Bachelor thesis Economie & Bedrijfskunde
The influence of oil price fluctuations on four
different industries.
Abstract.
Do the most recent oil price fluctuations have an impact on the oil, automobile, retail and airline industry? In this paper a research is conducted on 82
companies in the period January 2005 to January 2015. Monthly observations are taken to see whether the oil price influences the returns of these companies. Portfolios of the different companies are formed to see whether the complete industry is influenced. The oil price significantly influences the airline industry and oil industry. The retail and automobile industry are less dependent on the price of oil. After accounting for the lag of oil negative significant lag of oil coefficients were found for the retail and airline industry.
Name: Rick Waldron
Student number: 10373446 Date: January 4, 2016
Subject: Finance
Thesis supervisor: Doettling, R.J.
Statement of Originality
This document is written by Student Ricky Waldron who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision
of completion of the work, not for the contents.
Table of contents 1. Introduction 4 2. Literature review 7 3. Methodology 10 3.1 Hypothesis 10 3.2 Empirical models 11 3.3 Data 14 3.4 Summary statistics 15 4. Results 16
4.1 Results per company 16
4.2 Portfolio results 18
4.3 Portfolio results with the introduction of a lag 19
5. Conclusion 24 6. Reference list 27 7. Appendix 29
1. Introduction
As stated by Basher and Sadorsky (2006), oil is the lifeblood of modern economics; it is the most traded commodity in the world. While developing countries are growing more and more the demand for crude oil is on the rise. In the years 1994 to 2004 China’s oil consumption increased by 112.5% while India’s increased by 80.9% (Basher & Sadorsky, 2006). This increase in demand would suggest that the price should increase. Because higher demand normally results in higher prices and higher prices result in more profit for the fuel companies, because the wealth switches from oil consumers to oil producers (Nandha & Faff, 2008). But this has not been the case, in the last ten years the oil price has been fluctuating a lot and it did not steadily rise. This fluctuating oil price influences almost every industry. The effect of this fluctuating oil price on different industries is under investigation. Did the oil price affect the four industries that are researched?
Graph 1, the fluctuating oil price from 2005 to 2015
This is very interesting to investigate, because the oil price has been fluctuating a lot recently. All four industries that are under investigation are assumed to be dependent on the oil price. The basic idea behind this thesis is to see whether the industries benefit differently from the given oil price at that point in time. To see how the markets are influenced by the oil price, the stock
0 20 40 60 80 100 120 140 160 1-‐1 -‐2 00 5 9-‐1 -‐2 00 5 5-‐1 -‐2 00 6 1-‐1 -‐2 00 7 9-‐1 -‐2 00 7 5-‐1 -‐2 00 8 1-‐1 -‐2 00 9 9-‐1 -‐2 00 9 5-‐1 -‐2 01 0 1-‐1 -‐2 01 1 9-‐1 -‐2 01 1 5-‐1 -‐2 01 2 1-‐1 -‐2 01 3 9-‐1 -‐2 01 3 5-‐1 -‐2 01 4 1-‐1 -‐2 01 5
returns of different companies are being taken. The influence of the oil price can then be calculated by a regression.
The first market under investigation is the integrated oil and gas company market, which is directly related to the oil price. Al-‐Mudhaf and Goodwin (1993) find a positive impact of oil on 29 oil/fuel companies. The second market is the car market, which is also highly related to the oil price, since cars run on fuel and fuel is derivative of oil. Cameron and Schnusenberg (2009) found an inverse relationship between the oil price and the stock price of car manufacturers. The third market is the retail market. This market is chosen because it is important to also include a market, which is not directly linked to oil. Because it then is easier to compare the different industries and see what kind of impact oil actually has on industries dependent or not dependent on oil. But Nandha and Faff (2008) show that the retail market is dependent on the oil price. The final market under investigation is the airline industry. The airline industry is very dependent on oil since kerosene is a refined oil product and airplanes run on kerosene (Carter, Rogers & Simkins, 2006).
The first part of this research uses the Fama -‐ French three-‐factor model and adding the return of oil as another dependent variable. Around eighty regressions are run, a regression on every company to estimate how the stock returns of the companies respond to the return of the oil price. Most of the results from this part were not significant. Only in the oil industry some positive significant results are found.
