Page 1 Universiteit van Amsterdam
Author: Coen Binnerts Supervisor: Ieva Sakalauskaite
The effect of U.S. Presidential elections on
stock market volatility and return around
the world
Bachelor Thesis Economics and Finance
University of Amsterdam
Student ID: 10560998
January 2017
Abstract
There is substantial evidence that the U.S. Presidential election effects stock market volatility in countries outside the United States. Previous studies researched the effect of an election on the home
country. This study focuses on the effect of the U.S. Presidential election on foreign countries. I found that during the period of 1984-2016 there is a significant positive effect of the U.S. Presidential election on stock market volatility. This confirms the findings of previous studies. However excluding the election of 2008 a negative significant effect on stock market volatility is found. This contradicts previous studies.
This study confirms that a change in policy induce more volatility. Also this study finds no significant effect of the U.S. Presidential election on abnormal stock market return for foreign countries. This
Page 2 Statement of originality
This document is written by Student Coen Binnerts who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Page 3
Index
Abstract ... 1
Introduction ... 4
Literature review ... 6
The electoral system in the United States ... 9
Methodology and data ... 10
Data ... 10 Methodology ... 12 Volatility... 12 Abnormal return ... 14 Results ... 15 Conclusion ... 18 Bibliography ... 20 Appendix ... 22
Page 4
Introduction
On the 8th of November 2016 the new President of the United States of America, Donald J. Trump, was elected. Donald J. Trump is a wildly criticized Republican candidate. He is a businessman with a large company and no experience in politics. During his campaign he made various, for this decade,
remarkable comments. One of these comments is building a wall on the border of the United States and Mexico and cancel NAFTA, the trade agreement between the United States, Canada and Mexico. He also said that NATO is old-fashioned and that if he becomes President he does not know if he is going to help the European countries against Russian threats. He said he will retreat from negations about TTP on day one of his presidency. On the night of the election, 9th of November local time, the Nikkei 225 index in Japan dropped by 5.3%.
As I saw this I thought how much effect does the president of the United States have on other countries. How much does the rhetoric of Donald Trump effects the stock markets outside the United States. The largest economy in the world must have an effect on other countries which trade with the United States and depend on their consumers for their products. The U.S. President can, without much intervention from Congress, implement import tariffs and cancel trade agreements. These trade agreements are very important for small open economies. For instance the Netherlands exports 63% of its GDP to the world (Centraal Bureau voor de Statistiek, 2017). 3% of its GDP is exported to the United States. If large import tariffs are implemented in the United States it will be less attractive for U.S. consumers to buy Dutch products. Dutch producers will have less profits. The stock market will react to these changes and announcing such actions will cause uncertainty about future profits in the market. In graph 1 the effect around the Presidential election of 2016 is shown for a few country indices.
Page 5 Current literature about the effect of elections on stock market volatility and return studies the effect on the home country. A few of these studies are mentioned below.
Booth and Booth (2003) found that return in the first two years of a presidential cycle is lower and the last two years of the presidential cycle the return is higher in the United States. They found that the presidential cycle has an explanatory power beyond business conditions proxies which are important in explaining stock returns. They also found that small-cap stock excess return under Democratic
presidents are significantly higher than under Republican presidents.
Jinling and Born (2006) found that if the outcome of the U.S. Presidential election is uncertain the stock market return and volatility rise in the three months before the election day. If it is virtually certain that a candidate will win, the effect on stock market return and volatility is nonexistent.
Soikis and Kapopoulos (2007) found that changes in political regimes have a significant effect on the stock market volatility in Greece. They also found that the stock market volatility increases more during the pre-election period and when the right-wing government is in power.
Bialkowski et al. (2006) found compelling evidence that a national election has a significant effect on stock market volatility for all OECD countries. They argued that political uncertainty is reflected in the stock market volatility and that investors are still surprised by the distribution of votes. They also found that a rise in stock market volatility is effected mainly by parliamentary or presidential systems of a country and the margin of victory of the winner.
This study will examine, in contrast with previous studies, the effect beyond the borders of the United States and examine the effect of the U.S. Presidential election on the (semi) developed part of the world. The importance of this study is to understand the effect of the U.S. Presidential election on volatility and return for foreign countries. Volatility is one of the variables that changes prices in multiple derivatives, the most important one is options. It is important to know when it is likely there will be more volatility. Based on these findings one can trade and price the option as good as possible. To my
knowledge the following question has not been researched in the past, although as shown above, it is important for the pricing of derivatives. The goal of this research is to answer the following question: What is the effect of U.S. Presidential elections on stock market volatility and returns around the world and how can this be explained?
In this study daily stock market return will be used for 22 countries around the world. An OLS regression will be used to predict the daily stock market return of a specific country based on the return of a proxy of the world market portfolio. A GARCH(1,1), generalized autoregressive conditional
heteroscedasticity, model will be used to forecast the stock market volatility of an index and the effect in the period surrounding the election. For the explanation of the multiplicative effect of the stock market volatility in the election period an OLS-regression will be used.
Using event study methodology I will test if the cumulative abnormal volatility in a country is significantly positive in the election period than outside the election period and how this effect can be explained using multiple variables. These variables will be the export as percentage of GDP to the United States, the GDP per capita, a dummy variable if a president is re-elected or not and the margin of victory. I use export as a percentage of GDP as a measure of exposure of a country to the United States. I will
Page 6 also test if there is an abnormal return in foreign countries in comparison with the world market
portfolio.
The further structure of this study is as follows. Section 2 will discuss the previous studies related to this research. Section 3 will explain the electoral system in the United States Section 4 will explain the data and methodology used in this research. In section 5 the results will be stated and discussed. Finally, in section 6 a conclusion is given based on the results of this research and will aim at answering the research question. Additionally, the limitations of this thesis will be given and suggestions for future research.
Literature review
The efficient market hypothesis argues that all public information is imbedded in the price of an publicly traded assets (Fama, 1970). With regard to the presidential election in the United States, the public information about the U.S. Presidential election should also, according to this hypothesis, be incorporated in the price of stocks. This means that the probability of one of the candidates becoming the President must be incorporated in the price of an asset and also the probability of the candidate implementing changes. The moment one of the candidates is elected as the President there is no uncertainty about who is the new President. However the uncertainty about implementing the policy changes announced in the campaign stays. An investor does not know on the moment the winner is announced what he is going to do. In the few days after the election day the President-elect gives speeches where is policy plans become clearer. This uncertainty should be in the volatility of a specific country. In this section the previous studies investigating the effect of elections on stock market return and volatility will be discussed.
