The production policy of OPEC:

(1)

___________________________________________________________________________________

The production policy of OPEC:

An empirical analysis of the influence of OPEC on the European indices

and oil price

Author:

A.E.N. de Voogd

University of Groningen

Faculty of Economics & Management and Organization Business Administration, MSc Finance

Supervisor:

Prof. Dr. R.A.H. van der Meer

(2)

___________________________________________________________________________________

A.E.N. de Voogd University of Groningen 2

The production policy of OPEC:

An empirical analysis of the influence of OPEC on the European indices

and oil price

Abstract

The opinions about the influence of OPEC on the stock- and spot markets differ. Several researchers investigated the influence of OPEC on the U.K. and the U.S. stock markets. The relation however, has never been analyzed on a European level before. Therefore, in this research the Datastream European index will be investigated. This is tested by means of an event study methodology. The models used in this research are the mean adjusted model and the market and risk adjusted model. Moreover, different time periods in the event study analyzing OPEC have not been analyzed. Thus, the time periods 1986-2006, 1986-1996, and 1997-2006 will be analyzed. The period 1986-1996 is characterized by a steady price development of oil, whereas the period 1997-2006 is one of large increases in the oil price. Furthermore, the chemical industry and the oil and gas industry are investigated. The chemical industry is chosen because of the large share of oil in its total cost. Hence, the oil and gas industry is discussed due to its dependence on the energy prices. Periods of conflict and non-conflict are also analyzed because little is known about the influence of OPEC during these periods on the stock-and spot markets. This study consists of 56 samples for Europe and its industries, and 51 for the spot price Brent. The influence of OPEC is limited to Europe, the spot price Brent, and the oil and gas industry. OPEC has significant influence on the chemical industry during the first interval period. This applies to the market and risk adjusted model during the full period and periods of non-conflict.

A.E. . de Voogd

A.E. .de.Voogd@student.rug.nl Student number: 1525069

(3)

___________________________________________________________________________________

PREFACE

Throughout the last year I have been studying the MSc Finance at the University of Groningen. During the last semester of my master I have been writing my master thesis. The aim of this thesis is to highlight the influence of OPEC on the European indices and oil price. Handing in my thesis is a special moment since it means the ending of an era, as a student. Now a new era starts and hopefully this means being active in the energy market.

In the process of writing my thesis several people supported me. I would like to thank my supervisor, Prof. Dr. R.A.H. van der Meer. He gave me good advice about the energy market. Furthermore, I would like to thank Pim Breukelman, my sister’s boyfriend, who works for Nuon. He gave me beneficial insights in the energy market. Last but not least, I would like to thank my family and friends for their support.

(4)

___________________________________________________________________________________

TABLE OF CO TE TS 1 I TRODUCTIO ... 6

2 LITERATURE REVIEW ... 8

2.1 The rise of OPEC ... 8

2.2 OPEC and crude oil ... 8

2.2.1 OPEC from 1986-2006 ... 9

2.2.2 Empirical evidence in time ... 10

2.3 Production announcements ... 11

2.4 Conflict and non-conflict periods ... 12

2.5 Industrial effects ... 13

3 DATA A D METHODOLOGY ... 15

3.1 Construction of the sample ... 15

3.2 Normality test ... 16

3.3 Structure of an event study ... 17

3.3.1 The event day ... 17

3.3.2 The event window ... 17

3.3.3 The estimation window ... 17

3.4 Methodology of an event study ... 18

3.4.1 Mean adjusted returns ... 19

3.4.2 Market and risk adjusted returns ... 20

3.5 Testing ... .21

3.5.1 Parametric test ... 22

3.5.2 Non-parametric test ... 22

4 EMPERICAL RESULTS A D DISCUSSIO ... 24

4.1 Descriptive statistics ... 24

4.1.1 Europe and Brent ... 24

4.1.2 Industrial effects ... 25

4.2 Production quota announcements ... 25

4.2.1 Mean adjusted model ... 25

(5)

___________________________________________________________________________________

4.3 Conflict and non-conflict periods ... 29

4.3.2 Market and risk adjusted model ... 30

4.4 Industrial effects production quota announcements ... 31

4.5 Industrial effects conflict and non-conflict periods ... 34

4.6 Wilcoxon signed rank test ... 36

4.7 Discussion and interpretation ... 37

5 CO CLUSIO ... 40

5.1 Limitations of methodology ... 41

5.2 Recommendations for further research... 41

REFERE CES ... 42

WEBSITES ... 44

LIST OF TABLES ... 45

APPE DICES ... 46

APPENDIX 1 OVERVIEW LITERATURE... 47

APPENDIX 2 DEVELOPMENT SPOT PRICE BRENT ... 48

APPENDIX 3 OVERVIEW EVENTS ... 49

APPENDIX 4 DS-WORLD AND DS-EUROPE ... 51

APPENDIX 5 S&P GSCI ENERGY INDEX ... 52

(6)

___________________________________________________________________________________

1 I TRODUCTIO

At the moment the world demand for crude oil is 83 million barrels per day, and this will further increase to 97 million barrels per day in 2015.1 The content of one barrel is 42 gallons (159.02 litres). The world demand for oil is currently 3.486.000.000 gallons per day. The high crude oil prices since 1999 partly contributed to the economic downturn in 2000-2001. Fears of supply cuts from the Organization of the Petroleum Exporting Countries (OPEC), political tensions in Venezuela, and tight shocks have driven up international crude oil prices even further (IEA, 2004). As the IEA states OPEC partly contributes to the economic downturn in 1999.

OPEC is an organization that consists of twelve countries which are dependent on oil revenues. The main task of the OPEC is to stabilize the oil market and help oil producers to achieve a reasonable rate of return on their investments. OPEC owns 75% of the proven crude oil reserves in the world (OPEC statistic bulletin, 2005); this is approximately 900 billion crude oil barrels of the total 1200 billion crude oil barrels in the world. The OPEC Member Countries produce approximately 42.8% of the world’s crude oil and for the internationally traded oil this figure is 51%. Representatives of OPEC Member Countries (Heads of Delegation) meet at the OPEC Conference. They hold between two and four scheduled meetings a year, at times it announces extraordinary meetings when it believes market conditions are particularly uncertain. The Conference is the supreme authority of the organization and consists of the members and the Ministers of Oil, Energy, and Mines. The meeting operates at the principle of one member one vote.

Loderer (1984) states that the announcement of OPEC policy decisions affects oil prices during 1981-1983. Thus, what is the influence of an OPEC production quota announcement on the stock markets when it decreases the production of oil? Hyndman (2005) states that when OPEC reduces the production quota it produces significant positive cumulative returns for the sub indices which he analyzes. Guidi et al (2006) examine U.K. and U.S. stock markets. They find positive returns for U.S. stock market and negative returns for the U.K. stock market when OPEC announces to reduce production. Jones and Kaul (1996) explain that stock markets rationally reflect news concerning oil shocks.