The second part of this thesis researches whether these different companies combined, are affected by the oil price. By creating four portfolios containing the biggest listed companies in every industry based on revenue. The influence of the oil price on these portfolios is based on two regressions for each portfolio. The first regression is based on a weighted average return regression based on market value. The second regression is based on a panel regression. The regressions are run from January 1st 2005 to January 1st 2015, this is done so
that the outcomes are really recent and there are also no other similar researches conducted within this timeframe. Within this part the weighted average return results are positively significant in the oil industry and negatively in the airline industry. Within the panel regression the oil and auto industry
were positively influenced by the oil price, while the airline industry is negatively influenced by the oil price.
The last part in this paper is on the lag of oil. Companies and investors are not able to react immediately after an oil shock or price change (Driesprong, Jacobsen & Maat, 2008). Introducing a lag of oil might increase significance of the results. The results are dependent on the lag sizes; the two-‐month lag did not add a lot of value, while the one-‐month lag gave interesting results. The airline and retail industry are both negatively influenced by the one-‐month lag oil price. This research paper consists of four other parts and continues the
following way. The second part contains information on existing literature on the same subject. The third part consists of multiple hypotheses, data and
information about the empirical models used. Part four gives the results of the research and part five concludes this research and elaborates on next possible researches.
2. Literature Review
Oil and its derivatives are the most traded commodities in the world (Basher & Sadorsky, 2006). It has an impact on almost every industry; therefor it is highly correlated with economic growth (Nandha & Faff, 2008). There have been multiple studies about the effect of oil on all industries. Narayan and Sharma (2011) found that the oil price affects firms differently across different
industries. Nandha & Faff (2008) researched the oil price on 35 industry sectors and concluded a negative return on 33, except for the mining and oil & gas industry. The crude oil price and international stock markets are related, as the oil price decreases, the stock market prices increase (Miller & Ratti, 2009). Sadorsky (1999) concluded that oil prices and oil price volatility have an important effect in economic activity, but economic activity does not have an important effect on the oil price.
The oil and gas industry is highly dependent on the price of oil. A higher demand for oil should lead to an oil price increase. This would imply that an integrated oil company would benefit from a higher oil price. A positive relation between the oil price and an oil company’s stock price was found by Jin and Jorion (2006). A study done in the UK has shown that the oil price is positively related to stock returns of oil and gas companies (El-‐Sharif et al, 2005). Two articles found a positive relationship between the oil price and oil companies stock returns in Canada, Sadorsky (2001) and Boyer & Filion (2007). In this research similar results have been found in the oil and gas industry.
Oil is the biggest determinant of the petroleum price. Most cars produced run on petroleum, so this is why the car market is influenced by the oil price. Some cars are very economical while other cars, sports cars or SUV’s for example, need more petroleum to drive for the same distance. Cameron and Schnusenberg (2009) found almost no relationship in the stock price of
passenger car manufacturers and the oil price. This result corresponds with the result of this research. But on the other hand they found a negative relationship of SUV manufacturers and the oil price, this means that the stock price of SUV manufactures would decline if the oil price rises. Hilliard and Danielsen (1984) researched the oil price on the car and oil market, they found a negative
relationship between the oil price and returns of the car market and a positive relation between the oil price and the oil market.
As stated before almost every industry is affected by the price of crude oil. The retail industry for example is influenced in very indirect ways. For example if the oil price is low, shipping costs are low for retail stores, if oil prices are low consumers have to spend less on fuel and other products but have more money to spend in retail stores. Nandha & Faff (2008) found a negative return of the oil price on the retail industry. The result of this research does not correspond exactly with their conclusion. Although a negative lag of oil coefficient has been found, thus the retail industry is negatively influenced by a higher oil price in the previous month.
These three markets are under investigation because they are totally different from each other. As stated by Nandha & Faff (2008) the retail industry is slightly negatively influenced, the oil and gas industry is positively influenced and the automobile industry can be negatively or positively influenced. The fourth market under investigation is the airline market; this market has been heavily researched for the effects of oil and there has been found a big negative impact in prior researches (Carter et al, 2006). Since the three other markets do not have a clear and significant negative impact of oil this is also an interesting industry to research. According to Carter et al (2006) the total operating cost of airlines consist of ten to twenty percent out of kerosene. A lower oil price leads to a lower kerosene price and thus would decrease the costs for airlines a lot. This is consistent with the results of this research, where as a negative impact of the oil price has also been found in this research.