Regarding the effect of national elections on stock market volatility in general previous studies found the following results. Bailowski, Gottschalk and Wisniewski (2006) investigated the question if elections induce higher stock market volatility. They found that investors are still surprised by the outcome of an election. They also found strong evidence that volatility increases during an election period and during closely contested races volatility rises, measured by the margin of victory. Jinling and Born (2006) confirms this finding based on their research using poll data as a measure of closely
contested races. They found that volatility rises if it is too close to call to accurately state that one of the candidates will win. If it is virtually certain one candidate will win volatility is the same as in the non-election period. This indicates that on a day-to-day basis stock market volatility is strongly influenced by politics. Bittlemayer (1998) confirms this conclusion in an investigation of German stock prices that large political changes have a significant effect on stock market volatility during 1880-1940.
Siokis and Kapopoulos (2007) found in their research about volatility in Greece that politics strongly influences stock market return and volatility, this confirms the findings of Bialkowski, Gottschalk and Wisniewski (2006). The studies of Bialkowski, Gottschalk and Wisniewski (2006), Jinling and Born (2006) and Siokis and Kapopoulos (2007) all use the GARCH(1,1) model or a variation of it to predict volatility and measure the effect of an election on stock market volatility. They all find a significant positive effect of elections or large political changes on volatility.
Based on the above mentioned statements the following hypotheses is formed:
Page 7 Hypothesis 1:
The null hypothesis is that the cumulative abnormal volatility (CAV) is zero in the election period. The alternative hypothesis will be that the CAV will be higher in the election period than outside the election period. This can be expressed as:
𝐻0: 𝐶𝐴𝑉i= 0
𝐻1: 𝐶𝐴𝑉𝑖 > 0
Where CAV is the cumulative abnormal volatility in the election period and the i is the country. As opposed by Bailkowski, Gottschalk and Wisniewski (2006) the CAV can be explained for a country by the margin of victory, compulsaroy voting laws and presidential or parliamentary elections. They found that a larger margin of victory has a significant negative effect on the cumulative abnormal volatility. This is inline with the theory that more uncertainty causes more volatility. Additionally they found that mandatory voting reduces the election surprise. Voters with extreme political views show an above-average incentive to vote and can disort the electional outcome. This contributes to the
unpredictablility of the electoral outcome. Jinlang and Born (2006) also find evidence that if it is virtually certain that a candidate will win the CAV will be 0. Their research used poll data as an estimation of the likelihood that a candidate will be President.
The above mentioned statements are the base for the following hypothesis: Hypothesis 2:
The null hypothesis will be that the margin of victory has no effect on the CAV. The alternative hypothesis is that the margin of victory will have a negative effect on the CAV. This can be expressed as:
𝐻0: 𝛽ℎ= 0
𝐻1: 𝛽ℎ< 0
Where 𝛽ℎ is the effect of margin of victory in percentage points on CAV.
There have been several studies regarding the effect of poltical regime changes and elections on stock market volatility. Bealieu et al (2005) studied the impact of political risk on stock market volatility in Canada using a GARCH(1,1) model. They estimate the conditional variance on return. Following the dividend discounted model they hypothesized that the expected cash flows of firms are effected by political risk. The impact of political risk changes future dividends and future prices and makes the asset prices more volatile. The main conclusion of this study is that political risk effects volatility but can be diversified away by buying assets that are hardly effected by the political event. This study can be linked with the studies of Jinling and Born (2006) and Bialkowski, Gottschalk and Wisniewski (2006) which find that if the winner of the election is virtually certain the effect on volatility decreases. This means that political risk decreases.
Pastor and Veronesi (2012) found in their study about the effect of policy changes on stock market return that a change in policy overall lowers return. The effect of a policy change is larger if uncertainty about the government policy is large. According to Pastor and Veronesi (2012) a change in policy increases volatility. In the case of the U.S. Presidential Election a large change in policy is a change
Page 8 of the President. Every President has its own ideas about policies and a large shift in this would,
according to the study of Pastor and Veronesi (2012), increase stock market volatility. A example of a policy changes that effects other countries are implementing import tariffs on certain products or on all products from a country. This policy will cause higher prices of these products in the U.S. market which leads to a lower demand of them. This effects the profits and exports of the foreign firms.
The above statements are the base of the following hypothesis: Hypothesis 3:
The null hypothesis is that if the current president is re-elected it has no effect on CAV. The alternative hypothesis is that if a current president is re-elected it has a positive effect on CAV. This can be expressed as:
𝐻0: 𝛽ℎ= 0
𝐻1: 𝛽ℎ> 0
Where 𝛽ℎ is a dummy variable with 1 if the elected President is the current President.
Giovanni and Levchenko (2009) examined how economic output volatility is related to trade openness. They found in their research that first, industries that are more open to trade are more volatile. Second, that trade is accompanied by more specialization. Third, sectors that are more open to trade are less correlated with the rest of the home economy. These three points together have all an effect on volatility and the relation between trade openness and overall volatility is economically
significant and postive. Based on the research above it is assumed that if a country is more open to trade its economic output and stock market is more volatile. A policy change in the foreign country effects the home country more if the country is more open to trade. This implies that a country that has a higher exposure to the U.S. market is effected more by a policy changes in the United States than countries that have less exposure to the U.S. market. This should also be observed in the stock market volatility. I measure exposure to the U.S. market by the export as percentage of GDP in the year of the election.
The above mentioned statements are the base for the following hypothesis: Hypothesis 4:
The null hypothesis is that a higher export to the United States of country i has no effect on CAV during the election period. The alternative hypothesis is that a higher export of country i has a positive effect on CAV during the election period. This can be expressed as:
𝐻0: 𝛽ℎ= 0
𝐻1: 𝐵ℎ> 0
Where 𝛽ℎ is the effect of export in percentage of GDP on the CAV in the year of the election.
There are several studies on the effect of elections on stock market return. Jingling and Born (2006) investigated the effect of U.S. Presidential elections on common stock market return in the United States. before the election day. They found that mean daily common stock return rises in the three months prior to the election when the outcome of the election is uncertain. If the outcome is virtually certain then the effect on stock market return is non-existent. This is consistent with the
Page 9 efficient market hypotheses that all public information is priced in the stock market. The probability of one of the candidates becoming President is not new news then so it is must already, according to the efficient market hypothesis, be in the price of a stock.
Pantzalis, Stangeland and Turtle (2000) found in their study by using weekly international stock market data that national elections have a positive abnormal return in the two weeks prior to the election week. Their research was based on 33 countries with 129 elections included. They found that the strongest excess return is found in less-free countries and in elections which were held early. They conclude that there is a corresponding increase in equity prices the moment uncertainty is resolved. This is the same conclusion as Jinling and Born (2006). They found no effect if the outcome of the election is virtually certain. Based on the research above it is expected that there will be a positive effect on abnormal return.
