OIL PRICE RETURNS ON STOCK MARKET PERFORMANCE: EVIDENCE FROM OPEC MEMBERS

(1)

OIL PRICE RETURNS ON

STOCK MARKET

PERFORMANCE:

EVIDENCE FROM OPEC

MEMBERS

10-08-2018

Mike van Oudshoorn

s2562898

Abstract

This paper studies the effect of oil price changes on stock market performance of nine OPEC member countries from 1990 to 2017. From a pooled OLS regression I find that OPEC member countries’ stock market performance is significantly positively affected by changes in oil prices in the financial crisis period from 10-2008 to 01-2015. Country-specific analysis explains that this result is mainly driven by Iran, Nigeria, Qatar, Saudi Arabia and the United Arab Emirates. The findings in this paper together with literature on OPEC’s ability to influence oil prices serves as a foundation for research on OPEC’s power to influence its own stock markets.

University of Groningen

(2)

1

1. Introduction

Oil has been a major factor in the world economy for many decades and still is today. Changes in oil prices influence economies all around the globe. In February 2016, the price of a barrel of crude WTI oil was approximately $29.44, whereas a barrel in February 2018 comes at a price of $65.45.1 This fluctuation in price levels might have a significant impact on international financial markets. Many studies aim to tie oil price changes to stock market returns. Examples are Basher & Sadorsky (2006), Fang & You (2014), Jones & Kaul (1996), Scholtens & Yurtsever (2012), Arouri et al. (2011) and Bjørnland (2009). These and many other studies review the impact of oil price changes on stock market performance of several countries or regions. To my knowledge there is no previous literature on the impact of oil price changes on the OPEC’s stock market performance as a region, whereas this region is responsible for a large part of the world’s crude oil production and exports. The Organisation of the Petroleum Exporting Countries (OPEC) is an organisation of 13 countries that are large producers of oil. The countries included in this organisation are Algeria, Angola, Ecuador, Indonesia, Iran, Iraq, Kuwait, Libya, Nigeria, Qatar, Saudi Arabia, United Arab Emirates and Venezuela. These countries combined were responsible for 56.6% of the world’s crude oil exports in 2016.2 OPEC has policy’s affecting all its member countries. Guidi et al. (2006) finds that these policy decisions significantly influence oil prices. Moreover, Kaufmann et al. (2004) finds that OPEC can directly influence oil prices by setting production levels and capacity utilisation in their oil plants. If these oil prices in turn affect OPEC’s stock market performance, these relationships would imply that OPEC is able to indirectly influence its own stock market performance. This information should be available to investors and academics to be taken into account when investing in or performing research on this region. Therefore this paper aims to model the effect of oil price changes on OPEC’s stock market performance.

The central topic in this paper is to study the impact of oil prices on stock market returns in OPEC member countries. The aim is to provide empirical evidence about the relationship between oil price returns and stock market returns in the OPEC region. This relationship adds to existing literature on oil price returns and stock market returns at country and region level. The key question that arises from this purpose is formulated as follows: “Is OPEC’s stock market performance influenced by oil price changes?” If oil price changes indeed influence OPEC’s stock market performance, then this result in combination with the findings of Kaufmann (2004) and Guidi et al. (2006) would imply that OPEC is able to influence its own stock market performance. To answer this question I examine the effect of WTI oil price changes on a common sample of OPEC country stock market returns using a multi-factor model inspired by Basher & Sadorsky (2006), Wang et al. (2013) and Kang et al. (2015). I use daily and monthly time series data on country stock market returns, oil prices and global stock market returns to determine the statistical effect of oil price changes on stock market performance. A yearly time series analysis also includes a measure for inflation and changes in production level. The research period ranges from January 1990 to October 2017. The modelled effect of oil price changes on OPEC’s stock market performance represents

(3)

2

the influence of oil price changes on stock market performance. Based on this impact the research question and therewith OPEC’s ability to influence its own stock markets are evaluated.

The remainder of the paper is structured as follows: Section 2 discusses existing literature relevant to this study. Section 3 explains the construction of the model. Section 4 introduces the data. Section 5 presents the results of the empirical analysis and Section 6 concludes.

2. Literature

There are many studies that aim to relate oil price shocks to stock market performance. Some focus on country level financial performance, others study the sensitivity of oil price changes on a variety of sectors within a country or region. Basher & Sadorsky (2006) study whether oil price risks affect emerging stock markets, whereas Faff & Brailsford (1999) research the effect of oil price risks on different sectors in the Australian stock market, and Scholtens & Yurtsever (2012) focus on 38 industries in Europe. Jones & Kaul (1996) is one of the pioneering studies on the relationship between oil and stock markets. In their study they explain the reactions of four OECD countries to oil shocks. They find that stock market reactions of the United States and Canada can fully be explained by the impact of these shocks on real cash flows. However, for the United Kingdom and Japan they find that changes in oil prices cause larger changes in stock prices than can be explained. Miller & Ratti (2009) add Germany, France and Italy to this sample and find evidence for a negative long-term relationship between oil price and stock market performance. Over the post-war estimation period they find heterogenic results of this relationship over time among the countries studied. They find sign changes and insignificance in periods from 1980 to 1988 and from 1999 onwards, which they explain as ‘breaks’. For a detailed description of these breaks see Miller & Ratti (2009). Sadorsky (1999) builds upon Jones & Kaul (1996), also presenting a clear effect of oil prices on stock returns. Kang et al. (2015) covers a more recent estimation period for their study on oil price shocks on U.S. stock market returns and volatility. Their paper covers 1973 to 2013 using monthly data to find that increasing U.S. oil-market specific demand is associated with negative effects on their stock markets. The U.S. is one of the largest producers of crude oil. However, as one of the largest consumers as well, importing significantly more oil than they export. With the U.S. as a significant net importer of crude oil, I expect an opposing effect for OPEC member countries, being large exporters of oil. Scholtens &Yurtsever (2012) study how oil shocks influence 38 different industries in Europe. They find that oil price shocks influence firm value by the impact on the discount rate and expected earnings. Their results suggest that positive oil price shocks in general negatively influence industries, but that these shocks are beneficial to oil intensive industries.

