The contagion effect of international crude oil price fluctuations on Chinese stock market investor sentiment

Applied Energy 187 (2017) 27–36

Contents lists available at ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

The contagion effect of international crude oil price fluctuations on Chinese stock market investor sentiment Zhihua Ding a, Zhenhua Liu a,⇑, Yuejun Zhang b, Ruyin Long a a b

School of Management, China University of Mining and Technology, Xuzhou 221116, China Business School of Hunan University, Changsha 410082, China

h i g h l i g h t s
Chinese stock market investor sentiment index is proposed.
Oil price fluctuations significantly Granger cause stock market investor sentiment.
Crude oil price has negative contagion effects on stock market investor sentiment.
Contagion delay of oil price fluctuation on stock investor sentiment is 8 months.

a r t i c l e

i n f o

Article history: Received 10 August 2016 Received in revised form 7 November 2016 Accepted 11 November 2016

Keywords: International crude oil price Stock market Investor sentiment Contagion effect

a b s t r a c t Given the close contact between international financial markets, the contagion effect across markets is becoming increasingly obvious. In this paper, which uses principal component analysis to build a Chinese stock market investor sentiment index and further applies a structural vector autoregression (SVAR) model, we analyze the contagion effect of international crude oil price fluctuations on Chinese stock market investor sentiment. The results show that international crude oil price fluctuations significantly Granger cause Chinese stock market investor sentiment; in the long term, if the international crude oil price fluctuates by 1%, stock market sentiment will negatively fluctuate 3.9400%. From the perspective of short-term efficacy, if the international crude oil price fluctuates by 1%, stock market investor sentiment in the same period will negatively fluctuate 1.0223%. International crude oil prices made a greater early contribution to investor sentiment and showed a rapid growth trend, with a contribution of 2.8076% in the first period and 8.1955% in the second. The growth rate then slows and eventually stabilizes at the 25% level; the average contagion delay for international crude oil price fluctuation to affect investor sentiment is 8 months. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction Oil has always been known as the ‘‘blood of industry”. As the upstream raw material of industrial production, oil has been playing an irreplaceable role. To meet the huge demand from oil consumption, countries worldwide typically import large quantities of crude oil. In 2015, for example, the United States imported 3.66 billion tons of crude oil, China 3.36 million tons, India 1.95 million tons, Japan 1.68 million tons, and so on. Oil price fluctuations affect a country’s economic development, social stability and the lives of its citizens. In view of the important role of oil in economic development, an increasing number of domestic and international scholars have been studying the international crude ⇑ Corresponding author. E-mail address: [email protected] (Z. Liu). http://dx.doi.org/10.1016/j.apenergy.2016.11.037 0306-2619/Ó 2016 Elsevier Ltd. All rights reserved.

oil price. At present, international crude oil price analysis generally takes two perspectives: first, the oil price is treated as a dependent variable to explore the impact of supply and demand factors, political factors, futures markets and other independent variables; second, the oil price is treated as an independent variable to explore its impact on gross domestic product (GDP), the consumer price index (CPI) and the producer price index (PPI), along with the impact of inflation and other macroeconomic variables. As a barometer of the macroeconomy, the stock market is affected by international crude oil price fluctuations [1–4]. Some scholars have pointed out that there is a strong two-way volatility spillover effect between the crude oil market and the stock market [5]. As an important factor affecting investors’ decision, investor sentiment can effectively account for many issues in stock markets. Because these problems are difficult to explain using traditional economic theory, investor sentiment has also attracted the attention of many

28

Z. Ding et al. / Applied Energy 187 (2017) 27–36

scholars [6–8]. Further research results show that there is a twoway Granger causality and two-way volatility spillovers between stock market returns and investor sentiment [9,8]. There are strong correlations between international crude oil price fluctuations and stock market, and so are stock market and investor sentiment. In this way, we wonder if there exist correlations between international crude oil price fluctuations and stock market investor sentiment? The descriptive analysis of the international crude oil price volatility and the stock market investor sentiment indicates that there is a correlation between them. At present, the international literature on investor sentiment mainly focuses on studying the influence of investor sentiment on stock market volatility, earnings and forecasts. The literature on the relationship between international crude oil price fluctuations and the stock market mainly explores the impact and spillover effects of international crude oil price fluctuations on stock market returns, as shown in Table 1. There are few studies addressing the effect of international crude oil price fluctuations on the stock market from the perspective of investor sentiment; thus, this study offers some originality. As China has ‘‘rich coal, poor oil, less gas” energy characteristics and is experiencing rapid economic growth, its demand for oil is increasing. To meet this demand, China continues to increase oil imports, which has led to an increase in China’s external oil dependency. In 2015, its dependence was projected to reach 60.6%. The importance of oil for China’s economic development is clear. Meanwhile, China is still in an ‘‘emerging plus transition” stage; many problems still exist in its stock market, such as a higher proportion of individual investors, immature investment philosophies, higher stock market turnover and the presence of a large speculative component. China’s market is also more prone to the ‘‘herding effect” and other excessive effects [10,11]. As raw material of industrial production, oil will have profound impacts on the oil-related downstream industries. Oil has not only the attributes of resources but also the attributes of financial. Especially with the improve-

Table 1 Related research literature on investor sentiment and the relationship between international crude oil price fluctuations and the stock market. Research classification

Research topic

Representative research

Related research literature on investor sentiment

Impacts of investor sentiment on stock market returns Forecasts of investor sentiment on stock market returns Impacts of investor sentiment on stock market volatility Relationship between investor sentiment and different types of stock markets

Corredor et al. [7], Ni et al. [13], He and Casey [14], Frugier [15] Vozlyublennaia [8], Kim et al. [16], Aissia [17]

Impacts of international crude oil price fluctuations on stock market returns

Zhang and Wei [1], Awartani et al. [2], Bouri et al. [26], Ahmadi et al. [27], Diaz et al. [28], Nejad et al. [29] Chang et al. [3], Broadstock and Filis [4], Kang et al. [32], Du and He [5], Khalfaoui et al. [33], Ewing and Malik [34], Liu et al. [35] Wen et al. [36], Zhu et al. [37], Chen and Lv [38]

Related research literature on the relationship between international crude oil price fluctuations and the stock market