In ordnance with the conclusions in previous studies this study expects the following result: Hypothesis 5:
The null hypothesis is that the cumulative abnormal return (CAR) is zero in the period surrounding the election. The alternative hypothesis is that the CAR is non-zero in the period surrounding the election. This can be expressed as:
𝐻0: 𝐶𝐴𝑅 = 0 𝐻1: 𝐶𝐴𝑅 ≠ 0
The electoral system in the United States
In order to determine the variables needed in this study the Presidential election process in the United States need to be clearly defined. In this part of the study the electoral system of the United States will be discussed.
The election system in the United States consist of several types. In this study I will use the Presidential Election. Every four years on the first Tuesday after the first Monday in November the Election day is held (Presidential Election Process, 2016). Article II of the U.S. Constitution states that the President and the Vice-President holds the office for four years (Presidential Election Laws, 2016). The 22nd amendment states that no President can be elected more than twice.
The election process begins with primary selections where at the nominee conventions each party selects its nominee. The nominees have campaigns across the country to tell their views and ideas to the people. Americans on election day head to the polls to cast a vote for the President. Presidential elections use the Electoral College to determine the outcome of the election (Presidential Election Process, 2016). This means that the people choose the representatives of their vote. Those
representatives are called electors. The amount of electors in a state is equal to the amount of Senators of a state. The winning party in a state gets every electoral (Presidential Election Process, 2016). The candidate can only win by receiving the majority of electoral votes, this is at least 270 electors. The popular vote does not count as in most countries. On the 20th of January the new President is inaugurated (Presidential Election Laws, 2016). For this study the majority to win is based on the majority of electoral vote because that determines the outcome of the election. The figure 1 shows the electors per state.
Page 10
Figure 1: Distribution of electoral voters per state
Source: www.usa.gov/election
In this study nine elections are included. The following table shows the details of those elections. The source for this data is the electoral college website (Historical Election Results, 2016) . A one in the colom of re-elect means that the winner of the election is the current President.
Table 1: Presidential elections from 1984-2016
Methodology and data
In this part the methodology and the data used in this study will be described.
Data
In this study stock returns are measured using market indices. Market indices report the value of a particular portfolio of securities (Berk & DeMarzo, 2014, p. 402) . The benefit of using market indices is that it does not contain firm specific risk but it is an approximate of systematic risk and the actual market portfolio. To estimate systematic risk in a country the local market index will be used. As an estimate for the world market portfolio the MSCI World Index will be used. This index includes Presidential election dates Re-elect Electoral vote Total in % Margin of victory
6-11-1984 1 525 538 97,6% 47,58% 9-11-1988 0 426 538 79,2% 29,18% 3-11-1992 0 370 538 68,8% 18,77% 5-11-1996 1 379 538 70,4% 20,45% 7-11-2000 0 271 538 50,4% 0,37% 2-11-2004 1 286 538 53,2% 3,16% 4-11-2008 0 365 538 67,8% 17,84% 6-11-2012 1 332 538 61,7% 11,71% 8-11-2016 0 306 538 56,9% 6,88%
Page 11 1.600 stocks from around the world, the distribution of countries represented in this index is shown in Figure 2 in the Appendix.
Thomson Reuters Financial Datastream (Datastream) will be used for the historical prices of the indices. By using the daily quotes from datastream the daily return is calculated by using the formula below:
𝑅𝑡 =
𝑃𝑡
𝑃𝑡−1
− 1
Where 𝑅𝑡 represents the daily return of the index on day 𝑡. 𝑃𝑡 is the quote of the index at day
𝑡 and 𝑃𝑡−1 is the quote at day 𝑡 − 1 . There are 22 countries used in this study. The countries
incorporated in this study can be found in table 2.
The political data is retrieved from the electoral college website. See table 1 for the details. The data for the GDP per capita is retrieved from the World Bank. The approximate for the GDP per capita in 2016 is the year 2015. A summary of this data is shown in table 7 in the Appendix.
The data for the export to the United States are retrieved from the website1 of the United States. commerce of chambers. I used the imports of the United States as the export of the country. The GDP per country is retrieved from the World Bank database. I use 1985 as an approximate of the export of 1984 and 2015 as an approximate of 2016. See table 8 in the Appendix for a summary of this variable.
The export as percentage of GDP is calculated with the following formula: 𝐸𝑥𝑝𝑜𝑟𝑡 𝑖𝑛 % 𝑜𝑓 𝐺𝐷𝑃 =𝐸𝑥𝑝𝑜𝑟𝑡𝑖
𝐺𝐷𝑃𝑖
∗ 100 Where export and GDP for country i is in current USD.
1 www.census.gov
Page 12
Table 2: Details of the countries included
In the first column the name of the country is mentioned, in the second column the starting date of the index is mentioned, in the third column the number of U.S. Presidential elections included in this index is mentioned, in the fourth column the average daily return is mentioned, in the fifth column the average daily standard deviation is mentioned.
Methodology
Volatility
In order to determine the effect of the U.S. Presidential election on volatility. In this study volatility is predicted by the conditional variance of the error term.
To explain the conditional variance of the return a GARCH(1,1) framework is used. A GARCH(1,1) model estimates future volatility as a function of prior volatility and the error term of the day before.
Mathematically this can be expressed in the following way:
𝑅𝑖,𝑡 = 𝑥 + 𝛽𝑅𝑡∗ + 𝜀𝑖,𝑡, 𝜀𝑖,𝑡 𝑁~(0, ℎ𝑖,𝑡) (1)
ℎ𝑖,𝑡 = 𝛾0+ 𝛾1ℎ𝑖,𝑡−1+ 𝛾2𝜀𝑖,𝑡−12 (2)
Where 𝑅𝑖,𝑡 is the return of country index i on time t and 𝑅𝑡∗ is the return on the MSCI World
index. 𝜀𝑖,𝑡 denotes the country specific part of the index returns and ℎ𝑖,𝑡 is its conditional variance.
Name Starting date Number of elections included Average daily return Average daily SD
Australia 29 May 1993 6 0,023% 0,95%
United Kingdom 01 January 1980 9 0,033% 1,09%
The Netherlands 03 January 1983 8 0,035% 1,32%
Japan 01 January 1980 9 0,020% 1,35% Germany 01 January 1980 9 0,041% 1,34% France 09 July 1987 7 0,024% 1,38% Switzerland 10 July 1987 7 0,029% 1,14% Spain 05 January 1987 7 0,027% 1,39% Italy 01 January 1998 5 0,004% 1,57% Portugal 01 January 1993 5 0,013% 1,16% Croatia 01 January 1997 5 0,031% 1,51%
Czech Republic 06 April 1994 6 0,007% 1,32%
Denmark 04 December 1989 7 0,036% 1,17% Finland 02 January 1987 7 0,040% 1,60% Poland 19 April 1994 6 0,025% 1,78% Romania 19 September 1997 5 0,052% 1,64% Bulgaria 01 January 2007 2 -0,027% 1,33% Russia 22 September 1997 5 0,095% 2,61% Turkey 04 January 1988 7 0,156% 2,56% Sweden 01 January 1986 8 0,045% 1,43%
South Korea 03 January 1990 7 0,029% 1,75%
Page 13 (1) and (2) are estimated jointly using the Maximum Likelihood method over a period immediately preceding the event window. The convention in the literature and as described by Brown and Warner (1985) is to use 250 daily returns to estimate the benchmark model. A year of daily observations, however may be too short to estimate a sufficient and accurate GARCH(1,1) model. On the other hand an over-expanding estimation period substantially cut the number of countries included in my sample. Following these practical guidelines and the results from Hwang and Pereira (2004) I have decided to choose an estimation period of 500 trading days.