The last part of this thesis looks at the lag of oil. Companies and investors are not able to react immediately after an oil shock, or price change (Driesprong, Jacobsen & Maat, 2008). Driesprong et al. (2008) find evidence that stock market returns and the relation with oil returns increases up to a lag of five trading days. After the sixth trading day the explanatory power decreases (Driesprong et al., 2008) Furthermore they also allowed for a lag of two months, which was not statistically significant (Driesprong et al, 2008). This is consistent with the findings in this research, although there were significant one-‐month lag values found. Their research was done with daily data, this research consists of monthly
data. To see whether the lag of oil has an impact in this research, is by taking lags up to two months.
3. Methodology
3.1 Hypotheses
This purpose of this research paper is to study whether the oil price has a significant effect on the stock returns of different industries and in which way this direction is. The four proposed industries are the oil and gas industry, the car industry, the retail industry and the airline industry. A regression is done for every company to estimate the effect of the oil price on every company. The first way to estimate the influence on the different portfolios is to use the weighted average market value of the companies and estimate a portfolio beta compared to the oil price. The second way to estimate a portfolio beta is to do a panel regression. The last hypothesis is about whether a lag of oil influences the return of the industry.
H0: The oil price does not influence the return of a company. H1: The oil price influences the return of a company.
H0: The oil price does not influence the return of the portfolios. H1: The oil price influences the return of the portfolios.
H0: The introduced lag of oil does not influence the returns of the portfolios. H1: The introduced lag of oil influences the return of the portfolios.
3.2 Empirical models At first the returns of the stock must be calculated, this can be done by
subtracting the price of the stock in period 1 minus the price of the stock in period 0, divided by the price of the stock in period 0. The adjusted monthly closing prices are used for this equation, this is done so every corporate action is taken into account, including dividends and stock splits. The corporate actions are thus already in the stock price. The same can be done for the explanatory variables. In equation form this can be seen as:
𝑅𝑒𝑡𝑢𝑟𝑛 =!!!!!
!! (1)
The first regression model used in this thesis is based on a similar regression model used in different researches, including El – Sharif et al (2005), Brailsford & Faff (1999) and Carter, Rogers & Simkins (2006).
𝑅!" = 𝛼!+ 𝛽!∗ 𝑅𝑂𝑖𝑙!+ 𝛽!∗ (𝑀𝑘𝑡!− 𝑅𝑓!) + 𝐵!∗ 𝑆𝑀𝐵!+ 𝐵!∗ 𝐻𝑀𝐿!+ 𝑢! (2)
Equation (2) is a multiple linear regression model. 𝑅!" is the return of company i on time t. 𝛼! is the intercept corresponding with time t. 𝑅𝑂𝑖𝑙! is the return of oil on time t. 𝑀𝑘𝑡!− 𝑅𝑓! is the return of the market on time t minus the risk free rate on time t. 𝑆𝑀𝐵! is the Small Minus Big value on time t and 𝐻𝑀𝐿! is the High Minus Low value on time t. The error term on time t is given by 𝑢!.
Equation (2) is going to be used for every company in all four industries. At the end there will be around eighty estimated oil betas, one for every company.
To measure the influence oil price has on the different industries we take the weighted average return of the portfolios based on market value of the companies. So in fact there are two different ways to see whether the oil price has an impact on the return of the companies. To measure market value of the company Datastream has been used. In Datastream there are two different kinds of market value; Market Value and Market Value of the Company. Market value of the company has been chosen, because in these industries companies tend to own other companies and in this way market value for every part of the
origin of the companies. Therefore exchange rates were needed to calculate the market values in Euros. The Euro currency was chosen because most companies used are European companies. These exchange rates were found on Datastream as well.
To calculate the weighted average returns of the portfolios based on market value, a couple of steps had to be taken. The first step is to find the monthly market value for the company. See what currency the market value is given in, when necessary transfer these values to values in Euros. The second step is to add up all market values for the companies, this will result in having a total market value for your portfolio. Dividing the market value per company by the market value of the portfolio, will give the weight per company for time t.
𝑊𝑒𝑖𝑔ℎ𝑡𝐶𝑜𝑚𝑝𝑎𝑛𝑦!" = 𝑀𝑣𝐶𝑜𝑚𝑝𝑎𝑛𝑦!"/𝑀𝑣𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜! (3)
Multiplying this weight by the stock returns of the companies for the same period will give a weighted average return based on market values.
