The effect of crude oil price changes on the
American stock market
Abstract
I examine the effect of crude oil price changes on the returns of S&P500 listed companies. For my analysis, I use a fixed effect regression model to analyze monthly data from the period 2003-2017. In contrast to prior research, I find a positive, albeit insignificant, relation between crude oil price changes and stock returns. This result is in contrast to the widespread view that crude oil price changes are significantly negative related to stock returns. The change to a positive relation can be explained by higher oil production, employment and lower import dependency in case the crude oil price increases (OPEC, 2014). The change to an insignificant relation can be explained by market return to be a better and more significant estimator for stock returns and the partial mediation of the effect by the systematic risk factor, Beta, in CAPM (Aspergis and Miller, 2009; Baron and Kenny, 1986; Hammoudeh and Huimin, 2005).
Economics & Business track: Finance & Organization
Author (student number): Lars Pierrot (10799443)
Credits: 12 European Credits
Date: January 31, 2018
Statement of Originality.
This document is written by Lars Pierrot, who declares to take full responsibility for the contents 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 completion of the work, not for the contents.
1. Introduction
Until 2040 the global energy needs will rise with 30% (International Energy Agency, 2017). This rise in demand will be met with greater efficiency, a rise in the supply of renewable energy and with an increase in crude oil production (International Energy Agency, 2017). The United States will be responsible for 80% of the increase in crude oil production until 2025, keeping downward pressure on crude oil prices (International Energy Agency, 2017). According to The Guardian (2016), low crude oil prices have large consequences for the stock market, and states: “Oil company profits are plummeting, so oil company shares are plummeting, and that is dragging down the whole market”. The whole market is affected by the low crude oil prices, because investors also sell shares of companies that may be exposed to the oil industry (The Guardian, 2016). Investors sell shares of these oil-related companies, because they see low crude oil prices as an indicator for slower global economic growth and for a decrease in the expenditures of customers and companies (The
Guardian, 2016). The statement of The Guardian (2016) is primarily driven on assumptions and that is why in this research I will examine the following research question: ”To what extent is the
American stock market affected by the price of crude oil in the period May 2003 -December 2017?” The common believe is also that changes in crude oil price have a negative effect on stock market returns, due to the real-effects, which involves the decrease of production of crude oil-based products and an increase in energy costs, causing a decrease in output of companies (Bohi, 1998; Bjørnland, 2009). Rational consumers in crude oil importing countries will also lower their consumption and investments (Bohi, 1989). These changes overall cause a decrease in aggregate demand and output (Bohi, 1989). Consistent with this idea, Hamilton (1983) shows that increasing crude oil prices contributed to at least seven of the eight post-war regression in the United States and other researches find the same negative effect of crude oil price changes on stock market returns.
The question is whether these findings still hold, especially since Driesprong et al. (2008) show that crude oil price changes have predictive power for stock returns; a 0.1% decrease in stock returns in case the crude oil price would rise by 10%. In contrast to the existing researches, I focus on a time period less than a decade ago and do not disregard the larger role the United States has and is going to have in the crude oil market. For my analysis, I conduct a fixed effect regression model to analyze data of 317 continuously S&P500 listed companies in the period May 2003 - December 2017.
In contrast to prior research I find a positive effect, but insignificant, of crude oil price on stock market returns. The change in sign of the effect of crude oil price changes on stock market returns can be explained by the larger role the United States has in the crude oil market and the higher production costs that are associated with horizontal drilling (OPEC, 2014). Due to the higher
production costs that are associated with horizontal drilling, the oil-industry needs higher crude oil prices to be profitable, from which the economy and the stock market benefits with higher
employment, investments and lower import dependency (OPEC, 2014). The insignificance due to market return being a better and more significant estimator for stock returns and the partial
mediation of the effect by Beta of the capital asset pricing model (Aspergis and Miller, 2009; Baron and Kenny, 1986; Hammoudeh and Huimin, 2005). Further, this research shows evidence for the Mundell-Tobin effect, meaning that nominal interest rates increases less than the inflation, because the public reacts to the inflation by holding less in money balances and more in other assets (Tobin, 1965).And, this research shows evidence for the net-import of crude oil to be an indicator for economic growth. This is because the United States is not capable to provide in its own daily crude oil consumption, yet, therefore extra demand for this widely used commodity in production
processes of companies indicates economic growth (Tobin, 1965; OPEC, 2016).
This research is structured as follows. In section two, I provide information about crude oil, along with a discussion of the revised literature. In section three, I discuss the research design, the sample and my expectations. In section four I discuss the main findings of my research and in section five I perform three sensitivity analyzes. Finally, in section six I draw my conclusion, discuss the limitations of my research and provide recommendations for future research. 2. Literature review
2.1 Crude oil information
Crude oil is liquid petroleum that is pumped from oil wells and is composed of 84 to 87% carbon and 11 to 13% hydrogen (The Balance, 2017). This liquid substance also contains varying amounts of helium, nitrogen, sulfur and oxygen (The Balance, 2017). Oil brands often name their crude oil after the geographical source, like West Texas Intermediate and Dubai (The Balance, 2017). There are different classifications of crude oil; sweet, sour, light or heavy, depending on physical
characteristics and chemical composition (The Balance, 2017). Depending on the classification, crude oil varies not only in price, but also in usefulness and environmental impact. Depending on the amount of sulfur, crude oil is classified as sweet or sour (The Balance, 2017). Sweet crude oil has a lower sulfur content than sour crude oil (The Balance, 2017). The density of the oil
determines whether crude oil is defined as light or heavy. The density is based on the American Petroleum Institute (API) Gravity. In case the API Gravity is greater than ten, the oil is lighter than water and floats on water. In case the API Gravity is less than ten, the oil is heavier than water and does not float on water (The Balance, 2017). Crude oils that are sweet and light, like West Texas Intermediate, are generally higher in price than sour and heavy crude oils, like Dubai (Driesprong et al., 2008). This is because lighter and sweeter crude oils contain less residuum (The Balance, 2017).
Residuum is a byproduct that has to be reprocessed or sold at a discount, giving lighter and sweeter crude oils a higher premium and thus a higher price (The Balance, 2017).
The most common known products that are produced with the use of crude oil are gasoline and asphalt (OPEC, 2016). Crude oil is also a component in the manufacturing of chemicals, plastics and medicines (OPEC, 2016). West Texas Intermediate, Brent and Dubai are the primary crude oils (OPEC, 2016). Dubai crude oil is since the 1980’s the main Asian benchmark, West Texas Intermediate the main benchmark for North-America and Brent the main benchmark for the North Sea (The Oxford institute for energy studies, 2014; OPEC, 2016). Around 50% of the daily crude oil production is Brent, 17% is Dubai and 14% is West Texas Intermediate (Driesprong et al., 2008). Several times a day, the total global crude oil consumption is traded on spot, futures and over the counter markets (Driesprong et al., 2008). As can be seen in figure 1 in the appendix, the prices of different crude oils are volatile, with fluctuations between $18.- and $145.- per barrel in the period in the period from 1999 until 2017. The biggest change in oil prices is in the period July 2008 - February 2009, where the prices drop from $145.- per barrel to $40.- per barrel. Although price differences exist between crude oils, oil futures and spot prices move closely together as can be seen in the figure. This view is in line with the findings of Driesprong et al. (2008).
