Impact of Oil Prices on Dutch Stock Returns
Name: Henry Christiaan (Denny) Sellis Student number: 11257938
Date: 15-07-2020
Statement of Originality
This document is written by Denny Sellis, who takes full responsibility for the content of this document.
I declare that the text and the work presented in this document are 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 the completion of the work, not for the contents.
Chapter 1: Introduction
Past research has shown that unexpected changes in the prices of crude oil can greatly depress the prices of assets and increase volatility (Park & Ratti, 2008). Most journalists, financial investors, and policymakers attribute the oil price shocks to various factors such as political unrest in oil-exporting economies, particularly in the Middle East. Researchers such as Kilian (2009) have found that crude oil prices are endogenous to macroeconomic and financial conditions, which are largely influenced by innovations in both the demand and supply side. Currently, the literature on oil price shocks and how they impact financial and macroeconomic variables continues to grow as more of such debate and research takes place.
One specific area of research that has received a considerable amount of scholarly attention is the relationship between oil returns and stock returns. Oil remains the most important commodity for contemporary economies (Huang et al., 1996; Kilian & Park, 2009). This substantial and unchallenged reliance on oil has attracted many researchers to investigate how oil price affects the behavior and tendency of stock markets. Therefore, my research aims to contribute to this literature by examining the relationship between oil prices and stock prices. To do so, I formulate the following research question: “Do oil prices have an impact on the stock returns in the Netherlands?”
The sample period chosen for this research is from 2005 until 2020. This is because it allows the capturing of economic shocks, such as the financial crisis of 2008. Moreover, my country of analysis is the Netherlands. This choice is motivated by the fact that the Netherlands is a special case when considering oil-exporting vs. oil-importing countries. Wang et al. (2013)
examined the difference in oil price effects on stock market returns in oil-importing and exporting countries. The Netherlands is considered to be somewhere in between the two categories, making its classification complicated. This complication makes the Netherlands an interesting sample country.
The remainder of this thesis is organized as follows. In chapter 2, I summarize the previous relevant literature that examines the impact of oil prices on stock returns. The chapter concludes with the hypothesis relevant to the research question. In chapter 3, I provide an overview of the data source, data collection, and the descriptive statistics of the important variables. Additionally, in this section, I also explain the methodology I use to answer the research question. In chapter 4, I provide the results of the statistical analyses performed. Finally, in chapter 5, I conclude with a summary of the findings and highlight its implications, along with certain limitations of the research and directions for future research.
Chapter 2: Literature Review
This section discusses the theoretical background on the topic of interest, i.e., the relationship between oil prices and stock returns along with the findings of prior empirical research. An overview of these findings is shown in Table 1. In the leftmost column of the table, the names of the authors are given. The following columns show the variables used, research methods, sample country, and the findings, respectively. Comparatively more researchers have found a negative effect of oil returns on stock prices. The positive effect from oil price changes on stock returns is only found when energy industries are examined (Broadstock et al., 2012) or when Narayan and Narayan (2010) found a positive correlation between oil returns and stock returns which is probably due to other factors than increasing oil prices.
Researchers make use of two types of statistical analyses, namely, the Ordinary Least Squares (OLS) regression and the Vector Autoregressive model (VAR). An OLS regression gives a linear relationship between the dependent and the independent variables. By including controlling variables in an OLS regression, the researcher can control for factors that have a systematic impact on the dependent variable. To increase the accuracy of the independent variables tested, control variables should be included. The regression coefficient suggests the significance and direction of the relationship analyzed. The second method, The VAR model, is also aimed at determining the linear relationship between the dependent and the independent variable. Apergis and Miller (2009), amongst others, used the VAR model. The main difference is that in VAR models, multiple time series are included in one model. Thus, the lagged effect is considered, which the OLS regression omits.
Arouri (2011) and Elyasian et. Al. (2011) made a distinction in their research for different industry sectors, to find the short-term effect of oil returns and sector stock returns in Europe. The dataset Arouri (2011) used, over the period from 1998 until 2010, contained weekly data of twelve European sector indices and the Dow Jones Stoxx 600. The DJ Stoxx 600 aims at mimicking the largest companies in the most important industries in Europe. Arouri (2011) found a positive relationship between the Oil & Gas sector and oil returns. This is interesting because most of the other sectors discussed in his research show a negative correlation between their stock returns and oil returns.
