OPEC’s influence on oil stock returns
Maarten Croese1Abstract
This study examines the influence of OPEC quota decisions on the stock price of 4 typical-listed- oil firms in Europe. In addition, I consider the influence on the WTI crude oil price. Using the event study methodology, 51 OPEC announcements are considered in the period 1991 – 2012. I find that OPEC quota decisions have a direct influence on both crude oil returns and oil firms’ stock returns. This influence is either positive or negative and large or small, depending on the type of decision and the size of the firms in terms of market capitalization. However, since the difference between the 2 small firms is also significant, I conclude that market capitalization alone is not a determining factor.
Key words: OPEC, oil firms, stocks, event study, Europe Supervisor: Dr. W. Westerman
JEL codes: C12, G14, L90
1 Maarten Croese (s1623699) is a Master student at the economics and business faculty of the University of
1 I Introduction
Oil is inevitably one of the most valuable natural resources our planet has, as it is the most widely used component of primary energies in the world (Mazraati and Tayyebi Jazayeri, 2004). This is acknowledged by Hamilton (1983), Gilbert and Mork (1984), and others, who state that oil plays an important role in the economy. The first even concludes that increases in oil prices are related to declines in real GNP in the US. Therefore, one could imagine that countries with oil in their ground are in a way very powerful because of the dependency of countries that don’t have oil as a natural resource (Guidi et al. 2006). Bredin and Muckley (2011, p. 90) state that in the future, global oil demand is expected to grow by approximately 1% per year and that Europe, as a net importer of oil, therefore has a strategic vulnerable position. In line with this reasoning, geopolitical tensions (conflicts in the Middle East) and uncertainty over supply and demand have had significant influences on the oil price in Europe in the last decades, resulting in heavily fluctuating prices. However in the current economic situation, a more structural issue drives up oil prices; Asia’s booming economy is demanding more oil for its production and it seems unlikely that this demand will flatten in coming years. For example, in 2005 China was the second largest oil importer worldwide and it accounted for 31% of global growth in oil demand (Zweig and Jianhai, 2005). Even though recent dips in China’s industry growth have temporarily depressed oil prices, the potential growth of the economy suggests future oil prices to boom. A lot of research has been done to examine the relation between oil spot and future prices and stock returns (Huang, Masulis and Stoll 2006, Sadorsky, 2001, Jones and Kaul, 1996, and others). Oil-rich countries in the Middle East have always had influence on oil prices. This has been a matter of supply and demand, which is why Western conflicts with the Middle-East could temporarily boost oil prices. Since oil is a commodity that is used in almost every product, but also as an input for almost every plant to be able to operate, the link to stock prices seems logical.
2 counties) having started the joint Organization of Petroleum Exporting Countries (OPEC).2
In the decades thereafter the composition of the organization changed (for example, Gabon terminated its membership in 1995) and currently the OPEC has 12 member states. These are the 5 initial member states, joined by Qatar, Libya, United Arab Emirates, Algeria, Nigeria, Angola and Ecuador. They supply around 40% of the world’s oil production (Simpson, 2008). The OPEC describes its central mission as ‘to coordinate and unify the petroleum policies of its Member Countries and ensure the stabilization of oil markets in order to secure an efficient, economic and regular supply of petroleum to consumers, a steady income to producers and a fair return on capital for those investing in the petroleum industry.’ 3 In order to achieve this mission, the OPEC has several meetings a year (OPEC conferences) to agree on the supply level of world oil production. These meetings generally result in an official statement (hereafter referred to as OPEC announcement) in which a declaration about the oil production level is made (either cut, increase or maintain). Prior to the announcement, speculation about the outcome is widespread because, as research confirms, the different policies seem to have influence on the crude oil prices. Therefore the OPEC conferences can be seen as an important source of information to the market. In addition, Mazraati and Tayyebi Jazayeri (2004), state that “decisions made by OPEC, and the way they are implemented, can greatly affect oil market sentiment and prices’’. Therefore they argue that OPEC, as a key market player, should be transparent and analyze the economic situation thoroughly, in order to take well-timed and sensible decisions. This paper is therefore potentially valuable for OPEC’s future strategies concerning oil production levels in giving a deeper understanding of the impact of their decisions. In addition, the paper might be valuable for investors in the energy sector in giving them a better understanding of the different mechanisms (i.e. OPEC announcements, oil prices and their interaction) that influence oil stock returns. Existing literature has focused upon the relation between OPEC announcements and crude oil prices, although there is ambiguity in the results. However, even though the relevance of OPEC announcements and oil prices is acknowledged by the large amount of literature, a gap exists when it comes to market reactions on OPEC announcements, with respect to oil producing firms. The main purpose of this research is to determine whether OPEC
2http://www.opec.org/opec_web/en/about_us/25.htm
3 announcements are related to stock price movements of firms in the oil industry. Moreover, by selecting four typical firms (two large and two small firms) this study is the first to investigate the relation between exogenous oil supply shocks (i.e. the OPEC announcement) and stock returns of different oil firms in Europe. By distinguishing between large firms on the one hand (Shell and BP) and small firms on the other hand (Premier Oil and DNO International), I investigate if firm specific characteristics (i.e. size in this study) play a role in this relation.
The relation between OPEC announcements and oil stock returns cannot be seen separately from the relation between oil prices and oil stock returns. Therefore, and in line with existing literature among this topic, I first examine the relation between OPEC announcements and crude oil prices. Moreover, since a significant relation is found between OPEC announcements and oil futures prices on the one hand (Lin and Tamvakis (2009), Hyndman (2008), Guidi, Russell and Tarbert (2006), Demirer and Kutan (2010)) and oil futures prices and oil stock returns on the other hand (Huang, Masulis and Stoll (2006), Sadorsky (1999), Jones and Kaul (1996), and others) one could expect a relation between OPEC announcements and oil stock returns.
The impact of OPEC announcements on an entire share price index (of the UK and US stock markets) is measured by Guidi (2006). In his paper, he particularly suggests further research to be done to investigate the influence of OPEC announcements on individual firms. This paper is, as far as I know, the first to do such research. With respect to the expectations of my outcomes, I look at the outcomes of Guidi (2006) and Huang (1996). The first compares the US- and UK stock market reactions on OPEC announcements and finds a significant negative reaction in the share price index of the UK, at the day of an oil production cut announcement (for the US, no significant results are found for either a production cut or increase). The reasoning behind this seems straightforward. Since even the largest oil producing firms are dependent on OPEC supplies,4 a production cut (less
supply of oil) implies that oil producing firms have to pay more for their oil (i.e. a higher price), which implies a negative influence on firm cash flow and thus stock returns.
4 However, Huang’s (1996) results (oil futures returns lead by one day on oil stock returns) show an opposite direction.
As stated by Simpson (2008), the OPEC cartel has been a frequent topic of discussion. Some argue that the cartel has initiated unlawful conspiracies with large oil firms, causing extremely high gasoline prices and large differences in regional oil prices. In this paper, I assume efficient markets, where new information to the market is directly reflected in market prices of oil production firms. As stock prices reflect the discounted value of future cash flows, I investigate how news concerning the oil production levels is translated by the market into future performance of the firm. A widely used method to measure the effect of unexpected news ‘shocks’ to the market is the event study methodology (Brown and Warner, 1980, 1985 and MacKinlay, 1997). Since the OPEC conferences are always concluded with an official press announcement about the oil production level, the unexpected character of the event is guaranteed. However, since the OPEC conferences may take some days, I will capture the pre-announcement days in my analysis to account for any leaking effects.
The rest of this paper is structured as follows; in section (2) I provide an extensive overview of exiting literature concerning the relationship between oil spot- and future prices, stock returns, and OPEC announcements and I develop the hypotheses. Next, in section (3) I discuss the data, followed by the methodology in section (4). Subsequently, section (5) provides an overview of the results. Finally, in section (6) the concluding remarks and limitations of the research are discussed.