In 2017 the global crude oil consumption was around 98 million barrels per day of which the United States was the largest consumer, with approximately 20%, and producer, with approximately 15% (U.S. Energy Information Administration, 2017). The expectations are that the United States will be the largest crude oil consumer until 2030, China will be the largest crude oil consumer from then on after (International Energy Agency, 2017).
The global energy needs will rise with 30% until 2040. This extra demand will be met with a rise in oil production, renewable energy production and natural gas production. The United States will account for 80% of the extra oil production, making the United States a net exporter of oil in the late 2020s (International Energy Agency, 2017). Due to new horizontal drilling techniques, to unlock new resources of oil and gas in layers of shale rock, the United States can increase the production of crude oil (Washington Post, 2017). In the last three years the price per barrel crude oil was around $40-60. OPEC, the Organization of Petroleum Exporting Countries, will try to increase the price per barrel to around $50-70 (International Energy Agency, 2017). This price range is not sufficient to trigger a major change of crude oil use (International Energy Agency, 2017).
2.2 The effect of oil price on macro-economic variables
Crude oil is not only one of the most used, but also on of the most traded commodities in the world (Driesprong et al., 2008). Due to this fact, crude oil is an important commodity on financial markets and in the economy. A large body of studies researched the effect of the crude oil price on
macro-economic variables. There are less studies that study the relation between crude oil price and financial markets.
Crude oil price changes typically have real effects (Bjørnland, 2009). An increase in crude oil prices affects the output of companies, because as crude oil prices increase the production of crude oil-based products decrease and energy costs increase, causing a decrease in output of
companies (Bjørnland, 2009; Bohi, 1989). Rational consumers in crude oil importing countries will also lower their consumption and investments (Bohi, 1989). These changes overall cause a decrease in aggregate demand and output (Bohi, 1989).
According to Bjørnland (2009), changes in crude oil prices affect net oil importing countries and net oil exporting countries differently. Two major effects are distinguished, the wealth effect and the negative trade effect. The first effect is the wealth effect, which is an immediate effect, transferring wealth from net oil importing countries to net oil exporting countries, in case the crude oil price increases. The effect on the medium to long-term depends on how this extra wealth is used in net crude oil exporting countries. If this extra wealth is used to purchase goods and services in their own country, a higher level of domestic activity is generated by higher crude oil prices (Bjørnland, 2009). The negative trade effect is the second effect. Due to higher crude oil prices, oil importing face a decrease in real output, causing less trading with the net oil exporting countries (Bjørnland, 2009). Whether a change in crude oil prices induces a positive or negative effect, on the medium or long-term, depends on which of these two effects is larger. Hamilton (1983 and 2003) showed that the gross domestic product of the United States, a net-importer of crude oil, is negatively correlated to crude oil price. Jones et al. (2004) share the opinion that crude oil price movements have a large contribution to changes in the gross domestic product, in terms of elasticity it is around -0.05 to -0.06.
2.3 The effect of crude oil price changes on stock markets
Because crude oil price changes can have real effects, as shown by Burdridge and Harrison (1983), Jones et al. (2004), Lescaroux and Mignon (2008), Bjørnland (2009), and others, several researches predict that crude oil price changes will trigger investors to update their beliefs regarding the condition of the economy and the stock market. In testing this prediction, most of these researches focus on the United States, but also on Canada, Japan, Norway and the United Kingdom.
Sadorsky (2009), Hamilton (1983 and 2003), Jones and Kaul (1999), Driesprong et al. (2008), Park and Ratti (2008), Narayan and Sharma (2011), Aspergis and Miller (2009) and Chen et al.(1986) study the effect of crude oil price on the stock market in the United States. Some of these researches consider also other countries. Bjørnland (2009), El-Sharif et al. (2005) and Hamao (1988) study the effect of crude oil price on stock markets in respectively Norway and Japan.
Sadorsky (1999) uses the S&P500 as benchmark for the American stock market and conducts a vector auto-regression. Analyzing data from the period 1947 until 1996, he finds a significant negative effect of crude oil price changes on the stock market, and an insignificant effect of stock market returns on crude oil price movements. Sadorsky (1999) also shows that the
economy is affected in an a-symmetric way by the crude oil price changes. In case the crude oil price increases, the economy shows a larger negative change in real output than a positive change in real output, in case the price of oil decreases.
Park and Ratti (2008) study the effect of oil price on stock returns in the United States and in thirteen European countries. Consistent with Bjørnland (2009), they also find a positive effect of crude oil price changes on the Norwegian stock market. But in contrast to Sadorsky (1999), Park and Ratti (2008) do not find an a-symmetric effect of positive and negative oil price shocks on real stock returns.
Consistent with Bjørnland (2009), Driesprong et al. (2008), who use data from the period October 1973 until April 2003, show that stock returns can be forecasted with the use of changes in crude oil price, especially in developed countries. They also show that a 10% change in crude oil price predicts a -0.01% return next month. The effect is not instant, because there is delay in the reaction of investors.
Jones and Kaul (1999) research the effect of crude oil price on the stock market in the United States, but also in Canada, Japan and the United Kingdom. They conclude that the stock markets in the United States and Canada are rational; the changes of stock market returns are due to the impact that oil price movements have on the current and the expected future cash flows. This rationality is not found on the stock markets in the other countries, because of measurement errors in inflation, crude oil prices and real cash flow variables that are used.
Narayan and Sharma (2011) examine, with the use of a multiple linear regression model, the effect of crude oil price changes on stock returns of 560 companies that are listed on the New York Stock Exchange. They conclude that companies in the transportation and oil and gas industry experience an increase in stock returns in case the crude oil price increases. They also show that companies in other sectors experience a decrease in stock returns in case the crude oil price rises. In the energy, supply, manufacturing, food, medical, transportation, banking and real estate sector, the majority of the companies are significantly, negatively or positively, affected by the oil price. Narayan and Sharma (2011) also find that returns of small firms are more likely to be positively related to crude oil price and this positive relation changes to a negative relation when the size of the companies increase. This is in line with the view of Fama and French (1992), who find that smaller companies have higher average returns than larger companies.
Bjørnland (2009), shows that there is a positive relation between the crude oil price and the stock market. She states that a 10% change of the oil price will cause the stock market return to change by 2.5%.