In this research, there will be investigated if a production quota announcement of OPEC leads to abnormal returns on the European indices and the spot price Brent . The European countries are highly dependent on crude oil; they consume and import respectively 14.7 million and 12.8 million barrels a year. When OPEC announces to decrease the production of oil one can think this has consequences on the European indices and the spot price Brent.

1

(7)

___________________________________________________________________________________

Previous studies focused on the influence of OPEC on the stock returns of energy companies and between the U.S. and U.K. stock markets. This study will broaden the scope on a European index level and different industries during different time periods. The research question is defined in the following way:

” Is there in Europe a relationship between the production policy of OPEC and the indices and oil price at different time periods during 1986-2006? ”

An event study methodology will test the above described research question. Moreover, this research question will be tested with six hypotheses explained in chapter 2. The methodology will be tested as described by Brown and Warner (1980 and 1985), and MacKinlay (1997). The event study methodology has the advantage that if the markets are rational a stock price fully reflects the occurrence of the event. This means that the influence of an unexpected particular event, in this research a production quota announcement from OPEC, can be measured on the indices and the spot Brent (MacKinlay,1997). The data for this research will be gathered from Datastream. Daily returns of indices and spot prices will be investigated. The research covers the period from January 1986 until December 2006. The investigated time periods are January 1986 until December 1996, and January 1997 until December 2006.

This paper has several novel contributions. Firstly, this study contributes to the existing literature because little is known of the influence of OPEC on different industries. Guidi et al (2006) state in their article this is an interesting subject to analyze. Secondly, the different time periods investigated in this research haven’t been subject to investigation in other event studies. Finally, this study broadens the scope by analyzing Europe rather than one European country.

This research continues as follows. In the next section a review on the literature of OPEC will be presented. The influence of OPEC in time will be discussed due to the different time periods investigated and the theoretical framework will be explained. Chapter 3 describes the data and methodology. The construction of the sample and the different models used in this research are thoroughly explained. The empirical results will be presented and compared with previous research in chapter 4. Finally, in chapter 5 the conclusions are summarized. Furthermore, limitations of the event study methodology and recommendations for further research will be elaborated.

(8)

___________________________________________________________________________________

2 LITERATURE REVIEW

In this chapter the literature about OPEC will be discussed. In the first paragraph the rise of OPEC will be discussed, what led to the establishment of this organization. In the second paragraph a thorough analysis of important papers will be given which give a good indication of the impact of OPEC on the stock- and spot markets during different time periods. In the third and fourth paragraph the theoretical models of Hyndman (2004) and Guidi et al (2006) will be introduced which are the basis of this research. The fifth paragraph gives a description of the industries which will be investigated. In appendix 1 a summarized overview of the outcomes of the different papers discussed in this chapter are given.

2.1 The rise of OPEC

Before 1960 the oil industry was controlled by seven oil companies, also known as the ”Seven Sisters”. Falola et al (2005) describe that the producing countries influence was restricted from having a say in the price of their petroleum products. A free market for crude oil did not exist at that time. Chalabi (1997) states that the ”Seven Sisters” used to ”post” a price for oil. This means that it is not an actual price for oil but rather a point of reference for paying taxes and royalties to host governments. The income per barrel was a fixed proportion of the price of crude oil. At the end of the fifties new U.S. and Western oil companies tried to obtain licenses in other producing countries. This because the oil-producing countries had more fiscal advantageous regimes for these companies. A consequence was that the new oil companies were able to reduce the price. For the ”Seven Sisters” this threatened their competitive position. Thus, they lowered the price of oil, but this resulted in lower revenues to the governments due to the lower income of taxes. Lowering the price of oil was strategically not a good decision of the ”Seven Sisters”. Representatives of Iraq, Iran, Kuwait, Saudi Arabia, and Venezuela gathered and founded OPEC in 1960. Chalabi (1997) mentions that the OPEC was in it early years a defensive instrument used by the producing countries to stabilize the market. Formerly the oil companies were in charge but this changed. The main reason for this development was the tight supply-demand conditions. When the negotiations in October 1973 failed OPEC took control.

2.2 OPEC and crude oil

This paragraph describes the role of several events on the OPEC and crude oil from 1986-2006. This in order to give an indication for the reader what the influence was of OPEC on the price of crude oil in

(9)

___________________________________________________________________________________

the time period 1986-2006. Moreover, the empirical results of previous research is described in the second sub paragraph.

2.2.1 OPEC from 1986-2006

The war between Iran and Iraq removed 4 million barrels per day off the market. To give an indication of this amount in 1980, this was 8% of the world demand. The oil shock did not last long due to the increasing production of oil by non-OPEC countries. As a result the market share of OPEC diminished to 30% and their revenues declined. In 1985 the price of oil dropped to $ 12 a barrel. Despite cuts in the production of oil the price remained low. During this period the debate started if OPEC was still able to control the price of oil.

The Persian Gulf War started in the beginning of 1990 when Iraq invaded its fellow OPEC Member Kuwait. Crude prices increased from $ 15 to $ 30 a barrel as the markets reacted to this news. The oil price did not remain long on this level because Saudi Arabia decided to increase the production of oil. The price of oil remained stable until the end of 1997.

When OPEC ministers met in December 1997 they decided to increase the output ceiling. They thought that the demand for oil would increase during the following years. The production increase was already above the ceiling and this decision was again strategically not an optimal decision. What happened was an oversupply of oil on the markets. This oversupply was caused by a decrease in demand from Asian countries due to the currency devaluations. Kohl (2002) mentions furthermore that an exceptionally warm winter diminished the demand for oil by about one million barrels a day. As a consequence the price of oil collapsed. Thus, the revenues of the OPEC countries diminished.

At the beginning of 1999 OPEC countries and several non-OPEC Members complied to undertake action. These countries agreed to lower the production of oil in order to increase the price of oil. At the end of the year 2000 the price of a barrel was $ 35. The price of oil increased rapidly and OPEC did not do anything for a steady price development. Although several non-OPEC countries decided to increase the production of oil the oil price remained at a high level. This because OPEC announced production decreases at the same time. In the following years the price of crude oil further increased due to several developments. First of all, oil demand rose contrary to the projections of OPEC which predicted a stable demand for oil. Secondly, market disruptions caused an upward pressure on the price of oil. The Venezuelan oil strike and the war in Iraq caused this upward pressure. Finally, excess capacity diminished from 6 million per day in 2002 to 1 million during 2004. According to Kohl (2005) OPEC lost control of oil prices in 2004 because the OPEC basket price remained above $ 40. This was caused by a reduction in oil production despite an increased demand, the increased instability of Iraq, and low inventories of oil. OPEC announced at the end of 2004 to increase production but this was not sufficient to reduce the price of oil. Furthermore, postponed decisions about production in oil have

(10)

___________________________________________________________________________________

destabilized the market and its credibility. In the following years until the present time the prices of oil reached record levels due to the Iraq war and foremost the increased demand for oil despite increased production of the OPEC. For an overview of the spot price development Brent see appendix 2.