𝑊𝑒𝑖𝑔ℎ𝑡𝑒𝑑𝐴𝑣𝑒𝑟𝑎𝑔𝑒𝑡𝑅!"𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 =
𝛴( 𝑊𝑒𝑖𝑔ℎ𝑡𝐶𝑜𝑚𝑝𝑎𝑛𝑦!! ∗ 𝑅𝑒𝑡𝑢𝑟𝑛𝐶𝑜𝑚𝑝𝑎𝑛𝑦!! + ⋯ + (𝑊𝑒𝑖𝑔ℎ𝑡𝐶𝑜𝑚𝑝𝑎𝑛𝑦!"#∗
𝑅𝑒𝑡𝑢𝑟𝑛𝐶𝑜𝑚𝑎𝑝𝑛𝑦!"#) (4)
The weighted average return for the portfolio on time t is calculated by adding up all the weights times the returns of every company on time t. After these calculations there will be 120 weighted average returns for the portfolio.
𝑊𝑒𝑖𝑔ℎ𝑡𝑒𝑑𝐴𝑣𝑒𝑟𝑎𝑔𝑒𝑅!𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜! = 𝑎!+ 𝛽!∗ 𝑅𝑂𝑖𝑙!+ 𝛽!∗ 𝑀𝑘𝑡!− 𝑅𝑓! + 𝐵!∗ 𝑆𝑀𝐵!+ 𝐵!∗ 𝐻𝑀𝐿!+ 𝑒! (5)
In equation (5), the weighted average return regression for portfolio i, 𝑎! stands for the intercept on time t. 𝑅𝑂𝑖𝑙! is the return of oil on time t. 𝑀𝑘𝑡!− 𝑅𝑓! is the return of the market on time t minus the risk free rate on time t. 𝑆𝑀𝐵! is the Small Minus Big value on time t and 𝐻𝑀𝐿! is the High Minus Low value on time t.
The error term on time t is given by 𝑒!. This will result in four estimated oil betas, each for every portfolio.
The second way to calculate the oil betas for portfolios was by running a panel regression in Stata. This can be done by creating an id for every company and thus grouping data per company. Using this id every observation will be taken into account, and thus the total number of observations will be a lot more. If for example a portfolio consists of twenty companies the total number of observations will be 2400 using ten years and monthly data.
𝑅!" = 𝛼!+ 𝛽!∗ 𝑅𝑂𝑖𝑙!+ 𝛽!∗ (𝑀𝑘𝑡!− 𝑅𝑓!) + 𝐵!∗ 𝑆𝑀𝐵!+ 𝐵!∗ 𝐻𝑀𝐿!+ 𝑢!" (6)
The first difference between this regression and regression (2) is first off all that the former is about portfolios. Most important every observation for every company is estimated in the four betas. Thus around 2400 observations are in beta 1, 2, 3 and 4. A fixed effect estimation has been used, since interest lies only in variables that vary over time. By using fixed effect estimation you suspect that companies have individual characteristics and you want to control for that. A fixed effects estimation removes the effect of time-‐invariant characteristics. The variable 𝛼! is the company specific intercept, that does not change over time, because then changes in the dependent variable are not based on fixed characteristics, but based on time varying characteristics.
The last part of this research is about whether the industries are affected by the lag of oil. As stated before, companies, investors and thus industries are not able to react immediately after an oil price change (Driesprong et al, 2008). This is why a lag in the price of oil is created. In this research monthly data is used, therefore creating a lag, which is comparable with the other data, periods of a month should be kept. The influence of the lag of oil is only calculated for portfolios, using the weighted average return portfolios and the panel regression portfolios.
𝑅!" = 𝛼!"+ 𝛽!∗ 𝑅𝑂𝑖𝑙!!!+ 𝛽!∗ 𝑅𝑂𝑖𝑙!+ 𝛽!∗ (𝑀𝑘𝑡!− 𝑅𝑓!) + 𝐵! ∗ 𝑆𝑀𝐵!+ 𝐵!∗
𝑅!" = 𝛼!"+ 𝛽!∗ 𝑅𝑂𝑖𝑙!!!+ 𝛽!∗ 𝑅𝑂𝑖𝑙!!!+ 𝛽!∗ 𝑅𝑂𝑖𝑙!+ 𝛽!∗ (𝑀𝑘𝑡!− 𝑅𝑓!) + 𝐵!∗
𝑆𝑀𝐵!+ 𝐵! ∗ 𝐻𝑀𝐿!+ 𝑢! (8)
Regression (7) is run to check whether introducing a lag of one month would have significant effects. The variable 𝑅𝑂𝑖𝑙!!! stands for the return of the oil price of the previous month. Regression (8) is done to check whether holding a lag of two months will give interesting results. The variable 𝑅𝑂𝑖𝑙!!! stands for the return of the oil price two months prior to the researched month. With these regressions you are calculating whether or not the stock returns of the portfolios are influenced by the returns of the oil price one and two months prior to the stock return.