Spillover effects of international crude oil price fluctuations on the stock market

Interdependence or contagion relationship between international crude oil prices and the stock market

Kumari et al. [18]

Baker and Wurgler [6], Kadilli [19], Liston [20]

ment of oil futures market and the use of petroleum derivatives, financial attributes of oil has become increasingly prominent, oil price fluctuations will have an impact on the real economy and virtual economy. Taking the Chinese market as an example, choosing the international crude oil price fluctuation as the independent variable, and the stock market investor sentiment as the dependent variable, this paper will explore the impact of international crude oil price fluctuations on investor sentiment in China’s stock market and provide a positive reference to allow China, along with other countries and regions, to minimize the risk of financial contagion and develop policies for stock market regulation. The method presented here is innovative because (1) at present, related domestic and foreign research mainly focuses on the influence of international crude oil price fluctuations on stock market volatility and returns, while this paper takes the perspective of investor sentiment to explore the relationship between international crude oil price fluctuations and China’s stock market investor sentiment, diverging from earlier research perspectives. Furthermore, (2) this paper measures the impact of international crude oil price fluctuations on stock market investor sentiment in China in terms of long-term and short-term efficacy and contagion delay to reveal the contagion effect from international crude oil price fluctuations on investor sentiment in China’s stock market.

2. Literature review Investors’ individual decision-making behavior will affect market judgment, which then causes sharp fluctuations in the stock market over the short term. It is sometimes difficult to explain this volatility from the traditional economic point of view. To address this phenomenon, behavioral finance introduced an assumption and refers to all investor expectations that cannot be explained by basic information as ‘‘investor sentiment”. Baker and Wurgler pointed out that investor sentiment is a belief based on investors’ expectations of an asset’s future cash flow and investment risk [6], although this belief does not reflect current facts. The researchers proposed the following proxy variables: discounts of closed-end funds, NYSE stock turnover rate, IPO numbers, IPO first-day average yield, the proportion of equity financing and dividend premiums, etc., and measured investor sentiment using these proxy variables. Using Baker and Wurgler’s index construction method as a reference, the Chinese scholars Yi and Mao integrated indicators that can reflect Chinese stock market investor sentiment and constructed a composite index to better measure it (their index is known as the CICSI) [12]. At present, there are many applications of investor sentiment on the stock market, mainly focusing the influence of investor sentiment on stock returns [7,13–15], the forecast of stock returns [8,16,17], impact on stock market volatility [18], and the relationship between investor sentiment and different types of stock markets [6,19,20]. Energy, as an important factor of production, and the effects of energy price fluctuations on the economy and society naturally attract the attention of many scholars. For example, Zhang et al. analyzed the relationship between speculative trading and WTI crude oil futures price volatility [21,22]. Their study of price discovery and risk transfer effects on the crude oil and gasoline futures markets showed that the crude oil futures market has a greater price risk transfer ability, while the risk transfer effect between the crude oil and gasoline futures markets is not obvious. Ju et al. studied the macroeconomic impact of oil price shocks [23]; their results show that oil price shocks have a positive impact on GDP and exchange rates and a negative impact on the CPI. Ju et al. also proposed an incentive-oriented early warning system for predicting co-movements between oil price shocks and the macroeconomy [24]. Sun et al. identified regime shifts in the US

29

Z. Ding et al. / Applied Energy 187 (2017) 27–36

electricity market based on price fluctuations [25]. As the interaction between international financial markets increases, the relationship between the international crude oil market and the stock market strengthens; thus, there is much research on the influence of the international crude oil price on the stock market. This research mainly concentrates on two aspects—the impact of international crude oil price volatility on stock market returns [1,2,26–29] and the spillover effect of international crude oil price volatility on the stock market—and draws varying conclusions. Some scholars believe that the mean spillover effect between the international crude oil market and the stock market is very weak and unstable [30,31]. However, others believe that there are significant spillover effects between international crude oil prices and the stock market [3–5,32–35]. Few scholars study the contagion relationship between international crude oil price volatility and stock market investor sentiment, though some have explored the relationship between the international crude oil price and the stock market. Wen and Wei used a time-varying copula correlation model to measure the contagion relationship between the energy (oil) market and the stock market during the economic crisis [36]. Their research confirmed the existence of a contagion effect and also found that the contagion effect in the Chinese market is weaker than that in the American market. Zhu et al. found that the dynamic dependencies between the crude oil price and the stock market were positive before the global financial crisis of 2008 (except in Hong Kong) [37]; however, after the crisis, the dependent relationship was significantly enhanced. Chen and Lv studied the contagion effect between the crude oil price and the Chinese stock market and found that, during a crisis [38], the contagion effect will be greatly enhanced, and after a crisis, the contagion effect between the two markets will be significantly reduced. To summarize, there are many domestic and foreign studies that focus on the dynamic relationship between crude oil prices and stock market returns and on the significant impacts of investor sentiment on stock market volatility, returns and prediction, and these abundant research results serve as significant references. However, there are still several aspects worthy of further study. First, current studies mainly focus on the relationship between the crude oil price and the stock market from the perspective of stock market volatility and returns, but there are few studies from the perspective of stock market investor sentiment. At present, domestic and foreign research focuses on investor sentiment as the independent variable and explores its impact on the other variables of financial markets, and studies taking investor sentiment as the dependent variable are very scarce. In this paper, based on a Chinese stock market investor sentiment index constructed using principal component analysis and a structural vector autoregression (SVAR) model, we analyze the contagion effects of international crude oil price fluctuations on Chinese stock market investor sentiment to minimize financial contagion risks in China and provide positive references for policy regulating stock markets. 3. Research method 3.1. Principal component analysis Principal component analysis is a statistical method that transforms indexes into a few unrelated comprehensive indexes; these are new variables obtained by the linear combination of multiple variables and can reflect the original information of the multiple variables. The general model of principal component analysis can be expressed by formula (1):

F 1 ¼ a11 X 1 þ a12 X 2 þ a13 X 3 þ    þ a1m X m F 2 ¼ a21 X 1 þ a22 X 2 þ a23 X 3 þ    þ a2m X m