Moreover, a difference exists in the effect that oil returns have on stock returns in developed countries compared to developing countries. Driesprong et. al. (2008) found a highly significant effect of oil returns on stock returns in developed markets. However, for the emerging markets, the results were less clear. The effect of West Texas and Brent oil on the stock market returns in developing countries was not significant. So according to Driesprong et. Al. (2008) oil returns would show a significant effect on the Dutch stock returns, as the Netherlands is a developed country. Other researchers such as Fang & You (2014) also found mixed results when regressing the effect of oil returns on Newly Industrialized Countries.
Table 1. summary of previous research on the effect of oil returns on stock market returns
Author(s), year
Variables Method Sample Effect oil price
on stock returns Jones and Kaul, 1996 Dependent variable: stock market returns Independent variables: Crude oil prices Ordinary Least Squares Regression in the form of: (rt = µ+ α1*rOilt-1 + εt)
Stock returns in the US, Canada, Japan, and the UK
Negative Sadorsky, 1999 Dependent variable: returns of S&P 500 Independent variable: producer price index for fuels Vector Autoregression model (VAR) Xt = ∑𝑝𝑖=1 𝐴𝑖∗ 𝑋𝑡−𝑖 + εt
US stock market returns Negative
Driesprong, Jacobsen & Maat, 2008 Dependent variable: stock market returns Independent variables: Crude oil prices Ordinary Least Squares Regression (rt = µ+ α1*rOilt-1 + εt) Developed vs developing countries Negative Apergis and Miller, 2009 Dependent variable: stock market prices Independent variable: crude oil prices
VAR model Developed countries Global demand shocks = positive Idiosyncratic oil price shocks = negative Kilian and Park, 2009 Dependent variable: U.S. real stock return Independent variable: the real price of crude oil imported by the United State
Structural VAR model
US stock market returns, Kilian also tests the effect different causes (supply vs demand shocks) of oil price shocks have. supply shocks = little effect, demand shocks = negative Narayan and Narayan, 2010 Dependent variable: Vietnamese stock prices Independent variable: oil prices
Ordinary Least Squares Regression
Vietnam stock market returns
Arouri, 2011 Dependent variable: EU stock market returns of different sectors Independent variable: Brent crude oil prices
Multiple
regression model:
Different European sector stock returns Negative (for the oil&gas, basic materials sector it was Positive Elyasian, Mansur, and Odusami, 2011 Dependent variable: the stock market return of 13 industrial sectors in the U.S.
Independent variable: crude oil prices Ordinary Least Squares Regression Fama-French model
13 U.S. industries Mostly negative Broadstock, Cao, and Zhang, 2012 Dependent variable: Chinese energy-related stock market returns Independent variable: European Brent crude oil prices
Ordinary Least Squares Regression
Energy sector stock returns in China Positive Asteriou and Bashmakov a, 2013 Dependent variable: emerging CEEC’s stock market returns Independent variable: West Texas Intermediate crude oil prices
Ordinary Least Squares Regression
Emerging capital markets of Central and Eastern EU countries stock returns
Negative Wang, Wu, and Yang, 2013 Dependent variable: stock returns of oilimporting and -exporting countries Independent variable: West Texas Intermediate crude oil prices
VAR model importing vs oil-exporting countries Oil exporting = positive Oil importing = no significant effect
Fang and You, 2014
Dependent Variable: stock returns (India, China, and Russia) Independent variable: Global oil demand shock Structural VAR Model Newly Industrialized Economies (NIEs) india = negative, China = not significant, Russia = negative
One element missing in most of the prior literature on the relationship between stock and oil returns is that a distinction in the sample between oil-consuming and oil-producing countries is not made (Wang, Wu, and Yang, 2013). To consider this aspect, and its influence on the relationship between oil and stock returns, their study highlights how this factor affects the performance of stocks (Wang et al. 2013). The results of Wang et al. (2013) highlight a positive correlation between oil and stock returns in exporting countries, but not among those in oil-importing countries. (Wang et al., 2013). According to the authors, there is no significant effect of changes in oil returns on stock returns in oil-importing countries. The Netherlands, being a special case, is a net importer of crude oil but one of the major natural gas exporters worldwide. Therefore, it is not certain what the effect of oil returns on stock market returns will be. However, based on the prior studies shown in Table 1, it can be hypothesized that the relationship between oil returns and stock returns is likely to be negative, while the relationship between oil returns and stock returns of the Oil & Gas companies is likely to be positive. Therefore, based on this expectation, I test the following hypotheses:
H1(a): Stock returns for the market index are negatively related to oil returns
In addition, the study will also reveal that a financial crisis has negative consequences on Dutch stock returns. Furthermore, there will also be a negative effect of the interest and exchange rates on the stock price returns.