2.2.2 Empirical evidence in time

Loderer (1985) examines in his article whether the high oil prices are the result of producer collusion. He analyzes the influence of the OPEC after they announce to change their policy. He observes the stock prices in the energy sector, the spot price, and the future price. The first period he analyzes is from 1974 until 1980. The second covered period is 1981-1983. The reason for the two investigated periods is because different patterns in the oil prices can be observed. He uses weekly and daily stock returns from the tapes of the Centre for Research in Security Prices at the University of Chicago for the 1974-1980 period. For the period 1981-1983 he uses weekly settlement prices for Heating Oil No.2 Futures traded on the New York Mercantile Exchange reported in the Wall Street Journal. The OPEC meetings are defined as week 0. The first investigated period does not create significant abnormal returns on the energy sector, spot price, and the future price. The influence of OPEC seems therefore limited. Even when excluding the small changes in the policy of OPEC this does not lead to significant abnormal returns. For the period 1981-1983 he differentiates between different meetings due to the introduction of the production quota in 1982. He distinguishes between good news, bad news, and neutral meetings. Good news meetings increase prices, bad news meetings decrease prices, and neutral meetings have no influence on prices. Thus, the period 1974-1980 with high oil prices indicate that OPEC did not influence the prices and is therefore not an effective cartel. The period 1981-1983 with softening oil prices indicate that OPEC did not influenced the prices. Hence, OPEC is an effective cartel during the second period.

Gülen (1996) examines whether a change in the policy of OPEC creates a change in the price of oil. The period of investigation is from January 1965 till February 1992. The Oil and Gas Journal Energy Database is the source for the West Texas Intermediate (WTI) crude oil price. Four different time periods are investigated in the paper. The first period is the full sample, the second from January 1965 till September 1973, the third from February 1974 till February 1992, and the fourth from January 1982 till February 1992. During the third period a change in the policy of OPEC creates a significant change in the price of oil. The other periods investigated did not influence the price of oil.

Kaufmann et al (2004) investigate if the influence of OPEC has diminished. This covers the period October 1986 until October 2000. In order to test this they created a model for real prices from quarterly data. This model consists of the OPEC production quota (quota), the difference between crude oil production and the production quotas (cheat), capacity utilization by OPEC, and war as a

(11)

___________________________________________________________________________________

dummy variable for the Persian Gulf War. OPEC was able to influence the real prices by means of production quotas, the extent to which OPEC exceeds these quotas, and capacity utilization.

Bina et al (2007) analyze if OPEC’s output decisions create significant oil prices. Furthermore, they examine how OPEC reacts to changes in oil prices. They study the impact of the OPEC decisions on the daily future and spot markets on the NYMEX. They also examine the spot price Brent Crude Oil traded on the International Petroleum Exchange (IPE) in London. The investigated period is from April 1983 until July 2005. Bina et al (2007) determine a new test in order to overcome the drawbacks in the study of Loderer (1985). This drawback was during the years of steady prices in comparison to high prices. Steady prices cover only three years, whereas the high prices are investigated during seven years. Furthermore, they emphasise there is strong evidence of heteroskedasticity in the oil price time-series. Biased conclusions can be drawn if this is the case. OPEC is not able to influence the price of oil during the period they investigated. When OPEC observes a decrease in the price of oil they decrease their output. An increase in the price of oil is accompanied with an increase in the production of OPEC.

2.3 Production announcements

Hyndman (2004) studies the effects of OPEC decision making behaviour and its effects on the energy sector and oil prices from January 1986 until September 2002. The oil data consists of the two-month forward price and the spot price of the West Texas Intermediate Crude Oil. The sample consists of 50 announcements. Three types of announcements are defined in his study, which are quota reductions, quota expansions, and status quo announcements. In the methodology a market and risk adjusted model is used to determine the influence of OPEC on the market. The stock market data contains indices for integrated companies, exploration & drilling companies as well as oil companies as a whole. The Standard and Poors 500 (S&P) is defined as market. Thus, abnormal returns arise if the indices significant out-or underperform the S&P 500. Hyndman (2004) uses the standard method of Campbell et al (1997) in order to calculate the normal returns and abnormal returns. An event window is chosen of 41 days and an estimation window is chosen of 150 days.

When OPEC takes no action (status quo) negative cumulative abnormal returns arise on the indices, the two-month forward price of oil, and the spot price of oil. Negative cumulative abnormal returns occur in the days before the production announcement of the OPEC. In the days after the announcement the returns become significant negative. The cumulative abnormal returns are significant at the 5% confidence level (one-sided test). Twenty days after the announcement the returns underperform the market between -2.5% and 3.5%. When OPEC reduces the quota the results are weaker for the indices, two-month forward price of oil, and the spot price of oil. Although the results are weaker, at the end of the event window they become significant. An interesting pattern in the

(12)

___________________________________________________________________________________

cumulative returns can be observed. The cumulative returns are positive until the third day before the announcement. After the third day the returns drop and significant negative cumulative returns arise. The returns are significant positive towards the end of the event window. Hyndman (2004) suggests that the market is hopeful of an agreement and positive returns arise. Due to the uncertainty of the outcome during the meeting the prices decline significantly, positive effects arise in the days after the reduction announcement of OPEC. Finally, when OPEC increases its production there are no significant cumulative abnormal returns. In the last part of his study he excludes extraordinary meetings. The choice of this decision is because the market does not expect extraordinary meetings and these meetings could have substantial influence on the results presented. The empirical findings correspond to the results when the extraordinary meetings are not excluded. In this paper the following hypotheses are formulated following Hyndman (2004):

Hypothesis 1, 2 and 3: 2

H₁_{: After a production quota announcement of OPEC the European index and spot price Brent differ} significantly from zero during 1986-2006

H₂_{: After a production quota announcement of OPEC the European index and spot price Brent differ} significantly from zero during 1986-1996

H₃_{: After a production quota announcement of OPEC the European index and spot price Brent differ} significantly from zero during 1997-2006

2.4 Conflict and non-conflict periods

Guidi et al (2006) examine the effects of OPEC policy decisions on the UK and US stock markets and spot oil markets from 1 January 1986 until 31 December 2004. The stock market indices are the All Share Value-weighted index for the UK and for the US Dow Jones All Share Value-weighted index. The spot oil prices used in their paper are the OPEC Basket, Brent Crude, and North Sea Crude. The mean adjusted model is used to examine if OPEC production announcements create significant abnormal cumulative returns. The difference with their research compared to other papers is that Guidi et al (2006) investigate also periods of conflict and non-conflict. Three conflicting periods are distinguished by Guidi et al (2006):

1. Iran-Iraq conflict, January 1985 until July 1988

2. Iraq’s invasion of Kuwait, Augustus 1990 until February 1991 3. US invasion of Iraq, March 2003 until December 2004

2

(13)

___________________________________________________________________________________

Guidi et al (2006) investigate the full period and the different production quota announcements for the stock markets and the spot prices. Production decrease announcements create significant positive cumulative returns for the different spot prices. There is no significant reaction on the U.S. and U.K. stock markets. Increase announcements of OPEC do not produce significant cumulative returns on the stock markets and the spot prices.