In a related study to Sadorsky (2001), El-Sharif et al. (2005) show that there is a significant effect of crude oil price changes on the stock market returns in the United Kingdom for companies that are active in the gas and oil industry. El-Sharif et al. (2005) only examine the oil and gas industry.
In contrast to the so far presented researches, there are also researches that do not find a significant effect of oil price on the stock market. Chen et al. (1986) conduct a multivariate regression model to examine the effect of crude oil price changes on stock market returns in the United States in the period from 1953 until 1983. But Chen et al. (1986) do not find a significant effect of crude oil price changes on stock market returns. Ciner (2001), studies the effect of crude oil price changes on the S&P500, but does not find a significant relation either. Ciner (2001), concludes that this insignificant relation might be, because he considers a non-linear model. In contrast to Jones and Kaul (1999), who find a significant relation between crude oil price and stock market returns in Japan, despite the irrationality of the Japanese stock market, Hamao (1988) does not find the same significant relation. Driesporng et al. (2008) find evidence that stock market returns can be forecasted with the use of crude oil price changes in seventeen of the eighteen developed countries. In undeveloped-countries they do find the same relation as in the developed countries, but the significance is not as strong. Aspergis and Miller (2009) do also find a significant relation between crude oil price changes and the stock markets in the countries they examine, but the effect is small. They think this small effect would become insignificant in case they would have incorporated more control variables, like exchange rates, interest rates and consumer durable spending.
2.4 Conclusion, hypothesis and the reason for this research
To summarize, prior research find a significant effect of crude oil price changes on stock market returns, macro-economic variables and monetary policies. In the United States the effect of crude oil price changes on stock market returns is in all researches negative, and predominantly
significant. These prior researches all focus on a time period more than a decade ago. With the desired higher crude oil price, the larger role the United States is acquiring in the crude oil market and the by Driesrpong et al. (2008) shown predictive power of crude oil price changes for stock market returns, the question rises whether these findings still hold (OPEC, 2016). To test this, I use data from the period May 2003 to December 2017 to examine the effect of crude oil price changes on stock market returns. This translates to the following hypotheses:
H": crude oil price changes do not have a significant effect on stock market returns in the period May 2003 - December 2017.
Vs.
H$: crude oil price changes do have a significant effect on stock market returns in the period May 2003 - December 2017.
3. Research design 3.1 Research method
The main focus of this research is to examine the relation between crude oil price and the American stock market. A fixed effect regression will be estimated to test the null hypothesis; ”crude oil price has not a significant effect on stock market returns, in the period May 2003 – December 2017”. The data are monthly, because the strongest relation between crude oil price and stock market returns in previous researches was found using monthly data and it is less noisy than daily data (Driesprong et al., 2008). The model looks as follows:
R'( = B$WTI(+ B/R0(+ B1R2(+ B3BTM'(+ B5S'(+ B7CPI(+ B:FC(+ B<NI(+ α'+ u'( i = 1, … ,317 t = 1, … ,176
In this model the dependent variable, R'(, is the adjusted for subsequent capital actions return at the end of month t of the S&P500 listed company i. The returns of the listed companies will be
considered and not the return of the S&P500 index, because crude oil price fluctuations have different effects on different companies (Naryan and Sharma, 2011).
Because the main focus of this research is to examine the relation between oil price changes and stock market returns, the main independent variable, WTI(, measures the return of West Texas Intermediate crude oil at the end of month t. Changes in West Texas Intermediate crude oil price is considered, and not another crude oil brand, because it is the benchmark for North-America (OPEC, 2016). In section 5.1, a sensitivity analysis will be performed, to see whether the same conclusions can be drawn in case price changes of another crude oil brand, Brent, is used as benchmark.
The variables R0(, BTM'( and S'(, are added to the model, because these variables have explanatory power for stock market returns. R0( measures the adjusted for subsequent capital actions return on the S&P500 at the end of month t. This variable is added to the model, because according to the Capital Asset Pricing model (CAPM) the expected return of an investment can be calculated by comparing it to an efficient market portfolio as benchmark. This translates to the following formula: R'( = R2( + B' (R0( – R2(). In this formula, R0 is the return on the market
measures the volatility of a security due to market risk. (DeMarzo, 2014). By adding R0 as independent variable, the regression coefficient will be the average B' of the companies in the sample. The view that stock returns of individual companies are affected by the stock market as a whole is shared by Sharif et al. (2005). So, according to CAPM there is a linear relation between expected stock returns and market risk, but research shows that B' cannot effectively explain stock market returns on its own (Fama and French, 1992). That is why the variables Size, S'(, and the book to market ratio, BTM'(, are added to the model. S'( measures the market capitalization, of firm i in month t in hundreds of millions of dollars at the end of month t. This variable is added to the model, because Fama and French (1992) find that smaller companies have a higher average return and Naryan and Sharma (2011) find that stock returns of larger companies react more negatively to crude oil price changes. BTM'( is the market value to the book value of company i at the end of month t. This variable is added, because Fama and French (1992) find that companies with higher book to market ratios have higher returns in the long term.
Two macro-economic variables, CPI( and R2(, are added to the model. CPI(, measures the amount of inflation, or deflation, at the end of month t and is measured by the percentage change in the consumer price index of the United States. This variable is added to the model, because
according to Bjørnland (2009) and Nandha and Faff (2008) higher crude oil prices cause higher levels of inflation, causing consumers and companies to lower their investments and consumption, due to lower disposable income. Therefore, inflation has a negative influence on stock returns. R2(, is the risk-free rate at the end of month t and is measured by the three-month T-bill. This variable is added to the model, because according to Sadorsky (1999), stock returns are affected in three ways by higher interest rates. First, due to the higher borrowing costs company profits are lower. Second, higher interest rates alter the ratio in which investors hold other competing financial assets, like bonds. Third, lower ability and desire to purchase stocks on margin thus lowering stock returns. Additionally, Bjørnland (2009) argues that stock returns are also negatively affected by higher interest rates, due to lower consumption and investments by consumers and companies due to the higher borrowing costs.
NI( measures the net-import/export of crude oil of the United States in hundreds of thousands crude oil barrels per day at the end of month t. This variable is added to the model, because the discussed researches disregard the increasing crude oil production of the United States and the estimation that the United States will be a net-exporter of crude oil in de late 2020’s
(International Energy Agency, 2017). According to Bjørnland (2009), a net-exporting country could benefit from increasing crude oil prices, due to the wealth effects being larger than the negative trade effects. The closer this variable gets to being zero or negative, the closer the United States get to be a net-exporter of crude oil, diminishing the negative wealth and trade effects.