Chapter 3: Research Method
In this section, I first explain how the data is collected and refined, and then mention the research methodology used.
Data
The study will rely on the monthly oil market and stock market data from 2005 to 2020. This duration is significant since it helps demonstrate the impact of various economic shocks in the market, such as the financial crisis of 2008. The choice of this sample period is also guided by the availability of data. Data will be retrieved from major financial databases, such as the OPEC website, and other reputable and peer-reviewed databases, such as Yahoo Finance.
For instance, oil prices are collected from the OPEC website. For international oil prices, the paper will rely on the OPEC Reference Basket (ORB), which reflects the average price of crude oil in OPEC member countries and includes as of 2005 eleven Member Country crude oil streams. Additionally, the Amsterdam Exchange Index (AEX-index) data is retrieved from Yahoo Finance. The AEX-index is used as a proxy for the Dutch stock market. The AEX-index consists of the 25 stocks with the largest market capitalization of the Amsterdam Stock Exchange.
The AEX-index also contains the four Oil & Gas companies used as an index for the Dutch Oil & Gas companies. These companies are Fugro, SBM Offshore, Royal Dutch Shell A, and Koninklijke Vopak. The data for energy companies is also collected from Yahoo Finance. Using the collected data, the monthly returns of the AEX-index and monthly returns on the oil prices are computed. The important variables in the analysis are explained in Table 2.
Table 2: Variable explanation
Table 3: Descriptive Statistics
Variable Number of
observations
Mean Std. Dev. Min Max
RttAEX 176 0.0024 0.049 -0.197 0.111 RttCompanyoil 176 0.0043 0.094 -0.293 0.161 RttOil 176 -0.0001 0.059 -0.422 0.271 GDPt 176 99.4047 5.047 91.424 111.499 Interest_ratet 176 1.0892 1.655 -0.418 5.113 Exchange_rate t 176 1.2667 0.127 1.055 1.576 Variable explanation
RttAEX Monthly return of the AEX-index
RttCompanyoil Monthly return of the Oil & Gas companies
RttOil Monthly return of the ORB
Financialcrisis1 Dummy variable of the financial crisis
GDPt Control variable of Dutch GDP
Interest_ratet Control variable of Dutch interest rate
In the dataset from 2005 until 2020 I have 176 observations for all of my variables. This can also be seen in the descriptive statistics table (see Table 3) which contains important information on the variables used in the analysis. As can be seen, the average return on the AEX-index (RttAEX) is 0.24%, with a high standard deviation of 4.9%, indicating a high fluctuation. The average return on oil companies (RttCompanyoil) is higher than the AEX-index but also has a higher standard deviation. The explanation for this is that there is more diversification in the AEX-index, and therefore a lower risk. The minimum return of these companies is -29.3% while the maximum return during the sample period was 16.1%. Return on oil (RttOil) is on average much less, around -0.01% with a standard deviation of 5.9%. The mean GDP is 99.4 10s of billions of Euro, with a comparatively lower standard deviation of 5.1 10s of billion Euro. The interest rate is 1.09% on average, ranging from -0.413% to 5.113%. The Euro/Dollar exchange rate has a mean of 1.27 with a low standard deviation of 0.127, indicating that the exchange rate has not fluctuated a lot.