II Literature review
5
OPEC and oil prices
Lin and Tamvakis (2009) investigate the influence of OPEC announcements on crude oil prices. They differ in ‘heavy’ and ‘light’ crudes (bad versus good quality) and use data of announcements between 1982 and 2008. They find that a quota cuts always result in positive and statistically significant returns (price increase). For quota increases they find opposite results, however not always statistically significant. When prices are already relatively high, they find no significant results. When they look at differences in types of oil (e.g. heavy versus light crudes) they find significant results only when a decision is made to leave quotas unchanged. Thereby, the low quality heavy crudes have bigger price losses than the high quality lighter crudes.
6 foreign policy and to stabilize the US energy market. In their research, Demirer and Kutan (2010) distinguish between OPEC and SPR announcements with respect to the oil production volume. Their dataset consists of all announcements within the period 1983 – 2008. They only find statistically significant reactions for announcements that the oil quota would be reduced. In line with the findings of Hyndman (2008), they find no significant reaction of oil prices when a quota in production is increased. They do find significant results when OPEC announces to maintain the status-quo, i.e. when OPEC does not change production. They find significant negative cumulative abnormal returns, indicating price decreases in respond to the announcement. Therefore they conclude that the market seems to be surprised by the oil cut- unchanged announcements. With respect to the SPR announcements (increase or decrease), they find no significant cumulative abnormal returns.
Bina and Vo (2007) perform an event study to investigate OPEC output decisions in the period 1983 – 2005. On contrary to the previously described research, they only find statistically significant results when an oil increase is announced. The unchanged ‘group’ as well as the oil cut group show no significant results in the event window. What is remarkable is that the significant results are found in the days prior to the announcement date, indicating that the news reached the market before the official statement.
Schmidbauer and Rösch (2011) use OPEC announcements in the period 1986 through 2009 and investigate price reactions on West Texas Intermediate (WTI), a grade of crude oil also known as Texas light sweet. It is used as a benchmark in oil pricing and it is the underlying commodity of the New York Mercantile Exchange (NYMEX) oil futures contracts. They find that OPEC production cut announcements result in an oil price decrease opposed to a ‘maintain’ and increase decision that both result in a price increase.
7 the OPEC price announcements, indicating that the pronouncements have already been incorporated in market expectations.
Oil futures- and spot market
The October 1973 Arab-Israeli War initiated an embargo of oil shipments to the US from OPEC countries. It changed the oil market dramatically and provided the initiation for a futures market to be developed. In the beginning the future contracts were not traded very often, and forward contracts remained the most frequently used form of contracts. However, with the 1979 oil shock as a follow-up of the Iranian revolution, oil spot prices rose sharply again. These disruptions made long-term forward contracts less and less reliable, while at the same time the use of futures contracts increased sharply. Nowadays, oil futures contracts are traded many times more than heating oil contracts (Huang et al., 1996). Investors (speculators, hedgers) are the main agents that demand and supply these commodity contracts and thus together determine the price. As new information reaches the market, agents immediately incorporate this information into the price and thereby trade their contracts until a new equilibrium price is determined. In many oil related research there is a lack of clarity as for what prices to use.
Both futures and spot prices react to new information. In addition, there is ambiguity among research as for what prices reflect better new information to the market. Therefore I assume OPEC announcements to have a simultaneous impact on both oil spot- and future markets and I will only investigate the effect of OPEC announcements on spot prices. The following hypothesis is being tested;
H1: OPEC announcements do not have an effect on the spot price of the crude oil commodity. Oil and stock prices
8 Stock prices are a reflection of discounted values of expected future cash flows. More formally,
p =
E(c)E(r) (1)where p is the stock price, r is the discount rate, c the cash flow stream and (E) the expectation operator. Stock prices are thus affected by either expected cash flows, or the discount factor. Guidi (2006) names various reasons how future oil prices can affect expected cash flows and discount rates. The first is straightforward and already mentioned shortly in the introduction. Since oil is a real resource which is necessary for the production of many goods, expected changes in oil prices causes changes in the expected costs of a firm. In that way the expected cash flow of a firm is directly affected and thus stock prices will move in opposite directions as energy prices do. In addition, the effect will also depend on whether the firm is a net producer or consumer of oil. For the world economy as a whole, oil is an input and an increase in oil prices will therefore depress stock prices and real output (Huang et al., 1996, Mork et al., 1994 and Hammoudeh and Li, 2005). Another way how expected oil prices can affect stock returns is via the discount rate, since the expected discount rate is composed of the expected real interest- and expected inflation rate. Both may vary as respond to different oil prices. Huang et al. (1996) explain this by considering the United States, a net importer of oil. Higher oil prices would negatively affect the US’ balance of payments, putting a downward pressure on the dollar’s foreign exchange rates and an upward pressure on the US’ inflation rate. This implies that a higher expected inflation rate (as a consequence of higher oil prices) is negatively related to stock returns, as it is positively related to discount rates. However, since oil is a commodity, oil prices also track the inflation rate and so expected changes in the oil price might serve as a proxy for the expected inflation rate. So if an increase in the inflation rate causes stock prices to decline, and oil prices to rise at the same time, the negative impact of inflation on stock prices might be overstated.
9 market indices like the S&P 500. However they did find significant results for the correlation between the stock prices of three oil producing firms. The stock returns of Exxon, Chevron and Mobil are found to be positively correlated with current and lagged oil futures returns. However, when they test for the economic significance of this result, Huang et al (1996) find that the bid-ask spread is too large relative to the movement of oil and stock prices so that investors are not able to profit of this small arbitrage opportunity. Dorsman et al. (2013) study the general movements of oil-, stock- and bond returns in order to determine whether the addition of oil to a traditional portfolio of bonds and stocks can improve the risk-return trade-off. They show that sometimes stock- and oil prices show co-movements (during financial crisis), while sometimes they move independently from each other. In general they conclude that there are no specific cyclical or counter cyclical patterns to be identified. However they do show that adding oil to a traditional portfolio improves the risk-return trade-off, i.e. oil serves as a hedge for both stocks and bonds. These findings show that oil prices and oil firms stock returns, plus their interdependent relation, are also dependent on exogenous economic circumstances.
Hammoudeh and Li (2005) examine the oil sensitivity of equity returns of 2 oil exporting countries (Norway and Mexico). In addition, they perform a sector analysis in which they test the sensitivity to oil prices to two oil sensitive industries, the US oil industry and the transportation industry. They acquire oil future prices from the NYMEX and use the Amex Oil Stock Index5 (AMEXO) as a representation of the US oil industry. They find that oil price
growth has a positive impact on oil-related stocks and that there is a negative relation between the oil price and the US transportation industry. In addition, the different impact of an oil price increase on different countries makes them conclude that investors should first invest in the US oil industry, than in the Mexican stocks before investing in Norwegian stocks to take an optimal profit of higher oil prices.
In a comparable study, Boyer & Fillion (2007) measure the stock sensitivity of Canadian oil and gas stocks to five common factors, namely the interest rate, market return, oil and gas prices and the Canadian exchange rate in relation to the US dollar. They find that all factors
10 are significant at the 1% level and therefore state that the crude oil price and natural gas price have a positive impact on oil and gas stock returns, whereas interest and exchange rates have a negative influence. These findings are in line with the findings of Sadorsky (2001), who shows that Canadian oil- and gas firms’ stock returns are positively related to changes in oil price. In addition, Boyer & Fillion (2007) differentiate between oil producers and integrated firms, in which the latter are firms that have upstream (exploration, development, production) and downstream (distribution, marketing, refinery) activities. They find only two factors to be significant (market return, natural gas price return) in relation to integrated firms’ stock returns. Also, they find the impact of oil and natural gas prices to be larger in producing firms than integrated firms. As an explanation for this they state that integrated firms have less risk exposure because of vertical integration.
Nandha and Faff (2006) investigate 35 industry indices and find that oil price increases have a negative impact on equity returns, except for the oil, mining and gas industry. They confirm the view that the increases in earnings from OPEC- and other oil exporting countries due to a higher oil price is more than outweighed by the negative impact this has on economic activity in importing countries.
In the above section, the relation between OPEC announcements and oil prices and between oil and stock prices has been described. This paper expands the current body of research by examining OPEC6 press announcements concerning the oil production level in
relation to the stock market. In particular, I focus on oil producing firms’ stock returns to determine how the market values news concerning oil production levels.