FC( is a dummy variable that takes the value one in the time period December 2007 until May 2009 and zero otherwise. The National Bureau of Economic Research (2010) marked this period as the official time period of the financial crisis. This variable is added to the model to control for the irrationality of the stock market during this time period (Ball, 2009). Due to the irrationality, the efficient market hypothesis was violated, which is the idea that competitive markets, in this case the financial market, all available information exploit to price securities (Ball, 2009; Berk and DeMarzo, 2014). In case of the financial crisis, the securities were priced lower than they should have been and thus causing lower returns (Ball, 2009).
The last two variables in the model are u'( and a'. u'( represents the error term for company i at the end of month t. a' is the company specific intercept for company i. The fixed effect model controls for company fixed effects that arise from omitted variable bias, that vary across companies, but not over time. a' can also be written as Z'+ c, where c is the general intercept and Z' the
company specific intercept (Stock and Watson, 2014).
I expect B$ to be negative and significantly different from zero at the 5% significance level, and thus rejecting the null-hypothesis. Therefore, the conclusion can be drawn that oil price changes have a significant negative effect on stock returns, with 95% certainty.
The reason why a fixed effect model is considered and not a random effect regression model or pooled ordinary least squares (OLS), is because of the outcome of the Hausmant-test. According to the Hausman-test, the independent variables are endogenous in case the null-hypothesis can be rejected, causing the fixed effect regression model to have a better fit than pooled OLS or a random effect regression model. The outcome yields a chi/-value (p-value) of 18.25 (0.0109) and thus the null-hypothesis is rejected at the 5% significance level. But to perform a fixed effect regression the Gauss-Markov assumptions need to be made (Stock and Watson, 2014). First, the conditional distribution of u'(, given the independent variables, has a mean of zero. Second, the dependent variable and the independent variables are independently and identically draws from their joint distribution. Third, the independent variables and the error terms are not correlated to each other. Finally, there is no perfect multicollinearity (Stock and Watson, 2014). In case these assumptions hold, the estimator is BLUE; the best linear unbiased estimator (Stock and Watson, 2014). Here, ‘best’ means the estimator with the lowest variance of the estimate in comparison to other unbiased, linear estimators (Stock and Watson, 2014).
3.2 Sample
I examine the relation between crude oil price changes and the stock market in the United States, because the United States is the largest producer and consumer of crude oil and has the largest
economy in the world (International Energy Agency, 2017). To conduct my analysis, I download data for all variables in the model from Thomson Reuters Datastream and construct it as panel data.
The timeframe that is considered, is from May 2003 until December 2017. There are
multiple reasons why this timeframe will be examined. The first reason why this timeframe is used, is because this research is an extension of the research of Driesprong et al. (2008) and their
timeframe ended in April 2003. This research is however of a smaller focus and scale, because I only examine the stock market in the United States. The second reason is, because December 2017 is the most recent data on the subject matter. Finally, because the timeframe includes the financial crisis and it is interesting to see whether the presumed irrationality of the stock market is also present in the S&P500 (Ball, 2009). Companies that were delisted during the studied time period from the S&P500 are removed from the sample, leaving 317 companies to be examined in 176 months. The tables below, table 1 and table 2, provide information regarding the descriptive statistics of the variables and the Pearson correlation among the variables.
Table 1. Descriptive statistics.
N mean std. dev. minimum maximum skewness kurtosis
R'( 55475 .0099303 .0955258 -.9333162 2.199991 1.33531 29.62996 WTI( 55475 .0088979 .0986002 -.2625956 0.2603808 -.1977752 3.180241 R0( 55475 .0068672 .0418925 -.2388363 0.1262441 -1.357752 9.030855 R2( 55792 .0117875 .0159694 0 0.0503 1.336527 3.327506 BTM'( 55792 3.501448 26.60292 -481.83 1352.19 16.77329 777.318 S'( 55792 324.438 546.1619 .4164 8932.157 4.537808 34.32357 CPI( 55475 .0017673 .0031838 -.0177055 0.0137685 -1.326185 11.81173 FC( 55792 .1022727 .3030093 0 1 2.625205 7.891702 IN( 55792 86.9904 31.33171 24.98 134.42 -0.2034058 1.586722
This table provides information regarding the descriptive statistics of the variables; count (N), mean, standard deviation (std. dev.), minimum values, maximum values, skewness and kurtosis. The meaning of the variables can be found in table 13 in the appendix.
The results from table 1 show that the average monthly return of the companies in the sample 0.0099% is and that the average monthly price change in West Texas Intermediate crude oil 0.0089%. The average monthly market return and the change in risk-free rate is respectively
0.0069% and 0.0118%. The average size of the companies in the sample is 324.348 hundred million dollars and the average book to market ratio is 3.50. The financial crisis covers 10.23% of the sample period. The inflation growth rate in the United States was on average 0.0018% per month. Finally, the net-import of crude oil was on average 8,699,040 barrels per day.
Table 2. Pearson correlation table. R'( WTI( R0( R2( BTM'( S'( CPI( FC( IN( R'( 1 WTI( .1341** 1 R0( .5288** .2392** 1 R2( -.0820** .0812** -.0079 1 BTM'( .0104* -.0061 .0081 .0029** 1 S'( .0063 -.0048 .0173** -.0148** .0141** 1 CPI( .0869** .4644** .1491** .2009** -.0018 .0039 1 FC( -.0933** -.0882** -.2489** -.0192** -.0195** -.0436** -.1248** 1 IN( -.0261** .0817** -.0921** .6489** -.009* -.1063** .1226** .2385** 1
This table provides information regarding the degree of correlation between the variables in the fixed effect model. The meaning of the variables can be found in table 13 in the appendix.
*Significant at the 5% significance level. **Significant at the 1% significance level.
Table 2 shows the Pearson correlation among the variables. All the variables, except S'( for ratio, show a significant correlation with R'( at the 5% significance level. The correlation between WTI( and R'( is in contrast to the revised literature positive, the same applies for CPI(. Further, especially the correlation between R0( and R'(, IN( and R2(, CPI( and WTI( are large.
There is no ign of multicollinearity in the table, because all correlations have a value of less than 1 (Stock and Whatson, 2014). But, to test for multicollinearity a variance inflation factor-test (test) is conducted, the results of this test are presented in table 3. In case the values of the VIF-test are greater than 5, there is sign of multicollinearity and these variables should be eliminated from the model (Akinwande, Dikko and Samson, 2015). But, as can be seen in the table, the VIF-values are all below five, so there is no sign of multicollinearity in the model. Therefore, the fourth Gauss Markov assumption is fulfilled.
Table 3. Variance inflation factor-test
Variable VIF 1/VIF
IN( 2.00 0.500295 R2( 1.89 0.529963 CPI( 1.33 0.750100 WTI( 1.33 0.750596 FC( 1.20 0.834623 R0( 1.13 0.886364 S'( 1.02 0.983593 BTM'( 1.00 0.999312 Mean VIF 1.36 -
This table provides information regarding the variance inflation factor-test, which tests for multi-collinearity. The meaning of the variables can be found in table 13 in the appendix.