The next step in their research is to divide the production announcements in periods of conflict and non-conflict. When OPEC reduces the production during conflict periods the US market reacts positive and the UK market reacts negative. None of the cumulative returns are statistically significant. The interpretation by them is that during conflict periods non-OPEC countries produce more oil to sustain the supply of the US. The spot prices produce significant positive cumulative returns. During non-conflict periods the US and UK markets produce negative cumulative returns, but the results are not significant. The different spot prices produce positive cumulative returns but none of the returns are statistically significant. When OPEC announces to increase the production of oil during conflict periods the US and UK markets react positive, but none of the results are significant. During non-conflict periods the US market reacts positive and the UK market reacts negative. An interpretation by Guidi et al (2006) is that oil companies have a large share in the index, albeit the markets do not produce significant results. The oil markets react negative during periods of conflict and non-conflict, but the results are not significant. The following hypothesis is formulated following Guidi et al (2006):

Hypothesis 4:

H₄_{: After a production quota announcement of OPEC the European index and spot price Brent differ} significantly from zero during periods of conflict and non-conflict

2.5 Industrial effects

Lee et al (2002) study the effect of oil price shocks on demand and supply in various industries. The period of interest is from January 1959 until December 1979 and January 1980 until September 1997. The choice of the two periods is because two oil price shocks occurred during 1973-1974 and from 1978 until 1981. To test the effect of oil price shocks on the industry-level output and price they use a VAR model with monthly data to test this. An oil-intensity indicator is used to measure the cost of oil for each dollar of sale. The chemical industry has the highest share of oil in its costs. These costs can be accounted for 18.1 cents to the chemical industry, of which 2.5 cents are direct costs. The metal industry is also energy intensive but coal is the main energy source. Yucel et al (1994) explain that coal is weakly correlated with the price of oil. The price of the chemical industry has significantly increased. Lee et al (2002) find economical explanation for their findings in trade magazines. The explanation for

(14)

___________________________________________________________________________________

the reduction in supply of the chemical industry is because of the low availability of energy. Furthermore, as mentioned in paragraph 2.3 Hyndman (2004) finds evidence that the energy sector is affected by production announcements of the OPEC. The following hypotheses are formulated following Lee et al (2002), Hyndman (2004), and Guidi et al (2006):

Hypothesis 5 and 6:

H₅_{: After a production quota announcement of OPEC the chemical industry and the oil and gas} industry differ significantly from zero during different time periods

H₆_{: After a production quota announcement of OPEC the chemical industry and the oil and gas} industry differ significantly from zero during periods of conflict and non-conflict

(15)

___________________________________________________________________________________

3 DATA A D METHODOLOGY

In this chapter the data and methodology will be discussed. The first paragraph describes how the sample is selected. The distribution of the data and the descriptive statistics to test this will be explained in the second paragraph. Thereafter, the structure of an event study will be thoroughly discussed. The mean adjusted model and the market and risk adjusted model will be explained in the fourth paragraph. The fifth paragraph describes respectively a parametric test and a non-parametric test.

3.1 Construction of the sample

The data of the indices and the spot price Brent will be gathered from Datastream. Datastream is a statistical data program which consists of stock, macroeconomic, and financial data. Daily returns will be gathered from January 1986 until December 2006. The Return Index (RI) is used to gather the indices and for the spot price Brent the Price Index (PI) is used.

The first step in conducting the sample for this research is searching for the announcements of the OPEC meetings starting from January 1986 until December 2006. In the statistical bulletin of OPEC an overview is given of the dates of the meetings.3 A total of 69 meetings were held from 1986 till 2006. The second step was to determine the announcement date of the production announcements because the OPEC meetings often last several days. The database used to search for the production announcements is the lexis-nexis newspaper database from the University of Groningen. For an overview of all events see appendix 3. An important aspect of the event study methodology is that the estimation period cannot overlap the event window. When other events overlap each other one can draw biased conclusions from the results. After checking this point of criteria for the 69 samples, 13 samples are excluded from the sample. A total of 56 samples are left of which 16 are production reductions, 19 increases, and 21 status quo.

The conflict periods are determined following Guidi et al (2006). There are three conflict periods defined in this research. For an overview of the conflict periods see Appendix 3. The first conflicting period is the Iran-Iraq war which covers the period January 1985 till July 1988. The second conflicting period is the Iraq invasion of Kuwait and is from August 1990 until February 1991. The third conflicting period is the U.S. invasion of Iraq from March 2003 until December 2006. There are 19 conflict periods and 37 non-conflict periods. Of the 19 conflict periods there are 5 production decreases, 7 increases, and 7 status quo. Of the 37 non-conflict periods there are 11 production decreases, 12 increases, and 14 status quo.

3

(16)

___________________________________________________________________________________

The third step is to conduct a list of industries. The criteria for selecting this list is described in paragraph 2.5. For the period of interest all the industries data are available. Thus, the sample still consists of 56 events.

The last and fourth step in the construction of the sample is to select the spot price Brent for the period of investigation. The Brent Crude oil covers only the period May 1987 until December 2006. Therefore, five events are excluded from the sample of which two are production decreases, two increases, and one status quo. When discussing the sample of conflict periods and non-conflict periods, 14 are conflict and 37 non-conflict.

3.2 *ormality test

To conduct different research methods on the gathered data we have to know if the data is normally distributed. Descriptive statistics will be used to make inferences about the distribution of the data. To test normality the Jarque-Bera (JB) test will be used. A normal distribution is not skewed and it has a value close to 0. The following hypothesis can be defined as:

H₀ = The data is normally distributed H₁_{= The data is non-normal distributed}

The skewness measures if the distribution is not symmetric about its mean value. The kurtosis measures the fatness of the tails from the distributions compared to the normal distribution (mesokurtic). A normal distribution will have a value close to three. A leptokurtic distribution is more peaked at the mean and has fatter tails compared to a normal distribution. A playkurtic distribution is less peaked in the mean and will have thinner tails. The JB test is specified as:

      − + = 4 ) 3 ( 2 2 K S S n JB (3.1)

The n is the total observations, S is the skewness, and K is the kurtosis. The calculated tests will be covering the estimation period of the two models which are used in this paper. The outcomes of the tests will be discussed in chapter four.