9 > > > > > > > =

 > > F m ¼ am1 X 1 þ am2 X 2 þ am3 X 3 þ    þ amm X m > > > > kk k1 k2 k3 Y ¼ Pm F 1 þ Pm F 2 þ Pm F 3 þ    þ Pm F k > ; k i¼1 i

k i¼1 i

k i¼1 i

ð1Þ

k i¼1 i

X 1 ; X 2 ; . . . ; X m are measured variables; F1, F2, . . . , Fm are main components; aij ¼ ði ¼ 1; 2; . . . ; m; j ¼ 1; 2; . . . ; mÞ are factor loadings; Y are prediction scores representing the contribution rate of the principal component i; Pkm1 F i is the selection of the principal component i; k i¼1 i

and k is the contribution rate of the selected principal component. The factor load pij is the load of measured variable j on the main component i, with a greater load indicating a closer relationship between them. 3.2. SVAR model The improved structural vector autoregression (SVAR) model resolves these problems that VAR model cannot solve by applying constraints [39,40]. Therefore, to build an SVAR model through an existing VAR model, considering the content of this study, we need to construct a VAR model that contains the international crude oil price and stock market investor sentiment.

Yt ¼ a þ

n X bi Y ti þ et

ð2Þ

i¼1

where Y t is a column vector containing LBrent and SMISrt, which will be defined later; a is the constant term matrix; et is the residual error of the model, which is used to reflect the impact of different vector interaction effects; and bi is the coefficient matrix. This paper only considers the impact of international crude oil price volatility on stock market investor sentiment in a market environment and further applies constraint conditions by formula (2) to construct the SVAR model:

A0 Y t ¼ A0 a þ sumni¼1 Ai bi Y ti þ ut where ut ¼

et ¼

A0 et ,

ebrent t esmis t

!

 ¼

a11

0

a21

a22



ubrent t usmis t

ð3Þ ! ð4Þ

In formula (4), ut is the white noise sequence used to measure the influence of international crude oil price fluctuations on stock market investor sentiment, and its covariance matrix is the unit matrix. ubrent and usmis respectively represent the impact of internat t tional crude oil prices and stock market investor sentiment. Because it is difficult for stock market investor sentiment to adjust in a timely fashion based on short-term changes in the international crude oil price, it is assumed that stock market investor sentiment will not respond to the current international crude oil price change; thus, a12 ¼ 0. 3.3. Impulse response function The impulse response function is a system’s reaction to a shock or a new variable. If the VAR model is reversible, it can be represented as a vector moving average.

Yt ¼ C þ

1 X ws ets

ð5Þ

s¼0

In formula (5), ws is the coefficient matrix; C is the constant term; and et is the error vector. For the impulse response function of the SVAR model, we first need to solve the problem of orthogo-

30

Z. Ding et al. / Applied Energy 187 (2017) 27–36

nalization for the VAR model impulse response function. By formula (5), we can obtain the following orthogonal impulse response function:

@Y i;tþq @ujt

ðqÞ

dij ¼

ð6Þ ðqÞ

In formula (6), dij reflects the disturbance of variable j at time t and adds a unit; other variables’ disturbances do not change if disturbances are constant during other periods; Y i;tþq is the response to a structural shock on ujt . 3.4. Dynamic variance decomposition The variance decomposition analyzes the contribution of each structural impact to the endogenous variables (usually measured by variance) and further evaluates the importance of different structural shocks. Therefore, variance decomposition gives important information about variables’ effects on random disturbances in the VAR model. The model is as follows:

yit ¼

k X ð0Þ ð1Þ ð2Þ ð3Þ ðaij ejt þ aij ejt1 þ aij ejt2 þ aij ejt3 þ   Þ

ð7Þ

j¼1

The contents of each bracket are the total impacts of the first j disturbance ej from the infinite past to the present time on yi . Seeking its variance and assuming that ej has no sequence correlation, we obtain ð0Þ

E½ðaij

2

ð2Þ ejt þ að1Þ ij ejt1 þ aij ejt2 þ   Þ  ¼

1 X ðqÞ 2 ðaij Þ rjj

j

q¼0

¼ 1; 2; . . . ; k

ð8Þ

Effects of the perturbation term j on variable i from the infinite past to the present time are evaluated using the variance. The variance of yi can be decomposed into k types of unrelated effects. To measure the contribution of each perturbation term to the variance of yi, the following scales are defined:

P1 RVC j>i ð1Þ ¼

ðqÞ 2 q¼0 ðaij Þ

varðyi Þ

rjj

P1 ¼P

ðqÞ 2 q¼0 ðaij Þ

rjj

P1 ðqÞ 2 k j¼1 f q¼0 ðaij Þ

rjj g

i; j

¼ 1; 2; . . . ; k

ð9Þ

RVC is relative to the variance contribution; that is, according to the relative contribution of the variance of the jth variable to the variance of yi based on the shock, the effects of the jth variable on the ith variable can be observed.

net mutual fund redemptions and the proportion of fund assets held in cash [42,46], etc. In addition, considering the investor sentiment is not only influenced by cognitive psychological factors, but also affected by macroeconomic factors [48,50]. Therefore, impacts of macroeconomic factors need to be eliminated in the investor sentiment index construction. Both direct and indirect indicators can only reflect investor’s psychological changes from one side, which will affect the accuracy and objectivity of sentiment measure. So, direct and indirect indicators should be taken into consideration in the investor sentiment index construction, building a composite sentiment indicator. Baker and Wurgler formed a composite index of sentiment based on the common variation in six underlying proxies for sentiment: the closed-end fund discount, NYSE share turnover, the number and average first-day returns on IPOs, the equity share in new issues, and the dividend premium, while controlling growth in the industrial production index, growth in consumer durables, nondurables, and services, and a dummy variable for NBER recessions [48]. Subsequently, Baker et al. constructed a global composite index with the volatility premium, turnover, the number and average first-day returns on IPOs [51]. With reference to Baker and Wurgler’s sentiment index research methods and Zhi Gao Yi, Mao Ning’s CICSI index construction method [6,12], and combined with the reality of Shanghai and Shenzhen stock market and the availability of data, we chose proxy variables that can measure China’s stock market investor sentiment. In particular, the China’s stock market is developing rapidly, new monthly investor account represents the degree of investors’ demand and participation for securities, which can reflect investor sentiments. In high spirits, the enthusiasm of entering the market is high, new monthly investor account will be lager, and vice versa [52]. Xue found that the consumer confidence index compiled by the National Bureau of Statistics can reflect China’s stock market investor sentiments [53]. Based on the above considerations, we selected the Shanghai and Shenzhen A-share stock markets’ closed-end fund discounts, trading volume, number of IPOs, IPO first-day returns, new investor account indicators and the consumer confidence index and took monthly data for the six indexes to conduct principal component analysis, excluding the indexes for producer prices of industrial products, consumer prices and macroeconomic boom-related variables. We then constructed a monthly index of Chinese stock market investor sentiment (SMISrt). As a benchmark for oil prices, Brent crude oil price can well reflect international oil market prices. Based on data availability, this paper selected monthly Brent crude oil price data and sentiment index data from February 2005 to April 2015. The investor sentiment index variable names and definitions are shown in Table 2.