The event-window will be ten days surrounding the election. This period is used because some countries are in different time zones so they know the outcome of the election only one trading day later in local time than in U.S. time. Also many Presidents hold speeches a couple days after the election where they clarify what there policy is going to be. This is still a factor in uncertainty since we do not know their rhetoric as President-elect.
To measure abnormal volatility, one has to consider the the varation in 𝜀𝑖,𝑡 around the event
date in relation to its regular level.
To measure the normal volatility the mean of the estimated volatity outside the event window is used. A multiple is used to describe the effect of the election on the volatility. Mathematically this can be expressed as:
𝑀̂𝑡 =
ℎ ̂𝑖,𝑡
(∑𝑖= 𝑡∗𝑡= 𝑡∗−500(ℎ̂𝑖,𝑡))/500 − 1
Where 𝑡∗ is the day prior to the first trading date in the event window. This multiple is not econometrically flawless but is a good representation of the abnormal volatility.
For an event window of (𝑛2, 𝑛1) the cumulative abnormal volatility (CAV) can be calculated as
𝐶𝐴𝑉(𝑛1, 𝑛2) = ∑ 𝑀𝑡 𝑛2 𝑡=𝑛1
The null hypothesis can be expressed in the following way: 𝐻0: 𝐶𝐴𝑉(𝑛1, 𝑛2) = 0
The alternative hypothesis can be expressed in the following way: 𝐻1: 𝐶𝐴𝑉(𝑛1, 𝑛2) ≠ 0
To test the hypothesis a student t-test is used. This can be expressed in the following way: 𝑡 = 1
√𝑛 − 1∗ 𝐶𝐴𝑉 𝑠(𝑀𝑖)
Where 𝑛 is the number of trading days in the event-window.
To explain the volatility a regression is used. The regression can be expressed as:
Page 14 Where LN(GDP per capita) is the natural logarithm of the GDP per capita. Export is the export to the United States as percentage point of GDP. Marginofvictory is the percentage point above 50 percent of the electoral votes, Reelect is a dummy variable where 1 is the winner of the election is the current President.
Abnormal return
In order to determine the effect of the U.S. Presidential election on stock market return the following OLS regression will be used.
𝑅(𝐼𝑛𝑑𝑒𝑥𝑖)𝑡 = 𝛼 + 𝛽𝑅𝑀𝑆𝐶𝐼 𝑊𝑜𝑟𝑙𝑑 𝐼𝑛𝑑𝑒𝑥,𝑡+ 𝜀𝑡 ~ 𝑁(0, 𝜎2)
The dependent variable 𝑅(𝐼𝑛𝑑𝑒𝑥𝑖)𝑡 is the return of the index of country i on time t. The independent
variable 𝑅𝑀𝑆𝐶𝐼 𝑤𝑜𝑟𝑙𝑑 𝑖𝑛𝑑𝑒𝑥 is the return of the MSCI World index on time t.
The event study methodology introduced by MacKinlay (1997) will be used.
The estimation period for this regression will be 500 trading days and the event-window will be ten trading days surrounding the election day.
The abnormal return 𝐴𝑅𝑖𝑡 will be calculated by using the following formula:
𝐴𝑅𝑖𝑡 = 𝑅𝑖,𝑡− 𝑥̂𝑖− 𝛽̂𝑖,𝑡𝑅𝑀𝑆𝐶𝐼 𝑊𝑜𝑟𝑙𝑑 𝐼𝑛𝑑𝑒𝑥,𝑡
The cumulative abnormal return for country i of election j is given by:
𝐶𝐴𝑅𝑖,𝑗= ∑ 𝐴𝑅𝑖,𝑗,𝑡 𝑇2
𝑡=𝑇1
MacKinlay (1997) states that if the sample is large enough, the variance of the cumulative abnormal return is given as:
𝑣𝑎𝑟(𝐶𝐴𝑅𝑖) = 𝜎𝑖2(𝑇2, 𝑇1) = (𝑇2− 𝑇1+ 1) ∗ 𝜎𝜀𝑖
2
Using the above, the variance of the cumulative abnormal returns 𝜎𝑖2(𝑇2, 𝑇1) will be estimated using the
variance of the error term 𝜎𝜀2𝑖. To test if the abnormal return differs in the event period. The following
test is used: 𝐶𝐴𝑅𝑖
𝜎(𝐶𝐴𝑅𝑖)
~ 𝑁(0, 𝜎𝑖2(𝑇2, 𝑇1))
In order to test if elections have an effect on abnormal returns the average abnormal returns are examined across all countries. The average abnormal return of N elections at time t is given as:
𝐴𝑅 ̅̅̅̅𝑡 = 1
𝑁∑ 𝐴𝑅𝑖𝑡
𝑁 𝑖=1
The average cumulative abnormal returns is given as:
𝐶𝐴𝑅
̅̅̅̅̅̅𝑡 = ∑ 𝐶𝐴𝑅𝑖
𝑇2 𝑖=𝑇1
Page 15 Under the null hypothesis as mentioned in hypothesis 2 the CAR is equal to zero. The alternative
hypothesis is that the CAR is non-zero. To test this hypothesis the following test statistic is used: 𝑡 = 1
√10 𝐶𝐴𝑅
𝑠 ~ 𝑡𝑁−1
Where 𝑠 = √𝑁−11 ∑𝑁𝑖=1(𝐶𝐴𝑅𝑖− 𝐶𝐴𝑅̅̅̅̅̅̅𝑖)2 , and N is the number of elections in the sample.
The MSCI World Index has a shortcoming. It consists for 54% of U.S. stocks that is not incorporated in this study. The rest of the MSCI World index consist of stocks outside the United States, mainly the United Kingdom, Europe ex U.K. and Japan. For the exact distribution see figure 2 in the appendix. This means that the indices used in this study are also incorporated in the MSCI World Index.
Results
In this section the results of this study are described and explained.
The results for the cumulative abnormal volatility are shown in table 3, first showing the CAV and robust standard error of the CAV, then the regression to explain the CAV is shown.
