excess returns. The contributions of oil price impulses to the changes in the excess stock returns are lowest in the Financials and Utilities Sectors, which are around 3% and 4.7% respectively. It is also shown that for most sectors, oil price shocks can cause more fluctuations in the excess stock returns than the shocks to the exchange rate. In addition, for all sectors except the Telecom sector, the shocks to stock market return explain the largest fraction of variance in the excess stock returns.
Based on the results from the Granger causality test, impulse responses and variance decomposition, it can be concluded that the impacts of oil price shocks are significant on the excess stock returns in the Energy, IT, Telecom and Industrials sectors. For the Energy, IT and Telecom sectors, the impacts are positive. For the Industrials sector, the impacts are negative.
4.4 Asymmetric effects of oil price shocks
As mentioned in the literature review, the previous studies of Mork(1989), Davis and Haltiwanger (2001), Sadorsky (1999), Park and Ratti (2008) and Aloui et al. (2012) all suggested that the oi price shocks have asymmetric effects on the macro economy and the stock markets. Most of them found that positive oil price shocks have larger impacts on the economy.
To study the asymmetric effects of positive and negative oil price shocks, I separate the variable into oil price increase and oil price decrease ∆ . They are
Table 6 (%)Variance decompostion of forecast errors in excess stock returns (10 months) Sectors shock to ∆OP shock to ∆EX shock to ERM own shocks
CONSUMERD 7.3171 3.02529 68.43259 21.22507 CONSUMERS 8.4227 2.49869 52.81912 36.25952 ENERGY 9.5964 7.64901 59.65708 23.09750 FINANCIALS 2.9164 3.08269 75.92941 18.07149 HEALTH CARE 7.0023 4.83819 84.82457 3.33498 INDUSTRIALS 10.4202 8.85012 67.61126 13.11843 MATERIALS 6.9231 3.34697 76.41304 13.31691 IT 15.1891 9.20464 37.97890 37.62737 TELECOM 20.8012 7.93257 34.61758 36.64868 UTILITIES 4.6748 6.49022 67.13014 21.70487
19 formulated as ) and ∆ . As a result, an alternative five-variable VAR model is constructed, in which the vector of endogenous variables is expressed as: . In this VAR estimation, the equation for the excess stock return of sector i is expressed as follows:
∑ ∑ ∑ ∑ ∑
P is the lag order, which is selected by using the AIC and SC criteria. To whether the impacts of and are significantly different, a Wald test on the coefficients of them is performed. The hypotheses are , , j=1,…,P. As the above results suggest that only for Energy, IT, Telecom and Industrials sectors, the impacts of oil prices shocks on the excess stock returns are significant. Hence, I only consider the asymmetric effects of oil price shocks in these sectors.
The results are provided in Table 6. For the Energy, Telecom and Industrials sectors, the null hypothesis is not rejected at both 5% and 10% level of significance. That means there is no statistical evidence for asymmetric effects of oil price shocks in these sectors. As for the IT sector, however, the null hypothesis is rejected at the 10% level, which means there are asymmetric impacts of oil price increases and decreases at this significance level. Generally speaking, the asymmetric effects of oil price shocks on the Chinese stock returns are not obvious as only the IT sector shows asymmetric responses.
Table 7 Wald test (asymmetric effects)
Varibles Chi-square P-values 5% significance level 10%
Energy 0,0092 0,9236 Not reject Not reject
IT 5,5465 0,0625 Not reject Reject
TELECOM 0,0037 0,9514 Not reject Not reject
20
5. Conclusion
This paper explores how do oil price shocks influence the Chinese stock returns at sector level. Ten sectors in the Chinese stock market are covered and the sample period is from 01-02-2009 to 31-03-2016. By performing the Unit root test and cointegration test on the properties of the variables, the Vector Auto Regression model is selected for the analysis and the estimation results are interpreted by the Granger causality test, impulse response functions, variance decomposition and the Wald test. The effects of oil price shocks on the stock returns of different sectors are considered in two aspects: on which sectors do oil price shocks have significant impacts and in which directions; whether the effects of oil price shocks are asymmetric for these sectors.
The first aspect is discussed based on the outcomes of the Granger test, impulse response functions and variance decomposition. The Granger test results show that at the 5% level of significance, the past values of oil price shocks are helpful in predicting the future values of the excess stock returns in the Energy, Information Technology and Telecom sectors. At the 10% significance level, the Granger causality between oil price shocks and excess stock returns also exists in the Health Care and Industrials sectors. For all the other sectors, the Granger causality is not supported by statistical evidence. The impulse response functions indicate that the excess returns of the Energy, IT and Telecom sectors have significant positive responses at 5% level to a given positive oil price shock within 2 months. The responses are positive but not significant at 5% level for the other sectors except the industrials sector, whose response is negative for most of the time within 6 months. The variance decomposition shows oil price shocks account for over 10% of the variance of the excess returns in the Telecom, IT and Industrials sectors over a 10-month period. Given the above results, it can be concluded that the effects of oil price shocks are significant on the excess stock returns in the Energy, Information Technology, Telecom and Industrials sectors. However, the impacts are positive for the Energy, IT and Telecom sectors and negative for the Industrials sector. Therefore, the hypothesis 1 only holds for the Energy sectors and is rejected for the Materials and Industrials sectors as the effects are insignificant for the Materials sector and are negative for the Industrial sectors. The hypothesis 2 is rejected because for all sectors except the Energy, Materials and Industrials sectors, the impacts of oil price increases on the excess stock returns are positive.