Now, in order to investigate whether there is a direct link between OPEC announcements and oil stock prices, the following hypothesis is being tested;
H2: OPEC announcements have no effect on the stock prices of oil producing firms.
The next hypothesis to be tested is a follow-up of the previous one and will only be tested when there is evidence that OPEC announcements do have an effect on the stock prices of oil producing firms (i.e. a rejection of H2). With this hypothesis, in line with the findings of
11 existing literature considering OPEC announcements and oil prices, I will test for asymmetry in the effects of different OPEC announcements on stock prices. Therefore, the following hypothesis is being tested;
H3: There is no asymmetry in the effect of different announcements on the stock price of oil firms.
Finally, as far as I am concerned, no research has ever differentiated between different oil firms and their stocks’ reactions on OPEC announcement. By investigating 2 very large and 2 relatively small firms (in terms of market capitalization, table 1 in the appendix), I want to investigate whether the market responds differently to OPEC announcements, with respect to differences in this firm characteristics. Therefore I develop the following hypothesis:
H4: There is no asymmetry in the effect of different announcements on the stock price of different oil firms, based on the firms’ market capitalization.
III Data
Daily price data of the WTI crude oil, traded on the NYMEX is analyzed for the period 1992 – 2012. WTI is high quality oil, ideally to process gasoline, kerosene and high-quality diesel. Therefore, its demand is high in industrialized countries. It is used as a benchmark in oil pricing and is also known as ‘light sweet’. WTI is the commodity that is used most by investors to track the oil price.7 Therefore, I use WTI futures as a benchmark for the oil
price reactions on OPEC announcements.
Furthermore, daily stock prices of the 4 largest- (Royal Dutch Shell and BP) and 2 of the smallest listed European oil firms (Premier oil and DNO international) are retrieved. The firms are selected in order to have a European representation, of both small and large firms. The choice for Royal Dutch Shell and BP in that light is straightforward. Next, a search for 2 small firms with similar activities that are both listed resulted in Premier Oil and DNO International. A list containing information of the oil firms is provided in table A1
12 of the appendix. I use the entire sample of large and small firms to test for market reactions on OPEC announcements. Furthermore, by comparing the largest and smaller firms, I want to investigate whether the market responds differently to these firms, with respect to the OPEC announcements. In addition I compare the different announcements (quota cut, increase, unchanged) to test for asymmetry in stock price reaction. The stock prices of each firm are requested around the days of the announcement. A more thorough description of the estimation- and event window will be described in the methodology section. The dates for the OPEC conferences and their press releases in the period 1992 – 2012 were found on the OPEC website8 and in an article by the OPEC secretariat (2003). In total 51 OPEC
announcements are considered in this study, of which 16 are quota ‘cut’ announcements, 12 are quota ‘increase’ announcements and 23 are ‘unchanged’ announcements. A list of the OPEC announcements is provided in table 1. The daily stock price- and WTI futures data is retrieved from Thomson DataStream.
Table 1: List of OPEC announcements.
Announcement date Decision Announcement date Decision Announcement date Decision
14-Jun-12 o 1-Jun-06 o 23-Mrt-99 -
14-Dec-11 o 8-Mrt-06 o 24-Jun-98 -
11-Dec-10 o 31-Jan-06 o 30-Mrt-98 -
14-Oct-10 o 15-Jun-05 + 1-Dec-97 +
17-Mrt-10 o 16-Mrt-05 + 26-un-97 o
22-Dec-09 o 10-Dec -04 - 28-Nov-96 o
10-Sep-09 o 03-Jun-04 + 7-Jun-96 +
28-May-09 o 10-Feb-04 - 22-Nov-95 o
17-Dec-08 - 24-Apr-03 - 22-Nov-94 o
24-Oct-08 - 12-Jan-03 + 26-Mrt-94 o
10-Sep-08 o 12-Dec-02 - 29-Sep-93 +
5-Mrt-08 o 17-Mrt-01 - 10-Jun-93 o
01-Feb-08 o 17-Jan-01 - 16-Feb-93 o
5-Dec-07 o 30-Oct-00 + 27-Nov-92 -
11-Sep-07 + 11-Sep-00 + 17-Sep-92 +
14-Dec-06 - 21-Jun-00 + 22-May-92 o
20-Oct-06 - 29-Mrt-99 - 15-Feb-92 -
• Note: Dataset comprises of normal (bi-annual) OPEC meetings, and the extraordinary OPEC meetings. The latter are not planned in advance but are organized due to -changing- economic circumstances which require a more rapid quota decision according to OPEC.
13 Figure 1 shows the monthly changes of the WTI crude oil price in the period 1992 – 2012.
Figure 1: WTI crude oil price per barrel (1992 – 2012).
As one can see in figure 1, the price of oil has changed radically over time. In the first ten years of the sample period, oil prices fluctuate around 20 to 30 dollars per barrel9. From
that point on, prices more than quadrupled with an absolute peak in July 2008. Thereafter prices fell sharply in 2009 to 40 dollars a barrel, while climbing back to the 100 dollar level in the years following. While world real GDP increased by 9.4% between 2003 and 2005, the next two years even had a 10.1% cumulative growth. Chinese oil consumption alone in that period increased by 870.000 barrels per day. Consistent with the laws of supply and demand, the economic downturn halfway 2008 caused the oil price to decrease dramatically. US petroleum consumption fell by 8.8% in that period.10 The relative repair of
the crisis in the years thereafter caused the crude oil price to climb again from March 2009.
Figure 2 shows the time series of the 4 stock prices researched in this study. Because of large differences in prices, the stock of BP is used as a base and the other stocks are rescaled to get a better oversight of how the different stocks moved over time.
9
1 barrel of oil contains around 158 liters of oil.
10 http://www.econbrowser.com/archives/2009/04/causes_of_the_o.html $0,00 $20,00 $40,00 $60,00 $80,00 $100,00 $120,00 $140,00 $160,00 1-1992 1- 2-1993 1- 3-1994 1- 4-1995 1- 5-1996 1- 6-1997 1- 7-1998 1- 8-1999 1- 9-2000 1- 10-200 1 1- 11-200 2 1- 12-200 3 1-2005 1- 2-2006 1- 3-2007 1- 4-2008 1- 5-2009 1- 6-2010 1- 7-2011
14 Figure 2: Time series of the different stock (1992 – 2012). Vertical axis only reflects BP stock price. Royal Dutch Shell, Premier Oil and DNO international stocks are rescaled with factor 15, 2 and 40 respectively to better show their relative movement over time.
In period 1992 – 1998, the small stocks (Premier oil, DNO International) show a clear relation with the crude oil price. The time series of the larger firms in this period suggest that larger firms are less dependent on the oil price in their stock performance. On the other hand, Royal Dutch Shell and BP tend to move in similar directions over time. Furthermore the economic crisis in 2008 is reflected in all stock movements, however in the recovery period, when oil prices increase again, only BP shows a remarkable drop in share price. This drop obviously holds relation with the BP oil disaster in the Gulf of Mexico on April 20th, 2010. During an explosion on the ‘Deepwater Horizon’ oil rig, approximately
750 million liters of oil came into the ocean and BP had to set aside 37.2 billion dollars for the costs of the spill11. Credit rating office Fitch immediately downgraded BP from a
healthy AA status to BBB, almost the junk status12. The jump in Premier oil shares in 2010
is a result of the discovery of new oil fields in the North Sea in June of that year13. The
irregularity of the BP oil disaster and the discovery of a new oil field for Premier oil could 11 http://www.ft.com/intl/cms/s/0/69771938-b184-11e1-9800-00144feabdc0.html#axzz22xcGz0B1 12 http://www.elsevier.nl/web/Nieuws/Buitenland/268385/BP-stelt-16-miljard-euro-beschikbaar-voor-schade olieramp.htm 13 http://www.guardian.co.uk/business/marketforceslive/2010/jun/28/premieroil 0 200 400 600 800 1000 1-1992 1- 2-1993 1- 3-1994 1- 4-1995 1- 5-1996 1- 6-1997 1- 7-1998 1- 8-1999 1- 9-2000 1- 10-200 1 1- 11-200 2 1- 12-200 3 1-2005 1- 2-2006 1- 3-2007 1- 4-2008 1- 5-2009 1- 6-2010 1- 7-2011 BP
15 bias the dataset, however no OPEC announcements are held during or around these two events.