(17)

___________________________________________________________________________________

3.3 Structure of an event study

MacKinlay (1997) explains that the first step in an event study is to define the period of interest, so one has to determine a time window. This in order to analyze and measure the data of the event. The time window consists of three pieces of information which are the event day, event window, and the estimation window and will be discussed in the next paragraphs.

3.3.1 The event day

The first step as mentioned is defining the event day, day 0. The event day is the day when a certain event occurs for a firm or index. In this research day 0 is the day when the production quota is announced by the OPEC. Guidi et al (2006) and Hyndman (2004) choose the same definition for the event day in their studies.

3.3.2 The event window

In the event window the indices and spot price will be examined for a specific period. An event window is chosen due to the fact that it is not certain that the event day, day 0, is captured. In order to capture the event day several days surrounding the event are chosen. In the studies of Guidi et al (2006) and Bina et al (2007) an event window is used of eleven days from t = -5 until t = + 5. The same event window will be used in this research.

3.3.3 The estimation window

Until now the event day and the event window are defined. The period prior to the announcement of the event is also of interest. With the use of the estimation period the parameters of the normal performance model can be estimated. This in order to make a comparison of the normal returns from the estimation window to the abnormal returns during the event window. MacKinlay (1997) describes that it is common that the event window and the estimation window do not overlap. This in order to prevent that estimators are influenced by the returns around the event. The abnormal returns are therefore better measured. Hyndman (2005) uses in his study 150 days as estimation window and Guidi et al (2006) 100 days. Bina et al (2007) use 30 days as estimations window. In this research an estimation window will be chosen of 30 days. This number is chosen to limit the loss of events. The figure on the next page shows graphically the time window of the event study in this research.

(18)

___________________________________________________________________________________

Figure I Time window

3.4 Methodology of an event study

According to MacKinlay (1997) an event study measures the impact of a specific event on the value of a firm or index. In this research the event study methodology will be used to determine the influence of a production quota announcement on the indices and spot price Brent. The event study methodology takes the Efficient Market Hypothesis as given. Fama (1970) explains that a stock price fully reflects all available information. MacKinlay (1997) describes that one has to make an estimation of the impact of the event to measure the abnormal return. A return is abnormal only when one compares it with an certain benchmark. This benchmark can be the stock itself or a market. Thus, in order to calculate the abnormal returns one has to define a model generating normal returns. A normal return is calculated over the estimation period. This in order to capture the possible abnormal returns in the event window. Formula 3.2 is the return of index i on time t:

it it

it K e

R = + _(3.2)

The

K

_itis the normal return and the

e

_it is the abnormal return. This can be specified as:

it it it

=

K −

R

ε

(3.3)

This means that

ε

_it is the difference between the real and expected return. To calculate the normal returns the following models will be used as in Brown and Warner (1980):

1. Mean adjusted returns

2. Market and Risk adjusted returns

T 0 = -35 T 1 = -5 T 2 = + 5

Estimation window Event window

(19)

___________________________________________________________________________________

3.4.1 Mean adjusted returns

The first model is the mean adjusted returns model. Brown and Warner (1980,1985) state that the mean adjusted model is a simple model but works well compared to the more sophisticated model which will be discussed below. The normal returns for the mean adjusted return model are given as:

Rit = µi +

ε

it (3.4)

E (

ε

_it

=

0

) var (

ε

_it

=

σ

_it2)

The abnormal returns (

ε

_it) are described as:

it

ε

= Rit – µi (3.5)

Where Rit is the return of the DS-European Return Index at time t and µi is the average return of the index and

ε

_it is the abnormal return for the DS-European index at time t. The Brent Crude Oil as previously mentioned will be used as spot price. For the European industries the Chemical and DS-Oil and Gas indices are used.

Guidi et al (2006) use in their study the Dow Jones All-Share Value weighted index for the U.S. and the U.K. All-Share Value weighted index from Datastream. For the effect on the spot prices they use the OPEC basket, Brent Crude Oil and the North Sea Crude. Hyndman (2004) uses in his study the (WTI) to determine the impact on spot prices. Bina et al (2007) take the WTI and the Brent Crude Oil to investigate the impact of OPEC output decisions.

Due to the limited availability of indices the Datastream indices are chosen. Datastream Global Indices draw on Datastream’s coverage of currently 40 equity markets. A representative sample of stocks has been chosen for each of the markets. Using FTSE Actuaries classifications, the stocks are allocated into industries/sectors. Indices are calculated on a representative list of stocks for each market. The number of stocks for the market is determined by the size of the market capitalization. Datastream covers three Global Indices which are Regional and World indices, market indices, and sector indices. For an overview of the 40 included countries in the World, and the 18 included European countries see appendix 4.

(20)

___________________________________________________________________________________

3.4.2 Market and risk adjusted returns

This model assumes that the CAPM generates expected returns (Brown and Warner, 1985). The normal returns are given in the Black (1972) two-parameter Asset Pricing Model:

( )

R

it

E

( )

zt

[

E

( )

R

mt

E

( )

R

zt

]

E

=

+

β

−

(3.6)

In this model the

R

_it is the return on a minimum variance portfolio of assets with risk, which is not correlated with the market portfolio. The risk measure

β

reflects the covariance of the return on the index of the stock with the return of the market.

β

can be estimated by means of:

R

_it

=

α

_i

+

β

_i

R

_mt

+

ε

_t (3.7)

The ex post abnormal returns (

ε

_it) is the difference between the return on the stock/index and that on the market portfolio. This is specified as:

mt it it

=

R −

R

ε

(3.8) E (

ε

_it

=

0

) var (

ε

_it

=

σ

_it2)

where

R

_it is the return of previous mentioned indices and

R

_mt is the return on the market portfolio for day t. Here, the DS-World Market Return Index is the market return to measure the impact of a production quota announcement on the DS-European Market Return Index. To measure the impact of a production announcement on the industries the market return is the DS-European Market Return Index. The spot price market return is the S&P GSCI Energy Return Index for the spot price Brent.4

The NYMEX and IPE index are not available in Datastream and therefore the S&P GSCI Energy Return Index is used in this research as market return. The S&P GSCI Energy index consists of 6 components. For an overview of these components and their weights see appendix 5.