(4)

3

prices are driven by Russian supply-shocks. OPEC member countries, as large producers of crude oil, are likely to be sensitive to supply driven oil price shocks as well. According to the findings from Fang & You (2014), the relationship seems to be dependent on whether a country is net oil-importing or oil-exporting.

Indeed, Bhar & Nikolova (2009) confirms that this relationship is highly dependent on whether the country is oil-importing or oil-exporting, in their study on the influence of global oil prices on stock price creation processes and volatility in BRIC markets. If the country is net oil-exporting, oil prices positively influence stock market returns in most cases. Subsequently, if the country is net oil-importing, the relationship between oil and stock market performance in general is negative. Bjørnland (2009) elaborates on oil price shocks and stock market returns in an oil exporting country; Norway. He finds that an increase in oil price of 10% leads to an immediate increase in stock market returns by 2.5%. OPEC, as net exporters of oil as well, are therefore expected to react positively to oil price changes.

There are a number of studies available in academic literature that focus on OPEC’s ability to influence the world’s crude oil prices. Kaufmann et al. (2004) studies the ability of OPEC to influence oil prices through production factors. The authors conduct both a VAR and OLS analysis to find a statistically significant relationship between oil prices, and OPEC capacity utilisation, stocks of crude oil, quotas and the degree of OPEC exceeding production quotas. Their results indicate that OPEC possesses a significant power over oil prices through these production policies. They conclude that these factors ‘Granger cause’ real oil prices but that these oil prices do not ‘Granger cause’ these factors. An increase in production quota or the exceedance of these quotas results in a decrease of oil prices, whereas an increase in capacity utilisation results in an increase of oil prices. This implies that OPEC is able to influence the world’s oil prices by setting these production factors. Moreover, Guidi et al. (2006) finds that OPEC policy decisions does significantly influence U.S. and U.K stock markets, both directly and through oil prices. The authors conduct an event study to find that oil markets provide a positive return to OPEC decisions to decrease oil production, and thus world oil supply. Kilian (2009) also implies that oil prices are driven by structural demand and supply shocks. These findings, in combination with this study and existing literature, suggest that OPEC is possibly able to influence its own stock markets by policies on production. Investors and academics should be aware of this when making capital investment decisions or assumptions. An important note here is that Kaufmann et al. (2004) is conducted approximately 14 years ago and that this OPEC market power might be different now than it was over a decade ago. Kaufmann et al. (2004) does argue that the extend to which the OPEC is able to influence oil prices seems to decline due to increasing power of other oil producers. OPEC’s market power may be negatively influenced by the increasing supply of oil by for example BRIC, and most notably Russia. Basher & Sadorsky (2006) follows up on this, since at the time their study was published Russia was a larger producer of oil than Saudi Arabia, even though Russian oil reserves were smaller at the time. In 2017 Russia produced 10.097 million barrels of crude oil per day, whereas Saudi Arabia produced 10.020 million barrels of crude oil per day.3 Following up on the importance of oil production to stock market returns as Guidi et al. (2006) discusses, Kang et

(5)

4

al. (2015) argue that an unanticipated reduction in oil production is results in an increase in implied-covariance of return and volatility. Moreover, they claim that global oil production predicts this relationship increasingly in recent years. They find that on average, shocks to crude oil production explain 12.4% of the variation on the implied-volatility of returns in the U.S..

A similar cooperation to the OPEC is the political and economic union Cooperation Council for the Arab States of the Gulf, also known as the Gulf Cooperation Council (GCC). Four of the largest OPEC oil producing countries (Kuwait, Qatar, Saudi Arabia and the United Arab Emirates) are also a member of this union. Arouri & Rault (2012) study the influence of oil prices on the members of this union using SUR methods and bootstrap panel cointegration techniques. They find that all of these GCC member countries’ stock markets positively react to increases in oil prices, except for Saudi Arabia. This is remarkable, as Saudi Arabia is the largest producer and exporter of crude oil. The authors explain this by claiming that the Saudi Arabian stock market is largely dominated by the financial industry, which is highly linked to the U.S. and European markets. Moreover, they argue that Saudi Arabia, as largest producer, also suffers the most from imported inflation and economic pressure. However, results differ among his own studies. In Arouri et al. (2011) he finds that only three out of six GCC countries experience significant volatility spillovers between oil and stock markets. This can be a consequence from using different empirical estimation methodology. Based on the literature above it is highly likely that OPEC member’s stock market performance positively depends on oil price returns. Stock markets of large producers and exporters of oil are expected to be positively influenced by oil price increases. However, it is possible that reactions to oil price changes vary among the member countries. It is plausible to expect that not all OPEC member’s stock markets are significantly influenced by oil price returns. According to Kaufmann et al. (2004) it is useful to highlight which member countries are significantly influenced and which are not. To study the relationship between OPEC’s stock market returns and oil price returns, the following hypotheses are constructed:

𝐻₀: 𝑇ℎ𝑒 𝑤𝑜𝑟𝑙𝑑′_{𝑠 𝑐𝑟𝑢𝑑𝑒 𝑜𝑖𝑙 𝑝𝑟𝑖𝑐𝑒𝑠 ℎ𝑎𝑣𝑒 𝑛𝑜 𝑠𝑖𝑔𝑛𝑖𝑓𝑖𝑐𝑎𝑛𝑡 𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑛 𝑡ℎ𝑒 𝑠𝑡𝑜𝑐𝑘 𝑚𝑎𝑟𝑘𝑒𝑡} 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑜𝑓 𝑂𝑃𝐸𝐶 𝑚𝑒𝑚𝑏𝑒𝑟 𝑐𝑜𝑢𝑛𝑡𝑟𝑖𝑒𝑠

𝐻1: 𝑇ℎ𝑒 𝑤𝑜𝑟𝑙𝑑′𝑠 𝑐𝑟𝑢𝑑𝑒 𝑜𝑖𝑙 𝑝𝑟𝑖𝑐𝑒𝑠 ℎ𝑎𝑣𝑒 𝑎 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑛 𝑡ℎ𝑒 𝑠𝑡𝑜𝑐𝑘 𝑚𝑎𝑟𝑘𝑒𝑡 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑜𝑓 𝑂𝑃𝐸𝐶 𝑚𝑒𝑚𝑏𝑒𝑟 𝑐𝑜𝑢𝑛𝑡𝑟𝑖𝑒𝑠

A rejection of the null hypothesis answers the key question of this research, and therewith builds upon the available literature on oil prices and stock market returns. Moreover, this answer sheds light on to what extend OPEC is able to influence its own stock market returns through setting oil production factors.