4. Data sources and investor sentiment index construction 4.1. Data sources and variable declarations At present, there are two main categories of investor sentiment measurement indicators, single sentiment indicators and composite sentiment indicators. Single sentiment indicators include direct and indirect indicators. Direct indicators: investor intelligence index [41,42], analyst recommendation [41], consumer confidence index [41,43], etc. Although direct indicators can directly reflect the investor’s sentiments or beliefs, in practice, the impact of sentiments on decision-making behavior varies from different people and circumstances [41]. Therefore, indirect indicators need to be considered, which are mainly derived from the public traded data in the capital market. Indirect indicators indirectly reflect changes in investor sentiment, such as closed-end fund discounts [42,44– 46], IPOs and first-day returns [47,48], trading volume [48,49],

Table 2 Investor sentiment index variable name and meaning. Variable name

Variable symbol

Meaning and calculation

Discount of closed-end fund Trading volume

DCEF

IPO number

IPON

IPO first-day return New investor accounts Consumer confidence index

IPOR

Weighted average discount of Shanghai and Shenzhen A shares of all closed-end funds listed on the last trading day of each month Shanghai and Shenzhen A-shares market turnover/Circulation market value Number of IPOs based on release date every month A weighted average based on weights for the number of shares issued and circulated Shanghai and Shenzhen A-share monthly number of new accounts National Bureau of Statistics compiled, indicating confidence strength

TURN

NIA CCI

31

Z. Ding et al. / Applied Energy 187 (2017) 27–36

trolling for macroeconomic factors, and the correlation coefficients of SMISrt. Each variable is shown in Table 5.

4.2. Investor sentiment index construction To eliminate the effects of differences in each variable’s units, each variable was normalized before the principal component analysis. As the impacts of different indicators on investor sentiment have hysteresis, t and t-1 period information will have an impact on the t period investment behavior. Therefore, the six indexes and the lag variables were analyzed by principal component analysis, and a new investor sentiment index (smist) containing 12 variables was constructed, which is beneficial to avoid the high degree of autocorrelation between investor emotional variables. It is important to note that in the process of smist calculation, using Yi Zhigao and Mao Ning’s calculation methods and strictly abiding by statistical standards, the cumulative explained variance rate reached at least 85% [12]. For each calculation, the weighted average of 1, 2, 3, 4 and 5 principal components were used to retain more information. Then, we respectively analyzed the smist and the six indexes in advance, lagged variables with correlation analysis, and accordingly selected six variables with a larger correlation coefficient to construct the composite sentiment source index (SMISt). The results are shown in Table 3. As Table 3 shows, each index passes the significance test; the correlation degrees of smist with TURNt1, DCEFt1, CCIt1, NIA, IPORt1, and IPONt1 are high. In addition, all indexes except for the number of new investor accounts reflect investor sentiment in advance. Next, we choose these six variables as the final source for SMISt construction. First, the six source index variables, TURNt1, DCEFt1, CCIt1, NIA, IPORt1 and IPONt1, were standardized and then analyzed using principal component analysis. The first to fifth principal components explained 97.37% of the cumulative variance. Table 4 shows that SMISt as constructed based on six variables has good characteristics: from a statistical point of view; there are positive correlations between investor sentiment and new investor accounts, trading volume, IPO first-day returns and consumer confidence index. The greater the discount of closed-end funds, the lower the investor sentiment. Considering the representation of Chinese macroeconomic cycle variables and the availability of (monthly) data, to represent production, consumption and economic boom, this paper selects the industrial PPI, the CPI and the macroeconomic climate index (MBCI) as proxy variables of these economic fundamentals to eliminate macroeconomic factors in the sentiment index effects. Each SMISt source indicator variable for TURNt1, DCEFt1, CCIt1, NIA, IPORt1 and IPONt1 is regressed with the three macroeconomic variables of PPI, CPI and MBCI (before regression, the variables were standardized), obtaining a residual sequence of TURNrt1, DCEFrt1, CCIrt1, NIAr, IPORrt1 and IPONrt1. Next, we brought the six residual variables of the above step (1) through the principal component analysis (principal components 1–5 explain a cumulative variance of 96.186%). Finally, we obtained the Chinese stock market investor sentiment index (SMISrt), con-

4.3. Robustness When testing the correlation between SMISrt and SMISt, by calculating the Pearson correlation coefficient, we can know that the correlation coefficient of SMISrt and SMISt is 0.86 (two tailed, 1% significance level), which indicates that there is a significant positive correlation between them. To further test the robustness of the selected proxy variables, we compared the correlation between the constructed sentiment index before and after reducing proxy variables. For example, to reduce the proxy variable, the closed-end fund discount, the correlation coefficient of the sentiment index before and after the reduction is 0.79 (two tailed, 1% significance level), indicating that the sentiment index constructed with above six proxy variables can effectively reflect China’s stock market investor sentiment. 4.4. Descriptive analysis of the international crude oil price and stock market investor sentiment As Fig. 1 shows, through a Pearson correlation test, we found that the correlation coefficient between the SMISrt and the Brent crude oil price is 0.512 (two tailed, 1% significance level), indicating that the influence of Brent crude oil price volatility on stock market investor sentiment cannot be ignored and that it maintains a trend contrary to the trend of the stock market investor sentiment index. This paper will further analyze the intrinsic relationship between international crude oil price volatility and Chinese stock market investor sentiment by constructing the SVAR model. 5. Data test and empirical analysis 5.1. Data test 5.1.1. Stability test Based on modeling needs, to eliminate seasonal factors and heteroscedasticity in the crude oil price time series, this paper adopted the CensusX12 method to adjust data seasonally, and the natural logarithm was taken for the seasonally adjusted series, the Brent (Brent) logarithmic series or LBrent. To guarantee the validity of the model and avoid ‘‘spurious regression”, an augmented Dickey–Fuller (ADF) stationarity test was conducted on the variables. Each variable is shown in Table 6. Test results show that the LBrent and SMISrt sequences have the same order one I (1) process and that there may be a cointegration relationship or a long-term stability proportional relationship between the two sequences. 5.1.2. Granger causality test Because the Granger causality test depends on the choice of lag period, to more clearly illustrate the relationship between them,