In this research a significant robust CAV is found for the all the countries together. The p-value of the CAV is smaller than 5%. This shows that around the U.S. Presidential election countries outside the United States are effected with the uncertainty that the election brings. These results confirm the expectation that the investors are still surprised by the outcome of the election. If the poll data suggests that it is very likely that someone wins their should be no volatility if the outcome is the same. This is inline with what Bialkowsi et al. (2006) found in their research.
However volatility of the election in 2008 strongly effects these results since the peak of the volatility was on 15 september 2008, the bankruptcy of Lehman Brothers, and the election of Barack Obama was on the 4th of November of 2008. The volatility and uncertainty about the economy is still in the market around the election of 2008. The volatility experienced at that time has more to do with the subprime debt crisis than with the U.S. Presidential Election. If you exclude 2008 from the results there is a negative significant effect on the overall cumulative abnormal volatility. Also the regression to explain this effect has not one significant variable.
Page 16
Table 3: Results for the CAV and explanatory variables
Test variable Intial
regression Excluding 2008 CAV 4.945** (1.347) -0.638** (0.213) Constant -46.058** (14.779) -3.318 (2.712) LN(GDP per capita) 5.660** (1.512) 0.224 (0.282) Export as percentage of GDP -0.029** (0.010) 0.001 (0.002) 1/0 Re-elect variable -13.788** (2.376) 0.562 (0.488) Margin of victory 0.504** (0.125) -0.019 (0.024) F-statistic 14.23 0.71 R2 0.2890 0.0232
Note the model that is used is:
𝐶𝐴𝑉 = 𝛼 + 𝛽1𝐿𝑁(𝐺𝐷𝑃 𝑝𝑒𝑟 𝑐𝑎𝑝𝑖𝑡𝑎) + 𝐵2𝐸𝑥𝑝𝑜𝑟𝑡 + 𝐵3𝑀𝑎𝑟𝑔𝑖𝑛𝑜𝑓𝑣𝑖𝑐𝑡𝑜𝑟𝑦 + 𝐵4𝑅𝑒𝑒𝑙𝑒𝑐𝑡 + 𝜀
Where the margin of victory is in percentage points. Re-elect is a dummy variable of 1 and 0. Export is in percentage points as the export to the U.S. as percentage of GDP.
** Significant at the 5% level.
The CAV can be explained using the regression described in the methodolgy section. The results of that regression are shown in table 3.
A richer country, measured by the natural logarithm of GDP per capita, is significantly more effected by the political uncertainty than poorer countries. This is because richer countries are often more globalised than poorer countries.
A country with more export to the United States, measured in percentage of GDP, are significantly negatively effected by the political uncertainty from the U.S. Presidential Election than countries that export more to the United States. This is not inline with the research of Giovanni and
Page 17 Levchenko (2009). They found that a more open country, measured through the percentage of export of GDP, is more volatile. However the effect of the export is so small that it does not seem to be very important.
A higher percentage point in margin of victory in this research indicates a higher volatility. This is result is not inline with previous studies. Bialkowski et al (2006) found that a higher margin of victory reduces volatility significantly and Jinling and Born (2006) found that if it is virtually certain a candidate wins, abnormal volatility does not exist. The result can be explained because using the margin of victory does not give a strong measure of uncertainty. Poll data might be a better measure of uncertainty. Poll data gives the expectation of the margin of victory. If the race is close to call the uncertainty would rise and the volatility might rise too. This is a limitation in this research. This result must be examined more in future research using poll data instead of a margin of victory.
If the winner of the election is the current president, the re-elect dummy variable in table 3, the volatility is significantly reduced. This is inline with previous studies which indicate that a change of policy increases political uncertainty which increases volatility (Pastor & Veronesi, 2012). If the winner of the election is the current president it is unlikely that there is a change in policy, since the President will most likely continue its current policy. In particular tax policies and trade policies that effects future dividends and company profits of foreign companies.
Overall using a F-test this model explains significantly the cumulative abnormal volatility in a country during a certain election. The probability of F is smaller than 1%.
The result for the effect of the U.S. Presidential election on stock market return is shown in table 4.
Table 4: Result of overall cumulative abnormal return
In table 4 the result for the cumulative abnormal return is shown. In this research there is no significant effect found of the U.S. Presidential election on stock market return of countries around the world. The p-value of the overall CAR is larger than 5%. This result is not inline with previous studies such as Jinling and Born (2006) and Pantzalis, Stangeland and Turtle (2000). However previous studies
researched the effect on the countries where the election occurs. In this research countries outside the United States are researched.
This results tells us also the following that other countries are strongly effected by the United States since 54% of the MSCI World Index consists of U.S. stocks which is not incorperated in this study.
_cons .0061908 .0039228 1.58 0.117 -.0015629 .0139446 cumulative~n Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .04724 R-squared = 0.0000 Prob > F = . F( 0, 144) = 0.00 Linear regression Number of obs = 145
Page 18 If for example the foreign countries do not react on the election, the change in stock price is zero
percent, but if the S&P500 goes up by 10%. The MSCI World Index should go up by approx. 5.4%. If someone made a portfolio of the foreign countries the return is zero. If an investor bought the MSCI World Index the return is approx. 5.4%. Their would be a negative abnormal return in the foreign countries compared to the MSCI World Index. In this research there is no abnormal return found, that suggests that foreign countries are strongly effected by the United States.
This outcome however is inline with the research by Levy and Sarnat (1970) that a well diversified international portfolio is only effected by systematic risk or worldwide risk and public information is incorporated in the prices and returns of stocks. Since a proxy for the global market portfolio is the MSCI World Index which consist for 54% of U.S. stocks one can say that U.S. policy has a strong effect on worldwide stock market return. In table 9 of the appendix the cumulative abnormal return per country and election are shown. It shows that there is at least in some countries a significant abnormal return but because one country experiences a positive effect and an other country experiences a negative effect from the event the total return of the internationally diversified portfolio is not
significant.
Conclusion
The main focus of this research was to determine the effect of the U.S. Presidential Election on stock market volatility and return around the world. In order to explain the relation between the election and stock market volatility a GARCH(1,1) model is used. The GARCH(1,1) model is used to forecast and predict volatility and the effect of the election is estimated in a multiple. In order to explain the relation between the election and the stock market return an OLS regression is used. This regression used a proxy for stock market return by using stock return of the country index against the MSCI World Index. Event study methodology is used to measure the difference between the election and the non-election period for both the volatility and the return. In this research 22 countries are used to estimate the relations mentioned above.
The overall effect of the U.S. Presidential election on stock market volatility for countries outside the United States is found to be positive and significant. This suggest that also countries outside the United States are effected by the uncertainty of the U.S. Presidential election. However without the election of 2008 the results find a negative significant effect of the U.S. Presidential elections on stock market volatility on foreign countries. This effect cannot be explained by the regression against the GDP per capita, export as percentage of GDP, margin of victory and re-election of the current President.
The overall effect on volatility is explained by using a regression on a few variables. It is found that richer countries are effected more by the U.S. Presidential elections than poorer countries.