3. Methodology

(6)

5

following Basher & Sadorsky (2006). In their study these stock market returns, as dependent variable, are explained by oil price returns as the independent variable.To accurately measure the effect of oil price changes on stock market returns Basher & Sadorsky (2006) incorporate a global market index to their model. They add the global Morgan Stanley Capital International (MSCI) World Index to their model as a benchmark for global economic trends. They find that this variable has a negative relation to the model significant at a 5% level, against their expectations. The returns in this model are nominal values. In a long-term perspective, these returns may be affected by inflation, as suggested by Wang et al. (2013). To correct for inflation bias in the data I decide to add the GDP deflator to the model. Following Kang et al. (2015) and Kaufmann et al. (2004) I decide to add production levels to the model, as these levels tend to influence returns according to their findings. To select a proper estimation technique I research whether the data satisfies the Gauss-Markov assumptions as described in Brooks (2014). With these assumptions satisfied, the Ordinary Least Squares (OLS) regression method qualifies as a proper estimator of the model. As these Gauss-Markov assumptions are indeed satisfied, the OLS estimator is the Best Linear Unbiased Estimator (BLUE). Therefore I will perform an OLS regression to estimate the model in this study. This estimation method is also in line with Basher & Sadorsky (2006), which the model is inspired by. The regression model used to access the statistical relationship is described in Eq. (1) below:

𝑅𝑒𝑡𝑖𝑡 = 𝑐(1) + 𝑐(2)𝑂𝐼𝐿𝑡+ 𝑐(3)𝑀𝑅𝑒𝑡𝑡+ 𝑐(4)∆𝐷𝑒𝑓𝑙𝑖𝑡+ 𝑐(5)∆𝑃𝑟𝑜𝑑𝑖𝑡+ 𝜀𝑖𝑡 (1) 𝑅𝑒𝑡_𝑖𝑡 represents the stock market index returns of country i at time t, 𝑂𝐼𝐿_𝑡 are WTI crude oil returns, 𝑀𝑅𝑒𝑡_𝑡 are the MSCI World Market Index returns, 𝐷𝑒𝑓𝑙_𝑖𝑡 is the country’s GDP deflator and 𝑃𝑟𝑜𝑑_𝑖𝑡 represents oil production levels. Country returns, GDP deflators and oil production data are at country level, whereas oil returns and global market returns are at a global level. The model presented above is inspired by Basher & Sadorsky (2006), Wang et al. (2013) and Kang et al. (2015). It covers several important factors that may affect the relationship between oil prices and stock market returns. Returns at country level as a dependent variable are explained by oil prices, global market returns, inflation and oil production levels. Following Basher & Sadorsky (2006) WTI crude oil prices are incorporated into the equation. I chose to use WTI crude oil prices as this is most commonly used in academic literature as a benchmark for oil prices, and is the most widely traded oil futures contract in the world according to their study.

(7)

6

prices on stock market returns. This regression will be structured similarly to the daily regression, first as a pooled regression, followed by regressions for each country separately. Lastly, the regression is performed at an annual frequency, also both pooled and for each country individually. These regressions include the long term factors inflation and production levels, as these are unfortunately only available at annual frequency. The daily and monthly regressions will therefore be conducted without those two variables. Note that these missing variables should be taken into account when analysing the results of those regressions. Following up on the suggestions of Kaufmann et al. (2004) regarding differences among countries, I add cross-section effects to the pooled regression as a check of robustness. This allows the constant term to vary cross-sectionally, thus over the member countries. From the pooled models I expect to find a positive relationship between oil prices and stock market performance in this particular area. In the country-specific models it is likely to expect that some countries do have a significant relationship, while others do not.

4. Data

The data used to conduct this study is time series data of daily, monthly and yearly closing prices of OPEC member countries stock market indices. This data is obtained from the Thomson Reuters DataStream database.4 The pooled OLS regression covers the period from January 1990 to October 2017. In this range there are several missing values in the dataset, therefore qualifying as unbalanced panel data. For country-specific datasets (country stock market index returns, GDP deflators and production levels) I add cross-section identifiers using Stata to be able to import the datasets into EViews as dated panel data. These country-specific datasets are obtained from a sample of 9 OPEC member countries, out of the 13 members. From all the member countries I excluded Algeria, Angola, Libya and Venezuela because of a lack of reliable stock market data. In the monthly and annual regressions I include Ecuador, Indonesia, Iran, Iraq, Kuwait, Nigeria, Qatar, Saudi Arabia and the United Arab Emirates. Daily stock market data is available on only 5 of these members. Therefore the countries included in the daily analyses are Indonesia, Nigeria, Qatar, Saudi Arabia and the United Arab Emirates.

𝑅𝑒𝑡𝑖𝑡 in Eq. (1) is computed from daily, monthly and yearly closing prices of stock market indices from the included OPEC member countries for the daily, monthly and annual regressions respectively. These indices are computed into continuously compounded returns as described in Eq. (2). An overview on which index is used per country can be found in Appendix C Table A4.

𝑅𝑒𝑡_𝑖𝑡 = 𝑙𝑛( 𝐼𝑁𝐷𝑖,𝑡

𝐼𝑁𝐷𝑖,𝑡−1 ) (2)

4_{Thomson Reuters DataStream website:}

(8)

7

The Morgan Stanley Capital International (MSCI) World Index is used to benchmark country returns to a global perspective in the model, as suggested in Basher & Sadorsky (2006). These monthly returns are computed into continuously compounded returns according to the formula as described in Eq. (2). Oil Returns are based on monthly WTI Crude Oil Prices, as this is a globally traded oil price and used in most academic literature on the relationship between oil and stock prices. Therefore this study can be compared to other studies in this field to open up discussions on this relationship. WTI Oil Returns are computed as described in Eq. (3):

𝑊𝑇𝐼𝑡 = 𝑙𝑛 ( 𝑃𝑅𝐼𝐶𝐸𝑡

𝑃𝑅𝐼𝐶𝐸 𝑡−1 ) (3)

These oil price returns and MSCI World Index returns may have structural breaks in their time series. When data series show signs of structural breaks, multiple regressions on subintervals could lead to better results than a single regression on the full time series period. Therefore I investigate this data using Eviews’ Chow Breakpoint Test. I will apply this test to daily and monthly frequency data. Annual data consists of too few observations to divide the sample period into subsamples, and therefore I do not perform this test at the annual data frequency of oil price returns and MSCI World Index returns.