Table 3 Correlation between smist and the 12 variables. Smist Variable DCEF TURN IPON IPOR NIA CCI Note: ⁄,

⁄⁄

Correlation coefficient 0.684 0.785⁄⁄ 0.035 0.281⁄⁄ 0.324⁄⁄ 0.630⁄⁄

⁄⁄

Sample size 122 122 122 122 122 122

Variable DCEFt1 TURNt1 IPONt1 IPORt1 NIAt1 CCIt1

respectively represent 5% and 1% significance levels (bilateral); the same is true for tables below.

Correlation coefficient 0.687 0.830⁄⁄ 0.079 0.294⁄⁄ 0.270⁄⁄ 0.639⁄⁄

⁄⁄

Sample size 122 122 122 122 122 122

32

Z. Ding et al. / Applied Energy 187 (2017) 27–36

Table 4 Correlation between SMISt and the 6 variables.

NIA DCEFt1 TURNt1 IPONt1 IPORt1 CCIt1 SMISt

NIA

DCEFt1

TURNt1

IPONt1

IPORt1

CCIt1

SMISt

1 0.362⁄⁄ 0.112 0.386⁄⁄ 0.427⁄⁄ 0.034 0.391⁄⁄

1 0.742⁄⁄ 0.387⁄⁄ 0.047 0.595⁄⁄ 0.576⁄⁄

1 0.367⁄⁄ 0.027 0.445⁄⁄ 0.848⁄⁄

1 0.229⁄ 0.099 0.030

1 0.363⁄⁄ 0.241⁄⁄

1 0.442⁄⁄

1

NIAr

DCEFrt1

TURNrt1

IPONrt1

IPORrt1

CCIrt1

SMISrt

1 0.189⁄ 0.109 0.384⁄⁄ 0.309⁄⁄ 0.138 0.419⁄⁄

1 0.666⁄⁄ 0.616⁄⁄ 0.181⁄ 0.602⁄⁄ 0.654⁄⁄

1 0.406⁄⁄ 0.129 0.563⁄⁄ 0.927⁄⁄

1 0.077 0.348⁄⁄ 0.236⁄⁄

1 0.326⁄⁄ 0.298⁄⁄

1 0.583⁄⁄

1

Table 5 Correlation between SMISrt and 6 variables.

NIAr DCEFrt1 TURNrt1 IPONrt1 IPORrt1 CCIrt1 SMISrt

From the results in Table 4, it can be seen that SMISrt retains the characteristics of SMISt.

Fig. 1. Change trend chart for the Brent crude oil price and the stock market investor sentiment index.

we conducted the Granger causality test on the LBrent and SMISrt for five periods. From the results of Table 7, we can see that at the 5% significance level, the international crude oil price fluctuation Granger causes change in the Chinese stock market investor sentiment, and the causal relationship is significant.

5.2. Contagious effect analysis of international crude oil price fluctuations on stock market investor sentiment 5.2.1. Long-term effectiveness measure According to AIC and SC criteria that determine the two optimal lags, in the second lag period, we tested the cointegration relationship between SMISrt and LBrent. The Johanson cointegration test shows that there is only one unique cointegration relationship

33

Z. Ding et al. / Applied Energy 187 (2017) 27–36 Table 6 ADF test results for each variable. Variable

Inspection Form

t-Statistic

1% level

5% level

10% level

Prob.

Conclusion

LBrent DLBrent SMISrt DSMISrt

(c, (c, (c, (c,

2.7083 8.1462 3.0062 13.6275

4.0370 4.0363 4.0356 4.0363

3.4480 3.4477 3.4474 3.4477

3.1491 3.1490 3.1488 3.1490

0.2353 0.0000 0.1349 0.0000

Non-stationary Stationary Non-stationary Stationary

t, t, t, t,

2) 0) 0) 0)

Note: (c, t, q) represents the sequence ADF test form, c, t, q respectively represent the constant term, time trend and lag order, q the optimal lag is determined by the AIC criterion and the SC criterion.

Table 7 Granger causality test. Null hypothesis

Lags

Obs

F-statistic

Prob.

Conclusion

A B A B A B A B A B

1 1 2 2 3 3 4 4 5 5

121 121 120 120 119 119 118 118 117 117

0.26499 4.50799 0.43856 3.27071 0.40099 2.50852 0.27137 3.88068 0.38822 3.03992

0.6077 0.0358 0.6460 0.0415 0.7526 0.0625 0.8959 0.0055 0.8559 0.0132

Accept the null hypothesis Reject the null hypothesis Accept the null hypothesis Reject the null hypothesis Accept the null hypothesis Accept the null hypothesis Accept the null hypothesis Reject the null hypothesis Accept the null hypothesis Reject the null hypothesis

Note: A represents that SMISrt cannot Granger cause LBrent, B represents that LBrent cannot Granger cause SMISrt.

between them. After the stability test, we found that the reciprocals of all of the roots in the VAR(2) model feature polynomials of the second lag with values less than 1 and are located in the unit circle; thus, the model satisfies the requirement of stability. According to the results of the Johansen cointegration test, there is a long-term equilibrium relationship between the international crude oil price and stock market investor sentiment. The cointegration equation is as follows:

SMIS ¼ 3:9400LBrent þ 17:4032

ð10Þ

From formula (10), it is concluded that if international crude oil prices fluctuate by 1%, stock market investor sentiment will negatively fluctuate 3.9400%. In the long run, the negative influence of international crude oil prices on stock market investor sentiment is remarkable. 5.2.2. Short-term effectiveness measure The error correction model reflects the approximation of the variable from a short-term disequilibrium state to the long-term equilibrium state. The error correction model can be regarded as an equation that measures short-term effects, and the short-term effect equation of the international crude oil price on stock market investor sentiment is as follows:

DðSMISÞ ¼ 0:3559ðLBrentð1Þ þ 0:2538  SMISð1Þ

tively fluctuate 1.0223% over the same period; that is, there is a short-term negative effect of international crude oil price volatility on Chinese stock market investor sentiment.