Table 4: Correlations Matrix
Variable RttAEX RttCompanyoil RttOil Financialcrisis1 GDPt Interest_ratet Exchange_ratet
RttAEX 1 RttCompanyoil 0.420 1 RttOil 0.081 0.086 1 Financialcrisis1 -0.260 -0.081 -0.055 1 GDPt -0.023 -0.130 -0.046 -0.238 1 Interest_ratet -0.197 -0.023 0.003 0.552 -0.691 1 Exchange_ratet -0.086 0.016 0.151 0.438 -0.640 0.629 1
The correlation between the various variables is described above in Table 4. In this table, a correlation of 0.420 between the AEX-index stock returns and the Oil & Gas company stock returns is shown. The Oil & Gas companies are also part of the AEX-index, thus it is expected that their returns would have a higher and positive correlation. Furthermore, I find a correlation of 0.081 between the oil returns and the AEX-index stock returns, and a correlation of 0.086 between the oil returns and the Oil & Gas companies stock returns. The correlations are relatively low, and the positive sign of both correlations is not suggested by prior literature. The correlation of the financial crisis variable shows a negative sign with all the following variables: the AEX-index return, the Oil & Gas companies' returns, and the oil returns. This is in line with expectations, as a financial crisis is likely to reduce the return on all investments. Moreover, the interest rate has a relatively high positive correlation with the financial crisis. This is counterintuitive because the central bank generally lowers the interest rates during financial
crises. The GDP has a negative correlation with all the returns (AEX-index, Oil & Gas companies, and oil). These correlations are an anomaly as, during GDP growth, returns are expected to be higher. The correlation between GDP and financial crisis is negative, as expected. Finally, the interest rate and the exchange rate are positively correlated. This is in line with economic reasoning, as a higher interest rate attracts foreign investment, thereby increasing the demand for the home currency and causing an appreciation.
I also visualize the most important variables in my data. Figure 1 shows the AEX-index trends during the sample period, while figure 2 shows the return on the Oil & Gas companies during the same sample period. Both of these graphs show a similar trend. There is high fluctuation during the sample period. Most importantly, a sharp decrease can be seen in both these graphs during two time periods, the 2008 financial crisis and the 2020 coronavirus crisis. During the financial crisis, the returns on the AEX-index and the Oil & Gas companies decreased by nearly 20%, while part of the impact of the coronavirus crisis on the Oil & Gas companies, that can be seen in my dataset, was more severe than the AEX-index.
Fig. 1. Trends in AEX-index Returns
Finally, figure 3 shows the returns on oil, which also follow a similar trend to the AEX-index. However, the fall in returns for oil had the highest magnitude, where oil returns reduced to nearly -40% during the financial crisis of 2008.
Methodology
To examine if oil prices have an effect on Dutch stock returns, review of the literature showed that there are two main statistical analyses used. The first is the OLS regression, which is used by various researchers, such as Driesprong et al. (2008) and Broadstock et al. (2012). The other method often used is the Vector Autoregression Model, as can be seen in the following studies: Sadorsky (1999), Kilian and Park (2009), and Fang and You (2014).
For this research, I use the OLS regression analysis, based on the work of Asteriou (2013). The model used to test whether the crude oil prices have any effect on the Dutch stock prices, is as follows:
𝑅𝑡𝑡𝐴𝐸𝑋= β0 + β1* 𝑅𝑡𝑡𝑂𝑖𝑙+ β2 * Financialcrisis1 + β3 * GDPt + β4 * Interest_ratet +
β5 * exchange_ ratet + 𝜀t (1)
Where 𝑅𝑡𝑡𝐴𝐸𝑋 is the return of the AEX-index at time t and 𝑅𝑡𝑡𝑂𝑖𝑙 is the monthly return of crude oil at time t. The constant is the intercept, which is the value of the dependent variable when all independent variables are equal to zero. The coefficient β1 gives the effect crude oil has on the AEX-index. Financialcrisis1 is a control variable that is assigned 1 when the time period
is between December 2007 and June 2009 and is assigned 0 otherwise. Goodman and Mance (2011) gave an overview of the financial crisis, in which they stated that the economy officially entered a recession in December 2007. This recession lasted until June 2009. The coefficient β2 gives the effect of the financial crisis on the relationship between oil returns and stock returns.
time t. Interest_ratet controls for the different interest rates in the Netherlands at time t. The last
control variable I test for is exchanges_ratet, which is the EURO-Dollar exchange-rate at time t.
These control variables are macroeconomic factors that affect stock price movements according to various researchers such as Nasseh and Strauss (2000) and Geetha et al. (2011). These variables may have a non-random influence on the stock returns, therefore it is important to control for them.