21 The second aspect is assessed by the Wald test on the asymmetric effects of oil price shocks. The results suggest that among the sectors significantly affected by oil price shocks, only the IT sector has asymmetric responses at the 10% level. Therefore, the asymmetric effects of oil price increases and decreases on the Chinese stock returns are not evident in general. As a result, the hypothesis 3 only holds for the IT sector at 10% level.
Due to the data accessibility, the research only uses the indices from the Shanghai Stock Exchanges since 2009. For future studies, it will be better to include also the stock market indices of the Shenzhen Stock Exchange and cover a longer period to see if these relationships still hold. It will also be interesting if future studies can dig into the potential reasons behind the significant positive correlations between the oil price shocks and the stock returns in the IT and Telecom sectors of China, because this result is totally different from that of many existing literatures on the developed countries.
22
6. References
Aloui, Nguyen, & Njeh. (2012). Assessing the impacts of oil price fluctuations on stock returns in emerging markets. Economic Modelling, 29(6), 2686-2695.
Basher, & Sadorsky. (2006). Oil price risk and emerging stock markets. Global Finance
Journal, 17(2), 224-251.
Bhar, R., & Nikolova, B. (2009). Oil Prices and Equity Returns in the BRIC Countries. World
Economy, 32(7), 1036-1054.
Brooks, C. (2008). Introductory econometrics for finance (2nd, [rev.] ed.). Cambridge [etc.: Cambridge University Press.
Cong, Wei, Jiao, & Fan. (2008). Relationships between oil price shocks and stock market: An empirical analysis from China. Energy Policy, 36(9), 3544-3553.
Davis, & Haltiwanger. (2001). Sectoral job creation and destruction responses to oil price changes. Journal of Monetary Economics, 48(3), 465-512.
Economistcom. (2016). The Economist. Retrieved 28 June, 2016, from
http://www.economist.com/news/leaders/21688854-low-energy-prices-ought-be-shot-arm-economy-think-again-whos-afraid-cheap
Eiagov. (2016). U.S. Energy Information Administration (EIA). Retrieved 29 June, 2016, from
http://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=RWTC&f=D
Eiagov. (2016). U.S. Energy Information Administration (EIA). Retrieved 29 June, 2016, from https://www.eia.gov/beta/international/analysis.cfm?iso=CHN
Granger, C. (1969). Investigating Causal Relations by Econometric Models and Cross-spectral Methods. Econometrica, 37(3), 424-438.
Huang, Roger D., Ronald W. Masulis, & Stoll, Hans R. (1996). Energy shocks and financial markets. Journal of Futures Markets, 16(1), 1.
Johansen, S., & Juselius, K. (1990). Maximum likelihood estimation and inference on cointegration—with applications to the demand for money. Oxford Bulletin of
Economics and statistics, 52(2), 169-210.
23
McSweeney, E. J., & Worthington, A. C. (2008). A comparative analysis of oil as a risk factor in Australian industry stock returns, 1980-2006. Studies in economics and
finance, 25(2), 131-145.
Miller, & Ratti. (2009). Crude oil and stock markets: Stability, instability, and bubbles.
Energy Economics, 31(4), 559-568.
Mork, K. (1989). Oil and the Macroeconomy When Prices Go Up and Down: An Extension of Hamilton's Results. Journal of Political Economy, 97(3), 740-744.
Nandha, & Faff. (2008). Does oil move equity prices? A global view. Energy Economics,
30(3), 986-997.
Nikbakht, L. (2010). Oil prices and exchange rates: The case of OPEC. Business Intelligence
Journal, 3(1), 83-92.
Park, & Ratti. (2008). Oil price shocks and stock markets in the U.S. and 13 European countries. Energy Economics, 30(5), 2587-2608.
Phillips, P. C., & Perron, P. (1988). Testing for a unit root in time series regression.
Biometrika, 75(2), 335-346.
Ratti, R., & Hasan, M. (2013). Oil Price Shocks and Volatility in Australian Stock Returns.
Economic Record, 89, 67-83.
Sadorsky, P. (1999). Oil price shocks and stock market activity. Energy Economics, 21(5), 449-469.
Sadorsky, P. (2001). Risk factors in stock returns of Canadian oil and gas companies. Energy
Economics, 23(1), 17-28.
Xiao Wang, & Chuanguo Zhang. (2014). The impacts of global oil price shocks on China's fundamental industries. Energy Policy, 68, 394-402.
Zhang, & Chen. (2011). The impact of global oil price shocks on China’s stock returns: Evidence from the ARJI( -h t )-EGARCH model. Energy, 36(11), 6627-6633.
Zhu, H., Guo, Y., & You, W. (2015). An empirical research of crude oil price changes and stock market in China: Evidence from the structural breaks and quantile