MacKinlay (1997) explains that the market and risk adjusted models has a potential improvement over the mean adjusted return model. The variance of the abnormal returns will diminish by removing that part of the return that is related to the variation in the market’s return. With the parameters of the

4

(21)

___________________________________________________________________________________

market and risk adjusted model,

α

_iand

β

_i, the abnormal returns can be measured and analyzed (MacKinlay, 1997). The alpha and beta are estimated values by means of Ordinary Least squares (OLS). The abnormal returns (

AR

_i_τ) in this model can be denoted as:

τ τ τ i

α

i

β

i m i

R

AR

=

−

(3.9) 3.5 Testing

It is important that abnormal returns are aggregated in the period of interest to draw conclusions. MacKinlay (1997) elaborates that this aggregation is through time ( t = -5 until t = +5) and across indices. This means that this is the sum of the abnormal returns between the event window and across observations of the event window. MacKinlay (1997) assumes there is no clustering. Clustering means there is overlap in the event windows of securities or indices. Otherwise biased conclusions can be drawn. The abnormal returns will be aggregated for each event period of the indices and the spot price. This is from τ, T1 until T2, for the events and is specified for period τ as:

∑

= = * i i AR * AR 1 1 τ τ (3.10)

and its variance is:

∑

=

* i i

*

AR

1 2 2

1 )

var(

_τ

σ

_ε (3.11)

The

AR

τ is the abnormal return at a given day for the different events. By means of these estimates the abnormal returns in the event period can be investigated. Cumulative returns will be analyzed to draw conclusions if the null hypothesis is significant or not. It captures the full effect of the event window by aggregating all returns. Which is in this research from day -5 until day + 5. Furthermore, to analyze where the possible effect of the production announcement of OPEC is the largest several interval periods are determined. For every interval in the event window, T1 until T2 counts the following cumulative abnormal returns:

(22)

___________________________________________________________________________________

∑

= = 2 1 ) , ( ₁ ₂ τ τ τ τ

τ

AR

CAR τ1 and τ2 are -1 +1; -3 + 3; -5 +5 (3.12)

with the following variability:

∑

= = 2 1 ) var( )) , ( var( ₁ ₂ τ τ τ τ

τ

AR CAR _(3.13)

In the next step we have to determine if the cumulative abnormal returns and abnormal returns can be approached with a normal distribution by analyzing the returns of the estimation period. As previous mentioned in paragraph 3.2 this will be done by means of the JB test. The results of this test will be discussed in the next chapter.

3.5.1 Parametric test

If the data is normally distributed a parametric test can be used to test the null hypotheses. A t-test will be used to determine if the cumulative returns differ significantly from zero. Furthermore, a second t-test will be conducted to measure the significance of the CAR at different interval periods as mentioned in formula 12. To draw conclusions about the cumulative abnormal returns the following null hypothesis is specified as:

))]

,

(

var(

,

0 [

~

)

,

(

t

₁

t

₂

*

CAR

t

₁

t

₂

CAR

(3.14)

MacKinlay (1997) states that an estimation has to be used to calculate the variance of the abnormal returns due to the fact that the variance of formula 11 is unknown. The null hypothesis is therefore estimated by means of formula 3.15:

)) , ( ( var ) , ( 2 1 2 1 τ τ τ τ CAR CAR t = ~ N (0,1) (3.15) 3.5.2 *on-parametric test

The previously described tests have specific assumptions about the distribution of the returns from the estimation period. The assumption is that they are normally distributed. Brown and Warner (1980) explain that less restrictive assumptions about the distribution of abnormal returns can be made by using non-parametric tests. Common tests used are the sign- and the Wilcoxon signed rank test. The

(23)

___________________________________________________________________________________

sign test is independent across the indices and the spot price Brent. The positive abnormal returns have a value of 0.5. MacKinlay (1997) explains that a weakness of the sign test is that it is not well specified when the distribution of the abnormal returns is skewed. A test which bypasses this problem is the Wilcoxon signed rank test. To use this test the abnormal returns of the indices and spot price Brent have to be ranked for the event period. The test statistic for event day zero of no abnormal return is:

∑

=       + − = * t i i s K L K * 2 / ( ) 1 1 2 0 θ (3.16) and 2 2 1 2 1 1 2 2 1 1 1 ) ( _             + − =

∑

= + = L K * L K s _i * i T T τ τ (3.17)

where

L

₂ is the length of the event window and

K

_i_τ is the rank of the abnormal returns of the DS- European index and spot price Brent for event period

τ

. The

τ

ranges from

T

₁

+

1

to

T

₂ and

τ

is the event day. This means from day –5 until day +5.

(24)

___________________________________________________________________________________

4 EMPERICAL RESULTS A D DISCUSSIO

This chapter describes the results of the European indices and the spot price Brent. In the first paragraph the descriptive statistics will be discussed of the mean adjusted model and the market and risk adjusted model. Thereafter, the cumulative returns of the production quota announcements for Europe and Brent will be thoroughly explained. The industry effects will be discussed in paragraph 4.4 and 4.5. The next paragraph deals with the data which are non-normal distributed by means of the Wilcoxon signed rank test. In the last paragraph the results will be elaborated.

4.1 Descriptive statistics

In appendix 6 from table 1 until table 4 the descriptive statistics are presented of the mean adjusted model and market and risk adjusted model for the first four hypotheses. As previously stated in paragraph 3.5 the data has to be normally distributed in order to apply a parametric test. Otherwise biased conclusions can be drawn if one uses a parametric test while the data is non-normal distributed. This is of particular concern for small samples, like in this research. The different tables show four different values which are the JB, probability, skewness, and the kurtosis. The p-value of the JB should be larger than 0.05 to not reject the null hypothesis of a normal distribution at the 5% level.

4.1.1 Europe and Brent

The descriptive statistics for the European index and Brent are shown in table 1. The full period is summarized in the first column. The next columns present the different production quota announcements during this full period. In the last two columns the time periods 1986-1996 and 1997-2006 are summarized.

If one analyzes table 1 of the mean adjusted model none of the announcements are significant to reject the null hypothesis at the 5% level. The announcements and time periods of Europe and Brent are normally distributed. Also the market and risk adjusted model do not show any significant values, and hence the null hypothesis is not rejected at the 5% level.

The conflict and non-conflict periods for both models are shown in table 2 for Europe and Brent Crude Oil. All the values for Europe and Brent have JB-values close to zero and do not produce significant probabilities, which hold for both models. Hence, the conflict and non-conflict periods will be conducted with a t-test.

(25)

___________________________________________________________________________________

4.1.2 Industrial effects

In tables 3 and 4 the descriptive statistics of the chemical, oil and gas industry are presented for the last two hypotheses. In table 3 the mean adjusted model shows one significant value when OPEC announces to increase its production. With a JB-value of 8.3551 and a probability of 0.0153 the null hypothesis is rejected. Thus, the production increase applicable to the chemical industry will be excluded from the t-test. It will be approached with the Wilcoxon signed rank test. All other values of both models presented in table 3 do not produce significant values. Hence, they will be approached with the t-test.