To take inflation into consideration, the GDP deflator for each country is included in the long-term analyses. This deflator is defined by dividing the nominal GDP by the real GDP, times 100. This deflator is determined on a yearly basis, and can therefore only be included in regressions with annual data. As a country-specific, oil related factor I include Oil Production levels per country into the model, as suggested by Guidi et al. (2006), Kang et al. (2015) and Kaufmann et al. (2004). These production levels are measured in 1.000 barrels per day. As the dependent variable represents continuously compounded returns, I decide to convert the GDP Deflator and Oil Production levels into changes in GDP Deflator and Oil Production levels. Using nominal values of these variables, the parameters will be close to zero, making them more complicated to interpret. Using changes in GDP Deflator and Oil Production levels rather than nominal values increases the values of the parameters. This makes the parameters easier to interpret. They are computed following the philosophy in Eq. (4). This way, if the parameter of Oil Production is for example equal to 0.14, this implies that a 10% change in Oil Productions levels leads to a 1.4% change in the dependent variable, thus stock market index returns. The descriptive statistics of the daily, monthly and annual data are presented in Table 1, Table 2 and Table 3, respectively.

𝐶ℎ𝑎𝑛𝑔𝑒𝑡= 𝑙𝑛 (

𝐿𝐸𝑉𝐸𝐿𝑡

(9)

8

Table 1: Descriptive statistics of the daily data on 5 OPEC member countries

Country Returns WTI Oil Returns MSCI World Returns Observations 26863 37025 37025 Mean 0.0004 0.0002 0.0002 Median 0.0000 0.0000 0.0005 Maximum 0.8447 0.1883 0.0910 Minimum -0.8581 -0.4069 -0.0733 Std. Dev. 0.0161 0.0243 0.0091 Skewness -0.3592 -0.7363 -0.3728 Kurtosis 641.3316 18.7318 11.3095

These statistics are taken from a sample of 9 OPEC member countries, from January 1990 to October 2017. Note that not all stock market data covers the entire estimation period, hence the lower amount of observations. Countries included in these daily statistics: Indonesia, Nigeria, Qatar, Saudi Arabia and United Arab Emirates.

Day-to-day closing prices of the member countries’ stock market indices increase on average 0.04%, where day-to-day oil prices increase 0.02% on average. The largest drop in oil price between two trading days is slightly over 40%. Note that there is a high peak in the distribution of the Country Returns, as indicated by the high Kurtosis estimator.

Table 2: Descriptive statistics of the monthly data on 9 OPEC member countries

Country Returns WTI Oil Returns MSCI World Returns Observations 1840 2997 3015 Mean 0.0073 0.0024 0.0041 Median 0.0062 0.0028 0.0106 Maximum 0.5080 0.3368 0.1035 Minimum -0.6758 -0.3837 -0.2113 Std. Dev. 0.0699 0.0960 0.0429 Skewness -0.6721 -0.1960 -0.8718 Kurtosis 13.8370 4.2164 5.1698

These statistics are taken from a sample of 9 OPEC member countries, from January 1990 to October 2017. Note that not all stock market data covers the entire estimation period, hence the lower amount of observations. Countries included in these monthly statistics: Ecuador, Indonesia, Iran, Iraq, Kuwait, Nigeria, Qatar, Saudi Arabia and United Arab Emirates.

(10)

9

whereas the maximum global return in a single month is 10.35%. Oil prices rise on average 0.24% per month in the sample period.

Table 3: Descriptive statistics of the annual data on 9 OPEC member countries Country Returns WTI Oil Returns MSCI World Returns ∆GDP Deflator ∆Production Observations 149 243 243 244 242 Mean 0.0879 0.0316 0.0417 0.0745 0.0177 Median 0.0773 0.0577 0.0912 0.0752 0.0121 Maximum 1.2736 0.7523 0.2686 0.5612 1.6936 Minimum -0.8232 -0.7692 -0.5461 -0.8111 -1.9094 Std. Dev. 0.3273 0.3533 0.1811 0.1402 0.2287 Skewness -0.0026 -0.2323 -1.3790 -0.9261 -2.7773 Kurtosis 3.9885 2.6573 4.9299 10.2773 51.6524

Note: These are statistics on a sample of 9 countries, from 1990 to 2017. 242 yearly observations of Oil Production is the sum of the 9 countries included in the regression. Per country Oil Production is approximately 27 years. As stock market indices from several countries do not cover the full sample period, the amount of Country Returns observations is lower. Note that ∆GDP Deflator and ∆Oil Production statistics are based on their annual changes, as described in Eq. (4).

Table 3 presents the descriptive statistics on the data used in the annual regressions. On annual basis, country stock market returns are roughly equal to 8.8%. The GDP Deflator increases approximately 7.5% each year, with a maximum of 56% from one year to another. Production is on average rather stable, but presents high peaks with increases and decreases ranging from 169% to -191%. Table 4 below presents the correlation between the variables in the model. These correlations are based on the annual data to include all the variables. Correlations tables from the daily and monthly data are presented in Appendix A.

Table 4: Correlations between the model variables using annual data frequency

Country returns WTI Oil Returns MSCI World Returns GDP Deflator Oil Production Country returns

WTI Oil Returns 0.4238

MSCI World Returns 0.3786 0.4436

∆GDP Deflator 0.3584 0.5254 0.2910

∆Oil Production 0.1597 0.2543 0.1939 0.0069

(11)

10

As OPEC member countries primarily focus on producing and trading crude oil, it is reasonable to expect that stock markets returns are correlated to oil returns to some extent. Table 5 below presents the correlations between individual country returns and oil returns.