3.3 Data
The companies used were based on the revenue in 2014, the listed companies with the biggest revenues were used. A total of 82 companies were researched, including 24 integrated oil companies, 18 companies from the automobile industry, 20 airline companies and 20 companies from the retail industry. The companies are shown in table 1, and that can be found in the appendix. The data of all these companies can be found on Datastream. This dataset consists of monthly observations over the time period January 1st 2005 to January 1st 2015.
As stated under section 3.2 the share price that has been used is the adjusted monthly closing price, because this one is adjusted for all corporate actions. The market value for the companies can also be found on Datastream. For the
integrated oil companies, market values for PetroChina, Hellenic Petroleum and Total SA could not be collected and thus were eliminated for regression 5. They have been used for regression 6, because market value was not needed for that regression. For the airline industry the first couple of market values and stock prices could not be found for Air China so this company is completely eliminated from all regressions.
The monthly oil price has been found on www.eia.gov, which stands for energy information administration. The Brent crude oil price has been taken because this is one of the benchmark prices for oil in the world. The values for
the three factor model; Rm-‐Rf, SMB and HML can be found on Kenneth French’s website (http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html). 3.4 Summary statistics
The means and the standard deviations of the explanatory variables can be found in table 2.
Table 2. Summary statistics of explanatory variables
Variable Mean St. Deviation Median
Roil 0.4453291 8.56936 1.105245 Mkt-‐Rf 0.545 4.701117 1.22 SMB -‐0.0148333 1.483753 -‐0.18 HML 0.0748333 1.602763 0.07 ROil!!! 0.6456084 8.318778 1.113647 ROil!!! 0.8335006 8.096697 1.244165
4. Results
4.1 Results per company
Table 3 gives the results of regression (2) for the individual oil companies. The company is the dependent variable, oil, Mkt-‐Rf, SMB and HML are the
independent variables. Table 3 can be found in the appendix. From all the companies, 21 out of 24 companies are positively influenced by the oil price, 8 are significant at the 1% level, 5 are significant at the 5% level, 2 at the 10% level and the remaining 6 are not significant but still positively influenced. The other 3 are negatively influenced by the oil price, but only 1 of these 3 is significant at the 5% level. These results are thus in line with previous research (Faff &
Nandha, 2008), (Boyer & Filion, 2006), (El-‐Sharif et al, 2005) & (Sadorsky, 2001). If the oil price goes up, consumers spent more on oil and therefor increase the revenue of the oil companies. If revenue goes up, the value of the company goes up and therefor stock returns are positive. That is why a positive oil beta was an expected outcome in the oil market and thus the results are in line with what was expected. Almost all the values for the market minus the risk free rate are significant and positive at the 1% level. This suggests that the stock returns of the oil companies are more influenced by the market than the oil price. If the market goes up by one percent than the returns of all the companies increase by a certain percentage, mostly between zero and one. If for example, the market is in an upturn, the economy is doing well and thus people have more to spend. Oil is something you would use more if you have more money, driving a car instead of biking and travel more are just examples. Therefore, these positive market values are in line with expectations.
In table 4 an overview of the coefficients of the major companies in the automobile market is given for regression (2), table 4 can be found in the appendix. Although there are no significant values for the oil coefficient, the results are quite interesting. Only 3 out of 18 companies have negative coefficients for oil, this was not expected according to previous research
(Nandha & Faff, 2008), but Cameron & Schnusenberg had similar results (2009), so these results are not uncommon. If the oil price is low, you might drive more but it does not mean that you would buy more cars. Therefor it does not
necessarily mean that the stock returns of automobile companies would go up. Again almost all the market minus risk free coefficients are significant. According to prior researches, the market variable is an important factor in equity returns. It is known that the market return influences equity returns of almost every industry, and the results from these regressions show that as well.
Table 5 shows the coefficients for the airline companies according to regression (2), table 5 can be found in the appendix. The coefficients for oil are interesting, although only 4 out of 20 values are significant, but all of them are negative. This is in line with existing literature (Carter, Rogers & Simkins, 2006). If the oil price is low, airline companies can make more profit, this increases market value. Thus the stock returns of the company will go up if the oil price is low, and that is exactly what the results display. A lot of the market minus risk free coefficients are again significant and they are all positive. This suggests that the market positively influences the stock prices of airline companies. Which is logical, because airline companies tend to make more profit when the economy is doing well.