3.3 Expectations
My expectations are in line with the discussed previous researches. Therefore, crude oil price changes (WTI() have a negative effect on stock market returns, due to the rational reaction of the stock market on the changes in the expected cash flows of companies (Jones and Kaul, 1999; Narayan and Sharma, 2011; El Sharif et al, 2005). Higher levels of market risk (R2() have a positive effect on individual stock returns (Berk and DeMarzo, 2014; El Sharif et al., 2005). Size (S'() is negatively correlated to stock returns (Fama and French, 1992; Naryan and Sharma, 2011). The book to market ratio ( BTM'() of a company is positively correlated to stock returns (Fama and French, 1992). Higher levels of inflation (CPI() and interest rates (R2() have a negative effect on stock returns, due to fewer consumptions and investments (Bjørnland, 2009; Nandha and Faff, 2008). Lower or negative levels of net crude oil import (NI() have a positive effect on stock returns, due to diminishing negative wealth and trade effects (Bjørnland, 2009). And finally, the financial crisis (FC() has a significant negative effect on stock returns, due to the irrationality of the stock market (Jones and Kaul, 1999; Ball, 2009).
4. Analysis
To conduct a fixed effect regression, the Gauss Markov assumptions, that are stated in section 3.1, need to hold. The first assumption is fulfilled, because the F-value (p-value), constraining all error terms to zero is 0.81 (0.9930). The null-hypothesis is therefore not rejected at the 5% significance level and the error terms have a mean of zero. The second assumption is fulfilled, because the
observations are obtained via random sampling. The third assumption is fulfilled, because the correlation between the independent variables and the error terms is insignificant at the 5% significance level. Finally, the fourth assumption is fulfilled, because the results of the VIF-test show that the independent variables have VIF- values below five, indicating that there is no perfect multi-collinearity in the mode. Because all assumptions are fulfilled, the estimator is BLUE. Table 4. Regression results
R'( Coefficient Std. Err. t P>t 95% conf. interval
WTI( .0029914 . 0075276 0.40 .691 -.0118192 .0178019 R0( 1.225555 .028565 42.90 .000 1.169354 1.281757 R2( -.1719853 .0290095 -5.93 .000 -.2290614 -.1149091 BTM'( .0000247 7.62E-06 3.25 .001 9.75E-06 .0000397 S'( .0006470 .000487 1.33 .184 -.00031 .0001604 CPI( .3278026 .1661227 1.97 .049 .0009564 .6546489 FC( -.0108749 .0012884 -8.44 .000 -.01341 -.00834 IN( .0001103 .0000161 6.87 .000 .0000787 .0001419
This table contains the regression coefficients, standard errors, t-values, p-values and the 95% confidence intervals of the variables in the fixed effect regression model. The meaning of the variables can be found in table 13 in the appendix.
According to the revised literature there is a significant negative effect of oil price changes on stock returns in the United States (e.g. Aspergis and Miller, 2009; Driesprong et al., 2008; Hamilton 1996; Jones and Kaul; Nandha and Faff 2008; Naryan and Sharma, 2011; Park and Ratti, 2008; 1999; Sadorsky, 1999). Most of these researches focus on stock markets in the United States and are conducted on market-level. However, this negative relation is not found in this research, as can be seen in table 4.
The regression coefficient of WTI( is positive; 0.00299. Indicating a 0.00299% rise in monthly stock returns in case the oil price increases by 1% that month. This coefficient is however insignificant at the 5% significance level. Thus, the null hypothesis is not rejected. This result is in contrast to the revised, which find predominantly a significant negative relation between crude oil price changes and stock market returns. There are two possible explanations for the change in sign of the effect of WTI( on R'(. The first explanation is provided by The Guardian (2016), because investors see higher crude oil prices as an indicator for economic growth, and thus an increase in crude oil prices leads to more positive expectations of the future states of the stock market. However, according to Casey and Ling (2015), price changes in WTI and Brent do not have a significant positive effect on investor confidence. The second possible explanation can be found in
the growth of the oil industry in the United States and the larger role of the United States has and is going to have in the crude oil market. As stated before, the growing global crude oil demand will largely be fulfilled by increasing crude oil production in the United States, due to the new
horizontal drilling technique (Washington Post, 2017). According to OPEC (2014), the United States need higher crude oil prices to compensate for the higher production costs that are associated with horizontal drilling. Therefore, higher crude oil prices would result in higher crude oil
production, support job creation and reduce import dependency. There are also two possible
explanations for the insignificance. The first explanation is provided by Aspergis and Miller (2009). Although they do find a significant effect of crude oil price changes on stock market returns, they doubt whether this significance would hold in case more significant control variables would be added to the model. To test this, a sensitivity analysis with fewer control variables will be
performed in section 5.1. The second possible explanation is, because the effect of WTI( on R'(is
indirect, meaning that the effect is mediated by another variable (Baron and Kenny, 1986). This explanation will be tested in section 5.3.
The joint F-test with clustered standard errors, which constrains all control variables to zero, has a value (p-value) of 335.12 (0.000) and is therefore rejected at the 1% significance level. The effects of R0( and BTM'( are as expected and significant at the 1% significance level. The coefficient of R0(, 1.226, suggest a 1.226% rise in monthly stock returns in case the monthly market return rises by 1%.
The positive effect of BTM'( is significant at the 1% significance level, but is low. In case the ratio between the book to market-value of company i goes up by one, the monthly stock return increases by 0.00247%.
In contrast to the effects of R0( and BTM'(, the effect of S'( is not in line with the
expectations. Suggesting that a 100.000.000 dollar increase in market capitalization of company i, increases the monthly return of this company by 0.0647%. However, this effect is insignificant at the 5% significance level. A way to explain this insignificance result, is the fact that Fama and French (1992) find the effect of S'( on R'(, by comparing small-cap portfolios to large-cap portfolios. Given the fact that the S&P500 consists of the largest companies in the United States, size does not have a significant effect on stock returns.
Based on the arguments of Bjørnland (2009) and Nandha and Faff (2008), I expected a negative effect of inflation on stock returns, due to lower consumption and investments of consumers and investors. This expectation is not supported by the data, because the positive regression coefficient, 0.3278, is significant at the 1% significance level and suggests that monthly stock returns increase by 0.3278% in case the monthly inflation rises by 1%. The Mundell-Tobin effect is a possible explanation (Tobin, 1965). The Mundell-Tobin effect suggests that nominal
interest rates rise less than the inflation, because the public reacts to the inflation by holding less in money balances and more in other assets. In other words, the real interest rate decreases, making it less desirable to save money instead of investing (Tobin, 1965).
In contrast to the effect of CPI( on R'(, the effect of R2( is as expected. The coefficient, -0.172, suggest a 0.172% decrease of R'( in case R2( increases by 1%. This can be explained by, lower ability and desire of investors to purchase stocks on margin, the higher ratio in which investors will hold other securities and the lower consumption, the lower company profits and investments of consumers and companies, due to the higher borrowing costs (Bjørnland, 2009; Sadorsky, 1999).