The descriptive statistics for the firms’ stock prices and the WTI crude oil daily prices are presented in the following tables. First, I present the descriptive statistics containing the whole dataset of events (table 2a) without distinguishing for the different announcements. Following, table 2b, 2c and 2d present the descriptive statistics for the quota cut, -increase or unchanged quota respectively.
Table 2a: Descriptive statistics in the estimation window (-65:-10) for the four stocks and WTI crude oil. All the OPEC announcements are considered (N=51).
Statistic Total (of 4 firms) Shell BP Premier Oil DNO Int. WTI oil
Mean 0,0000 -0,0002 -0,0001 0,0001 0,0002 0,0000 Median 0,0000 -0,0004 -0,0004 0,0003 -0,0001 0,0000 St. deviation 0,0023 0,0021 0,0025 0,0039 0,0059 0,0031 Minimum -0,0058 -0,0044 -0,0064 -0,0073 -0,0130 -0,0057 Maximum 0,0045 0,0051 0,0063 0,0097 0,0133 0,0071 Skewness -0,1915 0,4699 0,0683 0,0937 -0,0484 0,2157 Kurtosis 0,1402 -0,0266 0,7880 -0,5431 -0,2561 -0,3172 Jarque-Bera 70,7654 21,3416 10,4371 26,7516 22,5498 23,7791
• Note: Since the Jarque-Bera statistic is larger than 5,99, a normal distribution cannot be assumed. Therefore, the non-parametric Wilcoxon signed rank test is conducted in addition to the student’s t-test. More information on these tests is explained in the methodology section.
For a parametric student’s t-test to be accurate, the data in the estimation period has to be approximately normally distributed. Therefore, the Jarque-Bera statistic is calculated, using the following formula:
(2)
16 Table 2b: Descriptive statistics in the estimation window (-65:-10) for the four stocks and WTI crude oil, for the quota cut announcement (N=16).
Statistic Total (of 4 firms) Shell BP Premier Oil DNO Int. WTI oil
Mean -0,0016 -0,0004 -0,0005 0,0002 -0,0006 -0,0003 Median -0,0036 -0,0009 -0,0014 -0,0007 -0,0021 0,0000 St. deviation 0,0210 0,0045 0,0045 0,0085 0,0126 0,0072 Minimum -0,0461 -0,0099 -0,0086 -0,0240 -0,0386 -0,0152 Maximum 0,0471 0,0102 0,0101 0,0199 0,0267 0,0229 Skewness 0,3628 0,3403 0,5090 0,1736 -0,2788 0,5548 Kurtosis -0,2273 -0,4103 -0,3068 0,9215 0,7224 0,9691 Jarque-Bera 7,2948 8,0621 7,9806 2,9605 3,6656 11,3809
Table 2c: Descriptive statistics in the estimation window (-65:-10) for the four stocks and WTI crude oil, for the increased quota announcement (N=12).
Statistic Total (of 4 firms) Shell BP Premier Oil DNO Int. WTI oil
Mean -0,0004 0,0001 0,0002 -0,0006 -0,0001 0,0002 Median -0,0008 -0,0004 -0,0004 -0,0015 0,0007 0,0008 St. deviation 0,0162 0,0039 0,0043 0,0060 0,0111 0,0072 Minimum -0,0347 -0,0080 -0,0153 -0,0149 -0,0413 -0,0214 Maximum 0,0370 0,0076 0,0086 0,0166 0,0215 0,0182 Skewness -0,0924 0,1246 -0,6088 0,5537 -0,8137 -0,2711 Kurtosis -0,3226 -0,7928 2,1022 0,6275 2,6026 0,9207 Jarque-Bera 5,5370 7,2238 1,1444 3,4276 1,4030 9,8118
Table 2d: Descriptive statistics in the estimation window (-65: -10) for the four stocks and WTI crude oil, for the unchanged quota announcement (N=23).
Statistic Total (of 4 firms) Shell BP Premier Oil DNO Int. WTI oil
Mean 0,0010 -0,0002 -0,0001 0,0005 0,0008 0,0004 Median 0,0013 -0,0002 -0,0001 0,0004 0,0008 0,0003 St. deviation 0,0146 0,0027 0,0031 0,0048 0,0085 0,0037 Minimum -0,0322 -0,0076 -0,0074 -0,0081 -0,0178 -0,0075 Maximum 0,0471 0,0063 0,0064 0,0118 0,0289 0,0088 Skewness 0,1791 -0,1212 -0,2348 0,2873 0,4368 -0,0870 Kurtosis 1,4517 0,4416 -0,2588 -0,7299 1,2506 -0,3607 Jarque-Bera 2,4205 6,3288 10,3884 13,6485 3,6644 24,0647 IV Methodology
17 crude oil prices. In line with Schmidbauer and Rösch (2011), I use WTI as crude oil type in this paper as well. The event study methodology is particular useful to test the impact of unexpected events, that are not anticipated by the market. In this study the unexpected event is defined as the OPEC press announcement about the oil production. Should these press announcements have no impact on stock- or oil prices, than the average abnormal returns (AAR’s) in the event window are insignificant. In this study I use an event window of 21 days (-10:10). 10 days prior to the OPEC press announcement and 10 days after the event day (t=0). Lin and Tamvakis (2009) use a similar event window. The duration of some of the events is longer than one day, and therefore there is a bigger chance of leaking effects, prior to the official OPEC press announcement. With a long prior-event window I try to capture any possible leaking effects, which would be reflected by significant AAR’s in the days (-10:0) of the event window. The argument for the long post-event window (0:10) is given by Lin and Tamvakis (2009) that it takes longer for the market to absorb information, and that this new information is therefore not priced directly into stock prices. On contrary, they argue that when the event window is too long, other information than the OPEC announcements could have an effect. However Wirl and Kujundzic (2004) found evidence that from a variety of event windows, the most appropriate is an event window with 10 days before and after the event.
18 By comparing the average of returns in the estimation period by the returns in the event window, the impact of the event can be determined.
Brown and Warner (1980, 1985) describe several ways to compute the average abnormal returns in the event window, of which the mean adjusted return, is the most basic one. However they find it yields similar results as the more complex methods to calculate average abnormal returns. Therefore, in line with Bina and Vo (2007) and Lin and Tamvakis (2009) I use the mean adjusted return method as well in this paper. In addition, I show results of the market adjusted return (Brown and Warner, 1980, 1985), in which the mean adjusted returns are corrected for the movement of the market. In this paper I use the MSCI world index as a proxy for market return.
The mean-adjusted return method subtracts the average of the total returns µt in the estimation period from the actual returns Ri,t to calculate the abnormal returns in the event window. In formula:
𝐴𝑅𝑖, 𝑡 = 𝑅𝑖, 𝑡 − µ𝑡 (3)
In addition as described by Brown and Warner (1980, 1985) I calculate the cumulative abnormal returns (CAR) by adding up the abnormal returns in the event window. In formula:
𝐶𝐴𝑅𝑖 = ∑𝑇 ARi, t
𝑡=1 (4)
In line with Deaves and Krinsky (1992) and Demirer and Kutan (2010) and first described by Brown and Warner (1980, 1985) I also use the market adjusted return method in this paper. As stated before, this method makes the assumption that each security moves with the market. Therefore, abnormal returns are calculated by subtracting the market returns of the individual security returns. In formula:
19 In order to test if the calculated abnormal and cumulative abnormal returns are significant, I use the one sample Student t-test as discussed by Brown and Warner (1980, 1985). It tests if the Abnormal Returns in the event window defer significantly from zero. In addition, I perform the non-parametric Wilcoxon signed-rank test as an alternative since not all data is normally distributed. It tests the significant deviation of the median of the different abnormal returns from zero.