Table 4 presents the descriptive statistics during conflict periods and non-conflict periods. For both models of the chemical, oil and gas industry the values are not significant. Therefore, the null hypothesis is not rejected and the t-test will be conducted for both models.

4.2 Production quota announcements

The results of the first three hypotheses are presented in this paragraph for the mean adjusted model and the market and risk adjusted model. First, the cumulative outcomes of the European index and the Brent Crude Oil during the full period will be discussed. Furthermore, the results of the different production quota announcements will be discussed. Moreover, the results of the first period and second period will be presented. Different interval levels of the t-test are presented and explained at the end of each sub paragraph.

4.2.1 Mean adjusted model

The first hypothesis claims that production quota announcements have influence on the European index and Brent Crude Oil. The t-tests conducted are tested two-sided, which holds for all periods and production announcements. Table 5 shows the CAR and t-values of the European index. If one analyzes Europe in appendix 6, table 5, several inferences can be made. The cumulative returns during the full period follow a negative pattern from day -2 until day + 5. Albeit none of the cumulative returns are significant. The increase announcements are accompanied with negative returns from day -5 until day +-5 on the European index. None of these cumulative returns are significant. The results presented are in line to the results of Guide et al (2006). They find during the increase announcements of the full period no significant cumulative returns on the US and UK stock markets. When OPEC announces a decrease in its production the cumulative returns are positive during day 0 until day +5. Although none of the cumulative returns during these days are significant. The status quo announcements follow a negative pattern in the returns, since the returns are negative from day -2 until day +5. However, none of the returns are significant. The period 1986-1996 shows positive returns

(26)

___________________________________________________________________________________

from day 0 until day +5, but again not significant. The period 1997-2006 shows contrary to the first period negative results from day -4 until day +5. The returns are negative but not significant. Table 6 summarizes the cumulative returns at different interval periods. As can be seen the different interval periods do not produce any significant cumulative returns. Hence, the results indicate that the influence of OPEC is limited on the European index. The first three null hypotheses of the mean adjusted model for Europe are not rejected.

Table 6 Results production quota Europe mean adjusted

Interval period Full period df = 55 Increase df = 18 Decrease df = 15 Status quo df = 20 1986-1996 df = 28 1997-2006 df = 26 -1, +1 -0.0008 -0.2053 -0.0044 -0.6315 0.0016 0.3776 0.0001 0.0081 0.0007 0.1550 -0.0025 -0.5131 -3, +3 -0.0029 -0.3376 -0.0042 -0.2582 0.0017 0.1674 -0.0064 -0.3907 -3.6E-05 -0.0036 -0.0064 -0.5692 -5, +5 -0.0030 -0.2250 -0.0106 -0.4128 0.0072 0.4587 -0.0072 -0.2814 0.0021 0.1311 -0.0092 -0.5212

Although the results are not significant this does not implicate that the announcements of OPEC have no consequences in the near future. Driesprong et al (2005) state that investors underestimate the impact of oil price changes. An oil price change can influence future cash flows and thus the stock returns. If OPEC is able to influence the spot price of oil this might have consequences in the near future for the European index. Hence, this will be discussed in the following paragraph.

Table 7 summarizes the cumulative returns and corresponding t-values of the spot price Brent. The t-tests are tested two-sided because the empirical evidence is not univocal in the outcome of OPEC’s ability to influence the spot price. The investors seem to anticipate the news of OPEC during the full period. This because the cumulative returns are increasing from day -5 until day -1. With day -5 statistically significant positive at the 95% level. The decision of OPEC to increase the production does not affect the spot price of Brent. The cumulative returns do not follow a clear pattern during the event window. The evidence presented is in line with the results of Bina et al (2007) and Guidi et al (2007). Decrease announcements in the production of OPEC cause a significant positive return at day -5 at the 99% level. From day -3 until day -1 the cumulative returns increase. A drop in returns occur at the announcement day and increases further during the following days. The results presented are in contrast with the results of Guidi et al (2006). They find significant positive cumulative returns during day -1 until day +5. The results of Bina et al (2007) correspond to the results presented in this research. There is no relation in the cumulative returns of the status quo announcements. Furthermore, none of the cumulative returns are significant. The period 1986-1996 does not follow a certain pattern in its

(27)

___________________________________________________________________________________

cumulative returns. None of the cumulative returns are significant. In contrast with the first period the period 1997-2006 shows a relation in the cumulative returns with one cumulative return being significant. The returns increase from day -5 until day -1 and the markets seem to anticipate the news. Day -5 is significant positive at a 95% level. Explanations for this difference in the two periods might be the increased demand for oil and market disruptions (chapter two, page 10). Remarkable is that the first interval period (-1 until +1) in table 8 produces negative cumulative returns, except during the period 1997-2006. No significant cumulative returns occur during the interval periods. Hence, the first three null hypotheses of the mean adjusted model for the spot price Brent are not rejected.

Table 8 Results production quota Brent mean adjusted

Interval period Full period df = 50 Increase df = 16 Decrease df = 13 Status quo df = 19 1986-1996 df = 23 1997-2006 df = 26 -1, +1 -0.0071 -0.7468 -0.0084 -0.5840 0.0022 0.0944 -0.0097 -0.9777 -0.0145 -0.9999 0.0015 0.1188 -3, +3 0.0073 0.3293 0.0059 0.1759 0.0413 0.7657 -0.0145 -0.6276 -0.0062 -0.1833 0.0199 0.6712 -5, +5 0.0171 0.4878 0.0146 0.2767 0.0526 0.6207 -0.0071 -0.1964 0.0022 0.0421 0.0292 0.6275

4.2.2 Market and risk adjusted model

The cumulative returns of the market and risk adjusted model of Europe are presented in table 9. The t-tests of the different subjects are tested two-sided. From day -5 until day -1 the returns show an increase in the negative cumulative returns during the full period. During and after the days of the announcement the cumulative returns decrease. The decision of OPEC to increase the production of oil causes no significant reaction in Europe. From day -1 until day 4 we observe an increase in the negative cumulative returns. The decrease and status quo decisions of OPEC do not cause significant returns and do not show a relation in the cumulative returns. The period 1986-1996 does not produce any significant cumulative returns. The same result is found previously with the mean adjusted model. The cumulative returns of the period 1997-2006 follow the same pattern as with the mean adjusted model. From day -4 until day +5 the returns are negative. If we compare these periods the conclusion which can be drawn is that the first period produces positive returns and the second negative. None of the cumulative returns are significant. The conducted t-tests at different interval periods are summarized in table 10 on the next page. None of the cumulative returns produce significant t-values. Thus, the total picture is that the first three null hypotheses of the market and risk adjusted model for Europe are not rejected.