The coefficient of FC( is -0.01087 and significant at the 1% significance level. So, during the financial crisis, the stock market was irrational and the monthly returns were 1.087% lower during the financial crisis. This irrationality is in line with the findings of Ball (2009) and thus my expectations.
A higher net-import of crude oil was expected to have a negative effect on stock returns, due to the attribution to the negative wealth and trade effects (Bjørnland, 2009). But the coefficient, 0.00011, which is significant at the 1% significance level, suggests that a rise in the monthly net import of 100,000 barrels of crude oil per day increases stock returns by 0.011%. A possible explanation can be found in the inability of the United States to provide in its own daily
consumption of crude oil, making the larger demand for crude oil an indicator for economic growth (OPEC, 2016). Because the United States is not capable to provide in its own daily crude oil
consumption, yet, extra demand for this widely used commodity in production processes of companies indicates economic growth (OPEC, 2016).
5. Sensitivity analysis
I conduct three sensitivity analysis to see whether the same conclusions can be drawn in case another benchmark for crude oil price is used, in case the control variable R0( is dropped from the regression and to examine whether the effect of WTI( on R'( is indirect. For the first sensitivity analysis, the return on Brent crude oil price, Brent(, is used as main independent variable. For the third sensitivity analysis, the three steps approach of Baron and Kenny (1986) will be conducted.
5.1 Different benchmark for crude oil price changes
The focus of the first sensitivity analysis is to see whether the same conclusions can be drawn in case another benchmark for crude oil price changes is used; Brent instead of West Texas
Intermediate. Brent is the largest crude oil brand in the world, with a crude oil production of 40-50 million barrels per day, and is the benchmark for the North Sea (OPEC, 2016). I expect that the
drawn conclusion in section four will not be significantly affected, because prices of crude oil brands move closely together (Driesprong et al., 2008). The fixed effect regression that will be conducted looks as follows:
R'( = B$Brent(+ B/R0+ B1R2(+ B3BTM'(+ B5S'(+ B7CPI(+ B:FC(+ B<NI(+ α'+ u'( i = 1, … ,317 t = 1, … ,176
This is also fixed effect regression model, because the Hausman-test yields a chi/-value (p-value) of 18.32 (0.0106), and thus the Gauss-Markov assumptions apply here as well (Stock and Watson, 2014). The first assumption is fulfilled, because the F-value (p-value), constraining all error terms to zero is 0.81 (0.9930). The null-hypothesis is therefore not rejected and the error terms have a mean of zero. The second assumption is fulfilled, because the observations are obtained via random sampling. The third assumption is fulfilled, the correlation between the independent variables and the error term is insignificant at the 5% significance level. Finally, the fourth assumption is fulfilled, because there is no sign of perfect multicollinearity in the Pearson correlation table and the VIF-values in table 7 are below five (Akinwande, Dikko and Samson, 2015). Because the assumptions hold, the estimator is BLUE.
Table 5. Descriptive statistics sensitivity analysis regression 1
N mean std. dev. min max skewness kurtosis
R'( 55475 .0099303 .0955258 -.9333162 2.199991 1.33531 29.62996 Brent( 55475 .008942 .0867426 -.2514393 .2513944 -.3458195 3.351053 R0( 55475 .0068672 .0418925 -.2388363 .1262441 -1.357752 9.030855 R2( 55792 .0117875 .0159694 0 0.0503 1.336527 3.327506 BTM'( 55792 3.501448 26.60292 -481.83 1352.19 16.77329 777.318 S'( 55792 324.438 546.1619 .4164 8932.157 4.537808 34.32357 CPI( 55475 .0017673 .0031838 -.0177055 .0137685 -1.326185 11.81173 FC( 55792 .1022727 .3030093 0 1 2.625205 7.891702 IN( 55792 86.9904 31.33171 24.98 134.42 -.2034058 1.586722
This table provides information regarding the descriptive statistics of the variables in the first sensitivity regression analysis; count (N), mean, standard deviation (std. dev.), minimum values, maximum values, skewness and kurtosis. The meaning of the variables can be found in table 13 in
Table 6. Pearson correlation table sensitivity analysis regression 1 R'( Brent( R0( R2( BTM'( S'( CPI( FC( IN( R'( 1 Brent( .1822** 1 R0( .5288** .3196** 1 R2( -.0082 .096** -.0079 1 BTM'( .0104* -.007 .0081 .0029 1 S'( .0063 -.0049 .0173** -.0148** .0141** 1 CPI( .0869** .5142** .1491** .2009** -.0018 .0039 1 FC( -.0933** -.095** -.2489** -.0192** -.0195** -.0436** -.1248** 1 IN( -.0261** .0982** -.0921** .6489** -.009* -.1063** .1226** .2385** 1
This table provides information regarding the degree of correlation between the variables in the first sensitivity analysis regression. The meaning of the variables can be found in table 13 in the
appendix. *Significant at the 5% significance level. **Significant at the 1% significance level.
Table 7. Results variance inflation factor-test sensitivity analysis regression 1
Variable VIF 1/VIF
IN( 2.01 0.498605 R2( 1.89 0.529548 Brent( 1.50 0.668548 CPI( 1.42 0.705891 FC( 1.20 0.834649 R0( 1.19 0.843501 S'( 1.02 0.983596 BTM'( 1.00 0.999285 Mean VIF 1.36
-This table provides information regarding the variance inflation factor-test, which tests for multi-collinearity. The meaning of the variables can be found in table 13 in the appendix.
Table 8. Results sensitivity analysis regression 1
R'( Coefficient Std. Err. t P>t 95% Conf. interval
Brent( .0109229 .0092534 1.18 .239 -.0072831 .0291289 R0( 1.220969 .0282658 43.20 .000 1.165356 1.276582 R2( -.1691685 .0290183 -5.83 .000 -.226262 -0.1120751 BTM'( .0000249 .00000763 3.27 .001 .00000993 .00004 S'( .0000647 .0000486 1.33 .184 -.000031 .0001605 CPI( .2266609 .173787 1.30 .193 -.1152649 .5685867 FC( -.010859 .0012824 -8.47 .000 -.0133821 -.0083359 IN( .00000108 .00000016 6.74 .000 .000000764 .00000139
This table contains the regression coefficients, standard errors, t-values, p-values and the 95% confidence intervals of the variables in the fixed effect regression model. The meaning of the variables can be found in table 13 in the appendix.
As can be concluded from the table above, table 8, the results are not significantly different from the results in section four. Although there is an increase in effect, a 0.011% increase of stock returns if the crude oil price rises by 1% instead of 0.0029%, and significance of crude oil price on stock returns, it is still insignificant at the 5% significance level.