5.2.3. Contagion effectiveness analysis Combined with the results of the long- and short-term effectiveness measures, it can be concluded that in the long and short terms, international crude oil price volatility has a negative impact on Chinese stock market investor sentiment.

5.3. Contagious delay analysis of international crude oil price fluctuations on stock market investor sentiment 5.3.1. Establishment of the SVAR model We established an AB type bivariate SVAR model with model forms as follows:

A  et ¼ B  ut

In formula (12), et and ut are two-dimensional vectors, while A, B are a 2  2 matrix to be estimated. Based on the existing literature and economic reality, we set the following short-term constraint: there is no reaction to current oil price fluctuations by Chinese stock market investor sentiment. This means that the A, B matrix in the SVAR model is defined as follows:

 4:4171Þ  0:2535  DðSMISð1ÞÞ  0:0655

 A¼

 DðSMISð2ÞÞ  1:0223  DðLBrentð1ÞÞ þ 0:5035  DðLbrentð2ÞÞ þ 0:0201

ð12Þ

1

0

Cð1Þ 1



 B¼

Cð2Þ

0

0

Cð3Þ

 ð13Þ

ð11Þ

From formula (11), the model of the error correction coefficient is 0.3559; the error correction term is a negative feedback mechanism and statistically significant, in line with the correct meaning, indicating that there is a short-term effect of international crude oil prices on stock market investor sentiment. When the short-term fluctuations deviate from the long-term equilibrium, the adjustment dynamics (0.35) shift from the disequilibrium state to the equilibrium state and finally achieve a long-term equilibrium. Considering short-term effectiveness, when the international crude oil price fluctuates by 1%, stock market investor sentiment will nega-

where CðiÞ ði ¼ 1; 2; 3Þ are the coefficients to be estimated. By using the maximum likelihood estimation method to estimate the coefficients of the A, B matrix, the A, B matrix in the SVAR model can be constructed by LBrent and SMISrt and are shown as follows:

 A¼

1

0

1:8618 1



 B¼

15:4919

0

0

15:4919

 ð14Þ

Using the impulse response function and variance decomposition analysis of the constructed SVAR model, it is more accurate to observe the delay effect of the variables on structural shocks.

34

Z. Ding et al. / Applied Energy 187 (2017) 27–36

5.3.2. Impulse response function analysis The purpose of constructing the SVAR model is to analyze the impact of an endogenous variable on other endogenous variables; this further analysis requires the impulse response function. The impulse response trajectories of stock market investor sentiment to international crude oil price based on the SVAR model are shown in Figs. 2 and 3. From Figs. 2 and 3, we can see that international crude oil price fluctuations always have negative effects on Chinese stock market investor sentiment. The effects gradually increase from the beginning to a maximum of 0.1458% in the fourth period, and then begin to gradually decrease at a relatively flat declining rate. Generally speaking, for a non-oil producing company with crude oil as an input factor, the rise in the price of crude oil increases the company’s production costs, leading to a deterioration of cash flow and a decline in stock prices. However, the rise in the price of crude oil will cause an increase in the general price level and inflation; relevant departments will raise interest rates and take other measures so that it becomes more attractive to purchase bonds than stocks, and stock prices will decrease [54]. From previous literature, we know that a decline in stock prices will cause a decline in investor sentiment. Thus, the negative impact of international crude oil prices on stock market investor sentiment can be explained. 5.3.3. Dynamic variance decomposition analysis The impulse response function can analyze the impact of the disturbance as it spreads to each variable, and variance decomposition further evaluates the importance of different structural shocks based on the analysis of each structural shock’s contribution to changes in the endogenous variable. Table 8 shows the variance decomposition results for international crude oil price on stock market investor sentiment. From Table 8, we can see that in the initial period, the contribution rate of international crude oil prices on investor sentiment increases rapidly. In the first period, it is 2.8076% and in the second, 8.1955%, but then the growth rate slows and eventually stabilizes at the 25% level. However, the contribution rate of stock market investor sentiment on international crude oil price is always low, at 5%. 5.3.4. Contagion delay analysis Based on the above analysis, it can be concluded that the shock from international crude oil prices on Chinese stock market investor sentiment reaches a maximum in the fourth period, while the contribution rate of international crude oil prices on Chinese stock market sentiment changes reaches a stable state in the 12th per-

Fig. 3. Impulse response function of SMISrt to LBrent.

iod. Based on the above conclusions, the average value of the periods’ impulse response function is taken to find the peak and the variance decomposition to achieve stable periods. It can be concluded that the average contagion delay of international crude oil price fluctuation on Chinese stock market investor sentiment is 8 months. 6. Research conclusions and policy suggestions 6.1. Research conclusions (1) There is significant Granger causality between international crude oil price fluctuations and Chinese stock market investor sentiment, and there is a long-term equilibrium relationship between them. This conclusion also verifies the research results of Lee et al. [55]: the major global markets are all interrelated. It also expands the scope of research by finding that in addition to the existence of contagion effects between inter-regional markets, contagion effects also exist in cross-industry markets. (2) The international crude oil price has negative effects on Chinese stock market investor sentiment. In the long term, if international crude oil prices fluctuate by 1%, stock market investor sentiment will negatively fluctuate 3.9400%. In the short term, if the international crude oil price fluctuates by 1%, stock market investor sentiment will negatively fluctuate by 1.0223% over the same period. (3) The average contagion delay of international crude oil price fluctuation on Chinese stock market investor sentiment is 8 months. International crude oil price fluctuations always have negative effects on Chinese stock market investor sentiment; they gradually increase from the beginning of a period, reach a maximum of 0.1458% in the fourth period, and then begin to gradually decrease with a relatively flat decline rate. In the initial period, the contribution rate of international crude oil prices on investor sentiment increases rapidly; in the first period, it is 2.8076%, and in the second, 8.1955%, but then the growth rate slows and eventually stabilizes at the 25% level in the 12th period. 6.2. Policy suggestions (1) International crude oil prices can be used as a guiding index of Chinese stock market investor sentiment. Investors are encouraged to enhance energy awareness, as the risk of international crude oil price fluctuations can be an important factor in stock pricing. Using as reference the interna-

Fig. 2. Impulse response function of LBrent to SMISrt.