The same model is used to test the effect of crude oil prices on Oil & Gas companies, except now the dependent variable will be 𝑅𝑡𝑡𝐶𝑜𝑚𝑝𝑎𝑛𝑦𝑂𝑖𝑙 instead of 𝑅𝑡𝑡𝐴𝐸𝑋.
𝑅𝑡𝑡𝐶𝑜𝑚𝑝𝑎𝑛𝑦𝑂𝑖𝑙= β0 + β1 ∗ 𝑅𝑡𝑡𝑂𝑖𝑙+ β2 * Financialcrisis1 + β3 * GDPt + β4 * Interest_ratet +
Results
This section presents the results of the linear regression analysis conducted, as explained in the methodology section. The oil return will be the main independent variable and the stock return has to be explained so it’s the dependent variable. Other variables will serve as control variables. Before presenting the actual analysis, I examine the OLS regression diagnostics. These are some of the assumptions of linear regression, such as homoscedasticity, normality, and no multicollinearity.
I first conduct a Breusch-Pagan test to check for homoscedasticity. Homoscedasticity suggests the error term is evenly distributed around the mean of zero. When heteroskedasticity exists, these standard errors are not constant. The results of the test are shown in Table A1 (χ2 = 24.04, p<0.01). This suggests that the null hypothesis is rejected, and therefore the error terms are heteroskedastic. To account for this I will use robust standard errors in all of my regressions, thus keeping the error terms constant.
Furthermore, I conduct the Jarque-Bera Normality test to find out whether the sample data is normally distributed. The test checks whether the skewness and kurtosis match that of a normal distribution. The results of the test are in Table A2. All the variables have a significant χ2,
therefore suggesting that the null hypothesis is rejected. This implies that I cannot conclude that the variables follow a normal distribution. However according to the Central Limit Theorem (Kwak & Kim, 2007), when the number of observations is more than 30, the results of the regressions may be considered reliable. To make my data follow a normal distribution, I tried
taking the logs of my variables. But then the outcomes of the Jarque-Bera Normality test, as can be seen in Table A3, are still significant and therefore do not follow a normal distribution.
Finally, I test for multicollinearity, which is a problem that exists when conducting multiple linear regressions. Multicollinearity exists when the independent variables are highly correlated. This could result in unreliable regression estimates. Multicollinearity can be detected from the Variance Inflation Factors (VIF). According to Kumar (2020), a VIF score of higher than 5 is considered to be problematic. However, Stock & Watson (2019) stated that the VIF score has to be higher than 10 for it to be problematic. The VIF scores of my regression models are presented in Table A4. As can be seen, all individual VIF scores for the variables are below 5 and the mean VIF score is 2.02. Based on these figures it can be inferred that my regression models do not suffer from the multicollinearity problem.
The regression diagnostics presented above are for the first regression model. The diagnostics show that the model does not completely violate the assumptions of linear regression, except for the normality assumption. However, with large enough data, that is also not a very significant problem. Thus, I proceed with the results of the regression analysis. I also conduct the regression diagnostics for my second regression model, and the results do not differ from the diagnostics for the first regression as I use the same independent variables. Therefore, regression diagnostics for the second model are not reported here.
Table 5 Regression output Dependent Variable 𝑅𝑡𝑡𝐴𝐸𝑋 𝑅𝑡𝑡𝑂𝑖𝑙 0.031 (0.76) Financialcrisis1 -0.022 (-1.05) GDPt -0.003** (-2.10) Interest_ratet -0.008** (-2.17) Exchange_ratet -0.008 (-0.19) intercept 0.278* (1.86) 𝑅2 0.099 Observations 176
Note: * p < 0.1, ** p < 0.05, *** p < 0.01. The main coefficient is the regression coefficient (β), and parentheses contain the t-statistics. The regression output is with the dependent variable 𝑅𝑡𝑡𝐴𝐸𝑋 and all the control variables included.
The coefficient of each independent variable shows the impact it has on the dependent variable, while t-statistics and the p-values show the significance. A t-statistic with an absolute value higher than 1.96, or a p-value smaller than 5% shows that the relationship is significant. The R2 value states how much of the dependent variable is explained by the model. The R2 is always between 0 and 1 and the higher the R2 value, the more the model can explain the outcome of the dependent variable.