In this study is found that a higher margin of victory induces more volatility. This is not inline with previous studies which found that a higher margin of victory is induces less volatility. In this study it is also found that a president which is re-elected induces less volatility. This confirms previous studies that a change in policy induces more volatility and political uncertainty.
This study does find an significantly negative relation between export and stock market volatility. This contradicts a previous study which found that countries that are more open to trade induce more volatility and are more effected by foreign policy changes.
Page 19 The overall effect of the U.S. Presidential election on stock market return in other countries is found to be not significant. This confirms some previous studies but contradicts others. However this result confirms the findings by Levy and Sarnat (1970) that a diversified global portfolio is only effected by worldwide risk and effects of individual countries are diversified away. In future studies one can try to explain individual country effects by checking for a relation between abnormal returns and left or right political regimes or large changes in trade policy. This results also shows that the United States strongly effects worldwide stock market return.
In future studies the expectation of uncertainty about the winner might be better explained by using poll data than using the margin of victory. Since poll data gives the expectation of the winner of the election before the election. While the margin of victory is not known before the election and is an ex ante variable. In future studies it might also be worthwhile to examine the effect of large political announcements that effects foreign countries on stock market volatility of the foreign countries. This could be an indication on the reaction of foreign markets on U.S. trade policies.
The limitation of this research is that the model used to estimate the effect of the election on volatility is not econometrically flawless. It is suggested in future studies to use for example volatility indices to estimate the effect of an election on stock market volatility. However at the time of this study there are not enough volatility indices and elections incorporated in them that can accurately predict the effect of the election on volatility.
Page 20
Bibliography
Beaulieu, M.-C., Cosset, J.-C., & Essaddam, N. (2005). The impact of political risk on the volatility of stock returns: the case of Canada. Journal of International Business Studies, 36, 701-718.
Berk, J., & DeMarzo, P. (2014). Corporate Finance (Vol. Third Edition). Harlow, Engeland: Pearson Education Limited.
Bialkowski, J., Gottschalk, K., & Wisniewski, T. P. (2006). Stock market volatility around national elections. Journal of Banking & Finance, 32(9), 1941-1953.
Bittlingmayer, G. (1998). Output, Stock Volatility, and Political Uncertainty in a Natural Experiment: Germany, 1880–1940. Journal of Finance, 53(6), 2243-2257.
Booth, J. R., & Booth, L. C. (2003). Is presidential cycle in security returns merely a reflection of business conditions? Review of Financial Economics, 12, 131-159.
Brown, S., & Warner, J. (1985). Using daily stock returns: The case of event studies event studies. Journal of Financial Economics, 14(1), 3-31.
Centraal Bureau voor de Statistiek. (2017, 1 21). Cijfers. Retrieved from Centraal Bureau voor de Statistiek: https://www.cbs.nl/nl-nl/cijfers
Fama, E. F. (1970). Efficient Capital Markets: A Review of Theory and Empircal Work. Journal of Finance, 25(2), 383-417.
Giovanni, J., & Levchenko, A. (2009). Trade openness and volatility. The Review of Economic Statistics, 91(3), 558-585.
Historical Election Results. (2016, 12 11). Retrieved from U.S. Electoral College: https://www.archives.gov/federal-register/electoral-college/historical.html
Jinliang, L., & Born, J. A. (2006). Presidential election uncertainty and common stock returns in the United States. Journal of Financial Research, 29(4), 609-622.
Levy, H., & Sarnat, M. (1970). International Diversification of Investment Portfolios. The American Economic Review, 60(4), 668-675.
MacKinlay, A. (1997). Event studies in economics and finance. Journal of economic literature, 35(1), 13-39.
Pantzalis, C., Stangeland, D. A., & Turtle, H. J. (2000). Political elections and the resolution of uncertainty: The international evidence. Journal of Banking & Finance, 1575-1604.
Pastor, L., & Veronesi, P. (2012). Uncertainty about government policy and stock prices. Journal of Finance, 67(4), 1219-1264.
Presidential Election Laws. (2016, 12 11). Retrieved from U.S. Electoral College:
Page 21 Presidential Election Process. (2016, 12 11). Retrieved from USA.gov: https://www.usa.gov/election Presidential Election Process. (2016, 12 11). Retrieved from USA.gov:
https://www.usa.gov/election#item-36072
Siokis, F., & Kapopoulos, P. (2007). Parties, elections and stock market volatility. Economics & Politics, 19(1).
Valls Pereira, P. L., & Hwang, S. (2004). Small Sample Properties of GARCH Estimates and Persistence. The European Journal of Finance, 12(6), 473-494.
Page 22
Appendix
Table 6: Summary of the returns of the indices