(28)

___________________________________________________________________________________

Table 10 Results production quota Europe market and risk adjusted

Interval period Full period df = 55 Increase df = 18 Decrease df = 15 Status quo df = 20 1986-1996 df = 28 1997-2006 df = 26 -1, +1 0.0011 0.5251 -0.0020 -0.5241 0.0012 0.2902 0.0026 0.5871 0.0022 0.8629 -0.0003 -0.0887 -3, +3 0.0008 0.1673 -0.0029 -0.3329 0.0019 0.1953 0.0013 0.1261 0.0014 0.2416 -0.0003 -0.0363 -5, +5 -0.0003 -0.0438 -0.0034 -0.2457 0.0004 0.0282 -0.0017 -0.1036 0.0004 0.0415 -0.0016 -0.1379

The spot price Brent in table 11 presents the cumulative returns of the market and risk adjusted model. As can be seen the cumulative returns do not follow a certain pattern during the full period. When OPEC announces an increase in the production the cumulative returns do not respond significant to this announcement. A production decrease announcement is accompanied with significant positive returns with the mean adjusted model. The market and risk adjusted model produces no significant cumulative return. The only similarity between the two models is that they produce positive cumulative returns. From day -5 until day -2 the cumulative returns increase. The cumulative returns decline during the following days. The status quo announcements of OPEC do not create significant cumulative returns. Positive and negative cumulative returns arise, without a relation between them. Similar results are presented with the mean adjusted model. The period 1986-1996 is accompanied with negative cumulative returns. None of the cumulative returns are statistically significant during the event window of the period 1986-1996. The cumulative returns during the period 1997-2006 do not follow a certain pattern in the cumulative returns. Positive and negative cumulative returns arise during the event window. Table 12 on the next page presents the results of the cumulative returns conducted at different interval periods. The market and risk adjusted model produce no significant cumulative returns for the different time periods and the production quota announcements. Hence, the first three null hypotheses of the market and risk adjusted model for the spot price Brent are not rejected.

(29)

___________________________________________________________________________________

Table 12 Results production quota Brent market and risk adjusted

Interval period Full period df = 50 Increase df = 16 Decrease df = 13 Status quo df = 19 1986-1996 df = 23 1997-2006 df = 26 -1, +1 -0.0044 -0.6720 -0.0092 -0.8824 -0.0013 -0.1032 -0.0008 -0.1106 -0.0075 -1.0170 -0.0007 -0.0813 -3, +3 0.0019 0.1209 -0.0019 -0.0760 0.0178 0.5870 -0.0042 -0.2401 -0.0042 -0.2458 0.0084 0.4206 -5, +5 0.0034 0.1432 -0.0024 -0.0631 0.0188 0.3946 -0.0002 -0.0063 -0.0038 -0.1423 0.0110 0.3517

4.3 Conflict and non-conflict periods

The fourth hypothesis in this research will be discussed in this paragraph. Tables 13 until 16 present the results of the mean adjusted model and the market and risk adjusted model for Europe and the spot price Brent.

4.3.1 Mean adjusted model

Table 13 presents the results of Europe and Brent during conflict and non-conflict periods. The t-test values presented are tested two-sided. If we analyze the positive cumulative returns of Europe the returns show an increase from day -2 until day +2. The market still reacts to the news after the announcement of OPEC at day 0. Although none of the returns are statistically significant. The results presented are in line with the results of Guidi et al (2006). None of their results of the US and UK markets are statistically significant. During non-conflict periods negative cumulative returns arise. There is a relation between the negative cumulative returns. From day -5 until day +4 the returns are increasingly negative, but statistically not significant. Guidi et al (2006) find similar results in their research. The t-test values conducted a different interval periods presented in table 14 do not show significant results for Europe. The relation between the returns is that they are increasing during conflict periods and decreasing during non-conflict periods. Therefore, the null hypothesis that production quota announcements have no influence on Europe during periods of conflict and non-conflict is not rejected.

The cumulative returns of Brent during conflict periods show positive returns prior to the announcement. After the announcement of OPEC the returns drop. Guidi et al (2006) suggest due to uncertainty with interpreting the information from OPEC during conflict periods.5 The same finding applies to the cumulative returns during non-conflict periods. None of the cumulative returns produce significant t-values. The different interval periods of Brent do not produce significant cumulative

5

(30)

___________________________________________________________________________________

returns as can be seen in table 14. The conclusion we can draw is that the returns are increasing during conflict periods and decreasing during non-conflict periods for Europe. For Brent the cumulative returns are decreasing during conflict-periods and increasing during non-conflict periods. Thus, the null hypothesis that production quota announcements have no influence on Brent during periods of conflict and non-conflict is not rejected.

Table 14 Results conflict and non-conflict Europe and Brent mean adjusted

Interval period Conflict df = 18 Europe Non-conflict df = 36 Europe Conflict df = 13 Brent Non-conflict df = 36 Brent -1, +1 0.0028 0.4209 -0.0026 -0.4966 -0.0035 -0.1835 -0.0070 -0.5007 -3, +3 0.0052 0.3274 -0.0070 -0.5746 -0.0032 -0.0703 0.0117 0.3608 -5, +5 0.0113 0.4566 -0.0104 -0.5411 -0.0141 -0.1997 0.0281 0.5518

4.3.2 Market and risk adjusted model

Table 15 presents the cumulative returns of Europe and the corresponding t-values. During conflict periods the cumulative returns of Europe follow the same pattern as with the mean adjusted model. The returns are increasing from day -2 until day +2, although none are significant. The non-conflict periods produce a drop in the cumulative returns from day -5 until day -1. Although the cumulative returns are negative, none of them are statistically significant.. The cumulative returns conducted at different interval periods are summarized in table 16. The cumulative returns of the different interval periods correspond with the results of the mean adjusted model. During conflict periods the cumulative returns are positive. The returns are just as with the mean adjusted model not significant. The cumulative returns of the non-conflict periods are during the first interval period positive. The other periods are negative which corresponds with the results of the mean adjusted model. None of the cumulative returns are statistically significant. Therefore, the fourth null hypothesis of Europe is not rejected.

The cumulative returns of Brent during conflict periods are in line to the returns of the mean adjusted model. After the production announcement of OPEC the returns drop as can be seen in table 15. During non-conflict periods positive and negative returns arise. The cumulative returns performed at different interval periods are summarized in table 16 on the next page. The difference with the mean adjusted model is that Brent produces negative cumulative returns in relation to the market during conflict periods. The cumulative returns during non-conflict periods are similar to the returns of the mean adjusted model. Both models do not produce significant cumulative returns. Thus, the fourth null hypothesis of Brent is not rejected.