V Results
In this section I present the results for the mean- and market adjusted method. In addition, the Wilcoxon singed rank values are shown since not all sub-datasets (see table 2a, 2b, 2c, and 2d) are normally distributed. However, as stated before and in line with current research, the mean-adjusted method is a solid measure for calculating abnormal returns. Therefore, and to focus on results rather than methodological issues, I will limit my analysis primarily to the mean-adjusted method. Also when the data is not normally distributed, I assume normality because according to the Central Limit Theorem, the sample of independent random variables will be approximately normally distributed. A look at the Wilcoxon test values shows that in all cases these are less significant than the mean adjusted method. A possible explanation is that the estimation period is not long enough. To give an appropriate estimation of the normal returns, MacKinlay (1997) suggests an estimation period of 200 days. However, all the events (OPEC announcements) would then overlap in this study, which would bias the data. Therefore, I use a shorter estimation period.
(-20 10:10). Figure 1 shows the CAR values in the event window graphically for the mean-adjusted method.
Table 3: Student’s t- and Wilcoxon signed rank test values of the CARs of the WTI crude oil returns around the OPEC announcements
Quota cut Quota increase Quota unchanged
adjusted Mean- adjusted Market Wilcoxon value adjusted Mean- adjusted Market Wilcoxon value adjusted Mean- adjusted Market Wilcoxon value CAR-10 -1,259 -1,109 1,392 -1,259 -0,583 1,537 1,150 0,965 1,992** CAR-9 -0,633 -0,941 0,880 -1,553 -0,850 1,398 -0,802 0,035 0,228 CAR-8 -1,072 -1,628 0,767 -1,241 -0,651 0,349 0,622 0,707 0,471 CAR-7 -1,612 -2,182** 0,483 -1,150 -0,847 0,909 1,057 0,648 0,715 CAR-6 -2,196** -2,717** 0,369 -0,132 -0,262 0,559 -0,308 -0,627 0,228 CAR-5 -0,648 -1,519 0,199 -0,232 -0,286 0,489 -1,351 -1,512 0,106 CAR-4 -0,297 -1,345 0,426 -0,368 -0,325 0,629 -0,576 -0,849 0,198 CAR-3 -0,374 -1,270 0,312 -0,564 -0,329 0,978 -0,696 -0,854 0,259 CAR-2 -0,492 -1,301 0,312 -0,898 -0,257 0,629 -0,602 -0,747 0,106 CAR-1 -0,083 -1,304 0,483 -0,198 0,110 0,699 -1,022 -0,913 0,167 CAR 0 -1,255 -2,278** 0,028 0,909 0,853 0,489 -1,862* -1,656 0,137 CAR1 -2,323** -3,396*** 0,540 1,111 1,294 0,489 -2,258** -2,586** -0,015 CAR2 -2,287 -4,143*** 0,596 1,571 1,550 0,210 -1,891* -2,636** 0,015 CAR3 -1,077 -3,214*** 0,085 0,342 0,943 0,489 -2,137** -2,238** 0,624 CAR4 -0,939 -3,331*** 0,142 0,963 1,563 0,349 -1,484 -1,715 0,745 CAR5 0,346 -2,384** 0,653 0,089 1,069 0,629 -0,967 -1,205 0,350 CAR6 -0,215 -3,388*** 0,426 1,294 1,943* 0,349 0,367 -0,281 1,171 CAR7 1,276 -2,496** 1,164 1,361 1,953* 0,070 -0,587 -0,939 0,502 CAR8 0,138 -3,659*** 0,937 1,123 1,959* 0,280 0,094 -0,178 0,776 CAR9 0,033 -3,685*** 0,710 0,114 1,356 0,839 -0,744 -0,884 0,441 CAR10 1,221 -2,709** 1,164 -1,180 0,633 0,978 -1,867* -1,458 0,137
*= significant at 10% level, **= significant at 5% level, ***= significant at 1% level
21 Figure 1: Graphical representation of the WTI crude oil CARs in the event window, mean-adjusted method.
For the quota announcement being unchanged, the first 4 days after the event are negative significant at both the 5- and 10% level. In addition, on the last day of the event window (CAR -10:10) a negative significant return occurs, albeit at a 10% level. For the quota increase, I find no significant returns for the mean adjusted method. Especially the findings of the quota cut announcement are contradictive to existing literature. Guidi et al. (2006), Demirer and Kutan (2010) and Lin and Tamvakis (2009) find positive significant returns in most cases. However, Lin and Tamvakis (2009) provide an explanation for negative returns after quota cuts as the OPEC not being credible enough to enforce the announced cuts on its members and that the market perceived the cuts as not being far-reaching enough.
Although the results are not all significant, just after the event there are negative significant returns in 2 cases (i.e. a quota cut and unchanged decision) at a 5% level. Therefore I reject the first hypothesis that OPEC announcements have no influence on the spot price of the crude oil commodity.
Now I focus on the most important part of this study, the influence of OPEC announcements on oil companies’ stock prices. Again, I focus primarily on the mean-adjusted method. Table 4 shows the Student’s t- and Wilcoxon signed rank test values of the cumulative abnormal returns for the 4 oil companies taken together, so without distinguishing for small or large companies. The CAR’s of all companies taken together are shown graphically in figure 2. A list of the companies is presented in table A1 of the appendix.
-4,00% -3,00% -2,00% -1,00% 0,00% 1,00% 2,00%
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 Quota cutQuota increase
22 Table 4: Student’s t- and Wilcoxon signed rank test values in the event window for the different OPEC
announcements (N=51) and all companies taken together (N=4).
Quota cut Quota increase Quota unchanged
adjusted Mean- adjusted Market Wilcoxon value adjusted Mean- adjusted Market Wilcoxon value adjusted Mean- adjusted Market Wilcoxon value CAR-10 -0,969 -0,881 0,982 -0,443 -0,430 1,544 -0,805 -0,824 2,171** CAR-9 -0,253 0,650 0,784 0,229 0,652 0,395 -2,702** -1,942* 2,266** CAR-8 0,080 0,954 0,346 -0,831 -0,377 0,149 -2,261** -1,490 1,827** CAR-7 -0,385 0,530 0,537 -0,488 -0,043 0,138 -2,570** -1,807* 1,902* CAR-6 -0,735 0,212 0,859 -0,056 0,376 0,631 -1,825* -1,045 1,245* CAR-5 0,096 0,968 0,092 -0,601 -0,153 0,497 -1,350 -0,558 0,758 CAR-4 1,030 1,817* 1,208 -0,331 0,109 0,056 0,054 0,879 0,002 CAR-3 1,138 1,915* 0,928 0,032 0,461 0,149 -1,039 -0,240 -0,002 CAR-2 0,739 1,552 0,626 0,388 0,806 0,621 -2,313** -1,544 1,130 CAR-1 1,357 2,114** 1,195 0,829 1,233 0,600 -2,660** -1,899* 0,572 CAR 0 0,476 1,313 0,346 0,997 1,397 0,467 -1,875* -1,095 0,346 CAR1 -0,180 0,717 0,229 1,224 1,617 0,713 -1,245 -0,451 0,042 CAR2 0,222 1,083 0,140 1,473 1,859* 0,569 -0,642 0,167 0,291 CAR3 0,465 1,303 0,168 2,113* 2,480** 0,744 -1,648 -0,864 0,825 CAR4 1,007 1,796* -0,003 1,618 1,999* 0,364 -2,409** -1,642 1,011 CAR5 0,944 1,739 0,229 1,610 1,991* 0,538 -1,713 -0,929 0,730 CAR6 1,738 2,461** 0,168 2,314** 2,675** 1,123 -3,036*** -2,284** 1,502 CAR7 2,157** 2,841** 0,243 2,275** 2,637** 1,174 -0,289 0,528 0,461 CAR8 2,449** 3,107*** 0,455 1,653 2,033* 0,969 -1,020 -0,221 0,885 CAR9 1,878* 2,588** 0,250 1,229 1,622 0,631 -1,377 -0,586 0,746 CAR10 2,348** 3,015*** 0,661 0,825 1,230 0,190 -4,157*** -3,431*** 1,415
*= significant at 10% level, **= significant at 5% level, ***= significant at 1% level
Figure 2: Graphical representationof the whole dataset of the oil companies’ CARs in the event window,
23
Quota cut announcement
Table 4 shows some highly significant values for both the mean- and market adjusted method quota cut announcements. Especially at the end of the event window, starting on day 7 CAR (-10; 7), positive significant abnormal returns accrue for the mean adjusted method. In addition, for the market-adjusted method there are positive significant CARs on day 4 (CAR -10: 4), 3 (CAR -10:3) and 1 (CAR -10: -1) prior to the event, which could indicate possible leaking effects. The results are perfect in line with the findings of Hyndman (2008), who describes that: ‘positive abnormal returns accrue, peaking
approximately 3 days before the announcement and then make a significant fall, only to recover again at the end of the event window’. On contrary, Hyndman (2008) finds the same
positive significant returns for the crude oil price, which I don’t find.