MSCIWORLDU~X 12238 .0002678 .008389 -.0984436 .0952324 ATHEXCOMPO~X 7346 .0002724 .0185839 -.1623282 .1473912 KOREASEKOS~X 7018 .0002857 .0175363 -.1196093 .1572491 OMXSTOCKHO~E 8062 .0004499 .0142906 -.0842415 .1165331 BISTNATION~X 7540 .0015584 .0255719 -.1810933 .1945097 RUSSIANMIC~X 5005 .0009466 .0261322 -.2081284 .3165375 BELGRADEBE~X 2578 -.0002693 .0133065 -.102923 .1292748 ROMANIABET~X 5006 .0005193 .0164384 -.1229293 .1223723 WARSAWGENE~I 5900 .00025 .0178326 -.1320388 .1599253 IRELANDSEO~E 8843 .0004119 .0122153 -.130325 .1022251 OMXHELSINK~X 7801 .0003969 .0159557 -.1597345 .1567693 OMXCOPENHA~I 7040 .0003594 .011679 -.1106211 .0996188 PRAGUESEPX~X 5908 .0000679 .0132453 -.1494352 .131609 CROATIACRO~X 5192 .000305 .0150529 -.1252421 .190907 PORTUGALPS~X 6237 .0001306 .0115607 -.098587 .1073383 FTSEMIBIND~X 4933 .0000437 .0156661 -.1248101 .114905 IBEX35PRIC~X 7800 .0002716 .0138646 -.1235299 .1443495 SWISSMARKE~X 7412 .0002884 .0113913 -.1054459 .1139097 FRANCECAC4~X 7667 .0002407 .0137997 -.0964065 .1117617 DAX30PERFO~X 13542 .0003031 .0121836 -.1281161 .1140195 NIKKEI225S~E 17381 .0003739 .0119328 -.1490094 .141503 AEXINDEXAE~X 8845 .0003474 .013185 -.1199608 .1183107 FTSE100PRI~X 10129 .0003283 .0109051 -.1660253 .1293418 SPASX200PR~X 6391 .0002272 .0094721 -.0833623 .0589146 Dates 17381 8619.2 7024.65 -3546 20786 Variable Obs Mean Std. Dev. Min Max
Page 23
Table 7: Summary of the LN of GDP per capita per country
Croatia 6 9.122159 .4868514 8.500988 9.673688 KoreaRep 9 9.344066 .7913445 7.813782 10.21176 Finland 9 10.26574 .4967858 9.291143 10.88559 Australia 9 10.2221 .6130454 9.427795 11.12205 Sweden 9 10.44256 .4736839 9.466093 10.95316 Turkey 9 8.357535 .7811723 7.127856 9.262873 RussianFed~n 7 8.545124 .8207701 7.479631 9.626051 Bulgaria 9 7.971317 .751147 7.097425 8.900188 Greece 9 9.53883 .584543 8.487263 10.37341 Romania 8 8.106815 .9045274 7.004976 9.223895 Poland 7 8.823794 .6864208 7.788153 9.540091 Denmark 9 10.45785 .5418309 9.348226 11.06948 CzechRepub~c 7 9.231386 .7182465 8.113323 10.02789 Spain 9 9.732845 .6378804 8.408238 10.4795 Portugal 9 9.358933 .7259548 7.833202 10.11923 Italy 9 10.0371 .5017947 8.950816 10.61251 Switzerland 9 10.73716 .5174932 9.708557 11.32911 Germany 9 10.22153 .5113256 9.135394 10.72984 France 9 10.16773 .4874758 9.151894 10.72356 Japan 9 10.33115 .4290932 9.286077 10.75152 Netherlands 9 10.27984 .5590353 9.195209 10.94956 UnitedKing~m 9 10.15596 .5691921 9.009349 10.71875 Variable Obs Mean Std. Dev. Min Max
Page 24
Table 8: Summary of the export as % of GDP per country
Croatia 6 .6670156 .3152524 .3007009 1.183804 KoreaRep 9 6.229308 2.236501 3.754459 9.937849 Finland 9 1.795605 .4848537 1.052071 2.589535 Australia 9 1.155176 .3390237 .6222403 1.574126 Sweden 9 2.622899 .7607708 1.682409 3.693928 Turkey 9 .9536373 .2113729 .6355798 1.258091 RussianFed~n 7 1.454083 .8891141 .1045861 2.948883 Bulgaria 7 1.233883 .4808614 .7147961 1.944135 Greece 9 .4796276 .203926 .2818191 .8253722 Romania 7 .7960362 .3190613 .3479069 1.262603 Poland 7 .6754676 .2929389 .3997346 1.187474 Denmark 9 1.805058 .5768339 1.091684 2.674998 CzechRepub~c 6 1.568205 .61859 .7224256 2.473797 Spain 9 .8639661 .2837554 .4770961 1.39499 Portugal 9 1.221567 .4214402 .6173638 2.013387 Italy 9 1.696512 .4839155 .9358819 2.433321 Switzerland 9 3.154229 .8674528 2.081851 4.723166 Germany 9 2.524431 .7737596 1.357448 3.719575 France 9 1.537273 .3735636 1.050397 2.177672 Japan 9 3.034727 .7737423 2.447567 4.967952 Netherlands 9 2.114281 .495402 1.47701 2.873954 UnitedKing~m 9 2.219328 .4259609 1.703305 3.052862 Variable Obs Mean Std. Dev. Min Max
Page 25
Table 9: Cumulative abnormal return per country and election
369102. 57 Germany 11/8/2016 -.0039115 -.1928486 358427. 56 Germany 11/6/2012 .0193931 1.198458 347752. 55 Germany 11/4/2008 .0169196 .2270577 337077. 54 Germany 11/2/2004 -.0322751 -1.135146 326407. 53 Germany 11/7/2000 .003922 .1986061 315732. 52 Germany 11/5/1996 -.0085697 -.3481012 305057. 51 Germany 11/3/1992 -.0036033 -.1000962 294388. 50 Germany 11/9/1988 -.0378577 -1.069051 283712. 49 Germany 11/6/1984 .0047443 .2392229 282432. 45 Japan 11/8/2016 .0119395 .1268751 271757. 44 Japan 11/6/2012 -.0204831 -.7150694 261082. 43 Japan 11/4/2008 .1746127 1.155224 250407. 42 Japan 11/2/2004 -.0014345 -.0574508 239737. 41 Japan 11/7/2000 .0168141 .3286963 229062. 40 Japan 11/5/1996 -.0246941 -.7667097 218387. 39 Japan 11/3/1992 -.0036485 -.1293997 207718. 38 Japan 11/9/1988 .0221229 1.588888 197042. 37 Japan 11/6/1984 -.0071072 -.3078619 195762. 33 NL 11/8/2016 -.014538 -.7770448 185872. 32 NL 11/6/2012 .0381824 2.091844 175982. 31 NL 11/4/2008 -.0053056 -.0475771 166092. 30 NL 11/2/2004 -.0269101 -1.013949 156207. 29 NL 11/7/2000 .0028065 .1532273 146317. 28 NL 11/5/1996 -.0092713 -.3894717 136427. 27 NL 11/3/1992 .0086254 .4173267 126543. 26 NL 11/9/1988 -.0604138 -1.785737 125002. 21 UK 11/8/2016 -.0312337 -1.310784 114327. 20 UK 11/6/2012 .016288 .9220513 103652. 19 UK 11/4/2008 .0088381 .1137941 92977. 18 UK 11/2/2004 -.0125768 -.8932025 82307. 17 UK 11/7/2000 .005672 .2253868 71632. 16 UK 11/5/1996 -.0430259 -2.520024 60957. 15 UK 11/3/1992 .0251812 1.325167 50288. 14 UK 11/9/1988 -.0335079 -1.237094 39612. 13 UK 11/6/1984 .0325931 1.915123 38332. 9 Australia 11/8/2016 -.0010451 -.0237628 30896. 8 Australia 11/6/2012 -.0127615 -.4328623 23460. 7 Australia 11/4/2008 -.004371 -.0445801 16024. 6 Australia 11/2/2004 .031717 2.208105 8593. 5 Australia 11/7/2000 .0050468 .2432258 1157. 4 Australia 11/5/1996 -.0166482 -.872144 group_id name eventdate cumulat~n test