Table 5: Correlation coefficients of Oil Returns and country-specific stock market index returns, based on monthly stock market returns and monthly oil price returns.

Ecuador Indonesia Iran Iraq Kuwait Nigeria Qatar

Saudi Arabia United Arab Emirates Oil Returns 0.019 0.050 -0.051 0.096 0.248 0.282 0.234 0.393 0.213

Saudi Arabia, as largest producer of crude oil, shows the highest correlation coefficient between stock market performance and oil returns. Iran, also a significant producer of crude oil, shows a slight negative correlation between stock market returns and oil price returns. A possible explanation for this could be that sometimes oil prices are increased due to U.S. sanctions against Iran involving for example the Obama nuclear Iran deal, which has recently been argued.5 This table serves as an example to highlight differences among the countries, as Kaufmann et al. (2004) also implied. From country to country there may be differences in stock market returns. An ANOVA test highlights these differences by assessing if means and variances significantly differ from one country to another. Therefore I perform Mean Equality tests and Variance Equality tests in EViews. This test presents whether the mean returns from Saudi Arabia are significantly different from for example the mean returns of Nigeria. The results of these ANOVA tests are presented in Appendix D Tables A5 and A6, respectively. Also, for some countries there are signs of heteroskedasticity in the data, whereas the rest of the data is homoscedastic. To assure that heteroskedasticity does not bias the results of this study, I perform the regressions using White standard errors and covariances.

5. Results

5.1 Empirical results

In this section I first analyse variations of Model (1) using daily data. After discussing these results I present the estimation outputs of the analysis using monthly data. Thereafter I analyse the results obtained from the yearly analysis, in which the two variables ‘∆GDP Deflator’ and ‘∆Production’ are added. These regressions are for each data frequency presented in a pooled regression first, followed by regressions for each country individually. Table 6 below presents the results of a pooled regression analysis on daily stock and oil markets data.

(12)

11

Table 6: Pooled OLS regression on the daily OPEC stock market returns from 01-01-1990 to 18-05-2018

Model 1 Model 2 Model 3

Coefficient Prob. Coefficient Prob. Coefficient Prob. Constant 0.0004 0.0017 0.0003 0.0025 0.0003 0.0026 WTI Returns 0.0335 0.0000 0.0162 0.0041 MSCI Returns 0.1831 0.0000 0.1732 0.0000 Adj. R-squared 0.0024 0.0121 0.0127 F-stat 66.79 0.0000 332.32 0.0000 173.66 0.0000 DW-stat 2.1046 2.1289 2.1303

This table presents the results of the pooled OLS regression on five OPEC member countries over the full sample period. The regression is conducted using White cross-section standard errors & covariance to account for heteroskedasticity in the data. The countries included are: Indonesia, Nigeria, Qatar, Saudi Arabia and the United Arab Emirates.

(13)

12

Table 7: Pooled OLS regression on the daily OPEC stock market returns from 01-01-1990 to 22-09-2008

Coefficient Prob. Coefficient Prob. Coefficient Prob. Constant 0.0006 0.0002 0.0006 0.0002 0.0006 0.0003 WTI Returns 0.0151 0.0250 0.0148 0.0259 MSCI Returns 0.0899 0.0000 0.0896 0.0000 Adj. R-squared 0.0003 0.0017 0.0022 F-stat 5.42 0.0199 25.64 0.0000 15.45 0.0000 DW-stat 2.2057 2.2109 2.2116

This table presents the results of the pooled OLS regression on five OPEC member countries over the sample period 01-01-1990-22-09-2008. The regression is conducted using White cross-section standard errors & covariance to account for heteroskedasticity in the data. The countries included are: Indonesia, Nigeria, Qatar, Saudi Arabia and the United Arab Emirates.

These pooled regressions provide insight on how OPEC’ stock markets, considered as a group, reacts to oil price changes. Regression analyses on each country individually highlights how each country responds to oil price changes. Table 8 below presents these regressions. These regressions are based on Model 3, including all of the variables available at the daily frequency.

Table 8: Country specific OLS regressions on daily OPEC member stock market returns from 01-01-1990 to 18-05-2018

Model 3 Indonesia Nigeria Qatar Saudi Arabia

United Arab Emirates

Variables Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob.

Constant 0.0003 0.0578 0.0004 0.0239 0.0003 0.2929 0.0003 0.1484 0.0003 0.1322 WTI Oil Returns 0.0225 0.0017 0.0103 0.2298 0.0094 0.5590 0.0302 0.0017 0.0111 0.3655 MSCI World Returns 0.2976 0.0000 0.0049 0.7878 0.1584 0.0001 0.2342 0.0000 0.1281 0.0000 Adj. R-squared 0.0410 0.0000 0.0045 0.0330 0.0095 F-Stat 159.13 0.0000 1.02 0.36 12.78 0.0000 88.03 0.0000 22.22 0.0000 DW-Stat 1.7748 1.9274 2.4464 1.9343 2.1538

(14)

13

From these individual analyses I find that all countries’ stock market returns are significantly influenced by changes in the MSCI World Market Index returns, except Nigeria. Only Indonesia and Saudi Arabia are also influenced by changes in oil prices at a significant confidence level. With parameters of 0.02 and 0.03 for Indonesia and Saudi Arabia respectively, and in combination with the results on the pooled regression in Table 6, I find that the oil price changes do not have a considerably large impact on OPEC’s stock market returns when assessing daily returns. These results are not in line with the expectations that OPEC stock market returns are immediately influenced by fluctuations of oil prices. Most studies in this field conduct similar regressions using monthly data. The results of the analysis using monthly data frequency is presented in Table 9 below. These results are based on a pooled regression which consists of 9 OPEC member countries, as monthly stock market data is available from more OPEC member countries than daily data.