Table 5 shows the regression output (F(5,170) = 1.73, p=0.13), with an R-squared of 0.099, where AEX-index returns is the dependent variable. This implies that the model can explain only 9.9% of the variation in the dependent variable. In the OLS regression, I find the
effect oil returns have on the stock market prices is positive, as seen with the positive coefficient of 0.031. This indicates that when the ORB price increases by 1% the AEX-index increases with 0.031%. This finding is not in line with H1(a). Based on the literature, I hypothesized a negative relationship. Compared to prior literature this effect is smaller and positive instead of negative (Driesprong et. al. 2008). However, the coefficient of 𝑅𝑡𝑡𝑂𝑖𝑙 is not significant, meaning it does not differ significantly from zero (p=0.446). As a result, I reject H1(a) as I do not find a negative relationship between AEX-index stock returns and oil returns. Moreover, the financial crisis variable has a negative coefficient of -0.022. This is in line with the expectations, as a financial crisis is expected to lower the economic return on assets. However, this coefficient is also not significant.
Besides, the variable GDPt had a coefficient of -0.003. This negative coefficient was not expected as stock returns tend to go up when GDP increases. Moreover, this coefficient is significantly different from zero with a p-value of 0.031, so it can be concluded that GDP had a negative influence on the stock market returns in our sample. The Interest_ratet has a coefficient of -0.008 which differs significantly from zero with a p-value of 0.037. The negative coefficient is intuitive and as expected, as higher interest rates would motivate investors to save rather than investing, and therefore stock returns will decrease. The last variable Exchange_ratet has a coefficient of -0.008. This negative coefficient was not expected, because an appreciation of the home currency will in general attract foreign investments in stocks. However, the coefficient of Exchange_ratet found in my regression is not significant.
Table 6 Regression output Dependent Variable 𝑅𝑡 𝑡 𝐶𝑜𝑚𝑝𝑎𝑛𝑦𝑜𝑖𝑙 𝑅𝑡𝑡𝑂𝑖𝑙 0.052 (0.89) Financialcrisis1 -0.004 (-0.20) GDPt -0.003** (-2.05) Interest_ratet -0.006 (-0.173) Exchange_ratet -0.031 (-0.64) intercept 0.392* (1.91) 𝑅2 0.050 Observations 176
Note: * p < 0.1, ** p < 0.05, *** p < 0.01. The main coefficient is the regression coefficient (β), and parentheses contain the t-statistics. The regression output is with the dependent variable 𝑅𝑡𝑡𝐶𝑜𝑚𝑝𝑎𝑛𝑦𝑜𝑖𝑙 and all the control variables included.
Next, using the Oil & Gas companies’ returns as the dependent variable, A significant regression was not found (F(5,170) = 0.96, p=0.4426) with an R-squared of 0.05. This implies that my model is only able to explain 5% of the variation in the dependent variable. In the OLS regression (Table 6), I find that the effect oil returns have on the stock market returns is positive, as seen with the positive coefficient of 0.052. This indicates that when the ORB price increases by 1% the return on the AEX-index increases with 0.052%. This is in line with prior literature, and this finding lends support to H1(b). However, the coefficient of 𝑅𝑡𝑡𝑂𝑖𝑙does not differ significantly from zero (p=0.375), thus a relationship is not found. Compared to prior literature,
this effect is smaller, as it is not significant. Moreover, the financial crisis variable has a negative coefficient of -0.004. This is in line with the expectations, as a financial crisis is expected to lower the economic return on assets. However, this coefficient was not found to be significant.
Also, the variable GDPt had a coefficient of -0.003. This negative coefficient corresponds with the GDP coefficient found in the previous regression which is likely because the two dependent variables are relatively highly correlated. This coefficient is again found to be significant thus it’s safe to say that GDP had a negative influence on the stock market returns in our sample. Moreover, the Interest_ratet has a coefficient of -0.006 which does not differ significantly from zero. This negative coefficient is also intuitive, as seen before. The last variable Exchange_ratet has a coefficient of -0.031 which is negative and not as expected. However, the coefficient of Exchange_ratet found in my regression is not significant.