In table 6, the coefficients for the retail market for regression (2) can be found; table 6 can be found in the appendix. For the retail companies 11 out of 20 are negatively influenced by the oil price and the other 9 are positively
influenced, only 4 oil coefficients were significant. According to Nandha & Faff (2006) the retail industry was negatively influenced by the oil price but this result cannot be concluded from this analysis. The effect of the oil price on the stock returns of retail companies is not clear, and moves in both ways, negative or positive. All the market coefficients are significant and positive; the market has thus a significant positive influence on the stock returns of the retail companies.
After looking at the results for the individual companies, there are not many significant values found. By looking at the whole industry instead of one company hopefully more significant results can be found, because the explaining power is bigger. In the next part of the results section the portfolio results are showed, where the four whole industries are looked at.
4.2 Portfolio results
Based on formulas (3) and (4), and on regression (5) the results for the weighted average return based on market value are given in table 7.
Table 7. Weighted average return coefficients based on market value Significance is given with a *, ** or *** for 10%, 5% and 1% respectively.
Portfolio Oil Mkt-‐Rf SMB HML Intercept
Automobile Industry -‐0.0013 (0.0796) 0.9091*** (0.1291) -‐0.1521 (0.2981) 0.2748 (0.4013) 0.2642 (0.4588) Oil Industry 0.1370*** (0.0497) 0.8529*** (0.1031) -‐0.3177 (0.2503) 0.0318 (0.2701) 0.0028 (0.3541) Retail Industry -‐0.0373 (0.0361) 0.6142*** (0.0543) 0.5941*** (0.1735) 0.0453 (0.1525) 0.2545 (0.2476) Airline Industry -‐0.1433** (0.0598) 0.9821*** (0.0951) 0.0141 (0.2606) -‐0.0245 (0.2607) 0.0368 (0.3576)
The results are interesting and in line with existing literature. The oil coefficient is negative and significant in the airline industry, the same holds in Carter et al. (2006). The oil industry is positively significant influenced by the oil price, as expected (Nandha & Faff, 2008). The automobile and retail industry are both negatively influenced by the oil price, but not significant. This is in line with Cameron & Schnusenberg, 2009, but not with Nandha & Faff, 2008). The oil price clearly has an effect on the oil industry. If the oil price goes up the weighted average return based on market value goes up. This is very logical since oil companies market value goes up if the oil price increases, because their revenue goes up. The airline industry is negatively influenced by the oil price, as said before if the oil price goes up the cost for the industry goes up and thus the profit will go down. The automobile industry is hardly influenced by the price of oil according to this regression. The retail industry is very slightly influenced by the price of oil, but not significant and thus the influence is not significantly different from zero. Although it is technically insignificant it appears that oil still
negatively influences the industry. If the oil price increases the average return goes down, this might be because transportation costs increases and thus profit goes down, this might be the cause for the slight negative influence.
The second way used to calculate portfolio returns was by panel regression, a fixed effects panel regression was done, and table 8 shows the results.
Table 8. Fixed effects regression with panel data
Significance is given with a *, ** or *** for 10%, 5% and 1% respectively.
Portfolio Oil Mkt-‐Rf SMB HML Intercept
Oil Industry 0.1491*** (0.0353) 0.7928*** (0.0676) -‐0.0379 (0.1039) -‐0.0430 (0.1138) 0.2770*** (0.0996) Automobile Industry 0.0622*** (0.0134) 1.1855*** (0.1250) -‐0.0592 (0.1482) 0.1939 (0.1385) 0.7237*** (0.0721) Airline Industry -‐0.1510*** (0.0293) 0.9867*** (0.0555) 0.0336 (0.1513) 0.2502* (0.1449) 0.2757 (0.2196) Retail Industry -‐0.0105 (0.0207) 0.7032*** (0.0727) -‐0.4144*** (0.0929) 0.1182 (0.1487) 0.3000*** (0.0360)
The fixed effects panel regression finds interesting coefficients. Significant values are found with the sign as expected for the oil and airline industry. A small but significant positive sign for the automobile industry, which is not what, was expected beforehand and what was not in line with existing literature. The cause for the positive influence is not clear. But because 15 out of 18 of the individual companies were positively influenced by the oil price this result is not
unexpected after regression (2). Because there are more observations the
regression is more powerful, therefore the results have become more significant. The retail industry has a small negative value, but not significant, this is in line with existing literature. The market clearly positively influences every industry. This is an expected outcome; the market is a big influence in the stock return of all industries. There are a few industries that benefit if the market is in a
recession, but none of these four industries is one of them. This can also be seen in the results.