35

Z. Ding et al. / Applied Energy 187 (2017) 27–36 Table 8 Variance decomposition of stock market investor sentiment. Period

SMISrt standard deviation

Contribution of LBrent to SMISrt (%)

Period

LBrent standard deviation

Contribution of SMISrt to LBrent (%)

1 2 3 4 5 6 7 8 9 10 11 12

0.4411 0.5363 0.6162 0.6732 0.7172 0.7514 0.7785 0.8003 0.8180 0.8323 0.8442 0.8539

2.8076 8.1955 11.6822 14.4826 16.6501 18.4143 19.8766 21.1095 22.1590 23.0578 23.8301 24.4948

1 2 3 4 5 6 7 8 9 10 11 12

0.0792 0.1278 0.1631 0.1889 0.2084 0.2233 0.2349 0.2441 0.2515 0.2575 0.2624 0.2664

0.0000 0.2169 0.4183 0.6532 0.9016 1.1572 1.4121 1.6606 1.8983 2.1222 2.3305 2.5222

tional crude oil price fluctuations and the stock market return rate and according to the financial situation of listed companies, a comprehensive analysis of business indicators and fundamentals can facilitate investors’ decisions. (2) Although the contagion delay of international crude oil prices on Chinese stock market investor sentiment is longer, its relationships mainly occur in the earlier periods. This tells government policy makers that when international crude oil prices show high fluctuations, response measures should be taken quickly to stabilize stock market sentiment, such as using policy instruments (adjusting interest rates and information disclosure, etc.) to adjust the stock market in time to reduce the probability of extreme risk. China also should establish the domestic crude oil futures market, strive for crude oil pricing power, to form a completed domestic oil market system, which will avoid the risks of international crude oil price fluctuations and enhance China’s voice in the international oil trade. (3) Reducing dependence on crude oil can weaken the risks caused by contagion effectiveness between international crude oil price fluctuations and stock market investor sentiment. With the increase of China’s external dependency on crude oil and considering China’s economic development strategy and national energy security, China should establish a sound strategic oil reserve system and advocate energy conservation to lessen its external dependency on oil. In addition to cultivating energy conservation awareness and improving energy efficiency, China should actively develop alternative energy sources. China has abundant natural gas resources, in recent years, with the development of science and technology, natural gas reserves have increased substantially, while speeding up the development of new energy sources, such as wind power, hydrogen fuel, deepsea and permafrost gas hydrate (combustible ice), optimize its energy consumption structure, and gradually reduce fossil energy consumption. Meanwhile, it is important to actively encourage residents to save energy, such as purchasing new energy vehicles and using clean energy, to reduce dependence on fossil energy.

Acknowledgement This work was supported by the National Natural Science Foundation of China (71573255), the Teaching Education Reform and Practice of Jiangsu Province (Grant No. JGZZ16_078), the Innovation Team Program of the China University of Mining and Technology (No. 2015ZY003). We are gratefully to acknowledge the respected editors and the anonymous referees for their suggestions in this article.

References [1] Zhang YJ, Wei YM. The dynamic influence of advanced stock market risk on international crude oil returns: an empirical analysis. Quant Finance 2011;11 (7):967–78. [2] Awartani B, Maghyereh AI. Dynamic spillovers between oil and stock markets in the Gulf Cooperation Council Countries. Energy Econ 2013;36:28–42. [3] Chang C, McAleer M, Tansuchat R. Conditional correlations and volatility spillovers between crude oil and stock index returns. North Am J Econ Finance 2013;25:116–38. [4] Broadstock DC, Filis G. Oil price shocks and stock market returns: new evidence from the United States and China. J Int Financ Market Inst Money 2014;33:417–33. [5] Du L, He Y. Extreme risk spillovers between crude oil and stock markets. Energy Econ 2015;51:455–65. [6] Baker M, Wurgler J. Investor sentiment in the stock market. J Econ Perspect 2007;21(2):129–52. [7] Corredor P, Ferrer E, Santamaria R. Investor sentiment effect in stock markets: stock characteristics or country-specific factors? Int Rev Econ Finance 2013;27:572–91. [8] Vozlyublennaia N. Investor attention, index performance, and return predictability. J Bank Finance 2014;41:17–35. [9] Dong XW, Zhang XD, Liu WQ. A study of binary relationship between investor sentiment and stock market returns——study using quantile regression. Econ Manage J 2013;06:103–11. [10] Jiang YM, Wang MZ. Investor sentiment and Stock Returns: an empirical study of the overall effect and cross section effect. Nankai Bus Rev 2010;13:150–60. [11] Demirer R, Kutan AM, Chen C. Do investors herd in emerging stock markets: evidence from the Taiwanese market. J Econ Behav Organ 2010;76(2):283–95. [12] Yi ZG, Mao N. The study of investor sentiment measurement in Chinese stock market: the construction of CICSI. J Financ Res 2009;11:174–84. [13] Ni Z, Wang D, Xue W. Investor sentiment and its nonlinear effect on stock returns—new evidence from the Chinese stock market based on panel quantile regression model. Econ Model 2015;50:266–74. [14] He LT, Casey KM. Forecasting ability of the investor sentiment endurance index: the case of oil service stock returns and crude oil prices. Energy Econ 2015;47:121–8. [15] Frugier A. Returns, volatility and investor sentiment: evidence from European stock markets. Res Int Bus Finance 2016;38:45–55. [16] Kim JS, Ryu D, Seo SW. Investor sentiment and return predictability of disagreement. J Bank Finance 2014;42:166–78. [17] Aissia DB. Home and foreign investor sentiment and the stock returns. Quart Rev Econ Finance 2016;59:71–7. [18] Kumari J, Mahakud J. Does investor sentiment predict the asset volatility? Evidence from emerging stock market India. J Behav Exp Finance 2015;8:25–39. [19] Kadilli A. Predictability of stock returns of financial companies and the role of investor sentiment: a multi-country analysis. J Financ Stab 2015;21:26–45. [20] Liston DP. Sin stock returns and investor sentiment. Quart Rev Econ Finance 2016;59:63–70. [21] Zhang Y, Wang Z. Investigating the price discovery and risk transfer functions in the crude oil and gasoline futures markets: some empirical evidence. Appl Energy 2013;104:220–8. [22] Zhang Y. Speculative trading and WTI crude oil futures price movement: an empirical analysis. Appl Energy 2013;107:394–402. [23] Ju K, Zhou D, Zhou P, Wu JM. Macroeconomic effects of oil price shocks in China: an empirical study based on Hilbert-Huang transform and event study. Appl Energy 2014;136:1053–66. [24] Ju K, Su B, Zhou D, Zhang YQ. An incentive-oriented early warning system for predicting the co-movements between oil price shocks and macroeconomy. Appl Energy 2016;163:452–63. [25] Sun M, Li J, Gao C, Han D. Identifying regime shifts in the US electricity market based on price fluctuations. Appl Energy 2016. doi: http://dx.doi.org/10.1016/j. apenergy.2016.04.032.