Table 9: Pooled OLS regression on monthly OPEC member stock market returns from 02-1990 to 10-2017

Variables Coefficient Prob. Coefficient Prob. Coefficient Prob. Constant 0.0068 0.0011 0.0067 0.0033 0.0064 0.0022

WTI Oil Returns 0.1401 0.0000 0.1380 0.0000

MSCI World Returns 0.1272 0.0611 0.1137 0.0435

Adj. R² 0.0381 0.0055 0.0424

F-Stat 72.33 0.0000 11.08 0.0009 40.87 0.0000

DW-Stat 1.6965 1.6776 1.7187

These regressions are conducted using White cross-section standard error & covariances. The countries included in this regression are: Ecuador, Indonesia, Iran, Iraq, Kuwait, Nigeria, Qatar, Saudi Arabia and theUnited Arab Emirates.

(15)

14

trial and error. I find that there is no significant effect of oil returns from 01-1990 to 09-2008. For the estimation period from 10-2008 to 01-2015 I find a highly significant effect of Oil Price returns on OPEC stock market returns. From the period 02-2015 to 10-2017 the effect again is insignificant. Note that the insignificance in the period from 02-2015 to 10-2017 may be a result of a low number of observations, as this period only covers less than three years of monthly observations. The results of the period from 10-2008 to 01-2015 are presented in Table 10 below.

Table 10: Pooled OLS regression on monthly OPEC member stock market returns from 10-2008 to 01-2015

WTI Oil Returns 0.2761 0.0000 0.2593 0.0000

Adj. R² 0.2053 0.0513 0.2265

F-Stat 161.65 0.0000 34.65 0.0000 92.07 0.0000

DW-Stat 1.7837 1.7220 1.8223

This table presents the pooled OLS regression outputs of the sample of nine OPEC member countries during the crisis and post-crisis period, 10-2008 to 01-2015. The models are variations of Model (1) as described in Eq. (1). The results are computed using cross-section White standard errors & covariances to account for heteroskedasticity in the data.

(16)

15

Table 11: Individual OLS regressions on monthly OPEC member stock market returns from 10-2008 to 01-2015

Model 3 Ecuador Indonesia Iran Iraq Kuwait

Variables Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob.

Constant -0.0021 0.6510 0.0097 0.0948 0.0234 0.0006 -0.0055 0.6508 -0.0115 0.1849 WTI Oil Returns -0.0177 0.6806 0.0191 0.7166 0.2037 0.0008 0.0777 0.6920 0.1077 0.1037 MSCI World Returns 0.0123 0.8396 0.9535 0.0000 0.2443 0.0165 -0.0725 0.8653 0.6846 0.0615 Adj. R-Squared -0.0250 0.5398 0.1695 -0.0595 0.2525 F-Stat 0.0839 0.9196 44.98 0.0000 8.65 0.0004 0.19 0.8319 13.67 0.0000 DW-Stat 2.2097 1.4956 1.3168 2.1129 2.3042

Model 3 Nigeria Qatar Saudi Arabia

United Arab Emirates Variables Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob.

Constant 0.0059 0.4083 0.0054 0.5112 0.0039 0.5321 0.0022 0.7011 WTI Oil Returns 0.2070 0.0312 0.4032 0.0000 0.4014 0.0000 0.2786 0.0000 MSCI World Returns 0.4794 0.0018 0.2274 0.2605 0.1059 0.3802 0.1977 0.1098 Adj R-squared 0.1742 0.3052 0.3971 0.3018 F-Stat 7.33 0.0015 17.48 0.0000 25.70 0.0000 17.21 0.0000 DW-Stat 1.7102 2.2266 2.3162 1.7806

These regressions are performed using White-Hinkley heteroskedasticity consistent standard errors and covariance.

(17)

16

the Saudi stock market being largely dominated by financial industry. Over the years past their study, the Tadawul All Share has grown in components, possibly decreasing the dominance of the financial industry in the index. In contrast to Saudi Arabia, stock markets in Ecuador, Iraq and Kuwait appear not to be significantly influenced by oil price changes. To extend the regression model with two long-term variables, I perform pooled OLS regression analyses on annual basis. This frequency allows to incorporate the changes in GDP Deflator and changes in Oil Production levels into the model. The regressions consist of variations of Model (1) as described in Eq. (1), by excluding and including several variables. The outcomes from these regressions are presented in Table 12 below.

Table 12: Pooled OLS regressions on annual OPEC member stock market returns from 1990 to 2017

WTI Oil Returns 0.0013 0.2743 0.0017 0.1714

MSCI World Returns 0.0000 0.8285 -0.0001 0.5316

GDP Deflator Oil Production Adj. R² 0.0048 -0.0062 0.0049 F-Stat 1.68 0.1975 0.14 0.7119 1.34 0.2646 DW-Stat 2.2455 2.3584 2.2883

Variables Coefficient Prob. Coefficient Prob. Coefficient Prob. Constant 0.1225 0.5311 0.0123 0.9472 0.0209 0.9031 WTI Oil Returns 0.0018 0.1481 0.0010 0.4039 0.0011 0.3777 MSCI World Returns -0.0001 0.4346 0.0000 0.8299 0.0000 0.7193 GDP Deflator 0.9623 0.0346 0.9538 0.0240 Oil Production 0.8355 0.1248 0.8059 0.0696 Adj. R² 0.0253 0.1158 0.1350 F-Stat 2.21 0.0903 7.11 0.0002 6.43 0.0001 DW-Stat 2.2330 2.2996 2.2754

(18)

17

From these analyses it appears that OPEC stock market returns are not significantly influenced by oil price changes on an annual basis. Neither in Model 1, nor the full Model 6, oil seems not to be significantly impacting the stock market returns. On an annual basis, the GDP Deflator appears to be of importance to stock market returns with a parameter of 0.96, significant at a 5% level. This relationship is in line with expectations of Wang et al. (2013). The stock market index and oil price returns are in nominal values, and are therewith expected to be prone to inflation in the long term. Oil Production also seems to have an impact on OPEC member stock market returns, which is in line with findings of Guidi et al. (2006) and Kaufmann et al. (2004). The countries’ stock market returns individually are also not influenced by oil price returns, except for Ecuador at a 10% significance level. The results of these individual OLS analyses are presented in Appendix B. A possible reason for insignificance in the results obtained from annual data is that within a year’s time stock market performance cannot be linked directly to a factor such as oil price changes, as too many aspects influence these markets in a large time period. This large time period makes it difficult to monitor the origin of the change in stock market returns. Another reason could be that the observations per country are too low to provide a clear relationship, because of a lack of available data.