4.3 Portfolio results with the introduction of a lag.
For the third and final hypothesis of this paper a lag has been introduced. The two ways to calculate the returns of the portfolios were again run. Hereby introducing the new lagged oil return variable. In table 9 the results of regression (7) are shown.
Table 9. Fixed panel regression with first lag variable.
Significance is given with a *, ** or *** for 10%, 5% and 1% respectively.
Portfolio 𝐑𝐎𝐢𝐥𝐭!𝟏 ROil Mkt-‐Rf SMB HML Intercept
Oil Industry .0159 (.225) .1421*** (.031) .8008*** (.068) -‐.0973 (.097) -‐.0797 (.1061) .2247*** (.049) Auto Industry -‐.0341 (.032) .0712*** (.016) 1.2006*** (.128) -‐.0558 (.147) .1761 (.139) .7112*** (.071) Airline Industry -‐.1973*** (.045) -‐.0924** (.034) 1.0407*** (.094) .1827 (.185) .2478 (.232) .3335*** (.052) Retail Industry -‐.0780*** (.0248) -‐.0121 (.019) .7273*** (.075) -‐.3659*** (.091) .1091 (.148) .3120*** (.036)
After the introduction of the lagged variable, results have changed. The oil coefficients that were significant in the prior fixed effects panel regression stayed significant. The industries where the lagged oil coefficients are significant are the airline and retail industry. They are both negatively influencing the industries this is what was expected. Apparently the oil price of the month before has a bigger influence on the stock returns of these industries.
Economically this is quite logical, if the oil price were high in the previous month, your profits as airline company and retail store would go down, because of higher costs. Your sales also might go down because of higher airfares and higher cost of transport for consumers. This profit decline can be better seen in the next month due to fewer sales and larger costs in the previous month. Fewer profits go hand in hand with lower stock returns and this is what can be concluded from this regression. This is not the case for the auto and oil industries, according to the results. They are influenced slightly in the direction that was expected, but no significant results.
The results of regression (8) are given in table 10.
Table 10. Lag panel regression with fixed effects.
Significance is given with a *, ** or *** for 10%, 5% and 1% respectively.
Portfolio 𝐑𝐎𝐢𝐥𝐭!𝟏 𝐑𝐎𝐢𝐥𝐭!𝟐 ROil Mkt-‐Rf SMB HML Intercept Oil Industry .0411 (.025) -‐.0725 (.020) .1393*** (.031) .8126*** (.068) -‐.1288 (.096) -‐.071 (.110) .2387*** (.042) Auto Industry -‐.0409 (.032) 0.0171 (.02) .0749*** (.015) 1.1961*** (.126) -‐.0466 (.151) .1802 (.144) .7187*** (.072) Airline Industry -‐.187*** (.047) -‐.0288 (.031) -‐.0943** (.035) 1.0458*** (.095) .1697 (.185) .2497 (.234) .3363*** (.054) Retail Industry -‐.0682** (.027) -‐.0343* (.018) -‐.0184 (.02) .7287*** (.074) -‐.3765*** (.088) .1285 (.144) .3455*** (.035)
In regression (8) a second lag variable is introduced, so that there are in fact three oil coefficients in regression (8). The three oil coefficients remained significant and the two one-‐month lagged oil variables also stayed significant. What is interesting is that for the retail industry the new two-‐month lagged variable is also negatively significant. Apparently the oil price two months prior to the stock returns also has a negative impact. It is a small influence, but if the oil price is high two months and one month prior to the researched month the stock returns are lower. Again, this might be because of higher transportation costs and thus decreasing profit. A higher cost of living for consumers, because of a higher oil price, might also be the cause. The other three industries are not significantly influenced by the oil price of two months ago.
The next two tables contain the results of the weighted average
regression based on market value with the two lagged oil coefficients. In table 11, regression (7) is done with the weighted average returns.