The significant and negative coefficient of GDP in both models is especially interesting as a higher GDP is expected to increase stock returns, as suggested by prior literature. I plotted the price trend line of both GDP and the AEX-index returns in one graph (fig. 4), this gave me insights into how they correlate over time. One possible explanation of this negative coefficient I found is the high volatility of the AEX-index and the financial crisis. The AEX-index returns decreased a lot during the 2008 financial crisis, as mentioned earlier the decrease was nearly 20%. During this time GDP remained quite constant and thus did not decrease, therefore the effect of GDP on the AEX-index can differ from what I found in previous literature.
Fig. 4. AEX-index price in € and GDP price in 10s of billions of € through time.
Conclusion
This study sheds light on the relationship between oil returns and Dutch stock market returns, as well as the relationship between oil returns and stock returns of Oil & Gas companies. I use data from August 2005 until April 2020, comprising 176 observations for the various variables of interest such as stock returns, oil returns, GDP, interest rate, and exchange rate. I used the OLS regression analysis to answer my research question: “Do oil prices have an impact on the stock returns in the Netherlands?” The results from the OLS regressions where insignificant so they suggest that a price change of the ORB will not have a significant impact on stock returns for both the Oil & Gas companies and the market index. Therefore, to answer my research question, I conclude that oil prices do not have an impact on the stock returns in the Netherlands.
Prior literature concludes that there is a difference in the effect oil prices have on stock returns in oil-exporting and oil-importing countries (Wang et al., 2013). Their study highlights a positive correlation between oil prices and stock returns for oil-exporting countries. However, for oil-importing countries, Wang et al. (2013) found a non-significant effect. If we consider the Netherlands as being an oil-importing country, we can conclude that the findings of this thesis are in line with the work of Wang et al. (2013). However, a majority of other papers discussed in the literature review found negative effects of oil returns on stock returns. My results are contradictory to these papers, as I do not find a negative relationship. This also highlights the contribution of my study, as prior literature has not examined the Dutch AEX-index returns as the dependent variable. The focus on the Netherlands, as mentioned previously, is due to the special position it has concerning oil-importing countries. Also, I find that the oil prices do not
significantly affect Dutch Oil & Gas companies, which is not in line with the positive correlation Arouri (2011) found.
Moreover, this research will allow Dutch Oil & Gas companies to understand that a very strong relationship between oil returns and their stock returns may not exist. This may be due to some sort of diversification strategy or perhaps because of the special position of the Netherlands in regards to importing vs. exporting oil. The companies in this sector, therefore, do not have to value the oil prices in the Netherlands as much, regarding their stock returns.
One of the limitations my research suffers from is the relatively smaller sample size. Secondly, my regression may suffer from an omitted variable bias, as other researchers used the dividend yield, and government bond yields, among others, as control variables. Thirdly, because of the special position of the Netherlands concerning importing and exporting oil and gas, my results may not be generalizable to other countries, that are for instance net importers.
My recommendations for future research build on the limitations of my research. I would suggest that further literature would use daily data to account for this limitation. The number of observations will increase, and as a result, the findings may become significant as suggested by prior literature. Additionally, further research can examine the price changes of gas as an independent variable. This would be interesting because as mentioned before, the Netherlands is a net exporter of natural gas. Finally, I would recommend other researchers to check if the Dutch stock market reacts efficiently to oil price changes by examining the lagged effect. Given the fact that the effect of current versus lagged oil returns, changes significantly on stock returns. A more explicit model that accounts for the lag must be used to determine whether the Dutch stock market reacts efficiently on oil returns. For this future study, a VAR model should be considered
as many other researchers used this model. However, the overall results of this thesis suggest that oil prices have no significant effect on stock returns in the Netherlands.
References
Apergis, N., & Miller, S.M., (2009). Do structural oil-market shocks affect stock prices? Energy
Economics, 31(4), pp. 569-575.
Arouri, M.E.H., (2011). Does crude oil move stock markets in Europe? A sector investigation.
Economic Modeling: 28(4), pp. 1716-1725.
Asteriou, D., & Bashmakova, Y. (2013). Assessing the impact of oil returns on emerging stock markets: A panel data approach for ten Central and Eastern European Countries. Energy
Economics, 38, pp. 204-211.
Broadstock, D.C., Cao, H., & Zhang, D. (2012). Oil shocks and their impact on energy related stocks in China. Energy Economics, 34(6), pp. 1888-1895.