36

Z. Ding et al. / Applied Energy 187 (2017) 27–36

[26] Bouri E, Awartani B, Maghyereh A. Crude oil prices and sectoral stock returns in Jordan around the Arab uprisings of 2010. Energy Econ 2016;56:205–14. [27] Ahmadi M, Manera M, Sadeghzadeh M. Global oil market and the U.S. stock returns. Energy 2016;114:1277–87. [28] Diaz EM, Molero JC, Perez De Gracia F. Oil price volatility and stock returns in the G7 economies. Energy Econ 2016;54:417–30. [29] Nejad MK, Jahantigh F, Rahbari H. The long run relationship between oil price risk and tehran stock exchange returns in presence of structural breaks. Proc Econ Finance 2016;36:201–9. [30] Liu XY, Zhu CM. The risk spillover effect of international oil price fluctuations on Chinese stock market. J Guangdong Univ Finance 2011;26(56–71):91. [31] Zhu HM, Dong D, Guo P. The relationship between international crude oil price and stock market returns based on Copula function. Theory Pract Finance Econ 2016;2:32–7. [32] Kang W, Ratti RA, Yoon KH. The impact of oil price shocks on the stock market return and volatility relationship. J Int Financ Markets Inst Money 2015;34:41–54. [33] Khalfaoui R, Boutahar M, Boubaker H. Analyzing volatility spillovers and hedging between oil and stock markets: evidence from wavelet analysis. Energy Econ 2015;49:540–9. [34] Ewing BT, Malik F. Volatility spillovers between oil prices and the stock market under structural breaks. Glob Finance J 2016;29:12–23. [35] Liu X, An H, Huang S, Wen S. The evolution of spillover effects between oil and stock markets across multi-scales using a wavelet-based GARCH-BEKK model. Physica A 2017;465:374–83. [36] Wen X, Wei Y, Huang D. Measuring contagion between energy market and stock market during financial crisis: a copula approach. Energy Econ 2012;34:1435–46. [37] Zhu H, Li R, Li S. Modelling dynamic dependence between crude oil prices and Asia-Pacific stock market returns. Int Rev Econ Finance 2014;29:208–23. [38] Chen Q, Lv X. The extreme-value dependence between the crude oil price and Chinese stock markets. Int Rev Econ Finance 2015;39:121–32. [39] Kilian L. Not all oil price shocks are alike: disentangling demand and supply shocks in the crude oil market. Am Econ Rev 2007;99(3):1053–69.

[40] Kilian L, Cheolbeom P. The impact of oil price shocks on the U.S. stock market. Int Econ Rev 2009;50(4):1267–87. [41] Fisher KL, Statman M. Investor sentiment and stock returns. Financ Anal J 2000;56(2):16–23. [42] Brown GW, Cliff MT. Investor sentiment and the near-term stock market. J Emp Finance 2004;11(1):1–27. [43] Qiu LX, Welch I. Investor sentiment measures. SSRN Electron J 2004;117 (35):367–77. [44] De Long JB, Shleifer A, Summers LH, et al. Noise trader risk in financial markets. J Polit Econ 1990;98(4):703–38. [45] Lee CMC, Thaler RH. Investor sentiment and the closed-end fund puzzle. J Finance 1991;46(1):75–109. [46] Neal R, Wheatley SM. Do measures of investor sentiment predict returns? J Financ Quant Anal 1998;33(4):523–47. [47] Ljungqvist A, Wilhelm WJ. IPO pricing in the dot-com bubble. J Finance 2003;58(2):723–52. [48] Malcolm B, Jeffrey W. Investor sentiment and the cross-section of stock returns. J Finance 2006;61(4):1645–80. [49] Baker M, Stein JC, Wurgler J. When does the market matter? Stock prices and the investment of equity-dependent firms. Nber Work Papers 2002;118 (118):969–1005. [50] Kumar A, Lee CMC. Retail investor sentiment and return comovements. J Finance 2006;61(5):2451–86. [51] Baker M, Wurgler J, Yuan Y. Global, local, and contagious investor sentiment. J Financ Econ 2012;104(2):272–87. [52] Wu YR, Han LY. Imperfect rationality, sentiment and closed-end-fund puzzle. Econ Res J 2007;03:117–29. [53] Xue F. An empirical test of investor sentiment index selection in China. World Econ Situation 2005;14:14–7. [54] Lao JN. Does the price of oil affect the Shanghai Composite Index? – An empirical analysis based on the data of 2000–2007. World Econ Situation 2008;5:71–6. [55] Lee Y, Tucker AL, Wang DK, et al. Global contagion of market sentiment during the US subprime crisis. Glob Finance J 2014;25(1):17–26.