Figure 3:Graphical representation of the oil companies’ CARs in the event window, for the quota cut
(mean-adjusted method)
Figure 3 shows the CARs of the 4 different oil companies. The different Student-t and Wilcoxon signed rank values of the different CARs around the event window are shown in table A2, A3, A4 and A5 of the Appendix. Again, I will limit my analysis mainly to the mean-adjusted method. However, the market-mean-adjusted values can be found in the above named tables in the Appendix.
When we look at the values of Shell and BP we see a very similar pattern. In line with the general findings described above of a quota cut for all companies taken together, positive significant values occur. In the prior event window, this occurs on day -4 (CAR-10: -4) and day -1 (CAR -10: -1) for Shell and on day -4 CAR (-10: -4) and day -3 CAR (-10: -3) for BP. In addition, for both Shell and BP the more significant values (on the 5- and 1% level) occur
24 for the first time on day 3 CAR (-10: 3). Then, from day 6 CAR (-10: 5) and day 5 CAR (-10: 6) on for Shell and BP respectively, positive significant returns occur until the end of the event window.
When looking at the CARs of Premier oil (Table A4 of the Appendix), I find different values than what one would expect. The values are negative during the whole event window and significant values occur, starting from day -9 (CAR -10: -9) until day 5 (CAR -10: 5) of the event window. In addition, on day 7 (CAR -10: 7), a negative significant value occurs, albeit on a 10% level. The CARs reach the lowest values on day 1 (CAR -10:1), 2 (CAR -10:2) and 3 (CAR -10:3) of the event window (all significant at 1% level). However, also in the prior event window on days -6 (CAR -10: -6) and -5 (CAR -10:-5), negative significant values of the 1% level occur. When we look at figure 3 we see that in general, the patterns of the 4 companies are not completely random. A certain pattern with increasing CARs on day 1 for Shell, BP and DNO International and day 2 for Premier Oil seems to occur. However, the negative values during the whole event window for Premier Oil give the impression that investors of this firm have different interests concerning a quota cut announcement than for the other 3 firms. Since Premier Oil is the smallest of the 4 firms, assuming higher oil costs because of the quota cut announcement might indicate higher relative costs for this firm. As DNO International also has only upstream activities, the size of the firm (market capitalization) seems to be a determining factor here.
25
Quota increase announcement
Now we look at table 4 for the values of the aggregate CARs of the 4 companies in case of a quota increase announcement. Again, the different Student-t and Wilcoxon signed rank values of the different CARs around the event window are shown in table A2, A3, A4 and A5 of the Appendix. For the mean adjusted method, significant returns occur on day 3 (CAR -10: 3) at a 10% level, and on day 6 (CAR --10:6) and 7 (CAR --10: 7) at a 5% level. Although I find some positive significant returns, the impact is less than for the quota cut announcement. Guidi (2006) argues that quota increase announcements don’t affect oil prices and therefore don’t affect oil companies’ stocks. In addition, Hyndman (2008) argues the same because OPEC always increases the quota in good economic times. Therefore, the market fully anticipates this. I do find light significant values for the WTI crude oil commodity for a quota increase announcement, but only on a 10% level for the market adjusted method (table 3). Therefore, one would expect only light significant values for the oil firms’ stock returns in case of a quota increase, as is the case.
The rationale behind positive significant values to occur could be that OPEC had not increased the quota during economic good times. Then, the market would never be able to fully anticipate the decision of OPEC anymore. As an increase would be required by the market, positive abnormal returns would occur if that decision has been made. It is not certain whether OPEC has made such a decision, as one can never know if the market and OPEC always have a similar view on when economic times are good or bad.
Figure 4:Graphical representation of the oil companies’ CARs in the event window, for the quota increase
26 Figure 4 shows the CARs of the 4 oil companies separately in case of a quota increase announcement. Immediately we can see that the paths of Shell and BP show a similar track, just as was the case for the quota cut announcement. On contrary, the 2 small firms Premier Oil and DNO International show opposite CARS during the event window. Considering the prior event window for Shell, on day 6 (CAR 10: 6) and day 3 (CAR 10: -3), significant returns occur at a 10% level. For BP, no significant returns occur in the prior event window. In the post event window, significant CARs on the first 3 days occur for Shell; on day 1 (CAR -10: 1) at a 5% level, day 2 (CAR -10: 2) at a 1% level and on day 3 at a 10% level for the mean adjusted method. For BP, only on day 2 (CAR -10; 2) and day 3 (CAR -10:3) positive significant values occur at a 5% level.
For Premier Oil positive significant values are found for the first time on day 3 (CAR 10: -3), at a 10% level. Then, from the day of the event (CAR -10: 0) until the last day of the event window, significant positive returns occur. They start at a 10% level on day 0 (CAR -10: 0) and 1 (CAR --10: 1), than at a 5% level on day 2 (CAR --10: 2), to return to a 10% significant level on day 9 (CAR -10: 9) and 10 (CAR -10: 10). For DNO International, there are even less expected results. As can be seen in figure 4, opposed to the other 3 firms, DNO international has negative returns almost during the whole event window. These are significant at a 5% level on days -5 (CAR -10: -5) until -3 (CAR -10: -3), and 2 days after the event (-10: 2). As was the case for Premier Oil for the quota cut, now the market seems to have different interests and beliefs for DNO International when a quota increase occurs than for the other 3 firms.
Quota unchanged announcement
-27 10: 10) at a 1% level. Lin and Tamvakis (2010) find mixed results for the unchanged quota announcement. Although their results are not all significant, they are all negative, as is the case in this study. In addition, Hyndman (2008) finds similar results with strong negative abnormal returns, significant at a 1% level.
Figure 5: Graphical representation of the oil companies’ CARs in the event window, for the unchanged quota announcement (mean-adjusted method)
When we look at figure 5, the CARS in the event window of the 4 companies show that in this case there is a visible difference between CARS of the large (Shell, BP) and small (Premier Oil, DNO International) firms. Shell and BP seem to follow a very similar track, with both negative and positive returns in the prior event window (non-significant). At the end of the event window, positive significant returns occur for Shell at a 5% level on days 7 (CAR -10: 7), 9 (CAR -10:9) and 10 (CAR -10: 10) and on a 10% level on day 8 (CAR -10: 8). For BP, on day 2 (CAR -10: 2), day 3 (CAR -10: 3) and day 7 (CAR -10: 7) positive significant returns occur at both the 5- and 10% level. These results are contrary to what one would expect. Hyndman (2008) finds cumulative abnormal returns to around -3,5% where Lin and Tamvakis (2010) even find cumulative abnormal returns that in some cases reach values around -10%. However the results are based on the crude oil returns rather than oil firms’ stock prices.
For Premier Oil, significant negative returns in the prior event window occur in days -7 (CAR -10: -7) at a 5% level and on day -5 (CAR -10: -5) at a 10% level. Then, as can be seen in figure 5, the CARS are declining from day 2 (CAR -10: 2) on, becoming significant on day
28 4 (CAR -10:4) until the end of the event window (Table A4 of the appendix). For DNO international, one day prior to the event there are negative cumulative returns that are significant at a 5% level. Then, as can be seen in table A5 of the appendix, the values stay significant until day 4 (CAR -10: 4). In addition, day 6 (CAR -10: 6) and day 10 (CAR -10: 10) show significant negative returns at the 5- and 1% level respectively. Figure 5 illustrates the drop in the end of the event period.