Driesprong G., Jacobsen B., and Maat B., (2008). Striking Oil: Another puzzle? Journal of
Financial Economics, 89, pp. 307–327
Elyasiani, E., Mansur, I., Odusami, B. (2011). Oil price shocks and industry stock returns.
Energy Economics, 33(5), pp. 966-974.
Fang, C., & You, S. (2014). The impact of oil price shocks on the large emerging countries’ stock prices: Evidence from China, India and Russia. International Review of Economics
and Finance, 29, pp. 330-338.
Geetha, C., Mohidin, R., Chandran, V., & Chong, V. (2011). The relationship between inflation and stock market: Evidence from Malaysia, the United States, and China.
Goodman, C., & Mance, S. (2011). Employment loss and the 2007–09 recession: An overview. Monthly Labor Review, pp. 3-12.
Huang, R.D., Masulis, R.W., and Stoll, H.R. (1996). Energy shocks and financial markets.
Journal of Futures Markets, 16 (1), pp. 1-36.
Jones, C., & Kaul, G. (1996). Oil and the Stock Markets. The Journal of Finance, 51(2), pp. 463-491. DOI:10.2307/2329368
Kilian, L., & Park, C. (2009). The impact of oil price shocks on the US stock market. International Economic Review, 50(4), pp. 1267-1287.
Kumar, K. N. R. (2020). Econometrics. Taylor & Francis, pp. 514-585.
Kumar, S., Managi, S., Matsuda, A. (2012). Stock prices of clean energy firms, oil and carbon markets: A vector autoregressive analysis. Energy Economics, 34(1), pp.215-226.
Kwak, S. G., & Kim, J. H. (2017). Central limit theorem: the cornerstone of modern statistics.
Korean Journal of anesthesiology, 70(2), pp. 144–156.
Narayan, P.K., & Narayan, S. (2010). Modeling the impact of oil prices on Vietnam’s stock prices. Applied Energy, 87 (1), pp. 356-361.
Nasseh, A. & Strauss, J. 2000. "Stock prices and domestic and international
macroeconomic activity: a cointegration approach," The Quarterly Review of
Economics and Finance, Elsevier, vol. 40(2), pp. 229-245.
Park, J., & Ratti, R.A. (2008). Oil price shocks and stock markets in the U.S. and 13European countries. Energy Economics, 30(1), pp. 2587-2608.
Sadorsky, P. (1999). Oil price shocks and stock market activity. Energy Economics, 21, pp. 449-69.
Stock, J. H., & Watson, M. W. (2019). Introduction to Econometrics, Global Edition (4th edition). Pearson Education Limited.
Wang, Y., Wu, C., & Yang, L. (2013). Oil price shocks and stock market activities: Evidence from oil-importing and oil-exporting countries. Journal of Comparative Economics, 41(4), pp. 1220-1239.
Appendix
Heteroscedasticity test
The following hypothesis tested: H0: no heteroscedasticity H1: heteroscedasticity Table A1: Chi-squared (χ2) 24.04*** Prob>χ2 0.000 Note: * p < 0.1, ** p < 0.05, *** p < 0.01.
Jarque-Bera test for normality
The following hypothesis tested: H0: normally distributed
H1: not normally distributed Table A2:
Variable Chi-square Prob>Chi-square
𝑅𝑡𝑡𝑂𝑖𝑙 23.83*** 0.000
Interest_ratet 22.41*** 0.000
GDPt 14.30*** 0.001
Exchange_ratet 10.47*** 0.005
Note: * p < 0.1, ** p < 0.05, *** p < 0.01.
Jarque-Bera test for normality, with logs of all the variables.
The following hypothesis tested: H0: normally distributed
H1: not normally distributed Table A3:
Variable Chi-square Prob>Chi-square
𝑅𝑡𝑡𝑂𝑖𝑙 48.46*** 0.000
Interest_ratet 43.11*** 0.000
GDPt 12.94** 0.001
VIF scores to check for multicollinearity
Table A4:
Variable VIF score
𝑅𝑡𝑡𝑂𝑖𝑙 1.05 Interest_ratet 2.83 GDPt 2.45 Exchange_ratet 2.13 Financialcrisis1 1.64 Mean VIF 2.02