The joint F-test with clustered standard errors, which constrains all control variables to zero, has a value (p-value) of 338.12 (0.000) and is therefore rejected at the 1% significance level. The only main difference with results in section four is the insignificant effect of inflation on stock returns. A possible explanation is provided by Feldstein (1980), he finds that share prices tend to rise in times of steady inflation to maintain a constant ratio of price-earnings.
5.2 Fewer control variables
In section four, the effect of crude oil price changes had an insignificant effect on stock returns. The insignificance could possibly be due to the addition of more significant control variables to the model (Aspergis and Miller, 2009). To test this explanation, the control variable R0( is dropped from the model. The results of this regression are presented in table 9.
Table 9. Results sensitivity analysis regression 2
R'( Coefficient Std. Err. t P>t 95% conf. interval
WTI( .1128908 .0082407 13.70 .000 .0966773 .1291043 R2( -.1452403 .0296389 -4.90 .000 -.2035547 -.0869258 BTM'( .0000346 8.44E-06 4.10 .000 .000018 .0000512 S'( .0000753 .0000545 1.38 .168 -.0000319 .0001826 CPI( .8340573 .1673389 4.98 .000 .5048181 1.163296 FC( -.0245334 .0010257 -23.92 .000 -.0265514 -.0225154 IN( 7.58E-07 .0000171 0.04 .965 -.0000328 .0000343
This table contains the regression coefficients, standard errors, t-values, p-values and the 95% confidence intervals of the variables in the fixed effect regression model of the second sensitivity analysis. The meaning of the variables can be found in table 13 in the appendix.
As can be seen in the table, the effect of WTI( is now significant at the 1% significance level, indicating a 0.113% rise in R'( in case WTI( rises by 1%. However , the R/-value drops from 0.2806 to 0.0234, making the model less suitable to explain R'(. This indicates that the explanation of Aspergis and Miller (2009) is valid, making market return a better estimator for stock returns than crude oil price changes. So, the conclusion in section four is unchanged; crude oil price changes do not have a significant effect on stock returns at the 5% significance level.
5.3 Testing for mediation
The effect of WTI( on R'( is insignificant in section 4. However, WTI( on R'( are significantly
correlated, as can be seen in table 2. And, the sensitivity analysis in section 5.2 shows a significant effect of crude oil price changes on stock returns, but a too low R/-value to be a good estimator for R'(. The effect of crude oil price changes on stock returns could therefore be already incorporated in the systematic risk factor, Beta; the coefficient of R0(. To test this, the three-step approach of Baron and Kenny (1986) will be conducted in this section.
The first step consists of showing that the independent variable, WTI(, is significantly correlated with the mediator, R0(. This is done by regressing R0( on WTI(.
Table 10. Step 1. Testing for correlation between the independent variable and the mediator.
R0( Coef. Std. Error t P>t 95% conf. interval
WTI( .1016474 .0024509 41.47 0.000 .0968435 .1064512
The result is consistent with the result table 2, so there is a significant correlation between the independent variable and the mediator at the 1% significance level.
The second step consists of showing that the causal variable, WTI(, is significantly
correlated with the dependent variable, R'(. This is done by regressing R'(on WTI(. This relation is significant at the 1% significance level, as can be seen in table 11.
Table 11. Step 2. Testing for causality between the independent and the dependent variable.
R'( Coef. Std. Error t P>t 95% conf. interval
WTI( .1299122 .0051499 25.23 0.000 .1198184 .140006
c .0087744 .0004097 21.42 0.000 .0079713 .0095774
The third and final step is to establish whether R0( mediates the effect of WTI( on R'(, this is done by regressing R'( on WTI( and R0(. In case the effect of WTI( becomes zero and
insignificant, the effect is totally mediated. In case the effect becomes insignificant, the effect is partially mediated (Baron and Kenny, 1986).
Table 12. Step 4. Test for (partial) mediation.
R'( Coef. Std. Error t P>t 95% conf. interval
WTI( .0077889 .0042213 1.85 0.065 -.000485 .0160627
R0( 1.201441 .0131915 91.08 0.000 1.175586 1.227296
c .0016105 .0003552 4.53 0.000 .0009143 .0023067
The results in table 12 show evidence for partial mediation of R0( between WTI( and R'(, because WTI( is insignificant at the 5% significance level. Meaning that the information regarding crude oil price changes is partially mediated by the systematic risk factor, Beta.
6. Conclusion
In this research, I examine the research question: ”To what extent is the American stock market affected by the price of crude oil in the period May 2003 -December 2017?”. My analysis is based on the return of 317 continuously listed companies on the S&P500 during the stated timeframe. Using data from these companies, I estimate a fixed effect regression model. The control variables that are added to the model are: market return, risk-free rate, the book to market ratio and size of companies, inflation, a dummy variable on the period of the financial crisis and the net-import of crude oil.
In contrast to prior research, I find a positive effect, albeit insignificant, of crude oil price changes on stock market returns. A possible explanation for the change in sign of the effect of crude oil price changes on stock returns, is because the United States needs higher crude oil prices to compensate for the higher production costs that are associated with horizontal drilling. Therefore, higher crude oil prices would result in higher crude oil production, support job creation and reduce import dependency (OPEC, 2014). The insignificance of the effect can be explained by market return to be a better and more significant estimator for stock returns than crude oil price changes and the partial mediation of the effect by the systematic risk factor or Beta in the capital asset pricing model (Aspergis and Miller, 2009; Baron and Kenny, 1986).
These conclusions do not change in case another benchmark for crude oil price changes is used. The only difference was the insignificance of the effect of inflation, which could be due to the commensurate rise of stock returns and inflation (Feldstein, 1980).
This research suffers from limitations and can be improved in future. In this research, variables that have a significant effect on stock returns are disregarded, like investor confidence and consumer durable spending. In addition, Driesprong et al. (2008) find a lagged response of stock markets on crude oil price changes, this lagged response is not considered in this research. And, I conduct a fixed effect regression model, while in most researches that find a significant effect of crude oil price changes on stock returns, a vector auto-regression is conducted. So, for future research, it would be insightful whether the same conclusions are drawn in case a vector auto-regression model is considered, including the variables investor confidence, consumer durable spending and the lagged effect of crude oil price changes. It would also be insightful to see whether the effect of crude oil price changes would be significant in case a stock market containing more oil-industry operating companies is used as benchmark.