5.2 Robustness

These monthly regressions present a clear relationship between oil returns and OPEC stock market returns. From the individual OLS regressions in Table 11 I find that there are many differences among these countries. Implementing cross-section fixed effects in the regression allows the constant in the pooled OLS to vary over the member countries. This implies that these fixed effects allow the constant to vary cross-sectionally. As a check of robustness, I regress the pooled OLS model using monthly data from 10-2008 to 01-2015, as discussed in Table 10, also using cross-section fixed effects. The results of this analysis can be found in Table 13 below.

Table 13: Pooled OLS regressions on monthly OPEC stock market returns from 10-2008 to 01-2015, using cross-section fixed effects

Variables Coefficient Prob. Coefficient Prob. Coefficient Prob. Constant 0.0062 0.0605 0.0021 0.6164 0.0051 0.1035 WTI Oil

Returns 0.2758 0.0000 0.2588 0.0000

Adj. R² 0.2175 0.0625 0.2393

F-Stat 20.2148 0.0000 5.6092 0.0000 20.5690 0.0000

DW-Stat 1.8350 1.7662 1.8773

(19)

18

By implementing cross-section fixed effects there is a minor change with respect to the original regression in Table 10, with a minor decline in stock market sensitivity to oil price returns, and a minor increase in the effect of MSCI World Market Index returns on OPEC stock market returns, from 0.1895 to 0.1906. Moreover, the adjusted R-squared increases by 1.5%. EViews’ Redundant Fixed Effects likelihood test indicates that these cross-section fixed effects are valuable to the model. The null hypothesis of this test that fixed effects are redundant is rejected at a 5% significance level.

Considering the empirical results from the monthly regression analyses, Tables 9, 10, 11 and 13, I am able to reject the null hypothesis that ‘The world’s crude oil prices have no significant effect on OPEC member stock market performance’ for the estimation period 10-2008 to 01-2015. Therefore I accept the alternative hypothesis that ‘The world’s crude oil prices have a significant positive effect on OPEC member stock market performance’. The rejection of the null hypothesis provides an answer to the main question of this paper. This study shows evidence for a statistical effect of oil price returns on OPEC member countries’ stock market performance. Therefore it is safe to say that indeed, OPEC’s stock markets are influenced by oil price changes. Increases in oil prices lead to increases in OPEC stock market performance, and vice versa. Moreover, comparing these empirical results to the findings of Kaufmann et al. (2004) and Guidi et al. (2006) is alarming. The evidence these studies provide on OPEC’s ability to influence oil prices through production factors, in combination with oil price influence on OPEC’s stock market performance found in this study, suggests that OPEC members are possibly able to influence their own stock markets. This is important news to investors and academics, and this relationship should be subject to further research in the near future.

6. Conclusion

(20)

19

the monthly analyses I find that mainly Iran, Nigeria, Qatar, Saudi Arabia and the United Arab Emirates tend to drive the relationship within the full sample of OPEC member countries included in this study. These results are in line with most academic literature in the field of oil prices on stock market performance. These results are consistent with the findings of Fang & You (2014), Bhar & Nikolova (2009) and Arouri & Rault (2012). This relationship in combination with literature on OPEC’s ability to influence oil prices raises questions of to what extend OPEC is able to influence its own stock market performance through production factors. The findings in this study contribute to existing literature on the relationship between stock market performance and oil prices. Moreover, the results that OPEC stock market performance was influenced by oil price changes during the 2008 financial crisis and its aftermath may be important information to investors, especially when new financial crisis’s may appear. Future research should investigate whether OPEC stock market performance is more influenced by oil prices when in times of crisis than in times of ‘regular’ economic state.

(21)

20

References

Basher, S. A., & Sadorsky, P. (2006): Oil price risk and emerging stock markets. Global finance journal, 17(2), 224-251.

Bjørnland, H. C. (2009). Oil price shocks and stock market booms in an oil exporting country. Scottish Journal of Political Economy, 56(2), 232-254.

Bhar, R., & Nikolova, B. (2009). Oil prices and equity returns in the BRIC countries. The World Economy, 32(7), 1036-1054.

Bloomberg Article by Fabiola Zerpa (18-01-2018): Venezuelan Hyperinflation Explodes, Soaring Over 440,000 Percent, https://www.bloomberg.com/news/articles/2018-01-18/venezuelan-hyperinflation-explodes-soaring-over-440-000-percent

Bloomberg’s Data on Oil prices: https://www.bloomberg.com/quote/CL1:COM

Brooks, Chris (2014): Introductory Econometrics for Finance. 3rd Edition

Faff, R. W., & Brailsford, T. J. (1999): Oil price risk and the Australian stock market. Journal of Energy Finance & Development, 4(1), 69-87.

Fang, C. R., & You, S. Y. (2014). The impact of oil price shocks on the large emerging countries' stock prices: Evidence from China, India and Russia. International Review of Economics & Finance, 29(1), 330-338.

Guidi, M. G., Russell, A., & Tarbert, H. (2006). The effect of OPEC policy decisions on oil and stock prices. OPEC Energy Review, 30(1), 1-18.

IMF, through the Federal Reserve Bank of St. Lewis

https://fred.stlouisfed.org/series/SAUNGDPMOMBD

Jones, C. M., & Kaul, G. (1996): Oil and the stock markets. The Journal of Finance, 51(2), 463-491.

Kilian, L. (2009). Not all oil price shocks are alike: Disentangling demand and supply shocks in the crude oil market. American Economic Review, 99(3), 1053-69.

Kang, W., Ratti, R. A., & Yoon, K. H. (2015). The impact of oil price shocks on the stock market return and volatility relationship. Journal of International Financial Markets, Institutions and Money, 34(1), 41-54.

Kaufmann, R. K., Dees, S., Karadeloglou, P., & Sanchez, M. (2004). Does OPEC matter? An econometric analysis of oil prices. The Energy Journal, 25(4) 67-90.

Nu.nl news article on oil prices Obama Nuclear Deal:

(22)

21

Reuters Website Article:

https://www.reuters.com/article/us-russia-energy-production/russian-daily-oil-output-edges-up-in-2017-to-30-year-high-idUSKBN1ER08T

Sadorsky, P. (1999). Oil price shocks and stock market activity. Energy economics, 21(5), 449-469.