Having seen the aggregate and separate CARS (figure 2 and table 3) of the OPEC announcements and their significance, I reject the second and third hypothesis. That is, I reject the hypothesis that OPEC announcements have no effect on the stock prices of oil producing firms, and that there is no asymmetry in the effect of different announcement outcomes on the stock price of oil firms respectively. In addition, looking at figure 2, 3 and 4 and tables A2, A3, A4 and A5 of the appendix I also reject the fourth and last hypothesis namely that there is no difference in oil firms’ stock reactions on OPEC announcements based on market capitalization. Moreover, the differences in reaction between Shell and BP on the one side and Premier Oil and DNO International on the other side address an interesting issue. As Shell and BP show very similar market reactions in all cases (see figure 3, 4 and 5), it seems that the market does not incorporate any more firm specific factors with respect to these firms and the different OPEC announcements. However, looking at the same figures we see that the returns of Premier Oil and DNO International move in separate ways. While these firms don’t have the exact same market capitalization, compared to Shell and BP they can both be categorized as ‘small’. The differences in their reaction imply that there are more firm specific issues to be taken into account.
VI Discussion/Conclusion/Recommendations
29 Table 5: Overview of the results on the four hypotheses tested in this study.
Hypotheses decision Quota Rejected yes/no significance Highest
level Sign positive/negative In line with literature yes/no OPEC announcements do not have an effect on
the spot price of the crude oil commodity. Cut YES 5% Negative NO
Increase NO - - YES
Unchanged YES 5% Negative YES
OPEC announcements have no effect on the
stock prices of oil producing firms. Cut YES 5% Positive YES
Increase YES 5% Positive NO
Unchanged YES 1% Negative YES
There is no asymmetry in the effect of different
announcements on the stock price of oil firms. Cut YES - - -
Increase YES - - -
Unchanged YES - - -
There is no asymmetry in the effect of different announcements on the stock price of different oil firms, based on the firms’ market
capitalization.
Cut YES - - -
Increase YES - - -
Unchanged YES - - -
• Note: For the third and fourth hypothesis, all the different quota announcements yield different results
for the four firms. The specific significance levels and the sign of the returns can be found in tables A2, A3, A4 and A5 of the appendix.
oil-31 producing firms are less consumers and more producers of oil, and a general increase in the level of crude oil prices would let oil’ producing firms benefit from higher selling prices. Firms from other industries obviously only use oil as an input. In addition, the argument Hyndman (2008) uses to explain the positive abnormal returns associated with a quota cut for the crude oil commodity, namely that the market does not fully anticipates a cut, also makes sense for the oil firms. Since the oil industry is better off with a cut in bad times, the (stock) market positively reacts on a quota cut, leading to positive stock returns.
For the quota increase announcement, I find some significant returns, albeit less than for the quota cut announcement. Hyndman (2008) describes that when economic times are good, OPEC always increases the quota. Therefore, the market already anticipates an increase decision and no abnormal returns occur. In light of the reasoning throughout this paper, a reasonable explanation for positive abnormal returns to occur is that OPEC, in the eyes of the market, has recently not increased the quota when economic times were good. Then, the increase announcement is unexpected for the market. This is already the case when OPEC does not increase the quota only once in a good economic period, as the market can never fully trust its own judgment anymore. In addition, information asymmetry between OPEC and the market might force OPEC to not increase the quota, even though times are good and the market expects an increase. For the unchanged quota, negative significant returns occur with the same reasoning as for the crude oil commodity.
Existing literature has focused primarily on the relation between OPEC announcements and crude oil returns. However a gap exists when it comes to OPEC announcements and oil stock returns. This paper presents new evidence to fill this gap; however the results imply that more work needs to be done. Firm specific characteristics, like size in this study, seem to play a role when oil stocks react to oil supply shocks.
32 insight in the European dependency on oil and the influence on Europe’s economy as a whole. Therefore further research should be done to investigate the inflcuence of oil supply shocks on firms. This research should control for firm specific characteristics, like firm activities, market to book values, geographical dispersion etcetera. In addition, the use of oil supply shocks other than OPEC announcements, like the discovery of new oil fields, could develop a broader view of the relation between oil supply shocks and stock performance.
33 Literature
Bina, C., Vo, M., (2007), “OPEC in the epoch of globalization: an event study of global oil prices,” Global Economy Journal 7, 1, 1-49.
Boyer, M.M., Fillion, D., (2007), “Common and fundamental factors in stock returns of anadian oil and gas companies,” Energy Economics, 29, 428-453.
Bredin, D. and Muckley, C. (2011) “The Price Forming Process in Energy Markets” in
Financial aspects in Energy. Ed. A. Dorsman, W. Westerman, M. Bara Karan & O. Arslan,
Springer-Verlag, Berlin Heidelberg, pp. 85-107.
Brown, S.J., Warner, J.B., (1980), “Measuring security price performance,” Journal of
Financial Economics, 8, 205-258.
Brown, S.J., Warner, J.B., (1985), “Using daily stock returns, the case of event studies,”
Journal of Financial Economics, 14, 3-31.
Buyuksahin, B., Brunetti, C., Robe, M.A., Soneson, K.R., (2010), “Do OPEC members know something the market doesn’t? ‘Fair price’ pronouncements and the market price of crude oil,” working paper http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1652834
Deaves, R., Krinsky, I., (1992), “The behaviour of oil futures returns around OPEC conferences,” The Journal of Futures Markets, 12, 563-574.
Demirer, R., Kutan, A.M., (2010), “The behavior of crude oil spot and futures prices around OPEC and SPR announcements: an event study perspective,” Energy Economics, 32, 1467-1476.
Gilbert, R.J, Mork, K.A., (1984), “Will oil markets tighten again? A survey of policies to manage possible oil supply disruption,” Journal of Policy Modelin, 6, 111-142.
34 Hamilton, J.D., (1983), “Oil and the macro economy since world war II,” Journal of Political
Economy, 91, 228-248.
Hammoudeh. S., Li, H., (2005), “Oil sensitivity and systematic risk in oil- sensitive stock indices,” Journal of Economics and Business, 57, 1-21.
Huang, R.D., Masulis, R.W., Stoll, H.R., (1996), “Energy shocks and financial markets,” The
Journal of Futures Markets, 16, 1-27.
Hyndman, K., (2008), “Disagreement in bargaining: an empirical analysis of OPEC,”
International Journal of Industrial Organization, 26, 811-828.
Jones, C.M., Kaul, G., (1996), “Oil and the stock markets,” The Journal of Finance, 51, 463-491.
Lin S.X., Tamvakis, M., (2009), “OPEC announcements and their effects on crude oil prices,”
Energy Policy, 38, 1010-1016.
MacKinlay, A.C. , (1997), “Event studies in economics and finance, Journal of Economic
Literature, 35, 13-39.
Mazraati, M., Tayyebi Jazayeri, S.M., (2004), “Oil price movements and production agreements,” OPEC Review 28, 207-226.
Nanda M., Faff, R., (2008), “Does oil move equity prices? A global view,” Energy Economics, 30, 987-997.
Sadorsky, P., (2001), “Risk factors in stock returns of canadian oil and gas companies,”
Energy Economics, 23, 17-28.
Schmidbauer, H., Rosch, A., (2012), “OPEC news announcements: effects on oil price expectation and volatility,” Energy Economics, 5, 1656-1663.
Simpson, J.L., (2008), “The effect of OPEC production allocations on oil prices,” 21st
Australian Finance and Banking Conference 2008 Paper,
35 Zweig, D., Jianhai, B., (2005), “China’s global hunt for energy,” Foreign Affairs, 84, 25-38. Dorsman, A., Simpson, J.L., Westerman, W., 2012, Energy economics and financial markets, Springer-Verlag, Berlin Heidelberg.
36 Appendix
Table A1: List of the 4 oil companies used in this study.
Company Country Market Cap. (EUR bn) Primary business Main upstream activity
Royal Dutch Shell NL 163,52 Upstream and Downstream North- and South America BP GB 105,07 Upstream and Downstream Angola, North Sea, Gulf of Mexico, Azerbaijan Premier Oil GB 0,11 Upstream North Sea, South East Asia, the Middle East, Africa,
Pakistan, Falkland Islands
DNO International NOR 1,31 Upstream Middel East
Table A2: Student’s t- and Wilcoxon values around the event window for Shell, mean adjusted method.