7. Appendix
Table 13. List of variables
Variable Meaning Bi Brent( BTM'( c Beta of company i
Return Brent crude oil at the end of month t Book to market ratio of company i at the end of month t
Intercept or constant in the regression
CPI( Percentage change of the U.S. consumer price
index at the end of month t
FC( Dummy variable, 1 during financial crisis
NI( Net-import of crude oil in hundreds of
thousands crude oil barrels per day R2(
R'( R0(
Risk-free rate at the end of month t Return of company i at the end of month t Return S&P500 index at the end of month t
S'( Size of company i at the end of month t,
measured by market capitalization in hundreds of millions of dollars
u'( Error term
WTI( a
Return West Texas Intermediate crude oil at the end of month t
Company specific intercept
Figure 1. Prices per crude oil brand
This figure shows the price changes of the crude oil brands WTI, Brent, Mars, Tapis and Dubai in the period 1999-2017.
8. Reference list
-
Aspergis, N., & Miller, S. M. (2009). Do structural oil-market shocks affect stock prices? Energy Economics, Vol. 31(4), P. 569-575.-
Akinwande, M. O., Dikko, H. G., Samson, A. (2015). Variance Inflation Factor: As a Condition for the Inclusion of Suppressor Variable(s) in Regression Analysis. Open Journal of Statistics, Vol. 5(7), P. 754-767.-
The Balance (2017, 9 July). The Basics of Crude Oil Classification. Retrieved from: https://www.thebalance.com/the-basics-of-crude-oil-classification-1182570-
Ball, R. (2009). The global financial crisis and the efficient market hypothesis: what have we learned? Journal of Applied Corporate Finance, Vol. 21 (4), P. 8-17.-
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, Vol. 51 (6), P. 1173-1182.-
Berk, J., & DeMarzo, P. (2014). Corporate Finance. Third edition. Boston, AM: Pearson.-
Bjørnland, H. C. (2009). Oil price shocks and stock market booms in an oil exporting country.-
Bohi, D. R. (1989). Energy Price Shocks and Macroeconomic Performance. Washington: Resources for the Future.-
Burbidge, J., & Harrison, A. (1984). Testing for the effects of oil-price rises using vector auto-regressions, International Economic Review, Vol. 25(2), P. 459-484.-
Casey, K.M., Ling, T. He. (2015). Forecasting ability of the investor sentiment endurance index: The case of oil service stock returns and crude oil prices. Energy Economics, Vol. 47, P. 121-128.-
Ciner, C. (2001). Energy Shocks and Financial Markets: Nonlinear Linkages. Studies in Nonlinear Dynamics & Econometrics, Vol. 5(3), P. 203-212.-
Chen, N., Roll, R., & Ross, S. A. (1986). Economic Forces and the Stock Market. The Journal of Business, Vol. 59(3), P. 383-403.-
Driesprong, G., Jacobsen, B., & Maat, B. (2008). Striking oil: Another puzzle? The Journal of Financial Economics, Vol. 89(2), P. 307-327.-
El-Sharif, I., Brown, D., Burton, B., Nixon, B., & Russel, A. (2005). Evidence on the nature and extent of the relation between oil prices and equity values in the UK. Energy Economics, Vol. 27(6), P. 819-830.-
Fama, E., & French, K. (1992), The Cross-Section of Expected Stock Returns. Journal of Finance, Vol. 47(2), P. 427-465.-
Faff, R. W., & Brailsford, T. J. (1999). Oil price risk and the Australian stock market. Journal of Energy Finance & Development, Vol. 4(1), P. 69 87.-
Feldstein, M. (1980). Inflation and the stock market. The American Economic Review, Vol. 70(5), P. 839-847.-
The Guardian (2016, January 21). Retrieved from:https://www.theguardian.com/business/2016/jan/21/oil-prices-stock-market-us-recession-economy
-
Hammoudeh, S., & Huimin, L. (2005). Oil sensitivity and systematic risk in oil-sensitive stock indices. Journal of Economics and Business, Vol. 57(1), P. 1-21.-
Hamilton, J. D. (1983). Oil and the macro-economy since World War II. Journal of Political Economy, Vol. 91(2), P. 228-248.-
Hamilton, J. D. (1996). This is what happened to the oil price-macroeconomy relation. Journal of Monetary Economics, Vol. 38(2), P. 215-220.-
Hamao, Y. (1988). An empirical examination of the Arbitrage Pricing Theory: Using Japanese Data. Japan and the World Economy, Vol. 1(1), P. 45-61.-
International Energy Agency (2017, November 14). World Energy Outlook. Retrieved from: https://www.iea.org/weo2017/-
Jones, M. C. & Kaul, G. (1996). Oil and stock markets. The Journal of Finance, Vol. 51(2), P.463-491.-
Jones D.W., Leiby P. N., & Paik I.K. (2004), Oil price shocks and the macro-economy: what has been learned since 1996. Energy Journal, Vol. 25 (2), P. 1-32.-
Lescaroux, F., Mignon, V. (2008). On the influence of oil prices on economic activity and other macroeconomic and financial variables. OPEC Energy Review, Vol. 32 (4). P. 343-380.-
Nandha, M., & Faff, R. (2008). Does oil move equity prices? A global view. Energy Economics, Vol. 30(3), P. 986-997.-
Narayan, P. K., & Sharma, S. S. (2011). New evidence on oil price and firm returns. Journal of Banking & Finance, Vol. 35(12), P. 3253-3262.-
National Bureau of Economic Research (2010, September 20). US Business Cycle Expansions and Contractions. Retrieved from: http://www.nber.org/cycles.html-
Organization of the Petroleum Exporting Countries (2014). World Oil Outlook. Retrieved from: http://www.opec.org/opec_web/static_files_project/media/downloads/publications/WOO_2014.p df-
Organization of the Petroleum Exporting Countries (2016). World Oil Outlook. Retrieved from: http://www.opec.org/opec_web/static_files_project/media/downloads/publications/WOO %20 2016.pdf-
The Oxford Institute for Energy Studies, (2014, October). Oil Markets in Transition and the Dubai Crude Oil Benchmark. Retrieved from: https://www.oxfordenergy.org/wpcms/wp-content/uploads/2014/10/Oil-Markets-in-Transition-and-the-Dubai-Crude-Oil-Benchmark.pdf-
Park J., & Ratti R., (2008). Oil price shocks and stock markets in the U.S. and 13 Europeancountries. Energy Economics, Vol. 30, P. 2587-2608.
-
Sadorsky, P. (1999). Oil price shocks and stock market activity, Energy Economics, Vol. 21(5), P. 449-469.-
Sadorsky, P. (2001). Risk factors in stock returns of Canadian oil and gas companies. Energy Economics, Vol. 23(1), P. 17-28.-
Stock, J. H. & Watson, M. W. (2015). Introduction to Econometrics. Updated third edition. Harlow: Pearson.-
U.S. Energy Information Administration (2017, December 12). Short-term energy outlook. Retrieved from: https://www.eia.gov/outlooks/steo/report/global_oil.cfm-
U.S. Energy Information Administration (2013, August 8). Oil and gas industry employment growing much faster than total private sector employment. Retrieved from:https://www.eia.gov/todayinenergy/detail.php?id=12451#