Page 26 617302. 129 Croatia 11/8/2016 .0043439 .2277187 611065. 128 Croatia 11/6/2012 .0042774 .216521 604828. 127 Croatia 11/4/2008 -.0556794 -.8808746 598591. 126 Croatia 11/2/2004 .0015179 .0408722 592359. 125 Croatia 11/7/2000 .0747067 1.428107 591342. 117 Portugal 11/8/2016 -.0608201 -2.341495 584060. 116 Portugal 11/6/2012 .0152514 .9225042 576778. 115 Portugal 11/4/2008 .0665731 .7804335 569496. 114 Portugal 11/2/2004 -.0030386 -.1539611 562219. 113 Portugal 11/7/2000 -.0037197 -.1153327 554937. 112 Portugal 11/5/1996 .0167459 1.461448 553920. 105 Italy 11/8/2016 -.0340661 -1.80664 547942. 104 Italy 11/6/2012 .0335097 1.439713 541964. 103 Italy 11/4/2008 .0080851 .0634799 535986. 102 Italy 11/2/2004 -.0097088 -.5778431 530013. 101 Italy 11/7/2000 .0157646 .8589497 529255. 93 Spain 11/8/2016 -.0575415 -2.415823 520410. 92 Spain 11/6/2012 .0248534 .7016094 511565. 91 Spain 11/4/2008 .0206761 .1771096 502720. 90 Spain 11/2/2004 -.0169854 -.8671879 493880. 89 Spain 11/7/2000 .000316 .0090813 485035. 88 Spain 11/5/1996 -.0034651 -.1625266 476190. 87 Spain 11/3/1992 .0095502 .4454347 474655. 81 Switzerland 11/8/2016 .0075551 .3401773 466198. 80 Switzerland 11/6/2012 .03324 2.183459 457741. 79 Switzerland 11/4/2008 -.0156715 -.1591431 449284. 78 Switzerland 11/2/2004 .0025366 .1208452 440832. 77 Switzerland 11/7/2000 .0215828 1.031549 432375. 76 Switzerland 11/5/1996 -.0013875 -.0948845 423918. 75 Switzerland 11/3/1992 .0200452 1.220593 422771. 69 France 11/8/2016 -.0039316 -.2112041 414059. 68 France 11/6/2012 .0343512 1.38732 405347. 67 France 11/4/2008 -.0066455 -.070394 396635. 66 France 11/2/2004 -.0200591 -.8509503 387928. 65 France 11/7/2000 -.0168583 -.8285685 379216. 64 France 11/5/1996 .0174635 .6351385 370504. 63 France 11/3/1992 .0272717 .8510823
Page 27 817067. 201 Romania 11/8/2016 .0066112 .4569533 811016. 200 Romania 11/6/2012 -.0073178 -.4928641 804965. 199 Romania 11/4/2008 .1705694 1.898242 798914. 198 Romania 11/2/2004 .0001616 .005716 792868. 197 Romania 11/7/2000 -.0266207 -1.020837 792037. 189 Poland 11/8/2016 -.0329176 -.9646277 785092. 188 Poland 11/6/2012 .0358822 1.32967 778147. 187 Poland 11/4/2008 .1026335 1.670175 771202. 186 Poland 11/2/2004 -.0007945 -.0195293 764262. 185 Poland 11/7/2000 .0591393 .8123713 757317. 184 Poland 11/5/1996 .0073529 .1409492 756637. 165 Finland 11/8/2016 -.0201731 -.8252897 747791. 164 Finland 11/6/2012 .0332517 1.449455 738945. 163 Finland 11/4/2008 -.0469742 -.5355562 730099. 162 Finland 11/2/2004 -.0039815 -.1251935 721258. 161 Finland 11/7/2000 -.0369852 -.4986375 712412. 160 Finland 11/5/1996 -.0090381 -.3933959 703566. 159 Finland 11/3/1992 .150668 1.817415 702030. 153 Denmark 11/8/2016 -.0673664 -2.216378 693945. 152 Denmark 11/6/2012 -.0028485 -.0486744 685860. 151 Denmark 11/4/2008 .0370481 .3590178 677775. 150 Denmark 11/2/2004 -.0139105 -.7079265 669695. 149 Denmark 11/7/2000 -.0523112 -2.324512 661610. 148 Denmark 11/5/1996 -.0118359 -.8066739 653525. 147 Denmark 11/3/1992 .0709162 2.418296 652750. 141 Czech Republic 11/8/2016 -.0215837 -.9209433 645797. 140 Czech Republic 11/6/2012 .0249792 1.230308 638844. 139 Czech Republic 11/4/2008 .1416652 1.020171 631891. 138 Czech Republic 11/2/2004 .003404 .0859126 624943. 137 Czech Republic 11/7/2000 -.071215 -1.872405 617990. 136 Czech Republic 11/5/1996 -.0658436 -3.274067
Page 28 1065072. 273 Greece 11/8/2016 -.0228988 -1.336598 1056681. 272 Greece 11/6/2012 -.0243981 -.2281307 1048290. 271 Greece 11/4/2008 .0811074 .7120245 1039899. 270 Greece 11/2/2004 .0295382 .8521171 1031513. 269 Greece 11/7/2000 -.0726723 -1.73143 1023122. 268 Greece 11/5/1996 -.0116159 -.5033576 1014731. 267 Greece 11/3/1992 -.0026837 -.0801212 1013650. 261 South Korea 11/8/2016 -.0291522 -.7885734 1005587. 260 South Korea 11/6/2012 .0071629 .2257067 997524. 259 South Korea 11/4/2008 .1214365 .898459 989461. 258 South Korea 11/2/2004 .0148262 .3276333 981403. 257 South Korea 11/7/2000 .0973097 1.129417 973340. 256 South Korea 11/5/1996 -.0177352 -.3149568 965277. 255 South Korea 11/3/1992 .1715822 1.955299 964524. 249 Sweden 11/8/2016 .006437 .3387766 955417. 248 Sweden 11/6/2012 .0218503 1.138121 946310. 247 Sweden 11/4/2008 .0379169 .4282855 937203. 246 Sweden 11/2/2004 .0052247 .1911834 928101. 245 Sweden 11/7/2000 -.0211396 -.7249986 918994. 244 Sweden 11/5/1996 -.0360424 -1.258372 909887. 243 Sweden 11/3/1992 .033332 .8667766 900786. 242 Sweden 11/9/1988 -.0233679 -.7217233 900028. 237 Turkey 11/8/2016 -.0466479 -1.290342 891443. 236 Turkey 11/6/2012 .0305826 .7181498 882858. 235 Turkey 11/4/2008 -.0148192 -.1423924 874273. 234 Turkey 11/2/2004 -.0011722 -.0255919 865693. 233 Turkey 11/7/2000 -.0404656 -.449961 857108. 232 Turkey 11/5/1996 -.0163401 -.3526419 848523. 231 Turkey 11/3/1992 -.0095868 -.1114101 847248. 225 Russia 11/8/2016 .0037676 .1101881 841198. 224 Russia 11/6/2012 -.0217696 -.7412286 835148. 223 Russia 11/4/2008 .1850678 .7892686 829098. 222 Russia 11/2/2004 .0042126 .070632 823053. 221 Russia 11/7/2000 -.1013982 -1.420508 822223. 213 Bulgaria 11/8/2016 -.0257432 -1.130552 818600. 212 Bulgaria 11/6/2012 .0525382 3.73392
Page 29
Figure 2: Distribution MSCI World Index