Quota cut Quota increase Quota unchanged
Mean-adjusted adjusted Market Wilcoxon value adjusted Mean- adjusted Market Wilcoxon value adjusted Mean- adjusted Market Wilcoxon value CAR-10 -0,729 -0,676 0,026 -0,404 0,310 0,431 -0,059 0,238 0,228 CAR-9 0,010 -0,337 0,026 0,294 0,751 0,196 -0,267 0,469 0,167 CAR-8 0,237 -0,450 0,491 0,215 0,470 0,275 -0,251 -0,065 0,411 CAR-7 0,672 0,336 0,388 0,474 -0,219 0,196 -0,659 -1,430 0,228 CAR-6 0,306 0,090 0,284 2,255* 1,449 1,137 0,227 -0,980 0,532 CAR-5 1,284 1,285 0,957 1,771 0,665 0,588 0,358 -1,270 0,441 CAR-4 2,043* 2,332** 1,370 1,234 0,091 0,745 0,040 -1,551 0,350 CAR-3 1,650 2,435** 0,491 1,840* 1,090 0,902 0,197 -1,272 0,745 CAR-2 0,831 1,720 0,233 1,152 0,992 0,510 0,301 -1,147 0,654 CAR-1 1,926* 2,660** 1,060 1,042 0,362 0,745 1,178 0,387 1,171 CAR 0 0,595 1,234 0,388 1,647 1,076 0,902 1,636 0,430 1,597 CAR1 0,321 0,753 0,284 2,911** 3,257*** 1,137 1,887 0,031 1,658* CAR2 1,616 0,930 1,008 3,223*** 3,021** 1,059 1,924 -0,410 1,840* CAR3 2,187** 1,622 0,698 1,853* 1,884* 0,902 1,687 0,264 1,475 CAR4 1,525 0,298 0,388 1,780 2,417** 0,902 1,215 -0,512 1,293 CAR5 1,912* 0,588 0,440 1,632 2,370** 0,745 1,552 0,353 1,293 CAR6 2,195** 0,013 0,646 1,658 2,474** 0,510 1,475 0,257 1,110 CAR7 2,838** 0,041 0,801 1,117 1,794 0,275 2,420** 1,390 1,506 CAR8 2,669** -0,237 1,215 -0,082 0,563 -0,039 1,979* 1,345 1,445 CAR9 2,353** -0,219 1,215 0,887 1,851* 0,118 2,499** 1,707 1,506 CAR10 2,580** 0,139 1,215 -0,570 0,419 0,039 2,251** 1,444 1,688
37 Table A3: Student’s t- and Wilcoxon values around the event window for BP, mean adjusted method.
Quota cut Quota increase Quota unchanged
adjusted Mean- adjusted Market Wilcoxon value adjusted Mean- adjusted Market Wilcoxon value adjusted Mean- adjusted Market Wilcoxon value CAR-10 -0,690 -0,548 0,905 -0,077 0,339 0,353 -0,584 -0,568 1,475 CAR-9 0,276 0,150 0,026 0,349 0,437 -0,039 -0,837 -0,359 0,624 CAR-8 0,667 0,327 0,078 0,222 0,101 0,039 -0,501 -0,334 0,106 CAR-7 0,594 0,395 0,801 0,047 -0,805 0,196 -0,985 -1,789* 0,411 CAR-6 -0,083 -0,235 -0,026 0,544 -0,702 0,510 0,106 -0,958 0,380 CAR-5 1,042 1,141 0,646 0,507 -0,945 0,667 0,108 -1,417 0,289 CAR-4 2,414** 3,023*** 1,370 -0,092 -1,609 0,039 0,087 -1,169 0,380 CAR-3 2,051* 3,189*** 1,008 0,219 -1,210 0,039 0,049 -1,158 0,198 CAR-2 1,143 2,423** 0,801 0,607 -0,414 0,275 0,078 -1,108 0,259 CAR-1 1,271 2,017* 0,646 0,696 -0,718 0,353 0,769 0,182 0,806 CAR 0 0,663 1,717* 0,491 1,146 -0,350 0,510 1,516 0,741 1,232 CAR1 0,146 0,971 0,026 1,551 0,378 0,588 1,554 0,034 1,232 CAR2 1,462 1,251 0,595 2,703** 1,119 0,902 2,331** 0,839 1,749* CAR3 1,881* 1,735 0,440 2,664** 1,341 1,137 2,068* 1,529 1,445 CAR4 1,510 0,951 0,388 1,639 0,561 0,745 1,569 0,748 1,201 CAR5 1,574 0,821 0,233 1,773 0,701 0,981 1,625 1,210 1,262 CAR6 2,043* 0,607 0,595 1,611 0,487 0,667 0,973 0,223 0,897 CAR7 2,181** -0,004 0,284 0,842 -0,420 0,196 1,906* 1,390 1,597 CAR8 2,865** 0,968 1,215 0,291 -0,931 -0,039 1,259 1,058 1,262 CAR9 2,468** 0,903 1,112 0,013 -1,256 0,196 1,320 0,719 1,171 CAR10 3,028*** 1,747 1,370 -1,628 -2,846** 0,824 0,968 0,331 0,806
38 Table A4: Student’s t- and Wilcoxon values around the event window for Premier Oil, mean adjusted method.
Quota cut Quota increase Quota unchanged
adjusted Mean- adjusted Market Wilcoxon value adjusted Mean- adjusted Market Wilcoxon value adjusted Mean- adjusted Market Wilcoxon value CAR-10 -1,184 -0,964 0,957 -0,241 -0,056 0,353 -0,409 -0,259 0,928 CAR-9 -1,858* -2,118* 1,163 0,039 0,103 -0,039 -1,342 -1,101 1,262 CAR-8 -2,045* -2,522** 1,474 0,119 0,111 0,039 -1,551 -1,559 1,810* CAR-7 -2,512** -2,854** 0,646 0,543 0,259 0,196 -2,119** -2,577** 2,114** CAR-6 -2,996*** -3,100*** 1,060 0,618 0,161 0,510 -1,703 -2,415** 1,627 CAR-5 -3,086*** -3,269*** 0,336 0,977 0,475 0,667 -1,737* -2,608** 1,232 CAR-4 -2,243** -2,269** 0,284 1,671 1,234 0,039 -0,579 -0,812 0,289 CAR-3 -2,295** -1,774* 0,491 1,969* 1,606 0,039 -0,456 -0,518 0,319 CAR-2 -2,808** -1,919* 0,388 1,615 1,410 0,275 -1,364 -1,646 0,715 CAR-1 -2,165** -1,543 0,491 1,726 1,363 0,353 -1,088 -1,098 0,137 CAR 0 -2,035* -0,929 0,284 2,029* 1,658 0,510 -0,990 -1,215 0,350 CAR1 -3,440*** -2,457** 0,957 1,876* 1,643 0,588 -1,081 -1,643 0,076 CAR2 -4,705*** -4,930*** 1,629 2,421** 2,035* 0,902 -0,330 -0,780 0,228 CAR3 -4,266*** -4,435*** 0,233 3,082** 2,892** 1,137 -1,570 -1,776* 1,141 CAR4 -2,039* -2,045* 1,939* 2,766** 2,700** 0,745 -2,351** -2,718** 1,384 CAR5 -1,945* -2,025* 0,078 2,553** 2,481** 0,981 -2,545** -2,694** 1,688* CAR6 -1,506 -2,081* 0,388 2,991** 2,955** 0,667 -3,025*** -3,209*** 2,144** CAR7 -1,785* -2,901** 0,905 2,692** 2,599** 0,196 -2,657** -2,806** 1,962** CAR8 -0,732 -1,630 0,595 2,449** 2,380** -0,039 -3,556*** -3,548*** 2,722*** CAR9 -1,320 -1,868* 0,491 1,998* 1,886* 0,196 -3,311*** -3,353*** 2,266** CAR10 -1,059 -1,457 1,215 2,068* 2,043* 0,824 -3,169*** -2,975*** 2,266**