Table 11. Results based on the weighted average regression based on market value with the lag oil variable. Significance is given with a *, ** or
*** for 10%, 5% and 1% respectively.
Portfolio 𝐑𝐎𝐢𝐥𝐭!𝟏 ROil Mkt-‐Rf SMB HML Intercept
Oil Industry .0186 (.047) .1274** (.054) .8693*** (.102) -‐.4140* (.238) -‐.0320 (.266) -‐.0854 (.351) Automobile Industry -‐.0307 (.081) .0078 (.079) .9174*** (.128) -‐.1285 (.308) .2747 (.403) .2736 (.475) Airline Industry -‐.1744*** (.052) -‐.0915 (.058) 1.030*** (.092) .1469 (.236) -‐.0259 (.257) .0888 (.348) Retail Industry -‐.0860** (.043) -‐.0120 (.034) .6386*** (.059) -‐.5322*** (.185) .0418 (.155) .2764 (.250)
Before adding the lagged oil variable, there were two oil coefficients significant, the one in the airline industry and the other in the oil industry. The coefficient in the airline industry became insignificant, but the lagged oil variable in that industry is significant. Apparently the oil price of the previous month explains the weighted average return of the market value more than the oil price of the same month. In the retail industry the same can be said, these results are similar with the results of the panel regression. Again the one-‐month lag results of the oil and auto industry are not significant.
In table 12 the results can be found for regression (8), where again the dependent variable is the weighted average return based on market value.
Table 12. Results based on the weighted average regression based on market value with the two lag oil variables. Significance is given with a *, **
or *** for 10%, 5% and 1% respectively.
Portfolio 𝐑𝐎𝐢𝐥𝐭!𝟏 𝐑𝐎𝐢𝐥𝐭!𝟐 ROil Mkt-‐Rf SMB HML Intercept Oil Industry .0543 (.047) -‐.0966** (.047) .1166** (.052) .8890*** (.097) -‐.4601 (.232) -‐.035 (.258) -‐.0923 (.34) Auto Industry -‐.0352 (.082) .0082 (.083) .0139 (.080) .9129*** (.131) -‐.1217 (.3077) .2854 (.408) .2924 (.492) Airline Industry -‐.1731*** (.046) -‐.0031 (.047) -‐.0927 (.0595) 1.0307*** (0.094) .1449 (.238) -‐.0277 (.256) .0855 (.358) Retail Industry -‐.0857** (.039) -‐.005 (.039) -‐.0072 (.035) .6366*** (.060) -‐ .5316*** (.188) 0.0524 (.154) .295 (.259)
The original oil variable is still significant for the oil industry, but now also the two-‐month lagged oil variable is also significant, but negatively. This is an
unexpected result, because this suggests that the oil price of two months prior to the month researched has a negative influence on the weighted average return of the market value of oil companies. Normally if the oil price goes up the returns of an oil company go up and thus the market value, but for some reason this is the opposite. A reason for this might be that companies try to protect themselves for a declining oil price. For the auto industry not one oil coefficient is significant. In the airline industry the one-‐month lagged oil coefficient is significant and the same holds for the retail industry. This is quite similar to the results of the panel regression.
5. Conclusion
The first hypothesis discusses the effect the oil price has on the different companies in each industry. In the oil industry 24 companies were researched, 21 of them were positively influenced by the price of oil. If the demand for oil goes up the oil price increases, this increases the revenue for the oil companies. The result of this first regression is thus economically what you would expect. Since 15 of the 24 companies are significantly influenced by the oil price, hypothesis one is rejected for the oil companies, and the oil companies are thus influenced by the oil price. In the auto industry 18 companies were researched, not one of these companies was significantly influenced by the oil price. From the companies only three of them had a negative oil coefficient. Most of the
companies thus benefit if the oil price goes up, although this effect is not
different from zero, but still not what you would expect. On the other hand if the oil price is low, consumers are not going to buy more cars and thus the stock prices of auto companies would not necessarily go up. Since there are no
significant results, the auto companies are not influenced by the price of oil and thus hypothesis one is accepted. In the airline industry 20 companies were researched, 4 out of 20 coefficients for oil were negatively significant. All the other companies also had negative oil coefficients. The oil price thus negatively influences the stock returns of airline companies, but not significantly. The reason for this is probably the increased variable cost, if the oil price increases. Since only 4 out of 20 were significant, hypothesis one is accepted for the airline companies and the companies are thus not significantly influenced by the oil price. In the retail industry 20 companies were researched, 4 of them had
significant oil coefficients. What interesting was that 9 were positively influenced and 11 negatively influenced by the price of oil. This result is interesting because this industry was chosen, because it was not directly dependent on the price of oil, and this result is thus in line with the expectations. It really depends on the company whether or not oil influences the stock returns of the retail companies negatively or positively. Since only 4 companies had significant oil coefficients hypothesis one is accepted for the retail industry and the companies are not
influenced by the oil price.