Scholtens, B., & Yurtsever, C. (2012). Oil price shocks and European industries. Energy Economics, 34(4), 1187-1195.

Thomson Reuters DataStream website: https://financial.thomsonreuters.com/en/products/tools-applications/trading-investment-tools/datastream-macroeconomic-analysis.html

(23)

22

Appendix A

Table A1: Correlations between the model variables of the daily frequency data Country returns WTI Oil Returns MSCI World Returns Country returns

Correlation coefficients of the variables from 5 OPEC member countries regressed in Table 6 and 7. These correlation coefficients are based on the daily data from the period 29-12-1989 to 18-05-2018.

Table A2: Correlations between the model variables of the monthly frequency data Country returns WTI Oil Returns MSCI World Returns Country returns

(24)

23

Appendix B

Table A3: Individual OLS regressions on annual OPEC stock market returns from 1990 to 2017

Model 3 Ecuador Indonesia Iran Kuwait

Variables Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob.

Constant -0.0182 0.8203 0.1591 0.0289 0.0355 0.8973 0.0506 0.3728

WTI Oil Returns -0.2969 0.0580 0.2280 0.1390 0.1698 0.6109 0.4339 0.1384

MSCI World Returns 0.2048 0.4552 1.1651 0.0006 0.4617 0.6142 0.2421 0.4769 GDP Deflator 2.4133 0.0973 -0.9940 0.0015 1.4374 0.3742 -0.1567 0.8016 Production 0.2909 0.7236 0.9697 0.3382 0.0328 0.9671 0.7802 0.4908 Adj R-Squared 0.2424 0.4931 0.1954 0.2620 F-Stat 2.60 0.0756 7.08 0.0009 1.49 0.3553 2.86 0.0555 DW-Stat 2.4283 2.1968 2.8078 0.9967

Model 3 Nigeria Qatar Saudi Arabia

United Arab Emirates

Variables Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob.

Constant 0.0975 0.7358 0.0639 0.3440 0.0071 0.9320 0.0996 0.2640

WTI Oil Returns 1.2072 0.1848 -0.1761 0.6328 0.2885 0.3923 0.7085 0.2710

MSCI World Returns 1.9212 0.1809 0.0539 0.8692 0.4036 0.3943 0.7201 0.0560 GDP Deflator -2.5233 0.4278 0.9017 0.3009 0.6455 0.5912 -1.4436 0.5757 Production -3.2974 0.3049 1.0217 0.4820 1.0650 0.4460 1.0533 0.4591 Adj R-Squared 0.5003 -0.0124 0.2580 0.3123 F-Stat 2.50 0.3054 0.95 0.4675 2.48 0.0958 2.59 0.1014 DW-Stat 3.2795 2.0896 2.2221 2.0431

(25)

24

Appendix C

Table A4: Each Country with the corresponding stock market index used in this study and the data coverage

Country Stock market index Coverage

Ecuador S&P Ecuador BMI 01-1996 to 10-2017

Indonesia Jakarta Stock Exchange 01-01-1990 to 18-05-2018

Iran Teheran Stock Exchange 03-2007 to 10-2017

Iraq Iraq Stock Exchange Index 07-2012 to 10-2017

Kuwait Kuwait Securities Market 10-1994 to 10-2017

Nigeria Nigerian Stock Exchange 17-01-2000 to 18-05-2018

Qatar Qatar Stock Exchange 11-08-1998 to 18-05-2018

Saudi-Arabia Saudi Tadawul All Share 20-10-1998 to 18-05-2018

United Arab Emirates ADX General 02-07-1998 to 18-05-2018

(26)

25

Appendix D

Mean equality tests and Variance equality test on the OPEC member country stock market returns from the period 01-1990 to 03-2018.

Table A5: Mean equality test on the OPEC member stock market returns

Ecuador Indonesia Iran Iraq Kuwait

T-stat Prob. T-stat Prob. T-stat Prob. T-stat Prob. T-stat Prob. Ecuador Indonesia -0.4896 0.6246 Iran -1.6371 0.1024 -1.3250 0.1858 Iraq 1.0321 0.3028 1.4191 0.1566 2.6018 0.0100 Kuwait -0.3718 0.7102 0.1388 0.8897 1.5970 0.1111 -1.4923 0.1365 Nigeria -0.3745 0.7082 -0.0234 0.9813 1.2535 0.2113 -1.3958 0.1646 -0.1339 0.8935 Qatar -0.4460 0.6558 0.0062 0.9951 1.3172 0.1886 -1.4024 0.1618 -0.1244 0.9011 Saudi Arabia -0.3243 0.7459 0.1452 0.8846 1.4774 0.1404 -1.3541 0.1767 0.0188 0.9850 United Arab Emirates -0.4193 0.6752 0.0416 0.9668 1.4941 0.1361 -1.5327 0.1265 -0.0911 0.9274

Nigeria Qatar Saudi Arabia

(27)

26

Table A6: Variance equality test on the OPEC member stock market returns

Ecuador Indonesia Iran Iraq Kuwait

F-stat Prob. F-stat Prob. F-stat Prob. F-stat Prob. F-stat Prob. Ecuador Indonesia 1.1250 0.3086 Iran 2.2273 0.0000 1.9799 0.0000 Iraq 1.1994 0.3775 1.0662 0.7690 1.8569 0.0027 Kuwait 1.5497 0.0003 1.3775 0.0058 1.4373 0.0201 1.2920 0.1616 Nigeria 1.8178 0.0008 1.6159 0.0053 1.2253 0.2791 1.5155 0.0600 1.1730 0.3582 Qatar 1.1994 0.1551 1.0662 0.6010 1.8570 0.0001 1.0000 0.9712 1.2920 0.0415 Saudi Arabia 1.2952 0.0439 1.1513 0.2492 1.7197 0.0008 1.0798 0.6677 1.1965 0.1542 United Arab Emirates 1.6956 0.0001 1.5073 0.0015 1.3135 0.0943 1.4137 0.0703 1.0942 0.5003

Nigeria Qatar Saudi Arabia